world bank document****n*ir*** thailand social monitor poverty and public policy preface poverty...

0

co-r%I... ¢ a -...........l

_R - 0J--4

Y ~ ~ ~ ~ ~ rh S,_

mU~M

CZ-

75 j: *

B~~ H

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

World Bank Office BangkokBangkok, 14th Floor Tower A Country Director: Ian C. PorterNovember 2001 Diethelm Towers Comments to: L /T. 4

93/1 Wireless Road Anil B. Deolalikar. Chris ChamberlinBangkok 10330, Thailand Maryam SalimTel (662) 256-7792 I,,.,

[email protected] @worldbank.org

The views expressed in the Thailand Social Moniitor are entirely those of the authors and should rzot be cited withoutpriorpertnission. Theydo not necessarily represent the view of the World Bank, its Executive Directors, or the countries they represent. The material containedhlerein has been obtained from sources believed reliable but is not necessarily cornplete and cannlot be guaranteed.

Poverty and Public Policy

Table of Contents

Preface.I

Acknowledgements .III

Executive Summary. iProfile of Poverty in Thailand . iiPatterns of Inequality . viStrategic Issues . viiiPublic Programs and Poverty . xiiiImplications for Policy . xviii

Chapter 1 Concepts, Measurement, and Extent of Poverty in ThailandWhat Is Poverty? . 1Measurement of Poverty .1Extent of Poverty and Changes in Poverty in Thailand. 3Poverty and Unemployment .5International Comparisons. 7Overlap between the Income - Poor and the Consumption - Poor. 8Reliability of Survey - Based Estimates of Consumptionand Income .10Policy and Research Issues .11

Chapter 2 Who are the Poor of Thailand?Region of Residence .13Rural - Urban Differences .15A Provincial Poverty Map .15How Concentrated are the Poor Geographically? . 16Household Size and Composition .17Age and Sex of the Household Head .18Education of Household Head .19Occupation of the Household Head .20Ownership of Land .21Policy and Research Issues .22

Chapter 3 Inequality in ThailandOverall Trends .26International Comparisons .27Regional Variation .28Patterns of Inequality .29Income Inequality by Sources of Income .30Causes of Income Inequality in the Northeast and in Bangkok . 30Changes in Inequality between 1998 and 199 .32

Thailand Social Monitor


Changes in the Distribution of Education .............................. 33Research and Policy Issues ............................................... 35

Chapter 4 Strategic Policy Issues in Poverty and InequalityPoverty and Self Reliance ...................................... 36Structural and Chronic Poverty ...................................... 39Poverty and Environmental Resource Constraints .......... .......... 41Gender and Poverty ...................................... 43Correlates and Causes of Poverty ....... ............................... 46Poverty in the Northeast ................... ................... 49Economic Growth, Inequality, and Poverty Reductionacross Provinces ...................................... 50Policy and Research Issues ................................. ..... 55

Chapter 5 Empowering the PoorSupport from the Legal Environment ................. ................. 61Institutional Environment ............................................... 62Progress in implementing reforms mandated by the Constitution.. 62Decentralization as a tool for the poor? .............. ................. 63Empowering Communities .............................................. 65Programs aimed at supporting community initiatives ............... 66Institutional Support ............................................... 67Measuring Results and Other Challenges .............................. 68

Chapter 6 Protecting and Investing in the PoorGovernment Expenditure on Anti-Poverty Programs ................. 72Programs that Protect the Poor ..................................... 75Programs Investing in the Capacities of the Poor ...................... 82Policy and Research Issues ................................. .... 85



AnnexA Government Expenditure on Anti-Poverty Programs

(in millions of current bath), 1993 - 99 ................................. 88Boxes1 The Definition of Poverty ................................................ 12 Poverty is a Feeling ................................................ 23 The Thai Socio-Economic Surveys ..................................... 24 Dissemination of Information on Poverty .............................. 125 Village - Level Targeting and Village Databases ..................... 236 The Ultra Poor .................... ............................ 397 Environmental Capital ................................................ 418 The Nexus between the Environment and Poverty ........ ......... 429 "Causes" of Poverty ................................................ 4610 Giving the Power to the People-The Thai Constitution ............... 6111 The 9th National Economic and Social Development Plan

and Poverty Alleviation ............. ....................... 6312 Central Government Subsidies to Local Authorities .......... ........ 6513 The Remedy for Poverty .................................... 6614 Thailand Social Investment Fund: Security Through

Empowerment ................................. 7015 Impact of the Crisis: Vulnerability of the Poor ........................ 7616 Disparity in Education Budget Allocations ............................. 82

Figures1 Number of employed persons and unemployment rate,

third quarter, 1989 - 2000 ........................................... 62 Unemployment Rate by Region, third quarter, 1996 - 2000 ......... 73 Headcount ratio of consumption - poverty, 1987 - 99

selected Asian countries (poverty line defined at 1993PPP $1.50 per person per day). 7

4 Indices of mean monthly real income and consumptionexpenditure per capita, 1998 - 99 (1988 = 100). 8

5 Percentage of poor based on income and consumptionexpenditure, 1988 - 99 .9

6 Cumulative distribution of per capita income andexpenditure, 1999. 9

7 Agreement between income and consumption -povertymeasures, 1988-99 .1 0

8 Headcount ratio of income-poverty (%), by region,1988, 1996 and 1999 .13

9 Headcount ratio of income-poverty, by region, 1988 - 99(1988 poverty headcount = 100) .14

10 Headcount ratio of income - poverty (%), by rural/urban area,1988, 1996 and 1999 .15

11 Headcount ratio of income - poverty, by rural/urban areas,1988 - 99 (1988 poverty headcount = 100) .15

12 Incidence of income - poverty (%) by province, 1998 .16


Poverty and Puiblic Policy

Figures13 Contribution of the 17 poorest provinces to national poverty,

1998 .................................................. 1614 Cumulative distribution of poverty across villages, 1998 ............ 1715 Headcount ratio of income poverty, by household size, 1992-99 ... 1716 Headcount ratio of income poverty, by dependency burden,

1992 - 99 .1817 Headcount ratio of income poverty, by age and sex of

household head, 1992 - 99 .1918 Headcount ratio of income-poverty, by household head's

schooling, 1992 - 99. 1 919 Percent change in real income per capita, 1996 - 99,

by household head's schooling .2020 Headcount ratio of income-poverty, by household head's

occupation, 1992 - 99 ............................................... 2021 Composition of the poor by occupation, 1992 - 99 ........ ........ 2122 Headcount ratio of income-poverty among farm households,

by size of owned land, 1992 - 99 .2123 Percent change in rice yields, by region and by per capita land

quintiles (adjusted for land quality), 1995 - 96 to 1998 - 99 ....... 2224 Income Inequality, 1990 - 99 ........................................... 2625 Lorenz curves for Income, 1981 and 1999 ........... .............. 2726 Gini Coefficient of consumption, 1987 - 99, selected

Asian countries (%) ......... ........................ 2727 Mean per capita income, by region, 1990 - 99 ........................ 2828 Decomposition of overall income inequality into its various

causes, Northeast and Bangkok, 1998. 3029 Lorenz curves for per capita income, expenditure and

owned land, rural areas only, 1998 .3130 Mean household income per capita and share in total

population, by major occupation, Northeastern region .3231 Mean real household income per capita in 1998 and percent

real change in per capita income, 1988 - 99, by schooling ofhousehold head .33

32 Mean real household income per capita in 1998 and percentreal change in per capita income, 1998 - 99,by age of household head .33

33 Distribution of enrolled students at the primary and lowersecondary level, by real per capita income quintile, 1992 - 99 . 34

34 Distribution of enrolled students, by level and by realper capita income quintile, 1992 - 99 .34

35 The volume of income transfers in relation to total income,by real per capita income quintile, 1969 - 99 .37

36 Incidence of income poverty in the absence of transfers,1996 - 99 .37

37 Expenditure shares on selected items of expenditure bythe poorest 20% of household, 1996 and 199 .38



Figures38 Income-poverty incidence in selected provinces as

a ratio of national poverty incidence, 1944 - 99 ....................... 4039 Estimated effect of female-headedness on the probability

of being poor, by age of female head .............. ........... 4940 Growth rate in per capita GDP and Changes in incidence of

consumption - poverty, 1988 - 99 ................ .................. 5041 Shares of each region in total GDP and population increase

taking place in Thailand between 1987 and 1996 ..................... 5142 Changes in the income-poverty headcount ration and in

the number of income-poor decomposed into changes due togrowth and changes due to distributional shifts,1992 - 94 to 1998 - 99 .53

43 Government expenditure on anti-poverty programs, 1993 - 99.... 7344 The government expenditure and average household living

standards across provinces, 2000 ...................... 7445 Coverage and incidence of government health programs,

by real per capita consumption quintiles, 1999 ........................ 7646 Lorenz curves for beneficiaries of government health

programs, 1999 ......... 7747 Coverage and incidence of social security and social pension

programs, by real per capita consumption quintiles, 1999 ............ 7848 Incidence of income-poverty among individuals residing in

households headed by a person aged 60 years or older,by household size, 1998 and 1999 ........ ............... 79

49 Lorenz curves for beneficiaries of the social security andsocial pension for the elderly poor programs, 1999 ................ 80

50 Coverage and incidence of public employment generationschemes, by real per capita consumption quintiles, 1999 ....... 81

51 Lorenz curves for beneficiaries of public employmentschemes, 1999 ........ 82

52 Coverage and incidence of government educational assistanceprograms, by real per capita consumption quintiles, 1999 .......... 83

53 Coverage and incidence of government school nutritionprograms, by real per capita consumption quintiles, 1999 ........... 84

54 Share of each real per capita income quintile in totalhousehold (gross) borrowing from the Bank of Agricultureand Agricultural Cooperatives, 1996 - 99(agricultural households only) .................... 85



Tables1 Measures of Poverty 1988 - 99 ..................... 42 Aggregate household consumption and income, by source, SES

and national - Account estimates, 1998 (billions of current bath) ... 1 13 Estimated maximum likelihood probit equations for the

probability of being income-poor, by rural and urban areas, 1998.. 574 Estimated maximum likelihood probit equations for the

probability of being income-poor, Rural Northeast andRural Other Regions, 1998 .58

5 Cross-province regressions of incidence of income-poverty,1994-99 ................................................ 59

References ................................................................................ 89

ThailaM*****n*i**r*******



Preface

Poverty has been studied extensively in Thailand. It is not a new problem, but it is re-emerging in prominence and therefore the subject of both public and private concern.Civil society groups have been working hard to put the plight of the poor at the forefrontof the national agenda. The Government has also taken steps to raise the profile of thepoverty issue. Tackling poverty and promoting pro-poor growth is a high priority in theNinth National Economic and. Social Development Plan (2002-2006).

The objectives of this issue of the Thailand Social Monitor are modest: to provide freshinsights into the poverty dilemma; to strengthen the case for an active poverty reductionstrategy; to suggest some issues for further research. Poverty and Public Policy cannotaspire to be a comprehensive policy review, nor does it propose a detailed strategy onhow to tackle the poverty problem. This issue of the Thailand Social Monitor, therefore,is written for Thai policy makers, to provide them with the best evidence currentlyavailable on poverty and public policy, and to strengthen the foundation for anti povertystrategies in Thailand. The coverage of this report is thus broad, ranging from detailedanalysis of poverty dimensions, a review of strategic issues and finally, an assessment ofthe performance of some key policies and programs.

Poverty and Public Policy offers us a deeper analysis of poverty and in particular: the.profile of the poor and the causes of poverty; on its relationship with distribution andgrowth; and on the persistent challenges of the Northeast.

The process of preparing this Social Monitor has been one based on partnership from itsvery inception. Poverty and Public Policy is the product of a fruitful collaborationbetween the World Bank, the National Economic and Social Development Board(NESDB), the Thailand Development Research Institute (TDRI), Thai academic andsocial experts, and international specialists. We particularly wish to thank themanagement and staff of the NESDB for their indispensable contributions and support tothis report.

What next? Fighting poverty is a high priority for the Royal Thai Government. To thisend a Country-Development Partnership on Poverty (CDP-P) is currently underpreparation between the NESDB and the World Bank. The CDP-P will be aimed atsupporting the Government's efforts to formulate and implement effective povertyreduction policies.

Another key activity will be the forthcoming collaboration between NESDB, the NationalStatistical Office (NSO) and the World Bank in developing a Poverty Map for Thailand.The purpose of this activity will be to build government capacity to analyze the incidenceof poverty at the district and sub-district levels by combining household survey andcensus data to estimate consumption- or income-based welfare indicators. Such a mapcan become a key instrument for improving the targeting of poverty programs.

Thailand Social Monitor I


At the end of November 2001, TDRI will host a Conference that will bring togethergovernment, academia, civil society and the donor community to exchange views in thisimportant area of poverty reduction. This event will offer an important opportunity forThais from all walks of life to express their views and to build consensus on the wayforward.

We would like to express our thanks to the many experts from Government,development agencies, and civil society who provided generously of their time andinsights during the preparation of Poverty and Public Policy.

Ian C. PorterCountry Director

World Bank Office Bangkok

Thailand Social Monitor II


Acknowledgements

This report has been prepared by a core team of principal authors led by Mr. ChristopherChamberlin, Task Team Leader, World Bank Office Bangkok; Dr. Anil Deolalikar, leadconsultant, and Ms. Maryam Salim, World Bank Office Bangkok. Overall guidance forthe report was provided by Mr. Homi Kharas, Chief Economist, East Asia and PacificRegion. Key contributions and policy guidance were provided by the National Economicand Social Development Board (NESDB) and the Thailand Development ResearchInstitute (TDRI). The report was discussed with counterparts and civil society on severaloccasions, including full review workshops held in March and October 2001.

Analytical inputs were provided by Ms. Benu Bidani, World Bank, Washington DC; Ms.Khuankaew Varakornkarn and Mr. Avinash Kaza, World Bank Office Bangkok; Dr.Priyanut Piboolsravut, Director of the International Economic Research Unit, NESDB;Ms. Waranya Teokul and Ms. Lalita Chanwongpaisarn, Macro-Social Studies and SocialInvestment Policy Unit, NESDB; Mr. Somchai Jitsuchon, Thailand DevelopmentResearch Institute; Dr. Harold Coloumbe, consultant; Dr. Nanak Kakwani, consultant;Ms. Sauwalak Kittiprapas, consultant; Dr. Isra Santisart, consultant; and Dr. Peter Warr,consultant.

Valuable comments and suggestions on earlier drafts of this report were provided by Mr.Ian C. Porter, Country Director, Thailand; from other World Bank staff including: Ms.Tamar Manuelyn Atinc, Mr. Ejaz S. Ghani, Mr. Ijaz Nabi, Mr. Kaspar Richter, Mr.Patchamuthu Illangovan, Mr. Samuel S. Lieberman, Ms. Pamornrat Tansanguangwong,and Mr. J. Shivakumar; and from counterparts in Thailand including: Mr. PromboonPanichapakdi, Country Director, Care International; Dr. Amara Pongsapich, Director,Social Research Institute, Chulalongkorn University; Mr. Khantong Dalad, DuangPrateep Foundation; Ms. Elizabeth Morris, ILO; Dr. Juree Vichit-vadakarn, NationalInstitute of Development Administration; Ms. Jirawan Boonperm, National StatisticalOffice; Dr. Witit Rachtatanun, Assistant Secretary General, NESDB; Ms. SomsriLuangmanirut, Director of Economic Analysis and Projection Division, NESDB; Ms.Suwanee Kumman, Director of Social Projects Division, NESDB; Dr. ChalongphobSussangkarn, President, TDRI; Dr. Ammar Siamwalla, Distinguished Scholar, TDRI; Dr.Worawan Chandeowit, TDRI, Dr. Mehdi Krongkaew, Thammasat University; Ms.Netnarumon Sirimonthon, UNDP; and Mr. Gamini Abeysekera, Country Representative,UNICEF.

Peer reviewers for the report were: Mr. Jefffrey S. Hammer, Mr. Peter F. Lanjouw andMs. Dominique Van De Walle. Support from the Asia-Europe Meeting Trust Fund isgratefully acknowledged. Document processing was undertaken by Ms. TasaneeChokwatana.



Executive Summary

1. This is the sixth issue of the Thailand Social Monitor. In the post crisis environ-ment, the problem of poverty has taken on new characteristics and renewed importance,and the recently elected Government has identified the plight of the rural poor as a highsocial priority. This issue of the Social Monitor, therefore, is written for Thai policymakers, to provide them with the best evidence currently available. on poverty and publicpolicy, and to strengthen the foundation for anti poverty strategies in Thailand.

2. This report approaches poverty in Thailand from four perspectives:

the changing profile of the poor, who they are, where they live, their definingcharacteristics, so as to better understand the dimensions of the problem (Chaptersl and 2);the changing profile of income inequality in Thailand, to assess whether incomedistribution should be an issue for public policy (Chapters 3 and 4);some of the strategic, cross-sectoral issues that are at the center of public policydebates on poverty reduction in Thailand (Chapter 4);the performance of recent policies and programs to reduce poverty, so as to ap-preciate the strengths and limitations of public policy in this challenging domain(Chapters 5 and 6).

3. The coverage of this report is thus broad, ranging from detailed analysis of pov-erty dimensions, strategic issues and, finally, policies and programs. However, it is im-portant to note the report's limitations, particularly on the policy and program side.While chapters 5 and 6 analyze targeted poverty interventions, they do not capture thefull range of public efforts to reduce poverty, nor do they address the anti-poverty bene-fits of universal programs, such as education, health, transport or agricultural services. Inaddition, macroeconomic policies, which can exert a large influence on poverty via suchinstruments as tax policy and trade policy, are not covered. These and other issues leavemuch to be done in the analysis of poverty.

4. Poverty has been studied and policies put in place to deal with it over many yearsin Thailand. This analysis of poverty began with a review of the large inventory of exist-ing work, mostly by Thai scholars, and then has explored the vast household data setsavailable in Thailand, and assembled a team of experts and advisers to pull from the datafresh insights into the poverty problem. Poverty and Public Policy is thus the product ofa fruitful collaboration between the World Bank, the NESDB, TDRI, Thai academic andsocial experts, and international specialists. With such inputs, it might be expected thatthis report would go beyond analysis and into recommendations on strategy and a list ofpolicy proscriptions.

5. The approach taken in Poverty and Public Policy is to present evidence, link it topublic policy options, and indicate where there are gaps in information and analysis. Itdoes not select a strategy nor does it rank policies and programs that Thailand shouldadopt. That is a task for Thai institutions and leaders to undertake. Indeed, the 9 th Eco-

Thailand Social Monitor i

Poverty and Pwblic Policy

nomic and Social Development Plan of the NESDB has been directed to the povertyproblem, and the TDRI, among other Thai institutions, is undertaking extensive work onpoverty policy. Hopefully their efforts to deepen and extend the anti poverty strategy forThailand will find some useful information from this issue of the Thailand Social Moni-tor.

6. Poverty and Public Policy argues that the poverty problem in Thailand is re-emerging as one of the nation's most serious problems. Until the onset of the crisis in1997, high economic growth rates pulled poverty down from 32.6% in 1988 to merely11.6% in 1996. During this period, average incomes were growing rapidly; migrationfrom the poor areas of the Northeast to Bangkok and the Central region helped keep pov-erty rates in the Northeast below what they might have been; dynamic constructiongrowth fueled demand for low and semi-skilled workers; and school enrollments in-creased impressively, providing the poor with access to primary and lower secondaryschooling. The reduction of poverty was underway and at a fast pace.

7. But much of this came to a halt with the economic crisis in 1997. During the nexttwo years, poverty incidence increased sharply. The weight of evidence also suggeststhat the poor endured the brunt of the crisis, as their wages, profits, and employment fellmore than that of better-off Thai households. While it is possible that the poor will regainquickly what they have lost once growth resumes, the current outlook for economicgrowth in Thailand is guarded, given global uncertainties. Faced with the dual challengeof restoring economic growth and better serving the needs of the poor, the Government isgrappling with fundamental choices on the allocation of public resources to competingprograms and to competing segments of Thai society, including the poor. This report iswritten to strengthen the case for pro-poor public policy.

Profile of Poverty in Thailand

8. Poverty and Public Policy sets out in Chapters 1 and 2 the alternative measures ofpoverty using two different sets of household data, consumption (also termed "expendi-ture" data) and income. The various measures of poverty and data sources can producequite different results. Thailand's official poverty indicators are currently based on theincome data set and employ a special procedure to calculate a minimum income for eachhousehold participating in the survey. Both approaches are distinctive and make the Thaidata difficult to compare with other countries. Poverty and Public Policy utilizes theseconventions, but also points out some of the difficulties.

Thailand Social Monitor ii


9. The most common meas- Headcount ratio of income-poverty (%),

ure of poverty is the head-count 50 48 by region, 1988, 1996 and 1999

ratio, which is the proportion of _45

the population with incomes be- 01988 *1996 0 1999

low an established poverty line.15 ~~~~33 32The official poverty line for Thai- , 35 3 1

land was established by the , 30 27

NESDB in 1998.1 The head-count 25-

ratio of poverty, based on this e. 20 i

poverty line, fell continuously 1 5 i1

from 1988 to 1996, but then re- 10 I 8

versed itself and increased in 1998 5 6

and 1999. All other indicators of 0+ X_poverty, such as the poverty-gap Northeast South North Central Bangkok

ratio and the severity of povertyindex, also show an increase in the Figure 1 This Figure is a replica of Figure 8 in Chapter 2

two years after the onset of the economic crisis.2

10. Another striking feature of Headcount ratio of income-poverty, by region, 1988-99

poverty in Thailand is the large dis- (198 poverty headeount = 100)

parity between regions, with the s - looNortheast having the highest inci- 90 -0-N Ntheast

dence of poverty and Bangkok hav- 80 South 80

ing virtually no poverty (Figure 1). a N_anokth1= 70 ~ ~ ~~~~~~~~~-Bangkok 7

The disparity has widened over the 11years, with poverty declining at a °60 60

faster rate in the better-off regions , 50 50

than in the poorer regions (Figure 40 40

2). The economic crisis further ex- r 30

acerbated the regional disparities in =poverty, as the incidence of poverty 20 20

increased sharply in the Northeast 101 _between 1996 and 1999, somewhat 0 oin the South and Central regions, 1988 1990 1992 1994 1996 1998 1999

and actually declined in Bangkok Figure 2 This Figure is a replica of Figure 9 in Chapter 2

'The national average of the official poverty line was Baht 878 per person per month in 1998 and Baht 886per person per month in 1999. This means that the poverty line (converted to a daily basis) was signifi-cantly lower than the minimum wage, constituting less than one-fourth of the minimum wage in 1999. Incurrent dollar terms, the official Thai poverty line in 1999 was equal to approximately US$0.75 per day.2Income- and consumption-poverty show different trends during the immediate post-crisis period, withconsumption-poverty showing a much greater increase than income-poverty from 1996 to 1998. This islikely to be due to the fact that income data in the SES are collected for a 12-month reference period pre-ceding the survey, which is itself undertaken over the entire year. As such, income data for 1998 for nearlyone-half of the SES sample capture pre-crisis (pre-July 1997) income, which might explain why income-poverty from the SES shows a relatively modest increase in 1998 (from 1996 levels) and a large increase in1999. On the other hand, consumption data are collected for a reference week, which is probably why con-sumption-based poverty shows a sharp increase from 8.1% in 1996 to 14.4% in 1998 (and flatness from1998 to 1999). The trend in consumption-based poverty is consistent with movements in GDP per capita.

Thailand Social Monitor iii


and the North.

11. The provincial differences in poverty are even greater than the regional differ-ences. These are highlighted in Figure 3, which shows a 'poverty map' of the country in1998. While the highest rates of poverty are mostly found among the Northeastern prov-inces, there are pockets of high poverty in the Northern and the Southern regions as well(e.g., provinces of Mae Hong Son, Narathiwat and Yala).

12. Since many of the poorest provinces in theNortheast are also the most populated provincesin the country, the share of all of the Thai poorwho live in these provinces is also very large. AThe SES data show that the 17 poorest (out of atotal of 76) provinces in Thailand accounted for Inearly two-thirds of all the poor in Thailand in1998. Indeed, the four poorest provinces - Nak- -. 73hon Ratchasima, Surin, Sisaket and Roi Etthemselves account for a quarter of all the poor inthe country. The fact that poverty is so highlyconcentrated in a few provinces suggests the ex-treme importance - and relative ease - of provin-cial targeting of poverty interventions.3 [1m 0-5

13. In principle, targeting poverty interven- P 13 - 21tions to a smaller geographical unit, such as a vil- 21 - 31lage, could be even more effective. The SES data 31 - 46suggest that poverty is concentrated in certain vil-lages as well. A mere 10% of the poorest villagesin the country account for nearly a thirdof all the poor, while slightly more than aquarterlof the poore vilelage aount f Figure 3: This figure is a replica of Figure 12 in chapter 2quarter of the poorest villages account forall of the Thai poor (Figure 4).4

14. The Government is in the process of implementing its most crucial rural policyinitiative, the Village Fund Scheme. The concentration of poverty in a relatively smallproportion of villages points to the need for techniques to identify such villages for tar-geting purposes. The analysis in Figure 4 refers to sampled villages, but does not providea complete list. Such statistical techniques are available and are being explored by theNESDB.

30f course, targeting of poverty interventions is never easy. What is meant here is that geographical target-ing is administratively more feasible when the poor are concentrated in a few provinces than if the poorwere dispersed over a large number of provinces.4Since the size of the SES samples is not large enough for the SES data to be representative at the villagelevel, the findings in Figure 4 should be treated as illustrative.

Thailand Social Monitor iv


15. There are many other Cumulative distribution of poverty across villages, 1998

characteristics that are corre- 100

lated with poverty in Thai- 90

land, such as occupation, 80 /

size of household, ru- 70

ral/urban location, and oth- O 60

ers, but at the top of the list O 50

is education. Individuals ,residing in households E0

headed by persons with no u 20' /

or primary education are at /10significantly greater risk of /

poverty than those in house- 0 10 20 30 40 50 60 70 80 90 100

holds headed by a secon- Cumulative % of villages (ranked by poverty incidence)

dary- or post secondary-educated person. Not only are Figure 4 This Figure is a replica of Figure 14 in Chapter 2poverty rates higher amongthe less-educated, they experienced a larger increase in poverty during the crisis. Be-tween 1996 and 1999, for instance, the headcount ratio of income-poverty increased from21% to 24% for households headed by an illiterate person and from 12% to 19% amonghouseholds headed by a primary-educated person. In contrast, poverty incidence did notchange appreciably among households headed by persons with vocational and post-secondary education.

16. Ownership of land is typically correlated inversely with the poverty rate, but inThailand poverty among small farmers owning less than 5 rai of land was increasing evenbefore the economic crisis, with the result that a larger proportion of small landholderswere poor in 1999 than in 1992. While medium-sized farmers (those owning 5-19 rais ofland) fared somewhat better, the incidence of poverty among them was not much changedbetween 1992 and 1999, although nationally poverty declined from 23.2 to 15.9%.

17. What these results demonstrate is that small and medium landholders have beenmore-or-less excluded from the general economic growth that Thailand experienced dur-ing the 1990s. One reason for this increase in poverty among small and medium farmersmay be that these farmers have experienced stagnation or declines in agricultural produc-tivity over time (MERI and World Bank, 2000). This sub-group of rural Thai householdscultivating small plots of land could be an important target group for poverty reduction,but the reasons for their declining yields are not clear from the available data. The avail-able evidence does not measure natural resource degradation, rainfall and water access, oraccess to agricultural services and marketing infrastructure, nor the key factors of laborand capital inputs into these farms, each of which could play a role in explaining theirlow and declining productivity.

Thailand Social Monitor v


Patterns of Inequality

18. These summary findings on poverty are amplified by the patterns of inequality inThailand, the subject of Chapter 3. Inequality measures are of interest because Thailandhas long shown a pattern of high income and consumption inequality relative to othercountries in the region. Income inequality increased sharply from 1981 to 1992, begantrending down after that, only to reverse itself sharply in 1999. It is unclear whether theincrease in inequality from 1998 to 1999 was a transitory event caused by the economiccrisis or whether the country is once again entering a period of rising income inequality.

19. An analysis of the sources of inter-household income inequality in 1998 suggeststhat education has the largest explanatory power in accounting for inter-household varia-tions in per capita income. The explanatory power of education is greater in the North-east than in Bangkok, reflecting the fact that education is significantly more widespreadin Bangkok than in the Northeast. The next most important factor explaining incomeinequality is the demographic characteristic of families. Differences in family size, ageand sex composition of the household, as well as the age and sex of the household headexplain approximately 10-13% of overall income inequality. There is an almost negligi-ble contribution of land ownership to overall income inequality in the Northeast. Land isnot distributed equally in the Northeast, but this only affects farm households, leaving outothers among whom income inequality is even more severe.5

20. To better under-stand income inequality in Mean real household income per capita in 1998 and percent real change

in per capita income, 1998-99, by schooling of household headThailand, a good place to 10,000 T 17.5 20

begin is the sharp in- 9,000 Mean real income per capita 3.4 15

crease in income inequal- %8000 -%change, 1998-99 I 0

ity between 1998 and 67000 /5

1999. This was a direct E 5,000 + ... ... .. ... .... ... .... ....... .............. .. ... .. ...... .. .... . .0 Xresult of the economic cri- .6,00

sis, which had a dispropor- = 3,0 1 I

15.9 ~~~~~~~~~~~~~~~'-10 *tionately negative impact 2,000 ,

on the incomes and liveli- 1,.000 - -5

hoods of the poorer sec- + L -20tions of Thai society. This None Primary Lower Upper Vocational University

tions of Thai society. This s~~~~~~~~~~econdary secondary and technical

is seen clearly in Figure 5, Completed schooling of household head

which shows that thepoorest households in Figure 5 This Figure is a replica of Figure 31 in Chapter 31998 (i.e., those headed bya person with no schooling or primary schooling) experienced the sharpest drops in real

5 The results reflect the fact that, even in the Northeast, farming is not the only occupation, and land is notthe only source of income. Indeed, only one-third of individuals in the Northeast belong to households inwhich the main occupation of the head is farming (including farm work). Furthermore, farm operatorshave on average the lowest incomes of all occupational groups in the Northeast. Thus, while land owner-ship may be an important source of inequality of incomes amongfarm households, it does not explain agreat deal of overall income inequality, as some of the richest individuals in the Northeast own no land atall. Of course, land ownership is still a very important determinant of poverty.

Thailand Social Monitor vi


income between 1998 and 1999, while those that were relatively better-off (i.e., thoseheaded by post-secondary educated persons) actually experienced an increase in real in-come between the two years.

21. Given that education Distribution of enrolled students at the primary and lower secondary

has the largest explanatory level, by real per capita income quintile, 1992-99

power in explaining income El Richest 90% 13 2 14

inequality and given its 80%

powerful role in reducing U Fourth 70%

poverty, an important ques- o 60%

tion is whether the distribu- Q 50%

tion of education in the Thai *Third 40% 22 23 23 22

population has been improv- 30% 2220% 1ing or worsening over time. I Second 10% 26 27 27 19 22

Data from the various So- 0% 1 9 9 1 1999

cioeconomic Surveys show 1992 1996 1999 1992 1996 999

that that there has been a l Poorest Primary Lower secondary

remarkable expansion ofeducation in Thailand, andthat the poorest quintile has Figure 6 This Figure is a replica of Figure 33 in Chapter 3

increased its share of lowersecondary students impressively during the decade of the 1990s (Figure 6). While only12% of lower secondary students were drawn from the poorest quintile in 1992, as manyas 22% were in 1999. Chapter 3 also shows that the poorest quintile has increased itsshare of upper secondary students in Thailand - from about 10% in 1992 to 15% in 1999.The share of the poorest quintile among vocational students has increased even more -from a mere 2% in 1992 to 8.4% in 1999. However, the share of the poorest quintile inupper secondary and vocational education remains below their population share. In addi-tion, the percentage of university students drawn from the poorest quintile is negligible -a mere 0.9% in 1999.

22. Of course, an important reason for the low representation of the poorest quintile inuniversity education is that the entry of this group into lower and upper secondaryschools is just too recent for it to have a cohort of secondary-completed students that areeligible for university admission. As more and more students from the poorest quintileenter and graduate from secondary schools, one would expect the pool of poorest studentsin university education to increase. However, this is by no means an automatic process.The cost of university education in Thailand, as in other countries, is large, and this re-stricts the access of poor students, howsoever eligible they might be, to university educa-tion. There is thus a role for public policy to proactively target educational assistanceprograms, such as student loan and scholarship schemes, to students from the poorestbackgrounds.

23. If the recent progress in enrollment rates of the poor is sustained and expanded tohigher education, it is reasonable to expect that the stubbornly high rates of inequality inThailand might steadily improve over time. This will have cross benefits to the reduc-

Thailand Social Monitor vii


tion of absolute poverty as well. As in most developing countries, education is the keyto economic advance for poor individuals and their families.

Strategic Issues

24. Chapter 4 of this report turns to the strategic issues confronting Thai policy mak-ers as they build the policy foundations for a resumption of growth and the reduction ofpoverty. Traditionally in Thailand, anti poverty programs were associated with transfersto the indigent, the elderly, and disabled. Rapid growth and policies promoting growthwere seen as more than adequate to deliver substantial income gains for all Thais, includ-ing the poor. In recent years, these views have undergone some refinement and devel-opment, as Thai debates on public policy have intensified:

* Growth rates that were once the highest in the world are now expected to moder-ate over the medium term, thus focusing attention on public policies in addressingmore directly the problem of poverty;

* The Northeast of Thailand has become the regional focal point of poverty con-cerns, as it has continued to lag behind the rest of the country and suffered mostduring the crisis;

* The crisis impacts demonstrated that private, informal safety nets were effective,but not adequate to cushion the effects of the economic contraction, particularlyon the poor;

* Highly unequal income distribution, long masked by the benefits of rapid growth,has emerged as a policy issue;

* Poverty among certain sub-groups and in certain areas appears to be chronic andresistant to the benefits of growth, posing a possible "structural" dimension topoverty.

25. The strategic issues taken up in Chapter 4 are cross-sectoral and thematic: selfreliance and informal safety nets; structural poverty; causes of poverty and the case ofthe Northeast, natural resource degradation and poverty; gender and poverty; and the roleof growth vs. income distribution in provincial and national poverty outcomes. The list isfar from complete. Sectoral issues are addressed in the program analyses of Chapters 5and 6, but some strategic perspectives are not covered in the report, such as the future ofnatural resource pricing for the agricultural and industrial sectors and its effects on thepoor. Credit policies affecting various productive sectors are undergoing significant pro-poor modifications under the Government, but are covered only partially in this report.With these other omissions noted, the following sections summarize the main strategicfindings of Chapter 4.

26. Poverty and Self Reliance. There is nothing more distinctive in the Thai socialcontext than the value and emphasis on family self reliance and mutual community-basedsupport. Thai families, especially in the lower-income groups, took concrete steps tocushion the effects of the crisis and to protect their social assets, even as their incomescontracted more deeply than those of better-off families. This pattern of self-reliance andmutual assistance can be seen in the evidence on income transfers and consumption shifts

Thailand Social Monitor viii

Poveriy and Public Policy

at the household level. As Thai policy makers address demands for more official safetynet programs, an important consideration is the admirable and socially beneficial role ofsuch voluntary family-coping strategies, and the importance of sustaining them.

27. One way in which thepoor can cope with adverseciorcumsancpes wisthadvero The volume of income transfers in relation to total income,circumstances IS through by real per capita income quintile, 1996-99

greater reliance on income 12 11.7

transfers from friends, rela- , rtives and the government. In 9.8 01996 *1998 [21999

the case of Thailand, most of ' 89 91 9.3

the transfers are from relatives, , 85 8.4r 8.3

typically those members of the - 8 7 8 7.6 7.7

family who have migrated to 7

the urban areas (typically, 6

Bangkok) in search of better- 5.2 5

paying jobs. However, the 5

economic crisis of 1997 was 4 'Poorest Second Third Fourth Richest

unusual in that it first affected Real per capita income quintile

Bangkok, and this resulted in alarge number of migrant Figure 7 This Figure is a replica of Figure 35 in Chapter 4workers in the construction sec-tor in Bangkok being laid off. As a result, the ability of these workers to send remit-tances to their families in the rural areas, especially the Northeast, was curtailed. Despitethe income losses imposed by the crisis, transfers sustained their strong role in proppingup the incomes of the less well off. As seen in Figure 7, transfers are larger in relation tototal income for the poorest three quintiles relative to the richest two quintiles, even moreremarkably, the poorest three quintiles increased their reliance on transfers between 1996and 1999. Without those transfers to the lower quintiles, poverty would have risen byover 6 percentage points.

28. Strategically, this and other patterns of self reliance and community based supportmechanisms (World Bank, 2000d) are the foundation of social protection in Thailand.Efforts to expand official safety nets in Thailand have in some cases utilized communitybased institutions to manage and allocate resources to beneficiaries, thus strengtheningthese social assets, rather than substituting for them (see chapter 5). These findings alsosuggest the importance of contributory insurance mechanisms as consistent with princi-ples of self reliance.

29. Structural poverty. Structural poverty refers to pockets of poverty that are persis-tent over time and not readily reduced with economic growth. Structural poverty thusimplies that some poor families continue to remain poor for reasons that are not easilyunderstood, while others in 'transitory poverty' are able to climb out of poverty. HasThailand reached a stage, at poverty well below 20%, where poverty has become easy tofind but "hard to reach"?

Thailand Social Monitor ix


30. First, as noted in Chapter 2, certain groups, such as farm workers and farm opera-tors (in particular, small and medium landholders), large households, households with alarge number of dependents relative to earners, and households headed by individualswith primary or less education, have had stubbornly high rates of poverty over a long pe-riod. Indeed, poverty for some groups, such as small landholders, has increased even inabsolute terms during the decade of the 1990s. All of this evidence suggests that a sig-nificant portion of the poverty among these groups is likely to be structural.

31. Similarly, the ranking of regions by poverty incidence has remained remarkablystable over time. The Northeast was the poorest region in 1988, and it continues to be thepoorest region today. Of course, poverty levels have fallen in absolute terms in theNortheast, but they have fallen far more slowly than in other regions. This suggests astructural element to poverty in the Northeast.

32. Another indication of structural poverty is seen in the provincial data on povertyincidence. A ranking of provinces by poverty incidence in each of the four years - 1994,1996, 1998 and 1999 - shows that, of the poorest 17 provinces in Thailand in 1994, asmany as 8 (Mae Hong Son, Narathiwat, Kalasin, Sakhon Nakhon, Yasothon, Nongkhai,Nakhon Phanom and Sisaket) remained in the poorest-17 group in every one of the fouryears.

33. Further telling evidence of structuralism would be that the poorest provinces arefalling further behind the national progress in poverty reduction. Chapter 4 presents evi-dence for four such provinces from 1994 to 1999. Notably all became relatively morepoor than the national average from 1994 to 1999, and also between 1994 and 1996 -years that predated the crisis.

34. The difficulty with demonstrating structural constraints on poverty reduction isthat the analysis only identifies likely symptoms of structuralism, not all its causes. Thedata do not allow the report to address soil and rainfall constraints, cultural or ethnic di-mensions, or systematic patterns of exclusion. But, for some provinces, a structural di-mension seems to be in evidence, and suggests that growth alone will be insufficient tonarrow the gap between these provinces and the better-off areas of the country.

35. Causes of Poverty. Questions on the structural aspects of poverty turn back to theunderlying causes of poverty, and in particular if the determinants of poverty (using sta-tistical techniques) are different for the consistently poorest region, the Northeast, com-pared to other rural areas of the country.

36. The results suggest that the determinants of poverty in the rural Northeast differfrom those in the rest of the rural areas in the country in four ways. First, the amount ofland owned has a much stronger effect on reducing poverty in the rural Northeast than inother rural areas. Second, while being a tenant significantly increases the probability ofpoverty in the other rural areas of the country, it has no such effect in the Northeast, per-haps because tenancy is not as widespread in the Northeast as in the other parts of thecountry. Third, somewhat surprisingly, schooling (with the sole exception of tertiary or

Thailand Social Monitor x


university education) generally has a much smaller effect on reducing poverty in the ruralNortheast than in other rural areas of the country. This surprising result may most likelyreflect the fact that the main value of education in the Northeast lies in moving out of theregion to other regions of the country where higher-wage employment opportunities canbe found.

37. Fourth, while residence in a household headed by a female at any age reduces therisk of poverty in the rural areas of the rest of the country, the situation is very differentin the rural Northeast. There, the age of the female head makes a big difference to therisk of poverty, with households headed by younger women (those aged less than 45years) having a lower risk of poverty than male-headed households but householdsheaded by older women having a substantially higher risk than their male-headed coun-terparts. These findings could be related to migratory patterns of males out of the North-east.

38. These results hold some interesting insights into potential targeting strategies inthe Northeast: anti-poverty programs in the rural Northeast, as opposed to those nation-ally, would need to target small and marginal farmers (not tenants) and householdsheaded by older women. These results also indicate that the Northeast has distinctivecharacteristics and that public policies devised nationally may need to be altered whenapplied to this region, which accounts for nearly half of the country's poor population.

39. Growth, distribution, and poverty. At the center of many debates on poverty sitsthe question of the role of income growth vs. its distribution in a country's poverty situa-tion. Chapter 4 examines the role of income distribution changes and their importanceover time at the national and provincial level.

40. The empirical results point to one reason why poverty reduction has been sluggishin some of the poorer provinces in Thailand. The results indicate that increased income

inequality is posi- Shares of each region in total GDP and population increase takingtively and strongly place in Thailand between 1987 and 1996

associated with in-I 00% _

creases in poverty, i *Southeven after control- I 1M7 E Northeast

ling for the impact of 80% -8

economic growth. A 70% 1 * North 32

one-point increase in 60% M Central

the Gini index of in- 50% 0 Bangkok and

come inequality is 40% Vicinity 22

associated with a 0.8 30%

percentage point in- 20% 52 Icrease in poverty in- 10%,

cidence. Thus, wors-ening income ine- % share in GDP increase % share in population increase

quality can substan-

tially offset the bene- Figure 8 This Figure is a replica of Figure 41 in Chapter 4.

Thailand Social Monitor xi

Poverty and Ptublic Policy

ficial effect of economic growth on poverty reduction. In addition, the empirical resultsdemonstrate that initial levels of high income inequality in a province are associated withreduced growth performance of that province in subsequent years, and this has an addi-tional dampening effect on poverty reduction.

41. This is also analyzed nationally for Thailand. Just as income growth slowed downdramatically in the aftermath of the crisis, income inequality began increasing. This ac-centuated the increase in poverty. For instance, between 1998 and 1999, poverty inci-dence increased by 3 percentage points. Had the distribution of income remained neutralbetween 1998 and 1999, the incidence of poverty would have fallen by a modest 0.5 per-centage points - not increased by 3 percentage points, as it actually did. In terms of abso-lute numbers, the worsening inequality resulted in an additional 2.1 million persons be-coming poor between 1998 and 1999.

42. This is further demonstrated by Figure 8 which shows that the Northeast, whichhas seen its population grow by 32% since 1987 has only captured 11% of the increase inGDP over this period.

43. What this suggests is that covering income inequality can play a critical role inpoverty reduction, especially in a low-growth environment. Given that prospects for fu-ture growth are guarded in Thailand, it may be particularly important for poverty-reduction policies to focus on reducing income inequality - or at least prevent further de-terioration in inequality - in the country.

44. Gender and Poverty. Chapter 4 summarizes the available information from thereport and other sources on the significance of gender in understanding and addressingpoverty in Thailand. Over the last three decades, Thailand, has achieved remarkableprogress in its social indicators. The gender gaps have narrowed, especially in health andeducation. Enrollment of women has increased at all levels, and more women than mennow are enrolled in higher education. Female workers seem to have fared better thanmales during the crisis, and female labor force participation is high.

45. But in other areas, there are clear disparities that may trap women in poverty.More women than men are employed at below-minimum wages; average wages are lowerfor women than for men; older female-headed households are at a significant greater riskof poverty, especially in the Northeast; women are a small minority of elected officeholders; and poor women are still turning to commercial sex.

46. Thus, as in other areas of social development, the situation in Thailand yieldsmixed results. Thailand has done a remarkable job in increasing access to education andhealth services to women, but more needs to be done to increase the voice of women andto reduce their sense of vulnerability and insecurity. The importance of several genderissues in the fight against poverty needs to be better appreciated by policy makers.

47. Environment. The linkages between poverty and the environment are complexand not easily defined. Access to natural resources - i.e. water resources, land and soil,

Thailand Social Monitor xii


the forest and fisheries - is critical to the everyday functioning of rural households. Forsome poor households, limited or low quality natural resources endowments trap them inlow productivity livelihoods. For others, their poverty compels them to over-exploit thenatural resources they have, causing a steady erosion in their natural capital base and intheir incomes from it. And in other instances, the poor are helpless bystanders as othermore powerful interests expropriate their natural resources, leaving them with degradedand less productive forests, fisheries and land.

48. In Thailand, rapid development over the last 30 years has been accompanied bysignificant pollution and degradation of natural resources in rural areas. Domestic sew-age, industrial effluents, and agricultural run-off have contributed to increasing pollutionof coastal, surface, and ground water. Forest cover has decreased drastically from 53percent in 1961 to 28 percent in 1989, when a logging ban was enacted, to 25 percent in1998. Deforestation is particularly serious in the Northeast, where forest cover has de-creased by nearly 60 percent between 1985 and 1995.

49. There is an increasing awareness and demand for environmental protection acrossa broad section of society (see Box 8 in Chapter 4). Civil society representatives haveunderscored the need to address environmental problems, given the clear links with pov-erty. Consensus is emerging among the Government, private sector, and civil society onthe need for increased attention to preserving the environment through better manage-ment of natural resources, broader public participation, and effective enforcement ofregulations for Thailand. Indeed, the Ninth National Economic and Social DevelopmentPlan (9 th Plan), for 2002-2006 explicitly emphasizes that sound natural resources conser-vation practices are important for building a strong social foundation as well as sustain-able poverty reduction.

50. This "nexus" of poverty and natural resource degradation is an important area forfurther analytical work and policy development, particularly in the Northeast, wherenatural resource quality and access are vital assets of the poor and over which there hasbeen episodes of conflict.

Public Programs and Poverty

51. Governments can combat poverty with a wide range of interventions, some di-rectly beneficial to the poor and narrowly targeted, while others more indirect. Typically,interventions are grouped into one of three categories: opportunity (actions that improvethe capacities of the poor to take advantage of economic opportunities, such as access toeducation, land reform, or tax relief); security (actions that help families prevent or copewith sudden loss of income or assets, as in the onset of costly health care, unemployment,or old age); and empowerment (actions that increase the power of the poor to participatein public decision making, such as building social capital, decentralizing government, orreducing corruption of public officials).

52. This issue of Thailand Social Monitor presents available evidence on severalmore important anti-poverty programs and policies in Thailand. Using data from a spe-

Thailand Social Monitor xiii


cial module of the Socioeconomic Survey of 1999, the analysis attempts to determine thetargeting and coverage of each poverty program. Since only a few programs were in-cluded in the special survey, the scope of the analysis is obviously limited.

53. How large is the Total government expenditure and average household

Thai government's poverty living standards across provinces, 2000

effort? This can be gauged 10.0by the volume of resources 9.8 1

spent on all anti-poverty 19.6

programs. In 1999, totalexpenditure on all anti- 9.4

poverty programs was ap- 92 *

proximately Baht 35 bil- 9.0

lion, which constituted E 8.8

4.2% of total government 8.6

expenditure and 0.74% ofGDP. Aggregate govern- 9.8 10.0 10.2 10.4 10.6 10.8 11.0 11.2 11.4 11.6

ment spending on poverty Log real household expenditure per capita (Baht)

programs has increased sig-nificantly in real terms over Figure 10 This Figure is a replica of Figure 44 in Chapter 6time: from Baht 4.5 billionin 1993 to Baht 21.1 billion in 1999 (in constant 1988 baht).

54. While data on the provincial distribution of anti-poverty public spending are notavailable, it is possible to examine the relationship between overall government expendi-ture and provincial living standards. Figure 10, which plots total government spendingper capita in a province against average living standards per capita in that province,shows that better-off provinces tend to receive significantly larger allocations of per cap-ita public spending than poorer provinces. While this is not an unusual finding (and re-flects the fact that local governments in better-off provinces are able to raise more reve-nues and thus finance their higher expenditures), it does suggest that there is substantialscope for central government spending to be used more strategically toward poverty-reduction goals.

55. Empowering the Poor. Chapter 5 reviews the remarkable advances that have re-shaped the role of civil society and participation in Thai institutions and in developmentitself. These advances are well known to Thais, but when seen together in an empower-ment context, the impact of these advances cannot be underestimated. The 1997 Consti-tution and the NESDB's 8th Plan are the hallmark of these advances.

56. The Constitution provides the legal foundation for community empowerment.Many of its provisions - from the decentralization of power to the local level, to the rightof access to information, and the right to participate in the decision making process at alllevels - have become powerful tools for civil society in its efforts to reform the politicaland social environment of Thai society. Complementing the Constitution, the EighthPlan (1997-2001) emphasized an agenda of "people-centered development, focusing on

Thailand Social Monitor xiv


the poor and under-privileged, and with initiatives designed to empower local communi-ties through decentralization of government, public participation in the Government'sdecision-making process, and the building of local capacities and self-reliance. The gth

Plan takes the poverty focus of the 8th Plan to a more central strategic concern.

57. Chapter 5 reviews Coverage and incidence of government health programs, by real per

other key developments on capita consumption quintiles, 1999

the empowernent front, 60 0 Poorest quintile * Second quintile O Third quintile [ Fourth quintile * Richest quintile

including decentralization 50 48

initiatives, donor programsCOM 40 ~~~~~~~38that have supported co- 4034

munity based development 30 2 27 25 23

and empowerment, and 20 19 18

new Thai institutions, such 14 12 t4

as the Community Organi- 'O A 4 3 8

zations Development Insti- o * T -- - -

tutei. % of population % share among all % of population % share among allreceiving benefit beneficiaries of receiving b t b ficiaries of

program program

58. The agenda for em- Low-Income Health Card Voluntarv Health Card

powerment-based devel- Figure 11 This Figure is a replica of Figure 45 in Chapter 6opment is a broad one, andthree issues are mentioned in Chapter 5 - the need for better quantitative monitoring andevaluation of investments in empowerment; the opportunity to better utilize formal tar-geting concepts and mechanisms; and the challenges of protecting the poor and ensuringparticipation and local voice as the Government proceeds with its decentralization initia-tives.

59. Opportunity for the Poor. Chapter 6 presents the empirical evidence on the per-formance of various programs that aim or should aim to reach the poor. An importantelement of opportunity for the poor is access to health services. Figure 11 presents thepercentage of the population, by real per capita consumption expenditure quintile, par-ticipating in the low-income health card program, which is generally meant for the poorand needy families, and the voluntary health card, which is meant to serve the near-poor.The latter entitles cardholders to free medical care for an annual contribution of Baht 500.Figure 11 also presents the share of all beneficiaries in the two programs that are drawnfrom the different consumption quintiles. These two indicators are broad measures of thetargeting efficiency of the programs.

60. Interestingly, a significantly larger proportion of the poorest quintile report havingthe voluntary health card than the free low-income health card (50% versus 19%). Ad-ministrative data also show that demand for the voluntary health insurance card increasedsharply during the crisis period, even more so than that for the low-income health card(World Bank 1999). This suggests either that the poor have difficulty obtaining the low-income health cards, or that they deliberately opt for the more expensive voluntary healthcard in the belief that they are likely to obtain more and higher-quality medical benefitswith that card. Medical staff also may have encouraged voluntary card enrollments for

Thailand Social Monitor xv


hospital revenue-raising purposes. Possession of the low-income health card is biased to-ward the poor, as it should be, with 38% of beneficiaries belonging to the poorest quin-tile. However, over a third of the beneficiaries are drawn from the three richest quintiles,which suggests large leakages of benefits to the non-poor, particularly since the incomethreshold for eligibility for the low-income health card is only Baht 2,000 per month for asingle person and Baht 2,800 per month for a family.6 As expected, the distribution ofthe voluntary health card is somewhat less progressive.7

61. Perhaps of greater Coverage and incidence of government educational assistance programs,

concern is the fact that by real per capita consumption quintiles, 1999

nearly a third of the poorest 30 Poorest quintile * Second quintile C Third quintile C Fourth quintile * Richest quintile

quintile does not report hav- 25 -2 25 24

ing either the low-income or 20 1 19

the voluntary health card. 16 16 16

Since this group is unlikely I"to have other forms of medi- '° 6

cal coverage, such as civil 5 - 3 6 4

servants' medical benefits or o iaI i - I * .* | private health insurance, it % of upper- and post- % share among all % of secondary and I % share among all

secondary students beneficiaries of post-secondary beneficiaries ofimplies that a large number receiving benefit program students receiving program

I ~~~~~~~~benefitof the poor are vulnerable to bthe adverse economic effects Government Scholarship Program Student Loan Program

of a catastrophic illness.Hopefully, these uninsured Figure 12 This Figure is a replica of Figure 53 in Chapter 6

will take advantage of theGovernment's recently-launched '30-baht Health Scheme.'

62. A vital part of the Thai government's opportunity programs is access to educa-tion, supported by several targeted programs, of which the scholarship and student loanschemes are highlighted here (due to availability of data). Only students in primary andlower secondary school are eligible for the government scholarship program, while thestudent loan scheme is open only to students at the upper secondary, vocational, and terti-ary levels. As Figure 12 shows, levels of coverage are very low for both of these pro-grams, with fewer than 5% of the eligible population of students in the country receivingany benefits from these programs. The distribution of benefits is somewhat better tar-geted for the government scholarship program than for the student loan program, which isto be expected given that eligibility for the latter is restricted to students at the upper sec-ondary and tertiary levels (which have significantly fewer poor students). However,more than one-half of all scholarship recipients and two-thirds of student loan recipients,belong to the richest three quintiles, indicating high levels of leakage to the nonpoor.

6The leakages probably come about from the fact that determination of eligibility for the low-income healthcard is at the discretion of local community leaders (e.g., the village head), who may misuse this discretionfor personal gain.7The richest quintile typically has access to other forms of health insurance, including the civil servants'Medical Benefit Scheme, the compulsory health insurance program for formal sector employees, and pri-vate health insurance.

Thailand Social Monitor xvi


63. Another potential 'opportunity' intervention is micro or small-farmer credit fromgovernment lending institutions. The Socioeconomic Surveys routinely collect informa-tion on household borrowing from various sources, including the Bank of Agriculture andAgricultural Cooperatives (BAAC).

64. Relatively few households in the SES report borrowing from the BAAC. In 1999,only 2% of all agricultural households reported borrowing from the BAAC during themonth preceding the survey. For those that borrowed, the average amount borrowed waslarge, amounting to approximately 4.4% of monthly household income. More impor-tantly, better-off agricultural households appear to account for the bulk of BAAC credit.This was true in 1996, 1998 and 1999 (Figure 13). Indeed, the data suggest that the shareof the richest quintile of agricultural households in total BAAC financing increasedsharply from 25% in 1996 to 59% in 1999. At the same time, the share of the poorestquintile of agricultural households fell from 11% to 5%.

65. The Government's proposed new scheme to forgive BAAC debt to low incomefarmers is thus well intended and designed to target the poorer farmers, but is unlikely toreach a large number of poor farmers given their apparently low participation rate in theBAAC loan programs.

Share of each real per capita income quintile in total household (gross)66. Security for the borrowing from the Bank of Agriculture and Agricultural Cooperatives,

1996-99 (agricultural households only)Poor. hapter6 also ana- 70-

lyzes some programs in 60

the security area, includ- 0 1996 * 1998 0[1999

ing the cash transfer to the D 501

elderly and public em- 40

ployment programs to 2 31

provide income support to E 22 27 25

I I~~~~~ 19 20the unemployed. In the < 201 11

fers to the elderly, a famil- L K K Iiiar pattern is seen of low Poorest Second Third Fourth Richest

coverage of the population Real per capita income quintile (among agricultural households)

and a mildly pro-poor biasin the distribution of bene- Figure 13 This Figure is a replica of Figure 54 in Chapter 6

ficiaries, with substantialleakages to the non-poor. However, public employment schemes, used during the eco-nomic crisis under various initiatives, show a far more efficient targeting of the poor withabout 47 percent of the beneficiaries drawn from the poorest quintile and only 4 percentfrom the richest. The success of the public employment schemes in targeting themselvesto the poor suggests that the wage rates that were offered under these schemes wereprobably too low to be attractive to the non-poor, with positive results to the unemployedand poor victims of the crisis. The fourth Thailand Social Monitor reported that the em-ployment schemes launched during the crisis played an important role in halting and re-versing the increases in the unemployment rate.

Thailand Social Monitor xvii


Implications for Policy

67. While Thailand's record at reducing poverty during the last two decades has beenimpressive, the economic crisis has brought the issue of poverty back to the forefront.The incidence of poverty increased sharply in the immediate post-crisis period and mayhave persisted into 2000. Rapid economic growth has substantially reduced poverty inthe past, but at the relatively low levels of poverty now present in Thailand (about 16%)and with growth prospects less buoyant than in the past, the government is likely to facenew challenges in lifting the remaining poor above the poverty line.

68. The report presents evide-, e to support this contention, both from a structuralperspective and from an analysis oL the role of income distribution in affecting povertyoutcomes. Progress in reducing poverty will depend, certainly on a resumption of vigor-ous growth, but this will probably be insufficient to achieve large gains. Such progresswill also require action from Thai policy makers in curtailing any further increases in in-come inequality and in designing effective mechanisms to target the remaining poor, par-ticularly in the Northeast.

69. The report analyzes the recent performance of government expenditure and spe-cific programs in reaching the poor. It needs to be repeated that, on balance, Thailand'spolicies are moderately effective in comparison with those of other countries. The pro-vincial distribution of overall public spending is found to be modestly regressive (othercountries show much more pronounced bias toward better-off regions and provinces), andmany anti-poverty programs appear to be well-targeted to the poor. Some programs, suchas the low-income health scheme, public employment schemes, and transfers to the eld-erly, are strongly pro-poor, but BAAC credit, scholarships and student loans are not.Education remains one of the most important avenues for lifting people out of poverty,and programs to help the poor advance to upper secondary and university educationshould be a high priority.

70. Poverty and Public Policy is only a step toward a comprehensive anti-povertystrategy with many analytical issues still to be addressed. In addition to the points al-ready made above on education, credit and overall incidence of public expenditure, somefurther preliminary implications from this study are:

Poverty is highly concentrated in a few provinces in Thailand, most of which are inthe Northeast (and a few in the South and the North regions). While the availabledata are not suitable for estimating poverty at the village level, early and tentativeresults suggest that poverty is also highly concentrated in a relatively small numberof villages in the country. This means that geographical targeting of poverty inter-ventions (such as the Village Fund) to the poorest provinces and villages is likely tobe cost-effective in reducing poverty.

* The Northeast remains the central regional target for any anti-poverty strategy. Al-though migration appears to have mitigated the population pressures, it is probablethat the better-educated, less-poor Northeast residents have migrated out of the re-gion. Evidence on the unique causes of poverty in the Northeast, and the importance

Thailand Social Monitor xviii


of natural resource issues there, indicate that strategies to reduce poverty in theNortheast may need to be adapted to the special conditions there.

* The Thai government has an impressive number of anti-poverty programs in place,and public spending on these programs has increased sharply during the 1990s.However, many of these programs have low levels of coverage, and experiencesubstantial leakages of benefits to the non-poor. Improving the targeting of suchprograms is one means to reach the more stubborn pockets of poverty;

* The patterns of participation in the BAAC farm loan programs suggests that the-debt forgiveness moratorium, although well intended, will probably reach a smallproportion of poor farmers;

* Public employment programs, which were used widely during the economic crisisas a short-term safety net mechanism, appear to be the most successfully targetedsecurity interventions, with relatively high coverage and low leakage; they alsoseem to have played a role in reversing the growing unemployment during 1999,and might benefit from a more structured and transparent institutional setting; and

* The role of family and community self-reliance and mutual support has been a rela-tively under-emphasized Thai asset in the fight against poverty, both in managingthe risks that families face in difficult times and in bringing people together(empowering them) to increase their opportunities for growth.

71. In the post-crisis environment, the efforts of government to address poverty willbecome far more crucial than in the past. Public policy choices - such as which programsand policies to support and how best to target their benefits to the poor - will becomemore central to the budget allocation and program design process than ever before. The9th Economic and Social Development Plan sets out an admirable framework for includ-ing poverty within the planning process. The next stage will be the articulation of more-detailed operational strategies to build upon the gth Plan and to confront the problem ofpoverty in Thailand.

Thailand Social Monitor xix


Chapter 1. Concepts, Measurement, and Extent of Poverty in Thailand

What Is Poverty?

1. The most commonly-used notion of poverty is the inability of an individual to attain aminimal standard of living typically defined in terms of a food basket that delivers a biologicalcalorie minimum (e.g., 2,100 calories of energy per person per day) and a small allowance fornonfood consumption such as clothing and shelter. This is the notion of absolute poverty.

2. But poverty can also be relative. Even if an individual is able to attain a minimal stan-dard of living, he may feel deprived relative to others in a society that is otherwise affluent. Al-though both absolute and relative deprivation figure in common perceptions of poverty, the for-mer tends to dominate in developing countries. Poverty can sometimes take particularly viciousforms, and these manifestations are better captured as absolute deprivation. However, as shownby Sen (1976), a measure of relative deprivation can be combined with that of absolute depriva-tion.

3. Poverty can also be defined not in terms ofincome or consumption but instead in terms of dep- Box 1: The Definition of Poverty

rivation of basic needs, such as primary schooling, The poor can be identified via income. But in-nutrition, health care, safe drinking water and hous- come is not a genuine poverty-measuring tooling. From one perspective, the worst manifesta- since farm incomes in Thailand fluctuate consid-tions of poverty include the occurrence of prevent- erably. Rather, asset or consumption capacityable child deaths and the feeling of powerlessness should serve as a better tool. If not, there arestemming from illiteracy and debilitation owing to other aspects of poverty that should be taken into

consideration. These are happiness, social accep-ill health. The basic human needs model promotes tance, political choices (if not, at least a voice that

development efforts that are geared toward meeting can be heard), and spiritual aspects. Some or-these needs directly, through the provision of goods ganizations, such as the UNDP, state that theand services, and indirectly, through income and "right not to be poor" is regarded as a humanemployment generation. right.

Source: Participant comments at a TDRI Poverty4. In spite of the importance and value of Workshop, December 22, 2000.broader notions of poverty, the starting point of anypoverty assessment is generally the standard measurement of absolute poverty. This chapter gen-erally follows the conventional absolute poverty approach, based on an economic measure ofwelfare, while later chapters consider and discuss additional dimensions and perspectives onpoverty in Thailand. Data availability, however, limits a fully comprehensive assessment of pov-erty in this Social Monitor.

Measurement of Poverty

5. The measurement of poverty is a contentious science. It involves defining a minimalstandard of living, establishing an absolute poverty line or threshold that accurately reflects thisminimal living standard across disparate areas and regions, choosing between individual con-sumption and income as a metric of welfare measurement, and then aggregating poverty across

Thailand Social Monitor 1


individuals in some fashion. Each of these steps involves choices that are not always agreedupon. As a result, there are sometimes many different poverty rates for the same country at thesame point in time.

6. This issue is particularly relevant in Thai- Box 2: Poverty is a Feelingland where there have been a number of differentpoverty lines in use over the last twenty years. "I would like to point out that poverty is a feel-

ing, which is difficult to measure. It is completelyHowever, in 1996, the NESDB developed a new wrong if we measure it with income. There arepoverty line that was officially accepted by the some people who live a simple live and don't feelThai cabinet,' and which is now used as the stan- poor, but will be under the poverty line if wedard poverty line in most poverty analyses in the measure it with income. Poverty is a feeling andcountry. This official line is used throughout this highly individual. National policy to remedy pov-

erty thus cannot put an end to the problem partlySocial Monitor, unless otherwise noted. Not only because it deals with income and not consideringdoes this poverty line vary across rural and urban other aspects of the problem. The solution is toareas and across the five regions owing to spatial examine the capability of an individual invariations in the cost of living, but it also varies utilizing and controlling his resources. As thereacross households of different sizes and demo- are other social factors involved, it is hopeless toacross~.. have the economists study poverty .............."graphic composition, as subsistence needs are as-sumed to differ by age and sex of individuals. The Source: Participant comments at a TDRI Povertynational average of the official poverty line was Workshop, December 22, 2000.

Baht 878 per person per month in 1998 and Baht886 per person per month in 1999.

7. To provide Box 3: The Thai Socio-Economic Surveyssome perspectiveon the level of the The first Socio-Economic Survey (SES) was first conducted by the National Statisticalpoverty line in Office (NSO) in 1957. It was only in. 1988 that the survey began to be undertaken

'aa nder every two years (the interval was 5 years between 1957 and 1987). The SES is under-Thailand, conside taken throughout the year, and has a typical sample size of about 25,000 households.that, when annual- For instance, the SES 1998 surveyed a total of 23,549 households comprising of 86,058ized (Baht 10,632 individuals. In 1999, the NSO conducted a special (periodic) SES to gauge the effectsper capita), the of- of the economic crisis on household income and consumption. This survey had a muchficial poverty line smaller sample size than the usual SESs (7,789 households), as it was conducted onlywas only about during four months of the year (viz., June-September 1999).

one-seventh The SES typically obtains information on household income and household expendi-(14.4%) of GDP ture, household consumption patterns, changes in assets and liabilities, owne:Qhip ofper capita in 1999. durable goods, and housing characteristics. The SES uses a stratified two-stag ,am-Converted to a pling design, with the primary sampling units being blocks for municipal areas and vil-

lages for non-municipal areas and the secondary sampling units being householdsdaily figure, the within each of these blocks/villages.poverty line is sig-nificantly lower For most years, the SES contains enough observations at the level of the province tothan the minimum calculate provincial estimates with a high degree of precision. Due to the relativelywage, constituting small size of the sample in the SES 1999, however, provincial means of poverty rateswaes, constiuti and other variables are not estimated very precisely for that year.less than one-fourth of the

'The poverty line is based on a technique developed by Kakwani and Krongkaew (1996).



minimum wage in 1999 (which averaged nationally at about Baht 135 during that year). Finally,2the official Thai poverty line in 1999 translated into approximately US$0.75 per day.

8. Once a poverty line is determined and the poor identified, individual poverty needs to beaggregated in the population. By far the most common approach for aggregating poverty is byusing the FGT class of aggregate poverty indices (Foster, Greer and Thorbecke, 1984). Theseindices include the head count ratio, the poverty gap, and the distributionally-sensitive povertygap squared indicator (in which the deprivation of the poorest gets higher weighting). The FGTclass of poverty indices has two attractive features: (i) all three indicators are special cases of ageneral form; and (ii) all indicators are fully decomposable and additive (which is convenientwhen measuring poverty for different subgroups or time periods).

9. Besides the quantitative approach outlined above, there are other means of assessing pov-erty. The poor themselves often constitute a reliable source of information about their situation,wants and needs. One of the most important considerations, therefore, in measuring poverty isthe people's involvement in that process because they are the ones who live with poverty everyday. Focus-group discussions and open-ended questions can illuminate aspects of poverty thatremain hidden from view in large sample surveys. Such qualitative methods are used routinelynow to explore such issues as structural, physical and social barriers to the poor, people's percep-tions of poverty, and family histories illuminating patterns of poverty, among others. Qualitativefindings are increasingly used to restructure survey questionnaires to explore new aspects of pov-erty using quantitative methods. Qualitative work can also be done more rapidly than nationalsurveys, and can be used, as they were in Thailand, to pose initial hypotheses on the effects of amajor shock on the poor.

Extent of Poverty and Changes in Poverty in Thailand

10. Headcount Ratio. The simplest and most commonly-used method of measuring povertyis the head-count ratio, which is simply the percentage of individuals in the population whosehousehold income per capita falls below the poverty line.3

11. Data from the Socioeconomic Survey (SES) shows a sharp decline in the incidence ofincome- and consumption-poverty from 1988 to 1996 (Table I) (Box 3).4 The incidence of in-come-poverty declined from 32.6% to 11.4%, with the number of income-poor declining fromabout 18 million to 7 million. On average, 1.4 million persons were being lifted annually out ofincome-poverty between 1988 and 1996. The onset of the economic crisis put a halt to these im-

2Note that this figure is different from the commonly-used 'dollar-a-day' poverty line, as that line refers to a dollar in1993 purchasing power parity-adjusted terms. Thailand's official poverty line in 1999 was equivalent to 1993 PPP$1.60.3Obviously, the level of poverty incidence will be sensitive to the choice of the poverty line. If the poverty line for1999 were increased by 5% and 10%, the estimated incidence of poverty would be 17.3% and 19.2%, respectively(instead of 15.9%). Conversely, a poverty line that was 5% and 10% lower than the actual poverty line in 1999would result in a poverty incidence of 14.1% and 12.2%, respectively. The main use of an absolute poverty line liesin comparing poverty rates over time and across population subgroups.4Most of the analysis in this Social Monitor is based on primary analysis of data from the Socioeconomic Surveys(SES). See Box 3 for a description of the SES.



pressive rates of poverty decline. The head-count ratio of income-poverty increased from 1 1.4%in 1996 to 13% in 1998 and 15.9% in 1999.5 This meant that an additional one million personsbecame poor in 1998, with 1999 seeing another 2 million persons added to this number.6

Table 1: Measures of Poverty, 1988-99Headcount ratio*** Poverty gap ratio Severity of poverty ratio

Based on: Based on: Based on:Year Income Consumption Income Consumption Income Consumption1988 32.6 27.6* 10.4 6.8* 4.6 2.4*1990 27.2 19.5 8.0 4.2 3.3 1.41992 23.2 17.6* 6.8 3.7* 2.8 1.2*1994 16.3 9.8 4.3 2.0 1.7 0.61996 11.4 8.1* 2.8 1.6* 1.1 0.5*1998 13.0 13.6* 3.3 2.7* 1.3 0.8*1999** 15.9 14.3* 4.3 3.0* 1.8 0.9*Note: *Based on data obtained from a short consumption module. Consumption-poverty estimates for other years are based on a long consumption module. See footnote6 in text for more details.**Data for 1999 refer to two (second and third) quarters only, while estimates for allother years are based on all four quarters of data. See footnote 4 in text for more details.***The point estimate of the head-count ratio for each year is significantly differentfrom the estimates for every other year at the 0.05 level of significance.

12. Poverty Gap. A shortcoming of the head-count ratio is that it does not say anything aboutthe depth of poverty, viz., the extent to which the incomes of the poor are below the poverty line.

5 Since the Socioeconomic Survey (SES) of 1999 was a special survey that was conducted only during the fourmonths of June-September 1999, results from earlier SESs are strictly not comparable unless data from the earlierSESs are analyzed only for the second and third quarters (which cover the June-September months). However, itwas observed that the use of two or four quarters of data for earlier SES samples made little difference to the resultspresented in this Monitor. Therefore, unless otherwise noted, statistics presented in this Monitor for years other than1999 are based on data covering all four quarters, while those for 1999 are based on two quarters of data.6The analysis of income- and consumption-poverty during the crisis period is complicated by issues of timing in thecollection of SES data on income and issues of comparability of consumption data across different SESs. Incomedata in the SES are collected for a 12-month reference period preceding the survey, which is itself undertaken overthe entire year. As such, income data for 1998 for nearly one-half of the SES sample capture pre-crisis (pre-July1997) income, which might explain why income-poverty from the SES shows a relatively modest increase in 1998(from 1996 levels) and a large increase in 1999. In contrast, national accounts data show a large decline in GDP percapita in 1998 and a recovery in 1999. Since expenditure data in the SES typically refer to consumption over theweek or month prior to the survey (there are very few items, such as education, for which expenditures refer to thepast year), consumption-poverty estimates are more consistent with national accounts data - showing a large increasein poverty in 1998 and a much smaller increase in 1999. In addition, there are issues of comparability in the SESdata on household consumption expenditure as well. Every four years (i.e., 1990, 1994, 1998), the SES obtains de-tailed data on household expenditures on 140 or so food items. In these years, the NSO calculates aggregate con-sumption figures based on this 'long' food questionnaire. However, during the other years (e.g., 1992, 1996, 1999),only a short food questionnaire is fielded, which obtains expenditure on 15 major food categories, and consumptiondata reported by the SES are based on this questionnaire. This creates an inconsistency in the consumption-povertyseries from one SES to another, as consumption expenditure based on a short list of questions typically results in anunderestimate of true consumption. For this report, we obtained from the NSO consumption data from the shortmodule for 1998 (which was a long consumption module year), and the consumption-poverty estimates reported inTable 1 for 1998 are based on data from the short consumption module. This makes the 1996, 1998, 1999 consump-tion-poverty estimates comparable.



The poverty gap index fills this shortcoming by measuring the shortfall between the incomes orconsumption expenditures of poor households and the poverty line. The sum of all individualpoverty gaps in a sample can be interpreted as the minimum amount of income transfers neededto bring all of the poor just up to the poverty line in the presence of perfect targeting.7 A povertygap of 4.3% in 1999, as shown in Table 1, suggests that, if perfect targeting were possible, anincome transfer of Baht 38 (= 0.043 x national poverty line of Baht 886) per person per monthwould be required to eliminate poverty. The total volume of income transfers required to bringall of the poor to an income level just above the poverty line would then be Baht 28 billion (=Baht 38 x 12 months x 61.66 million persons).8

13. The poverty gap measure follows roughly the same pattern over time as the head-countratio. There was a small increase in the poverty gap between 1996 and 1998, but a much sharperincrease from 1998 to 1999.

14. Severity of Poverty. A problem with the poverty gap index is that it is not sensitive to thedistribution of income among the poor and hence to the severity of poverty. A redistribution ofincome from a destitute individual to someone much better off (but still under the poverty line)will leave the poverty gap index (and the head-count index) unchanged, although it is clear thatthis redistribution will have increased the severity of poverty in the population. The poverty gapsquared index, which belongs to the broader family of Foster-Greer-Thorbecke (FGT) indices, issensitive to the distribution of income among the poor, since it weights the shortfall between anindividual's income and the poverty line more heavily the further below the poverty line that in-dividual's income falls. A shortcoming of this index is that, unlike the head-count and the pov-erty gap indices, it does not lend itself to an easy interpretation. Its use, therefore, lies mainly incomparing the severity of poverty over time or across different subgroups in the population.9

15. Like the headcount ratio and the poverty gap index, the poverty gap squared index alsofell sharply from 1988 to 1996, but then reversed itself in 1998 and 1999. The increase in allthree measures of poverty in 1998 and 1999 was sufficient to erase virtually all of the progress onpoverty reduction that had been achieved since 1994. In this sense, the impact of the economiccrisis in Thailand on poverty was profound.

Poverty and Unemployment

16. Unemployment increased significantly during the crisis, more than tripling from 0.9% to3.4% between the third quarters of 1997 and 1998 (Figure 1). As the labor force gradually in-creases again - during the crisis, many young adults returned to school, women turned to raising

7 Perfect targeting implies that each individual below the poverty line would receive a transfer that would permit hisor her income or consumption to reach the level of the poverty line and that no one above the poverty line wouldreceive any transfers.8It should be realized that these are hypothetical numbers. Besides the fact that perfecting targeting is rarely possiblein practice, few developing-country governments would choose to continue making income transfers to the poor inperpetuity. In addition, there would be significant disincentive effects of making transfers to each poor individualbased on his or her income shortfall (relative to the poverty line).9AII of the three measures discussed here - viz., the head-count ratio, the poverty gap index, and the poverty gapsquared index - are fully decomposable across different subgroups in the population.



families, and older citizens retired early - the labor markets have done a modest job of absorbingworkers. Over 650,000 Thais entered the labor force between the third quarters of 1999 and2000, and the labor markets generated nearly one million jobs, lowering the unemployment ratefrom 3.0% to 2.4%. However, the May 2001 figures indicate the recovery may be slowing. Thelabor markets, unlike earlier in the recovery, were not able to absorb all the new entrants to thelabor force, and the unemployment rate rose slightly to 4.2% from 4.1% year on year. Unlike therecovery between 1991 and 1992, this recovery has been modest and protracted so far.

Number of employed persons and unemployment rate, thirdquarter, 1989-2000

33,500 4.0- No. of employed persons

z 33,000 - + Unemployment rate (%) 3.50, 32,500- _ 3.0-

0.~~~~~~~~~~~~~~~~~~~~~~~~~3

32,000 A 2-

° 31,500-0. 2.0 E

E 31,000 >

30,500 E

E 30,000 - 1.0'

Z 29,500 0.5

29,000 0.089 90 91 92 93 94 95 96 97 98 99 00

Figure 1

17. Throughout the crisis, labor force data (based on the Labor Force Surveys of the NationalStatistical Office) show that the lowest wage and poorly-educated workers have endured dispro-portionately large impacts. These patterns were analyzed in Thai Workers and the Crisis (WorldBank 2000) and still persist. This can be seen from regional data on the labor force. All the re-gions are rebounding with declines in unemployment (Figure 2), but the poorest region - theNortheast - is the slowest. While the other regions were recovering by 1999, the unemploymentrate peaked that year in the Northeast. The third quarter 2000 unemployment rate in the Northeastis still nearly five times that of the third quarter 1997 rate - viz., 3.2% and 0.7% respectively.

18. Another indicator of the impact of the crisis is the number of employed with a primaryeducation or less. The percentage of the workforce which has a primary education or less has de-clined throughout the past six years, falling from 76.7% of the workforce in the first quarter of1995 to 65.9% in the first quarter of 2001. During the crisis, this indicator dropped more steeplyfrom 75.2% to 71.4% in one year (between the third quarters of 1997 and 1998), as uneducatedworkers lost employment and dropped out of the labor force in large numbers. Labor force pat-terns thus reflect the poverty increases recorded in the income and consumption data, as lower-educated workers (a reliable proxy for income-poverty) tended to take the brunt of the crisis.



Unemployment Rate by Region, third quarter, 1996-2000

5

4.5

4 -4--- Bangkok3/5 wMetropolis

-11\-- Central3 / /t\ \ Region

s Northern

2.5 Region-/E--X Northeastern

2 Region-ME-Southern

1.5 Region

1

0.5

0~3Q96 3Q97 3Q98 3Q99 3Q00

Figure 2

International Comparisons

19. How does poverty in Thailand compare to that in the other countries of the Asia-Pacificregion? Obviously, to compare poverty across countries, one cannot use national poverty lines.A commonly-used international poverty line is $1.50 per day per person (expressed in 1993 pur-chasing-power parity dollars). Additionally, since comparable data are available for consump-tion- poverty for the other countries, we need to use the measure of consumption-based poverty.

20. Figure 3 shows the trend in consumption-poverty over the 12-year period from 1987 to1999 for four countries in the East Asia region. Thailand falls between Indonesia and Philippineson the one hand and Malaysia on the other in terms of the extent of poverty. During the early1990s, three of the four countries - Figure 3Indonesia, Malaysia and Thailand -

Headcount ratio of consumption-poverty, 1987-99, selected Asian countriesexperienced dramatic and compa- (poverty line defined at 1993 PPP$1.50 per person per day)

rable rates of reduction in poverty 5

incidence. (The Philippines was 45dones-

the only country among the four to 40 Philippines

experience virtually no change in =35 .. -- ,"

poverty between 1988 and 1994.) E3Oi '30 -

B25

21. In the aftermath of the eco- .B20 -- Thailand

nomic crisis, Indonesia, Malaysia °15 .5.L

and Thailand all experienced an \ Malaysia

increase in poverty. (Post-crisis 51data on poverty are not available 0 -

for the Philippines.) However, the 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999


Poverty and Plublic Policy

increase in poverty in Indonesia was significantly sharper and more pronounced (from about 26%in 1996 to 37% in 1999) than in Thailand. While Malaysia too experienced an increase in pov-erty in the aftermath of the crisis, its level of poverty incidence in 1998 was almost negligible(0.74%).

22. What these data indicate is that, despite some differences, Thailand's overall performanceon poverty reduction closely mirrors the experience of at least two other countries in the region -Malaysia and Indonesia.

Overlap Between the Income-Poor and the Consumption-Poor

23. In Thailand, the convention has been to measure poverty and inequality based on individ-ual and household incomes. This approach is different from that adopted in many other develop-ing countries, where poverty is typically measured by the shortfall of consumption expenditurerelative to the poverty line. The choice of income versus consumption expenditure as the metricwith which to measure poverty is important because poverty estimates can differ significantlyusing the two approaches. Which is the best measure for poverty analysis in Thailand and why?

24. Consumption is generally believed to be a better indicator of an individual's long-termwelfare than current income. This is because individuals and households tend to smooth con-sumption over time by saving when income is high and borrowing when income is low (relativeto 'permanent' income). In addi- Figure 4tion, income is relatively more sub- Indices of mean monthly real income and consumption

ject to measurement errors than expenditure per capita, 1988-99 (1988 = 100)

consumption, especially in a de- 200 2 7964

veloping country where a signifi- @ Income

cant portion of the income is from 1 - Consumption

self-employment activities (such as 63.6

farming). 160 -

25. At the same time, the use of 140 -8

consumption expenditure as a /measure of individual welfare is 120

also not free of problems. House- 126

hold expenditure on consumer dur- 1099 1990 1992 1994 1996 1998 1999

ables and other unexpected items(such as expenditures on a catastrophic illness) can cause considerable noise in the data. Addi-tionally, from a conceptual viewpoint, an individual's welfare should be based on his or her op-portunity for or 'capability' of consumption - not actual consumption - and income more accu-rately reflects these opportunities or capabilities.

26. How sensitive are poverty estimates in Thailand to the use of the income or consumptionmeasure? Figure 4, which shows indices of mean real income and consumption expenditure percapita over the 1988-99 period, indicates that living standards, whether measured by either in-come or consumption, improved dramatically between 1988 and 1996, but with income growing


Poverty and Putblic Policy

faster than consumption. With the onset of the economic crisis, income growth slowed downsharply while consumption actually declined. In principle, the difference between the incomeand consumption expenditure lines in Figure 4 reflects the amount of household savings per cap-ita, but, in practice, it also reflects errors in the measurement of consumption and income. Thisis particularly likely to be the case in Thailand, where the SES measures household income dur-ing the 12 months preceding the survey but consumption expenditure during the past week.

27. The poverty headcount ratio follows virtually identical patterns, depending upon whetherincome or consumption is used as the metric of welfare. By both criteria, poverty fell sharplybetween 1988 and 1996, and increased in 1998 and 1999 (Figure 5). What is interesting to note,however, is that the incidence of consumption poverty is typically lower than that of income pov-erty for most years (with the sole exception of 1998), even though mean income per capita istypically greater than mean con- Figure 5sumption per capita (Figure 5). This Percentage of poor based on income and consumption expenditure,

is explained by the fact that the dis- 1988-99

tribution of income is more unequal 35 32.6

than that of consumption expendi- 30 -*-Income

ture - i.e., consumption is greater 27.2 - Consumption

than income at the lower end of the 25 27.6 23.2

distribution while income exceeds \consumption at the upper end of the & 20 163

distribution. Since the poverty line 19.5 13 15.9

is typically based at the lower end of l5.414.3

the welfare distribution, estimated 10 13.0

consumption poverty is lower than 98 8.1

estimated income poverty. 5 iTT 1988 1990 1992 1994 1996 1998 1999

28. Figure 6 shows the cumula- Cumulative distribution of per capita income and expenditure, 1999

tive distribution of per capita income 1.0

and consumption expenditure in 0.9

1999. Because the poverty line oc- .os8curs below the point of intersection t 0.7 Per capita expenditure

of the two curves, a slightly larger 0 0.6-

proportion of the population is in- . 0.5- Per capita income

come-poor than consumption-poor. /

However, if the poverty line were to 0.3

the right of the intersection point, theproportion of consumption-poor 02

would be greater than the proportion 0.1

of income-poor. 0.0 - -5.0 5.5 6.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5 10.0 10.5 11.0 11.5 12.0

29. The fact that estimates of in- Figure 6 Log ofper capita income/consumption expenditure

come- and consumption-poverty have followed similar trends over time (Figure 7) may give theimpression that there is a large overlap among those identified as poor by both criteria. However,such an impression would be inaccurate. Figure 7 shows that, in 1999, only 62% of individuals



that were identified as poor by the consumption criterion were also poor by the income criterion.Of those identified as income-poor, only 56% were also consumption-poor. The degree ofagreement between the two measures of poverty is thus low, and has fallen since 1988, with asignificant increase during the crisis. Also shown in Figure 7 is the proportion of Thai peopleidentified as both income- and consumption-poor, and this too fell between 1988 and 1996, butthen increased in 1998 and 1999. Why is this so, and what does it mean for understanding pov-erty in Thailand?

Figure 7Agreement between income- and consumption-poverty measures, 1988-

ture measures long-run or perma- 80 72.5 72.899

nent income, it is tempting to think 70 68.8

of the consumption-poor as the 613 62.2

permanently- or chronically-poor 60 * 5 52 2 52.2 53 62.2056.0

and the income-poor as the transito- 50 4 '

rily-poor in Figure 7. However, as 43838.6 - -

noted earlier, consumption expendi- ------ ture and income are measured over 30 -Percentage ofconsumption-poor who ae also income-poorture over 30 ~~~~~~~~~~~~~~~~~~- a* Percentage of income-poor who are also consumption-poor

different intervals in the SES, with 20.0 -,-% of income- and consumption-poor

consumption being measured over a _4 2.1 8.9

reference week and income being '0 -

measured over 12 months. This 0--complicates the interpretation of the 1988 1990 1992 1994 1996 1998 1999

different trends observed in Figure5, as they could well be the result of differences in the temporal measurement of income and con-sumption.

31. What about the proportion of the Thai population that is both income- and consumption-poor? Figure 7 also shows the proportion of the population that is identified as poor by both cri-teria. This measure could serve as the lower bound on the incidence on poverty in the country.Thus, for 1999, one could say confidently that, at a very minimum, 8.9% of the Thai populationwas poor, although the corresponding rates of poverty by the income and consumption criteriawere 15.9% and 14.3%, respectively.

Reliability of Survey-Based Estimates of Consumption and Income

32. In addition to the low overlap between the income- and the consumption-poor, there isalso the issue of reliability of SES estimates of income and consumption expenditure. In otherwords, how accurately does the SES estimate income and consumption relative to their true val-ues? While true consumption and income are never observed, it is possible to compare the SESestimates of aggregate income and consumption expenditure to those obtained from the system ofnational accounts (SNA). Unfortunately, there is a large divergence between the SNA- and SES-based estimates of income and consumption expenditure, with estimates of both income and ex-penditure obtained from the SES data being 32-42 percent lower than those obtained from theSNA (Table 2). Further, the discrepancy between the two sets of estimates varies by the sourceof income. While wage income is underestimated by the SES by merely 16 percent (relative to



the national accounts estimates), farm and non-farm profits are underestimated by 44-53 percentand income from rent, interest, dividends and other sources is underestimated by 132 percent.This means that the incidence of poverty is likely to be overestimated with household survey datarelative to national accounts data. On the other hand, income inequality is likely to be underes-timated by household survey data, since income from interest, dividends and rent, which is oftenan important source of income inequality, appears to be heavily underreported in the SES.

33. Two points are worth mentioning here. First, this discrepancy between national accounts-and household survey-based estimates of income and consumption is not unique to Thailand. Infact, this is a common problem in many developing countries. Second, there is no assurance thatthe national accounts estimates of income are more reliable or accurate than survey estimates.Indeed, the national accounts estimates themselves are based on labor force, enterprise, and agri-cultural crop surveys as well as on administrative data on production and tax collections - all ofwhich are likely to be subject to measurement errors.

Table 2: Aggregate household consumption and income, by source,SES and national-account estimat s, 1998 (billions of current baht)

National % underesti-Measure SES accounts mation by SESHousehold expenditure 1,914 2,529 32.2Household total income 2,157 3,058 41.8Wages and salaries 1,142 1,326 16.1Farm profits 253 363 43.5Non-farm profits 502 768 53.0Rents, interest, dividend & other income 259 601 131.8

Policy and Research Issues

34. The economic crisis had a large impact on poverty in Thailand, especially when viewed inthe context of the strong poverty decline that had taken place in the 1988-97 period. Indeed,many of the gains in poverty reduction achieved since 1994 were erased during the crisis years.The analysis in this chapter suggests that although the choice of consumption versus income as ametric of living standards does not affect greatly the incidence or the trend in incidence of pov-erty, it makes a large difference to which households and individuals get identified as poor ornon-poor. This low overlap between the two measures of poverty means that any attempt to tar-get the poor will reach different groups of people depending on which measure is used. Policymakers have a choice of targeting the more volatile (transitory) income-poor, or the more long-term (perhaps structural) consumption-poor. Further analysis of this low overlap is thus an im-portant priority. In addition, the significant discrepancy between household survey- and nationalaccounts-based estimates of mean income is troubling and needs further exploration.


Poverty and Putblic Policy

Box 4: Dissemination of Information on Poverty

The Office of National Economic and Social Development Board (NESDB) has issued several developmentpublications each year, including the quarterly Economic and Social Development Journal, the monthlyDevelopment News Bulletin, and the Annual Report on National Income Accounts, including the newsletteron Prosperity Decentralization. The latter aims to disseminate information about program implementation,particularly on programs of rural development and redistribution of prosperity to the rural areas. Thesepublications are produced in Thai for disseminating development issues and outcomes to the general public.

Recently, the Development Evaluation Division (DED) of the NESDB received technical assistance fromthe Asian Development Bank to issue a quarterly newsletter in both Thai and English on "Indicators ofWell-Being and Policy Analysis." An initial aim of this newsletter was "... to disseminate information onrecent research on the development of indicators of the Thailand's 8th National Economic and Social De-velopment Plan..." Taking into account the holistic approach and main development strategies of the 8thplan, the first issue suggested seven components of well-being - health and nutrition, education, workinglife, family life, economic growth, poverty, inequality and economic welfare, environment and safety, andgovemance. Four of the issues of the newsletter presented substantive information about poverty, such asthe development of poverty lines based on nutritional requirements and spatial price variations, the inci-dence of poverty by geographical and demographic characteristics, the impact of the economic crisis onpoverty. These issues were based on household survey data from the Socioeconomic Surveys and the LaborForce Surveys. The response to these issues was overwhelming and they have been widely quoted and dis-cussed.

After completion of Asian Development Bank's technical assistance, the NESDB plans to finance this pub-lication out of its regular budget. Future publications were expected to address export, labor, health andnutrition, and education, respectively. During the 9th Plan, the newsletter will focus on analyzing the sevencomponents of well-being and monitoring development performance, using data from the SocioeconomicSurveys.

Source: NESDB.



Chapter 2. Who are the Poor of Thailand?

35. In order to target poverty-reduction programs to the poor, it is important to know their ob-served characteristics, such as their place of residence, the geographical regions in which they areconcentrated, their educational levels, their family size, and the usual occupations in which theycan be found. This profile of poverty can also serve as a baseline of information against whichthe impact of poverty programs can be measured. The profile can also be used to help estimate thecosts of targeting the poor for a specific program.

Region of Residence

36. Geographic targeting is by far the most commonly-used method of directing policy inter-ventions to the poor, as it is administratively easier to implement than other forms of targeting,such as household income-based targeting, or individual needs-based targeting. But the successof geographic targeting depends largely on whether the poor are concentrated or dispersed acrossgeographic regions.

37. In Thailand, there is an ex- Figure 8tremely strong regional dimension h50 by region, 1988, 1996 and 1999

to both poverty levels and the reduc- _tion of poverty over time. Histori- 01988 U 1996 451999

cally, the Northeast has been the 40

poorest region in the country, fol- 0 35 31 33 32

lowed by the South, North, Central E 3 0 1 r 27

region, and Bangkok (in that order). 2 25 i l

In 1988, for instance, nearly one- 20

half (48.4%) of all Northeastemers e 5 *

were poor, as compared to only 6%of Bangkok residents (Figure 8). By1996 - the year before the economic 5 *crisis - the incidence of poverty had Northeast South North Central Bangkok

declined in all regions, but theNortheast continued to be the poor-est region in the country. In the period after the crisis (1996-99), the incidence of poverty in-creased sharply in the Northeast, somewhat in the South and Central regions, but declined inBangkok and the North. The fact that the relative ranking of the regions by the incidence of pov-erty was roughly similar in all three years (1996, 1998 and 1999) suggests that the regions thatwere poor to begin with did not experience more rapid poverty reduction.

38. In fact, the evidence suggests exactly the opposite - viz., that poverty has declined morerapidly in the better-off regions than in the poorer regions. This is observed clearly in Figure 9,which shows the headcount ratios of poverty from 1988 to 1999 across the five regions, startingfrom a common base of 100 in 1988. Figure 9 shows that the Northeast had the smallest relativedecline in poverty between 1988 and 1996, with the incidence of poverty in 1996 being 40% of its1988 level. The South and the North were also relative laggards in poverty reduction, having



poverty rates in 1996 that were Figure 9about a third of the corresponding Headcount ratio of income-poverty, by region, 1988-99

(I1988 poverty hieadcount 1 100)1988 levels. Bangkok was the best 100 (188oeryha -c 100

performer, having achieved a pov- 90 -4-Northeast 90

erty headcount in 1996 that was 80 4-- South 80

only 10% of its 1988 level. -^-anko_Northo70 \ \\ Bangkok 70ll \ \ \ \ \ +~~~~~~~~Centmil,,~60 60

39. The regional disparity in 0 50

poverty reduction performance ac- ' 40 40

tually grew wider with the onset of .1 -0

the crisis. Between 1996 and 1999, 30\0

the incidence of poverty increased 20 20

sharply in the Northeast, somewhat 10 10

in the South and Central regions, 0 0 O

and actually declined in Bangkok 1988 1990 1992 1994 1996 1998 1999

and the North. Thus, by 1999, thepoverty headcount in the Northeast was nearly two-thirds of its 1988 level, while the correspond-ing ratio for Bangkok was a mere 3%. The North was somewhere in between, having a povertyheadcount in 1999 that was one-third of its 1988 level.

40. The sharply adverse impact of the crisis on poverty in the Northeast, the South, and theNorth may seem puzzling in view of the fact that the economic crisis of 1997 originated in thefinancial center of Thailand (viz., Bangkok) and that the rural areas in 1998 enjoyed favorableagricultural prices. However, several mechanisms seem to have transmitted the initial Bangkokfocus of the crisis to the rest of the country and particularly to the Northeast. As the financialshock spread to the real economy shortly after July 1997, aggregate demand fell, and firms re-sponded by cutting back on employment and reducing the wages paid to employees. This impactfell heavily on the low-wage sector and on the self-employed 'own-account' workers. This re-sulted in a modest decline in the amount of remittances that migrant workers in Bangkok andelsewhere sent to their families back in the Northeast. In addition, although there was no large-scale return migration of migrant workers out of Bangkok, the heavy volume of out-migrationfrom the Northeast, which had been a mainstay of Thai labor markets in the last two decades, es-sentially ceased during the crisis years, shutting down a major income source for the Northeastand its excess labor.'0 Thus, the collapse in demand, both consumption and investment, spreadquickly and hit the poor the hardest. Also, the decline in remittances and the reduced outflow ofmigrants from the poorer regions of the country to the better-off regions had adverse effects onhousehold incomes and poverty in the Northeast as well as in the North and the South.

1 0See World Bank (2000) for a discussion of the trends in migration in the aftermath of the crisis. Chapter 4 in thisMonitor also discusses the remittance and migration issues.



Rural-Urban Differences

41. The differences in poverty Figure 10across rural and urban areas mirror Headcount ratio of income-poverty(%),

those across regions. Poverty is by rural/urban area, 1988,1996 and 1999

most pronounced in the rural areas, 40

followed by sanitary districts and 35 | 1988 n 1996 01999

municipal areas, and this was true inevery one of the years 1988, 1996 30_ ' and 1999 (Figure 10). Figure 11 25 22

shows that the municipal areas ex- . 20

perienced the most rapid decline in r. 15

poverty between 1988 and 1996, while the rural areas enjoyed the 10 9 8

smallest decline. Between 1996 and 5

1998, the incidence of poverty in- *creased significantly in the rural ar- Rural Sanitary Municipal

eas and the sanitary districts, whileit continued to fall in the municipal areas.

42. What Figure 10 suggests is Headcount ratio of income-poverty, by rural/urban areas, 1988-99

that poverty is now largely a rural 100 (1988povertyheadcount=100)

problem in Thailand. Of course, 90

there is some poverty in the sanitary 80 -4-Rural

districts as well, but it is relatively a -\ Sanitary districts70 ~~~~~~~~~~~~~--Municipal areas

modest in comparison to that found \in the rural areas.' This means that '\policies to combat poverty in Thai- 5

land must focus largely on the rural n 40

areas. t 30

20

A Provincial Poverty Map lo

43. Large as the regional differ- 1988 1990 1992 1994 1996 1998 1999

ences in poverty shown in Figure 12 Figure 11are, they pale in comparison to the provincial variations in poverty incidence. For instance, anaverage poverty headcount ratio of 24% in the Northeast in 1998 masks the enormous disparity inpoverty rates across the two provinces of Mahasarakham (5%) and Surin (38%). Figure 12 high-lights these provincial differences by showing a 'poverty map' of Thailand.12 The Northeasternprovinces of Surin, Kalasin, Sakon Nakhon, and Roi Et, the Northern province of Mae Hong Son,

" Of course, the near-zero rates of income-poverty in municipal areas and sanitary districts depend upon the selectionof the spatial price index, which 'corrects' for the higher cost of living in urban areas. The use of a different spatialprice index might yield higher rates of income-poverty in the urban areas. Additionally, urban areas might have non-negligible rates of, say, nutritional-poverty or health-poverty (as opposed to income-poverty).12The poverty map is shown for 1998 instead of 1999, since the Socioeconomic Survey (SES) of 1999 was a specialsurvey that had a much smaller sample than the usual SES, and this sample was not representative at the provinciallevel.

Thailand Social Monitor 1 5


and the Southern provinces of Narathiwat and Yala, are observed to have some of the highestpoverty rates in the country. Poverty is also widespread in other Northeastern provinces and in afew of the Northern and Southern provinces.

How Concentrated are the Poor Geographically?

44. If the objective of policy is to try to lift the largestnumber of poor out of poverty with special interventions,it is more useful to look at the distribution of the pooracross provinces - not at the incidence of poverty acrossprovinces. The two measures will yield different answersif the provinces that have the highest rates of poverty in-cidence are generally not as heavily populated as the prov-inces having lower rates of poverty incidence. In the caseof Thailand, however, many of the Northeastern provinceshaving the highest incidence of poverty are also amongthe most populous provinces in the country. For instance,the province of Khon Kaen is the third most populous j 0- 5province in the country, and the provinces of Udon Thani, { ' 5 - 13Buri Ram, Si Sa Ket, Surin and Roi Et are all among the 13 - 21ten most populous provinces in the country. As a result, 21 - 31the two measures of poverty incidence and the distribution a 31 -46of the poor yield virtually identical results.

-4

45. Figure 13, which shows the in- Figure 12: Incidence of income-poverty (%) bydividual as well as the cumulative con- province, 1998tribution of the poorest 17 provinces tothe total number of poor nationally,indicates that only 17 (out of a total Contribution of the 17 poorest provinces to national poverty, 1998

of 76) provinces accounted fornearly two-thirds of all the poor in : C v.t

Thailand in 1998. In fact, the four < 60 * Contributionto poverty5o *~~~~~ Contribution to poverty ,

poorest provinces accounted for a a 50o

quarter of all the poor in the coun- 0 140 36

try. The fact that poverty is so O 1 28.3

highly concentrated indicates the 0 0 18.7

extreme importance - and relative i 20 t

simplicity - of provincial targeting E 10 t7.362 5.2 5.0 4.6 4.4 4.3 4.1 3.9 3.7 3.5 3.1 2.5 2.2 2.2 1.9 1 8

of poverty interventions. 0 . H

46. In principle, targeting pov- \& rOwerty interventions to a smaller geo- \graphical unit, such as a village, Figure 13could be even more effective. TheSES data are suggestive of poverty in Thailand being extremely concentrated in a relatively smallnumber of villages. For instance, in 1998, a mere 10% of the poorest villages in the SES sample



accounted for nearly a third of all the poor, and slightly more than a quarter of the poorest villagesin the sample accounted for all of the poor (Figure 14). Since the SES covers only a fraction ofall the villages in the country andthe number of sampled households Cumulative distribution of poverty across villages, 1998

in each village is too small to be 100 - -- _

representative, these numbers are 90

merely suggestive of possible pat- 80

terns. There are promising new s 70

methodologies available for more o 60

accurate identification of village < 50]

poverty on the basis of merged , 40

household survey and population E 30

census data (see Box 5). It would u 20 /

be worthwhile to explore the use of 10

such methodologies to identify the 0 -T T -

poorest villages in the country, so 0 10 20 30 40 50 60 70 80 90 100

that government initiatives, such as Cumulative % of villages (ranked by poverty incidence)

the Village Fund Scheme, could be Figure 14better targeted to needy villages.

Household Size and Composition

47. One near-universal identify- Headcount ratio of income poverty, by household size, 1992-99

ing characteristic of the poor is their 34

large family size. In Thailand, as Iwell, there is a very strong positive 30 u 1992 0 1996 Cl

relationship between poverty inci- 25

dence and household size, which | -seems to have become stronger over 8 20

time (Figure 15).13 For example, in 15

1999, the incidence of povertyamong one-person households was 107only 0.8%, while it was as high as 5 i27.4% among households with z L LU.seven or more members. ° -

1 2 3 4 5 6 7 rmore o

Household size (number of members)

48. The data also indicate pov- Figure 15erty declined much more rapidly dur-ing the 1990s among smaller households relative to larger households. For instance, poverty in-cidence among one-person households in 1999 was less than a third of its corresponding level in

13The strong relationship between poverty incidence and family size shown in Figure 15 may be overstated sincethere is no adjustment for economies of scale in household consumption in the construction of the Thai poverty line(although there is an adjustment for household age-demographic composition differences). To the extent that thereare economies of scale/size in household consumption, the basic consumption expenditure needs of successivehousehold members will be lower than indicated by the poverty line, and this will result in an overestimation of pov-erty incidence for larger households.



1992, but among households with seven or more members the incidence of poverty in 1999 wasas high as 82% of its 1992 level.

49. In addition to size, household composition is another important correlate of poverty.Households with a large dependency ratio - i.e., the ratio of the number of children under 15 andadults over 60 to the number of adults aged 15-60 years - are much more likely to be poor thanhouseholds with a smaller dependency ratio. In 1999, households with a dependency ratio ofmore than one were more than three times as likely to be poor than households with a dependencyratio of zero (Figure 16). As in the case of household size, poverty declined more rapidly duringthe 1990s among low-dependency households than among high-dependency households.

eHeadcount ratio of income poverty, by dependency burden, 1992-9950. Neither of these relation- 4542

ships requires much explanation. In 40

most cases, an increase in house- 35 0 1992 * 1996 O 01999

3'1hold size comes about due to an in- 30

crease in the number of children. 26 2625

Thus, larger households are also the . X9

ones with higher dependency ratios. 1520 18

As the number of dependents in- 15 14 13

creases relative to earners, there is 10 10

less income (and consumption ex- 5

penditure) available to each house- 0 -

hold member and thus more pov- 0 00-5 0.51-1.00 >1.00

erty. What the survey data addition- Dependency burden (Ratio of members aged 0-14 and 60 & over to those aged 15-59 years)

ally show is that large households Figure 16with a high dependency burden aremore likely to stay in poverty over time, with the result that poverty rates decline much less rap-idly for these households than for other types of households.

Age and Sex of the Household Head

51. Poverty typically shows a U-shaped relationship with the age of the household head inmost countries. However, in Thailand, there is an additional sex dimension to the relationship.Among male-headed households, the incidence of poverty is lowest among households headed byan individual under the age of 30, but it does not vary significantly across heads who are above 30years of age. The lower poverty rate among households with younger heads probably reflects thefact that young household heads are more likely than older heads to be better-educated. 14

14 Of course, the lower poverty rate among young adult male household heads could in part be due to such householdsbeing small in size. Since the poverty lines used in Thailand assume no scale economies, any household characteris-tic that is correlated with small household size is going to appear less poor.



52. In the case of female-headed Figure 17households, however, poverty has a Headcount ratio of income poverty, by age and sex of household head,

much sharper gradient with respect 1992-99

to the head's age. For instance, in 25 24 25

1999, the incidence of poverty was 2 5 01992 *1996 C11999 23

eight times greater among house- 20 18 1 18 1 8 1 9

holds headed by a woman aged 60 0 13 16 13

and over than among households 8 1 2 13

headed by a woman under the age & I 0 1 2

of 30 (Figure 17). 101 8 n853. However, a comparison of 02 | lpoverty rates among male- and fe-male-headed households does not under 30 I 3044 45-59 60 & over under 30 1 3044 45-59 60 & over

show higher poverty rates among Male-Headed (Age of head) Female-Headed (Age ofhead)

female-headed households, as iscommonly believed. Indeed, the incidence of poverty is typically lower among female-headedthan among male-headed households. 15 The reasons for this are not clear; perhaps, female-headedhouseholds are those in which the adult male has migrated temporarily to an urban area for betterwages. If so, female-headed households are more likely than male-headed households to receiveremittances that keep them out of poverty.

54. The data also indicate that households headed by young women (those under the age of 30years) saw the sharpest decline in poverty during the 1990s (Figure 17). The incidence of povertyin this group declined from 10% in 1992 to merely 2% in 1999. Indeed, this was the only demo-graphic group experiencing a decline in poverty incidence even during the crisis period (1996 to1 999).

Education of the Household Head

55. Another important and Figure 18universal predictor of poverty is Headcount ratio of income-poverty, by household head's schooling,

education. In Thailand, espe- 35 34 1992-99

cially, where there has been an 30 _

enormous expansion of education 26

in the last two decades, one would 25 21 _ 1O1992 0 1996 * 1999

expect to find sharp differences in 8 20 19

the incidence of poverty across 15 - 12

educational levels.10

56. This is observed to be the 5s 4 i44case in Figure 18, which shows LIU _ -°the headcount ratio of poverty by None Primary Lower Upper Vocational & University

the completed schooling of the Secondary Secondary TreaicnheCompleted school of head

15The exception is households whose heads are 60 years and older. In this group, the poverty rates among male- andfemale-headed households are similar.



household head. Individuals residing Percent change in real income per capita, 1996-99, by household head's

in households headed by persons with 15 schooling

no or primary education are at much 12

greater risk of poverty than those in 1 'households headed by a secondary- or ' 5

post secondary-educated person. Lia -2

57. Not only are poverty rates - None rimar Lower Upper Vocational & University

higher among the less-educated, Fig- 1l,0 S conda V Secondiny Teacher

ure 19 indicates that the less-educated , Training

experienced significantly larger de-clines in real income per capita than -20 19 -18

the better-educated during the crisis. Completed school ofhead

For instance, between 1996 and 1999, Figure 19real income per capita fell by 13-19%among households headed by persons with lower secondary schooling or less, while it increasedby 5-12% among households headed by a person with upper secondary and vocational education.Households headed by a university-educated person saw their real incomes per capita decline by amodest 2%. These results contradict the popular impression that the economic crisis of 1997-98largely affected the educated middle-class in Thailand.

58. What the data indicate is that in Thailand, as in other countries, education beyond the pri-mary level is a powerful means of climbing out of poverty. In addition, education reduces thevulnerability of individuals to income and consumption shocks induced by external events, suchas a macroeconomic crisis.

Occupation of the Household Head

59. How does poverty vary acrossoccupations? Figure 20 forcefully Headcount ratio of income-poverty, by household head's occupation,

occupations? Figure 20 forcefully ~~~~~~~~1992-99demonstrates, the point that poverty is I

Farm workerspervasive in the agricultural sector in 38

Thailand. Farm workers and farm op- Farm P"'ors(includi.g 138trunmts) -32

erators belong to the poorest occupa- e w _General w-ke 1

tions in the country, with more than 20=20

one-third of the persons engaged in Economically inactive 14 4 01992

these occupations being poor in Production and conruction _11 t 01996

1999.16 The incidence of poverty is workers ao

moderate among general workers and Enureprenears,tradey nd

economically-inactive households Clerical, sales and service 4

(20% and 14%, respectively), but it is wnrkers 4Professional, wtechcatr ?22significantly smaller among other oc- managerial W

cupational groups. Additionally, pov-Figure 20

16Farm operators include both owner and tenant farmers, while farm workers are agricultural laborers. The occupa-tions shown in Figure 20 refer to the household head's main occupation - i.e., the occupation in which the headearned the majority of his income/earnings.



erty rates among farm operator and farm worker households have remained stubbornly highbetween 1992 and 1999. This is surprising in view of the fact that Thailand's agricultural GDPgrew at an average annual rate of Figure 211.4% during the seven years between Composition of the poor by occupation, 1992-99

1992 and 1999.0 Professional, technical, 100%- '

managenai _ _90%-60. The extreme concentration of *Clericalsales and 10 10

the Thai poor in just a few economicsectors is apparent from Figure 21, * Genem1 workers 70% 12 13

which shows the sectoral distribution * Entrepreneurs, trade and 60% 13 14 15industry 1

of the poor in each of the years be- o Production and 50%

tween 1992 and 1999. Nearly 70% of constroctionwaken 40%a

the poor in Thailand in 1999 were M Economically inactive 3 63 59

drawn from the farming occupations O Famworkers 20% 52

(either cultivators or farm workers).Another 1 1% of the poor were eco- (Fincelradton`ants) 11%

nomically inactive population. Thus, 1992 1994 1996 1998 1999

about 80% of the poor belong to twooccupations - farming and economically inactive.

Ownership of Land

61. While the incidence of pov- Figure 22erty is high among farm households,it is also the case that these house- Headcount ratio of income-poverty among farm households,

holds form a rather large group in 60 by size of owned land, 1992-99

Thailand. In 1999, for instance, 01992 *1994 0[1996 0C1998 *1999

about 16.5 million persons belonged 50 45 46

to households headed by a person 41 44

whose main occupation was farm- 40 37 39

ing. From a policy point of view, it 31 31

is therefore important to know what 30 A

types of farm households are poorer 21

than others. 20 21

62. Since land is typically the 10 most important factor of production <5 mi 5-19 f i 20 & mare lai

in agriculture, agricultural incomes- and hence the' incidence of poverty- are likely to depend strongly on farm size. This is borne out by the data in Figure 22, whichshows the incidence of poverty among farm households by the size of owned land.17 While it isnot surprising that small- and medium-sized farmers face significantly higher rates of povertythan large farmers, it is remarkable that the incidence of poverty among small landholders in-

'7 The data obviously do not control for the quality of land. If corrections were made for land quality, it is likely thatthe relationship between size of owned land and poverty would be even stronger than shown in Figure 22, as a largefraction of land in the Northeast is hilly or otherwise of low quality.



creased from 1994 to 1996, even as the Thai economy and Thai agriculture were experiencing ro-bust growth (and medium and large landholders were experiencing declines in poverty). With theonset of the economic crisis in 1997, the incidence of poverty among small landholders continuedto increase into 1999, with the result that a larger proportion of small landholders were poor in1999 than in 1992. While medium-sized farmers (those owning 5-19 rais of land) fared some-what better, the incidence of poverty among them was not much changed between 1992 and 1999either, due primarily to 1999 being a bad agricultural year. What these results demonstrate is thatsmall landholders have been largely excluded from the general economic growth that Thailandexperienced during the 1990s, while medium landholders gained during the 1990s but lost it all inone bad year. At least two causes for these patterns can be proposed: first, the global shifts inagricultural prices that took a steep downturn in 1999, and, second, long-run agricultural produc-tivity trends, which are taken up below.

63. One reason for the increase Percent change in rice yields, by region and by per capita land63. One reason for the increase ~~~quintiles (adjusted for land quality), 1995-96 to 1998-99in poverty among small and me- 30 27.6 26.5

dium farmers during the 1990s may _be that these farmers have experi- o 20 16.7

enced stagnation in agricultural 12.1_

productivity over time. Indeed, a e, lostudy using data from two succes- 4.1

sive surveys of 1,626 agricultural -households collected during the 0 G D1995/96 and 1998/99 crop years X -10

finds just that.18 The study findsthat there was very uneven growth -'3.4in rice yields across farm sizes and -20 Per capita adjusted land quintiles Region

across regions during the threeyears spanning the two surveys Figure 23(MERI and World Bank, 2000).The largest farms recorded an increase of 28% in rice yields during the four-year period, while thesmallest farms saw their rice yields decline by 17% (Figure 23). Likewise, rice yields in theNortheast declined by 13%, while those in the Central region increased by 27%. The study wasunable to identify the reasons for the large decline in rice yields among the smallest two quintilesof land holdings.


64. The poor of Thailand are easily identified by some of their characteristics. The most im-portant among these are rural residence, geographical region of residence, education and farming.While these results are not in and of themselves very surprising, what is puzzling is the persis-tently high poverty rate among small landholders. The evidence reviewed in this Monitor sug-gests that not only did this group not participate in the boom period of the early and mid-1990s,

18Households were sampled in more than 400 villages in 9 provinces in the North (708 households), Northeast (642),and Central (276) regions. The same villages were sampled in both years (Mekong Environment and Resource Insti-tute and the World Bank, 2000).


Pover-ty and Public Policy

but it also lost ground during the economic crisis in the late 1990s. As a result, small landholdersas a group are poorer today than they were at the start of the decade.

65. The evidence that rice yields among small farmers have actually declined over time, whileconsistent with the rise in their poverty, is puzzling. Some possible explanations are the rise ofunproductive "retirement" farms, occupied by older family members, and supported by migrantremittances. Another set of causes may relate to the financial viability of such smallholdings inan increasingly mechanized sector. The causes of this decline in agricultural productivity need tobe explored, if policy interventions to reverse the trend are to be designed.

66. Village-level targeting is an important area to explore further, given the focus of the newgovernment on allocating a village fund to every village in the country. Since village differencesin income are substantial, even more so than regional differences, block grants and other anti-poverty interventions targeted to the poorest villages are likely to represent a better use of publicfunds than universally-distributed programs, at least from an equity perspective. A number ofvillage-level data sets have or soon will become available in Thailand, and these will permit betteridentification of the poorest villages in the country (see Box 5).

Box 5: Village-Level Targeting and Village Databases

It is possible that targeting poverty interventions to administrative units lower than the province - say, a districtor even a village - may achieve even better results. Unfortunately, identification of poor and non-poor villagesis difficult, since most household surveys are not sufficiently large to allow reliable poverty estimates at thelevel of the village. However, in some cases, village-level data from a village or population census can be com-bined with household survey data to identify poor and non-poor villages. The fact that an extensive data baseon all villages in the country exists in Thailand suggests that a village-targeting exercise could be done.

Additionally, the fact that the NSO conducted both a general population census as well as socioeconomic sur-vey in 2000 offers an important opportunity to conduct a village targeting exercise. Since the SES and thePopulation Census have a few variables in common - e.g., family size and composition, type and quality ofhousing, and access to drinking water and sanitation - it may be possible to estimate (or predict) household in-come or expenditure (and hence poverty status) for the Census households from regressions estimated with theSES data. Since the Census sample would be large enough to produce reliable estimates at the village level, thiswould permit estimation of mean income or expenditure (and poverty status) for every village in the country.Some countries around the world, such as Indonesia, are already using such methods to target policy interven-tions and block development funds to the poorest villages.

(Continued on the next page)



There are two sources of village data available in Thailand, both collected by the Ministry of Interior (MOI).The first is the National Rural Development Committee (NRDC)'s Village Census Database and the second isthe Basic Minimum Need (BMN) database. These two data sets provide extensive information on a village'sdemographic, physical, and economic conditions, more suitable for the development policy planning at the vil-lage level. The BMN database, which contains variables related to a household's basic need items, has a nar-rower focus than the NRDC database, which contains many more details on villages' overall economic situationand development levels. These two data sets are superior to the SES in terms of non-monetary variable inclu-sion and a census-type survey. A major shortcoming of these two MOI's databases is their reliability, in par-ticular their data collection processes. The following section provides an overview of the NRDC database anda preliminary comparison with the SES.

The NRDC Village Census Database is an official survey of villages outside municipal and sanitary areas inThailand. The census covers all of more than 60,000 villages and has been undertaken biannually since 1984 to1988/99. It provides a considerable amount of details including socio-economic, social capital, availability andquality of natural resources, public services, health, and labor mobility variables.

The Census is aimed to be a tool to measure living condition of people in the rural areas, and to be used for na-tional policy planning and rural development evaluation. The census helps local governments rank the devel-opment .level of villages in their jurisdictions. Local governments can appropriately set the target for villages'development in the annual and five-year sub-district (tambon), district (amphoe), and provincial plans. Collec-tion of the NRDC database is carried out by local government in a bottom-up process. Members of the TambonRural Development Support Committee and representatives from Tambon Council fill out a questionnaire re-garding their responsibility. All of these representatives investigate the questionnaire together and then appointthe village committee and village headmen to fill in the incomplete answers and to check the questionnaire.The completely-filled questionnaire is sent to the District Development Committee. The questionnaire is theninvestigated, and some questions are cross-checked with the official village database before being forwarded tothe survey coordinating committee at the provincial government level. The questionnaires from all districts arefinally compared with the provincial official village database.

There are some overlapping variables between the NRDC database and the SES that could be used to double-check the consistency of data between these two surveys. Such variables are demographic variables, number ofpeople in designated education levels, number of households with utilities, and number of vehicles such as pick-up truck, motorcycle, and bicycle. Selected variables with the same definitions from the 1996 databases areshown in the table below.

Both the numbers of household and population in the NRDC database are considerably lower than those fromthe SES (where only households in villages are included). Household size is lower in the SES. The proportionsof population and households in each region are distributed in fairly similar patterns across the two data sets.However, for each region, the SES shows a higher proportion of people above 60 years of age, while the NRDCshows a higher proportion of people in age group 12-18.

In addition to age variable, household ownership of bicycles and motorcycles are also shown in the table. Thesetwo types of vehicles are the important mode of transportation in rural Thailand. The number of householdswith bicycle and motorcycle in the SES are 40 and 54 percent higher than those in the NRDC database. Never-theless, the proportions of households with bicycle and motorcycle differ by only a few percentage points.

The major advantage of the NRDC is that it provides details on the type and sale prices of agricultural produc-tion, land used for each crop, usage of fertilizer, and average income from each crop in the village. Non-monetary indicators of development can also be evaluated from this dataset; for instance, types of public ser-vices available, adequacy of drinking water and water for farming, the number of orphans and homeless orabandoned youth, the distance from the village to the nearest public hospital and the number of illiterates. Sincethe NRDC database is a village-level survey, it can be used to target poor villages. However, due to the incon-sistency of the NRDC and the SES data as shown in the following table, one might need to randomly select afew villages in the data set and verify the accuracy of the data reported for these villages in the NRDC database.



NRDC 1996Population_by_a__e__ro ip Region

Population by age group Central N orth Northeast So uth Total

Below 3 yrs. 4.48 4.33 5.17 5.29 4.853-6 yrs. 5.45 5.27 6.12 6.45 5.836-12 yrs. 9.47 8.93 10.49 11.06 10.0112-14 yrs. 5.94 5.52 6.59 6.76 6.2414-18 yrs. 9.36 8.82 10.13 9.90 9.6418-50 yrs. 46.64 48.94 46.21 43.24 46.4850-60 yrs. 9.52 9.44 8.83 9.30 9.18Above 60 yrs. 9.15 8.74 6.46 8.00 7.77Total 100.00 100.0 0 100.00 100.00 100.00Population 7,750,233 8,062,764 15,150,764 5,234,755 36,198,516No. of households 1,677,191 1,883,454 2,984,970 1,055,090 7,600,705Household size 4.62 4.28 5.08 4.96 4.76% of population in each re- 21.41 22.27 41.85 14.46 100.00gion _ _ _ _ _ __ _ _ _ _

% of households in each re- 22.07 24.78 39.27 13.88 100.00gion% of households with bicycle 45.44 47.26 54.82 24.23 46.63% of households with motor- 57.85 57.21 40.00 64.01 51.54cycle

SES 1996Region

Population by age group Central N orth Northeast South Total

Below 3 yrs. 4.88 4.09 5.18 5.51 4.933-6 yrs. 5.61 5.43 5.74 5.85 5.666-12 yrs. 9.90 10.62 12.79 13.53 11.8212-14 yrs. 3.83 3.67 4.91 4.59 4.3614-18 yrs. 6.11 6.20 7.21 7.60 6.8218-50 yrs. 47.61 47.50 45.23 44.45 46.1150-60 yrs. 8.94 9.76 9.15 7.93 9.04Above 60 yrs. 13.11 12.72 9.79 10.54 11.25Total 100.0 0 100.00 100.00 100.00 100.00Population 9,180 ,785 9,008,509 17,615,156 6,530,547 42,334,997No. of households 2,247,100 2,453,400 4,156,500 1,545,400 10,402,399Household size 4.09 3.67 4.24 4.23 4.07% of population in each re- 21.69 21.28 41.61 15.43 100.00gion% of households in each re- 21.60 23.58 39.96 14.86 100.00gion% of households with bicycle 44.21 50.56 56.78 24.73 47.84% of households with motor- 62.68 62.32 48.63 69.43 57.98cycle

Source: Thailand Development Research Institute.


Povertv and Public Policy

Chapter 3. Inequality in Thailand

67. Trends in the incidence of poverty over time are essentially driven by two forces: the rateof economic growth (which is reflected by the growth in mean income) and changes in the distri-bution of income across different economic groups. It is thus possible for income or GDP in-creases to be accompanied by an increase in absolute poverty due to a marked deterioration in thedistribution of income. In such a case (relatively rare), income growth is concentrated among thebetter-off, with the poor gaining little or even losing their share of aggregate income. Likewise,it is possible for poverty to decline substantially during a period of low economic growth with amarked improvement in the distribution of income. It is important to understand the role of thesetwo forces in recent years and into the future, because the policy response to poverty can be muchdifferent than the response to high levels of income inequality.

68. Understanding inequality is thus important, both to appreciate the magnitude and direc-tion of the problem and to identify the causes of inequality. Such measures as the Gini coeffi-cient can measure overall inequality of income distribution while more specific measures of ine-quality, such as the percentage share of income held by the poorest 20% of households, can showmore directly how inequality is related to poverty. This chapter discusses both the evolution andcauses of income inequality in Thailand using SES data.

Overall Trends

69. As the economy grew rap- Figure 24idly in the 1980s, income inequality Income inequality, 1990-99 4.0

increased in Thailand. For in- 0.531 0.531

0.530 - * Gini index stance, the Gini coefficient of in- 3.-0 Shorrockndex

come inequality increased from 0.525 -3.0

0.453 in 1981 to 0.479 in 1988, ,5 ' 053.0

0.513 in 1990, and 0.531 in 1992. X

After 1992, however, income ine- : 0515 ' 2-5

quality began trending down (Fig- 0s51, 0.513

ure 24). This trend only continued 2.00.505 O1'

through 1998. The latest availabledata from the Socioeconomic Sur- 0.500 1.6

vey of 1999 indicate that income 0.495 .1 10

inequality has increased sharply be- 1990 1992 1994 1996 1998 1999*

tween 1998 and 1999. It is unclearwhy this might have happened, but one possible reason is that the economic crisis (whose impactshows up in the household survey data only in 1999) had a larger negative impact on the incomesof the poor relative to the better-off.



70. The result of these changes Figure 25 Lorenz curves for income, 1981 and 1999

is that, over a two-decade period 10o

spanning 1981 to 1999, the distri- 90bution of income has worsened 1

80

considerably (Figure 25). The 70

share of income of the poorest = 60

quintile (20%) has dropped from 950/

5.4% to 3.9%, while that of the /

richest quintile has increased from E 999

51.5% to 58.3%. /' 20

20

71. Consumption inequality has 0followed the same trends, although, 0 20 40 60 80 100

of course, consumption is more Cumulative % ofpopulation

equally distributed than income in Thailand, as in all countries.

International Comparisons

72. How does inequality in Figure 26Thailand compare to that in the Gini coefficient of consumption, 1987-99, selected Asian countries (%)

other countries of the Asia-Pacific 50 -

region? Since comparable data are 45

available for consumption inequal-ity for the other countries, Figure 26 40 - -

shows the trend in the Gini coeffi- -Thailand- - Philippines

cients of consumption over the pe- 35 Malaysia

riod 1987-99 for five countries in -Korea

the region.19 Up until 1992, Thai- 30

land had the highest consumptioninequality among all five countries, 25

although inequality was onlyslightly lower in the Philippines and 20 -

1987 1 198 9 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999Malaysia than in Thailand. Therelative position of Thailand (ascompared to Malaysia and the Philippines) in more recent years is not clear owing to the lack ofrecent data for Malaysia and the Philippines. What is clear, however, is that Indonesia and Koreahave consistently had a much more egalitarian distribution of consumption than Thailand (al-though inequality in Indonesia also increased sharply during the 1990s).

19 Of course, to the extent that household surveys in different countries often use different definitions of consumptionor income and varying periods of data recall, comparability of welfare measures across countries always poses someproblem.



Regional Variations

73. Oviouly, egioal vria- Figure 2773. Obviously, regional varia- Figure 27 Mean per capita income, by region, 1990-99

tions in income are one source of 9,000 - - 9,000

overall income inequality in Thai- 8,000 e Bangkok 8,000-* Vicinity

land. In 1999, for example, mean "000 -Central D /

per capita income in Bangkok was , South--00*--North t6,000

more than four times as large as I -Northeast

that in the Northeastern region. 5000000

Further, as Figure 27 indicates, 4,000 4.000

these large inter-regional variations 3000 3,000

have not narrowed at all during thedecade of the 1990s. 2,000 _,000

74. It is possible to decompose - --- -

the overall income inequality in the 1990 1992 1994 1996 1998 1999,

country into that attributable to differences in mean income across the regions ("between group")and that due to differences in incomes among persons residing within a region ("within group").Such an exercise shows that the between-group variation accounts for a relatively small propor-tion (less than 15%) of overall income inequality.20 Indeed, the explanatory power of the be-tween-group variation, which was in the 10-14% range from 1990 to 1998, fell sharply to only6.5% in 1999. This means that the overwhelming share of income inequality in Thailand arisesdue to income variations across individuals residing within the same region - not differences inmean income across regions. In 1998, the regions accounting for the largest within-group varia-tion were 'Bangkok and Vicinity' and the Northeast.2 1 Indeed, nearly one-half of overall incomeinequality in the country could be explained by variations in interpersonal income within thesetwo regions.

75. The implications for policy are obvious. Although reducing the income gap across re-gions will help, it will by no means eliminate overall inequality of incomes in the country. Moreimportant will be policies that reduce inequality across households and individuals within theNortheast and Bangkok. The sources of interpersonal income inequality in these two regions arelikely to be very different. In the Northeast, differences in income across households are likely tobe explained by three factors: land ownership, education, and the presence/absence of householdmembers employed outside the region (who send remittances back to the family in the North-east). In Bangkok, education and occupation are likely to be the main determinants of incomevariations across individuals and households. This issue is explored in more detail below.

20 This is a common finding of similar decompositions done for other countries.2'Data for 1998 - instead of 1999 - are used as they are more typical of the pattern observed during earlier years.These two regions account for such a large share of the overall within-group variation both because they have highlevels of income inequality and because they are populous regions.



Patterns and Sources of Inequality

76. In much the same way that overall income inequality was decomposed into that arisingdue to differences in mean income across regions and that due to differences in income amongpersons residing within a region in the previous section, it is possible to decompose inequalityinto between- and within-group components where the groups are differentiated by rural/urbanresidence and age, sex, occupation and education of the household head. This is what is at-tempted in this section in an effort to understand the sources of income inequality in Thailand.2 2

77. Rural/Urban Residence. As in the case of regions, differences in mean income acrossrural/urban areas account for about 15% of overall income inequality (with the remaining 85%accounted for by within-area differences). Approximately one-half of overall income inequalityis explained by interpersonal variations in income within rural areas, reflecting the fact that theseareas account for three-quarters of the national population and that they have the highest degreeof within-group inequality.

78. Age and Sex of Household Head. Virtually none (0.03 - 0.4%) of the overall incomeinequality in the country is accounted for by differences in mean income across male- andfemale-headed households and across households headed by persons of different ages, suggestingthat inter-household differences in the age and sex of a head are not important sources of incomeinequality.

79. Education of Household Head. As would be expected, education of the household headis significantly more important than his/her age and sex in explaining inequality. Differences inmean income across households headed by persons of different schooling attainment account for19% of overall income inequality. Most of the remaining 81% of inequality that is attributable towithin-group variation is due to interpersonal variations in income within two educational groups- those with primary and secondary education - reflecting the fact that they two groups accountfor 83% of the population and also have high levels of intra-group inequality.

80. Occupation of Household Head. Differences in mean income across households headedby persons of different occupations account for 15% of the overall inequality in incomes. Thegreatest degree of income inequality within occupations is found in 'entrepreneurs, trade and in-dustry' and farm operators. These two occupational groups account for approximately one-halfof the overall inequality of incomes in the country.

81. What these results indicate is that each of the factors considered above - region of resi-dence, rural/urban residence, and age, sex, education and occupation of the household head -does not by itself explain more than 15-19% of the overall inequality in incomes. That is not to

22All of the results in this section refer to 1998. Although data for a most recent year - 1999 - are available, theyshow a very different pattern from earlier years. It is not clear whether the results for 1999 reflect the smaller size ofthe sample that was used by the Socioeconomic Survey during that year, or whether they reflect real changes thatoccurred in the characteristics of the income distribution during that crisis year. In addition, it should be noted thatthe decomposition analysis presented here is essentially an analysis of variance, and does not indicate a causal rela-tionship between factors, such as occupation and rural residence, and income inequality.



say that these factors are unimportant or that, in combination with each other, they might not ac-count for a larger share of overall income inequality. But the results of the analysis do indicatethat a very large share of the overall inequality is "within-group" - differences among individualshaving the same occupation or residing in the same region.

82. Certainly one reason for the failure of the above factors to explain a great deal of the totalvariance in incomes is that the groups have been defined very broadly. In other words, there area very large number of individuals residing in the Northeast or in households headed by a pri-mary-educated person. Therefore, it is not surprising that there is considerable heterogeneityamong these households, and that intra-group (within-group) variance dominates between-groupvariance.

Income Inequality by Sources of Income

83. Another way of looking at income inequality is to decompose it by the sources of in-come. Although one-half of total personal income in Thailand is derived from wages and sala-ries, the contribution of inequality in wages and salaries to overall income inequality is only 27%,since wages and salaries are distributed much more equally than other sources of income. On theother hand, even though nonfarm profits (such as income for businesses and self-employmentactivities other than farming) constitute only 22% of total income, their contribution to overallincome inequality is significantly greater (56%). In 1998, the inequalities in nonfarm profits,wages and salaries, and farm profits made the largest contributions (in that order) to overall in-come inequality.

Sources of Income Inequality in the Northeast and in Bangkok

84. It was noted earlier that nearly one-half of the overall inequality in incomes in Thailandis attributable to variations in income across households within the two regions of the Northeastand 'Bangkok and Vicinity'. It would therefore be useful to know the main causes of incomeinequality in these two regions.

85. Using a statistical method- Figure 28ology developed by Fields (1993), Decomposition of overall income inequality into its various causes,

the total variance in incomes in Northeast and Bangkok, 1998

each of the two regions has beendecomposed into that attributable to Demographic factors 12.6

four factors: demographic variables Ocupaton

(such as family size, household age U.5 Northeast

and sex composition, and age and Education 28.3 I Bangkok and Vicinity

sex of the household head), occupa- R

tion (of the household head); educa- Rura esden _ 76

tion (both of the household head Land ownership and 1.0

and the best-educated adult in the tenancystatus

household); rural residence; land Unexplained residual 70.7

ownership and tenancy status; andProportion of overall income inequality explained by factor



an unexplained residual factor. The results of the analysis are shown graphically in Figure 28.

86. Education has the largest explanatory power in accounting for inter-individual variationsin income. The explanatory power of education is greater in the Northeast than in Bangkok, re-flecting the fact that education is significantly more widespread in Bangkok than in the North-east. In addition, the differing contributions of education to income inequality in the Northeastand Bangkok also reflect the fact that, in the Northeast, education is often a ticket to migrate toBangkok and earn significantly higher wages. However, there is no such option for Bangkokresidents, as Bangkok already offers the highest wages in the country.

Figure 2987. The next most important Lorenz curves for per capita income, expenditure and owned land,

factor explaining income inequality rural areas only, 1998

in both regions is the demographic 100characteristic of families. Differ- 0o

ences in family size, age and sex 0 80

composition of the household, as X 70-

well as the age and sex of the E 60 Perfect equality0

household head explain approxi- - Expenditure

30mately 10% and 13% of overall in- S 0 ' /_

come inequality in Bangkok and 1 30 o Ownce l

the Northeast, respectively.

88. In the Northeast, rural/urban 0 10 20 30 40 50 60 70 80 90 100

residence explains another 7.6% of Cumulative % of population

overall income inequality. This factor is, of course, not relevant in Bangkok, as the 'Bangkokand Vicinity' region is entirely urban (and suburban).

89. What is surprising is the almost negligible contribution of land ownership to overall in-come inequality in the Northeast. This does not reflect the fact that land is distributed equally inthe Northeast; indeed, as Figure 29 shows, the distribution of land is significantly more unequalthan that of income or consumption expenditure.

90. Instead, the results reflect the fact that, even in the Northeast, farming is not the only oc-cupation, and land is not the only source of income. Indeed, only one-third of individuals in theNortheast belong to households in which the main occupation of the head is farming (includingfarm work). Furthermore, as Figure 30 shows, farm operators on average have the lowest in-comes of all occupations in the Northeast. Entrepreneurs, trade and industry persons, whoconstitute 19% of the population, and managers, professionals and technical workers, who have apopulation share of 11%, enjoy per capita incomes that are about three times as large as the in-comes of farmers and farm workers. Thus, while land ownership may be an important source ofinequality of incomes among farm households, it is unlikely to explain a great deal of overall in-come inequality as some of the richest individuals in the Northeast own no land at all.23

23 This discussion does not in any way imply that land is not an important source of income and that land-ownershipis not associated with poverty.



Figure 3091. Finally, the decomposition Mean household income per capita and share in total population, by

results show that a very large pro- major occupation, Northeastern region

portion of overall income inequality 7,000 32.4 35

in both regions - one half or more - , 6,000

simply cannot be explained by any r: \ Meanincomepercapita5,000 - \% share in population 25 z

of the factors considered here. This 18.9

is a result that is commonly ob- 15 120 53,000 --

tained for other countries as well. It 0 '\ 10.9 15

reflects the fact that inequality of C 2,000-I0 0

income arises due to the unequal a 1,000 , LJ5distribution of factors and endow- 0 lments that are not easily observed ,,0I a t ,,- c

in survey data, such as inherited E

assets other than land, innate abil- n a

ity, 'social capital', and personaland family connections. In addition, of course, random luck often plays an important role in ex-plaining variations in economic success across individuals.

Changes in Inequality Between 1998 and 1999

92. As noted earlier, the period 1998-99 was one associated with a sharp reversal of the de-cline in inequality that had begun in 1992. One reason for this increase was that the economiccrisis had a disproportionately negative impact on the incomes and livelihoods of the poorer sec-tions of Thai society. The clearest effect of this can be seen in the changes in real income percapita during 1998-99 of households headed by persons with differing educational attainments.

93. As would be expected, higher levels of education of the household head are associatedwith higher levels of household income per capita. Individuals residing in households where thehead has tertiary schooling have more than five times the income of individuals in householdswhere the head has no formal education. Figure 31 shows that the effects of the crisis werehighly regressive in that households that were the poorest to begin with in 1998 (i.e., thoseheaded by a person with no schooling or primary schooling) experienced the sharpest drops inreal income, while those that were relatively better-off (i.e., those headed by post-secondary edu-cated persons) actually experienced an increase in real income between the two years.


Poverty and Ptiblic Policy

94. This highlights the specialnatur ofi edcaionligas athuma spciapi Mean real household income per capita in 1998 and percent real changenature of education as a human capi- Min per capita income, 1998-99, by schooling of household head

tal investment. Education does not 10,000 17.5 20

merely increase the average earnings 9,000 Mean real income per capita 13 15

of an individual over his or her life- 1,000 4-% change, 1998-99 lo .EB7,000

time, but it also reduces his or her Q 6,000

vulnerability to income fluctuations. 'E 3:oo0... .vulnerability to income fluctuations. E ~~~~~5,000 ..,....................................... .... ... , ........ ........... .. .. . 05

Even those better-educated individu- 4,0006

als in Thailand who may have lost , 3000

their jobs during the economic crisis, 2 2,000 -159-10

such as individuals working in the 1,ooo -15

finance sector, were able to find 0 - - - -20

other jobs and sources of income, None Primary Lower Upper Vocational Universitysecondary secondary and technical

and maintain their real standard of Figure 31 Completed schooling of household head

living. Indeed, the data in Figure 31indicate that individuals with upper secondary education and vocational and technical qualifica-tions actually improved their standard of living during 1998-99, in contrast to those with lowersecondary or lower levels of schooling, who suffered real income declines.

95. The same pattern is ob- Figure 32served with respect to age of the Mean real household income per capita in 1998 and percent real change

household head. Individuals resid- in per capita income, 1998-99, by age of household head

ing in households with older heads 3000 8.10

generally have higher income per 2 M 82,800 Mean real income per capita

capita than those in households --%change, 1998-99 / 6 =

headed by younger heads. But, in ' _ 5.2

general, households with older 'E

heads fared better between 1998 2 2\

and 1999 than those with younger =°0heads (Figure 32). The age effect 2 2,200 ~~~~~~~. .... ...... .. .......... ....... ........... . -... . .. . ....... ........... . ...... . ..... . 0

heads (Figure 32). The age effect 2,0 .0-0 reflects the importance of seniority, -2

especially in the Thai context. 2,000-- -+4- -

Younger workers were much more <20 21-30 31-40 41-50 51-60 >60

likely to have lost their jobs during Age of household head (years)

the economic crisis than olderworkers.2 4

Changes in the Distribution of Education

96. Given that education has the largest explanatory power in accounting for income inequal-ity and given its powerful role in reducing poverty, a natural question is what has been happeningto the distribution of education in the Thai population over time. If access to schooling has beenimproving rapidly for low-income groups, future prospects for an improvement in the distribu-

24This was documented extensively in the Social Monitor IV: The Thai Crisis and Workers.



tion of income would look bright.25 On the other hand, a worsening of the distribution of school-ing opportunities in the population would not bode well for the evolution of income distributionin the future.

97. Data from the various Socio- Distribution of enrolled students at the primary and lower secondarylevel, by real per capita income quintile, 1992-99

economic Surveys show that there 100%- has been an enormous expansion of Richest 90%

education in Thailand, and that the 80%

poor quintiles have increased their Fourth -? 70%0 60%

share of lower secondary studentsimpressively during the decade of * Third 40% 22 23 23

the 1990s (Figure 33). For example, 30% ___22

while only 12% of all lower- 20%

secondary students in Thailand were Second 10% 26 27 27 19 22

drawn from the poorest quintile in 0% 1992 1996 1999 1992 1996 1999

1992, as many as 22% were in 1999. O Poorest Primary Lower secondary

(No large changes in the share of Figure 33primary students are observed forany quintile, because all quintiles had already reached universal primary enrollment by 1992.)Correspondingly, the share of the richest quintile among lower-secondary students fell from 24%to 14%.

98. Figure 34 suggests that the Distribution of enrolled students, by level and by real per capita income

poorest quintile has increased its quintile, 1992-99

representation among upper- O Richest 100%

secondary students as well - from 80% 2 31 27

about 10% in 1992 to 15% in * Fourth -E 73 66

1999. This number should further m 60% 81

increase in the coming years, as T 40%

the number of years of compulsoryschooling has been increased by 20% 7 / /

the government just recently from 0 Second 0% 19 199 1 1 1992

9 to 11 years. 1992 1996 1 191969999 992 996 1999

O Poorest Upper secondary Vocational & technical University

99. The share of the poorest training

quintile among vocational students Figure 34has also increased impressively inrelative terms - from a mere 2% in 1992 to 8.4% in 1999. The only schooling level where thepoorest quintile has not managed to increase its representation appreciably is university educa-tion. The percentage of university students drawn from the poorest quintile barely increased dur-ing the seven years, remaining at a mere 0.9% in 1999. The domination of university educationby the top quintiles is observed in the fact that more than 90% of all university students in the

2 5 0f course, changes in the distribution of earnings and income will depend not only on changes in the distributionof schooling in the population but also on changes in the pecuniary returns to different levels of schooling.



country were drawn from the richest two quintiles in 1999 - down only slightly from 96% in1992.

100. Undoubtedly, an important reason for the low representation of the poorest quintile inuniversity education is that the entry of this group into lower- and upper-secondary schools issimply too recent for it to have a cohort of secondary-completed students that are eligible foruniversity admission. As more and more students from the poorest quintile enter and graduatefrom secondary schools, one would expect the pool of poorest students in university education toincrease. However, this is by no means an automatic process. The total (both direct and oppor-tunity) cost of university education in Thailand, as in other countries, is large, and this restrictsthe access of poor students, howsoever eligible they might be, to university education. It wouldtherefore be important for public policy to proactively target educational assistance programs,such as student loan and scholarship schemes, to students from the poorest backgrounds. How-ever, as is discussed later in chapter 6, this does not appear to be the case. The benefits of gov-ernment scholarship and student loan schemes are derived disproportionately by the upper in-come quintiles.

Research and Policy Issues

101. While regional inequalities - reflecting the Bangkok-centric bias in economic growth -are an important source of income inequality in Thailand, the evidence suggests that within-region variations in household income are even more important in explaining overall inequality.Education is one variable that accounts for a significant source of this intra-regional, interper-sonal inequality. This suggests that equality of opportunity in education will be an important pol-icy tool to reduce income inequality in Thailand (under the assumption that the returns to educa-tion do not change over time). The evidence does indicate a substantial improvement in the dis-tribution of primary and secondary education during the last decade. However, higher (college)education is still very unequally distributed, with the poor having almost no representation in in-stitutions of higher education. While this is also likely to change over time, as a cohort of poorstudents graduates from secondary schools and becomes eligible to enter universities, policy willneed to play an important role in order to make this happen. The high cost of university educa-tion typically restricts the access of poor students to higher education. The government thereforehas an important role in subsidizing the cost of higher education for students from the poorestbackgrounds. However, as discussed later in chapter 6, the benefits of the limited governmenteducational assistance programs in Thailand do not appear to be well targeted to the poorest stu-dents; indeed, the overwhelming majority of the beneficiaries of these programs are studentsfrom the richest two quintiles.

102. The evidence from the SES also suggests that, after several years of improvement in theearly and mid-1990s, income inequality increased sharply in the aftermath of the crisis. One rea-son for this was that the crisis had a stronger negative impact on the earnings and incomes ofless-educated and younger workers, who typically are poorer than better-educated and olderworkers. As more recent post-crisis data become available, it would be important to explorewhether the increase in inequality in 1999 was a transitory (crisis-related) event or whether thecountry is once again entering a period of rising income inequality.



Chapter 4. Strategic Policy Issues in Poverty and Inequality

103. Chapter 4 of this report turns to the strategic issues confronting Thai policy mak-ers as they build the policy foundations for a resumption of growth and the reduction ofpoverty. Traditionally in Thailand, anti poverty programs were associated with transfersto the indigent, the elderly, and disabled. Rapid growth and policies promoting growthwere seen as more than adequate to deliver substantial income gains for all Thais, includ-ing the poor. In recent years, these views have undergone some refinement anddevelopment, as Thai debates on public policy have intensified:

Growth rates that were once the highest in the world are now expected tomoderate over the medium term, thus focusing attention on public policies in ad-dressing more directly the problem of poverty;

The crisis impacts demonstrated that private, informal safety nets were effec-tive, but not adequate to cushion the effects of the economic contraction, particu-larly on the poor;

Highly unequal income distribution, long masked by the benefits of rapidgrowth, has emerged as a policy issue;

Poverty among certain sub-groups and in certain areas appears to be chronicand resistant to the benefits of growth, posing a possible "structural" dimension topoverty.

104. The strategic issues taken up in Chapter 4 are cross sectoral and thematic: self re-liance and informal safety nets; chronic, structural poverty; causes of poverty and thecase of the Northeast, natural resource degradation and poverty; gender and poverty; andthe role of growth vs. income distribution in provincial and national poverty outcomes.The list is far from complete. For example, taxation policy and its effects on poverty arenot addressed, nor are trade policies, both of which might be considered strategic issues.

Poverty and Self Reliance

105. The crisis clearly imposed hardships on poor, marginal, vulnerable families, be-fore Government or community programs could reach them. Incomes and wages fell,jobs were lost, and unemployment rates rose sharply. Social problems also increased,with higher rates of crime, drug use and arrests and suicides. Thai children seemed tohave suffered the most - the number of infants abandoned after birth and children placedin orphanages increased.

106. Despite these worrisome signs, there were also clear signs emerging from thehousehold level that families took concrete steps to cushion the effects of the crisis and toprotect their social assets. This pattern of self-reliance and mutual assistance can be seenin the evidence on income transfers and consumption shifts at the household level. AsThai policy makers address demands for more official safety net programs, an importantconsideration is the admirable and socially beneficial role of such voluntary family cop-ing strategies.



107. One way in which the poor can cope with adverse circumstances is throughgreater reliance on income transfers from friends, relatives and the government. In thecase of Thailand, most of the transfers are from relatives, typically those members of thefamily who have migrated to the urban areas (typically, Bangkok) in search of better-paying jobs. However, the economic crisis of 1997 was unusual in that it first affectedBangkok, and this resulted in a large number of migrant workers in the construction sec-tor in Bangkok being laid off. As a result, the ability of these workers to send remit-tances to their families in the rural areas, typically the Northeast, was curtailed.

108. There are two observations The volume of income transfers in relation to total income,

that can be made about transfers by real per capita income quintile, 1996-99

from Figure 35, which shows 12 11.7

transfer income as a percentage of I01996 rl998 01999

total income by real income quin- 10 9.8

tiles for 1996, 1998 and 1999. 8 9 8.9 8. 9.1 39 8.5~~~~~~~~~~~~~~~.

First, transfers are larger in rela- 7.8 .. 7

tion to total income for the poorest -three quintiles relative to the rich- 7

est two quintiles. Second, the evi- 6 5 2.0

dence also indicates that, while the 5

poorest three quintiles increased 4

their reliance on transfers between Poorest Second Third Fourth Richest

1996 and 1999, the richest two Real per capita income quintile

quintiles experienced a decline in Figure 35the proportion of total income derived from transfers.

109. Another means of exam- Incidence of income poverty in the absence of transfers, 1996-99

ining the importance of transfersto the poor, especially during the 25 E Actual incidence of income poverty 22.5

crisis, is to calculate what the O1 Incidence of income poverty in the absence ofremittances andassistance

incidence of income poverty 20 18.0

would have been in the absence .6.6 15.9

of transfers. 26 In all three years 8 13.0

- 1996, 1998 and 1999 - trans- < 114

fers played a very important role 10

in keeping poverty significantlybelow what it might have other- 5

wise been (Figure 36). How-ever, the difference between ac- 0 _ 1918

tual poverty and the poverty that 1996 1998 1999

might have been present in the Figure 36

26Actually, the counterfactual of what incomes would have been in the absence of transfers cannot simplybe obtained by subtracting transfers from income, since household behavioral responses intervene. Forexample, it is likely that in the absence of transfers, households might exploit other sources of supplement-ing their incomes, in which case the numbers in Figure 36 overstate the effect of transfers in reducing pov-erty.



absence of transfers is much greater for 1999 than for the earlier two years (6.6 versus 5percentage points). Thus, transfers are an important coping mechanism of the poor inordinary times, and even more so during periods of crisis. Thus, despite a substantial re-duction in wages, families continued to support one another, pulling together to help eachother cushion hardships and relying on one another.

110. Another manner inExpenditure shares on selected items of expenditure by the poorest 20%

which the poor adjust to nega- of households, 1996 and 1999

tive income shocks is to E -

change their consumption pat- e 16

tems. Families can and do > 14 0* * * = ~~~12 _El 1996 * 1999make conscious decisions to 12

E 10cut down on those items that E.

are of less importance to the , 6 hoverall well-being of the fai- 4 j - cX

ily. Figure 37, which shows 0the proportion of household K

expenditure spent on selected J o / F

items of expenditure in 1996 , o 9

and 1999,27 suggests that c, Figure 37some of these adjustments didtake place in Thailand. The poor cut back their share of expenditure on tobacco and al-cohol. They also appear to have switched from expensive sources of calories, such asmeat and poultry, to cheaper sources, such as grain and cereal products.28 Figure 37 alsoshows a shift in consumption away from fruits and nuts, which typically are expensive,toward vegetables that cost less.

111. More disconcerting is the fact that the share of expenditure on medical visits andmedicines were cut significantly. However, this does not necessarily mean that health-care visits fell. Other evidence suggests that the number of health-care visits were notchanged, but that households shifted from more expensive private health care to publichealth care providers (World Bank, 1999). An expansion of the public assistance pro-grams - in particular, the low-income card and the voluntary health insurance card -could account for this. It is hard to validate these hypotheses with the SES data.

112. Surprisingly, the data suggest that the share of expenditure on education increasedsignificantly (albeit from a small base) - from 1.6 to 1.9 percent. It is not clear why thiswould have happened, although evidence from other countries does suggest that school-ing enrollments are often counter-cyclical with respect to the economy - i.e., the demandfor schooling rises during times of recession since the opportunity cost of attending

2 7 Consumption expenditures are compared only across 1996 and 1999 for reasons of data comparability.Expenditures on food are typically obtained in the SES from a 'short' questionnaire that asks respondentsabout expenditures incurred on 15 major food categories. In 1998, however, food expenditure data werecollected on 140 detailed food items. Hence food expenditures are not measured in a comparable wayacross the 1996 and 1998 surveys.28This need not necessarily be a negative outcome. For example, it is possible for the shift from more ex-pensive to cheaper sources of calories to actually improve nutritional or health outcomes.



school is lower (because of lower wages and higher rates of unemployment in the labormarket).

Structural and Chronic Poverty

113. Structural poverty is often re-ferred to as poverty that is persistent Box 6: The Ultra Poorover time and not easily amenable toeradication by the usual instrument of A study conducted by a team of research-economic growth. Indeed, even direct ers led by Professor Medhi Krongkaew used anpolicy interventions sometimes are un- innovative technique of asking village leaders toable to reach and to benefit the structur- identify the five poorest households in their sam-ally poor. Structural or chronic poverty pled villages. These households, who were labeledmaly peotheor. sult of many different fac- as the 'ultra poor' by the researchers, were inter-may be the result of many different fac- viewed about their socioeconomic characteristicstors, including social exclusion (due to and their methods of coping with abject poverty.ethnic identity, class or caste), geo- The study found that abject poverty was associatedgraphical isolation (e.g., residence in with large household size, a large ratio of depend-remote mountainous regions that are dif- ents to earners, heads with little or no schooling,

and lack of ownership of land. Even though some officult to access), and physical location the abject households had little or no land to work,(e.g., residence in arid regions having they were too poor to move out of the village tovery low levels of soil fertility or low seek other gainful employment. Other correlates ofrainfall). In other cases, a 'culture' of abject poverty were widows as household heads, thepoverty may have evolved among cer- critical burden of having to look after a chronically

ill or disabled person in the family, and the failuretain population groups - perhaps in re- to receive any benefits from a public assistancesponse to their being poor for so long - scheme. The latter finding is particularly importantthat makes them resist government poli- as it suggests that public social assistance schemescies that would lift them out of poverty. do not reach the ultra or abject poor.

Whatever its causes, chronic poverty is a Source: Krongkaew (2001).considerably more difficult problem toaddress with the standard policy pre-scriptions and tools.

114. In contrast to chronic poverty, transitory poverty refers to those individuals orgroups who move in and out of poverty, largely as a result of income shocks in the formof natural disasters, catastrophic illness, macroeconomic recessions, or a collapse in, say,agricultural commodity prices. Transitory poverty may also arise due to a lack of accessto assets (e.g., land), opportunities (e.g., schooling and employment), and markets (e.g.,credit, product and labor markets). If access to these factors is enabled, the transitorilypoor, at least in principle, can manage to lift themselves out of poverty.

115. It is possible to distinguish the transitorily-poor from the chronically-poor withlongitudinal survey data that tracks the same households and individuals over a numberof years. However, in the absence of such data, it is very difficult, if not impossible, toseparate out the chronically- from the transitorily-poor. For Thailand, there is no nation-ally-representative, longitudinal household survey data available, and this makes it diffi-cult to estimate the incidence of structural or chronic poverty.



116. Some indications of chronic or structural poverty can be obtained by examiningtemporal changes in poverty incidence for different population subgroups defined on thebasis of characteristics that are not likely to change much from year to year. These wouldinclude gender, education (of adults), region of residence, and land ownership. Thisanalysis has already been done and presented in Chapter 2. There, it was noted that cer-tain groups, such as farm workers and farm operators (in particular, small and mediumlandholders), large households (viz., those with 6 or more members), households with alarge number of dependents relative to earners, and households headed by individualswith primary or less education, have had stubbornly high rates of poverty over a long pe-riod. Indeed, poverty for some groups, such as small landholders, has increased even inabsolute terms during the decade of the 1990s. All of this evidence suggests that a sig-nificant portion of the poverty among these groups is likely to be structural.

117. Likewise, the ranking of regions by poverty incidence has remained remarkablystable over time. The Northeast was the poorest region in 1988, and it continues to be thepoorest region today. Of course, poverty levels have fallen in absolute termns in theNortheast, but they have fallen far more slowly than in other regions. Again, this sug-gests a strong structural element to poverty in the Northeast.

118. Perhaps, the strongest evidence of structural or chronic poverty is seen in the pro-vincial data on poverty incidence. A ranking of provinces by poverty incidence in eachof the four years - 1994, 1996, 1998 and 1999 - shows that, of the poorest 17 provincesin Thailand in 1994, as many as 8 (Mae Hong Son, Narathiwat, Kalasin, Sakhon Nakhon,Yasothon, Nongkhai, Nakhon Phanom and Sisaket) remained in the poorest-17 group inevery one of the four years. Another 4 provinces (Roi Et, Buriram, Pattani, and Nan)ranked among the poorest 17 provinces in three of the four years, while 3 provinces(Loei, Mukdahan, Chaiyaphum) ranked among the poorest 17 provinces in two of thefour years. Only two provinces (Krabi and Ranong) that were among the poorest 17 in1994 did not rank among the poorest 17 in any of the other years.

119. Figure 38, which Income-poverty incidence in selected provinces as a ratio of national

shows the incidence of inpoverty incidence, 1994-99

come-poverty in four of 3.5 3.5

these provinces in relation 3.2 1994 [1996 C11998 1999

to the national poverty in- 3 3.0 3.0

cidence, indicates that these -8 2.8 2.8 2

provinces, which were * 2.6 2.6

among the poorest prov- 2.5 2.42.3

inces in the country in *_1994, have performed *worse than the national av- 20 1.8

erage in bringing their pov- * 3 2 2* *

Even before the onset of the Narathiwat Sakon Nakhon Sisaket Nakhon Phanom

economic crisis in 1997Figure 38



(i.e., from 1994 to 1996), all of these provinces had experienced an increase in the ratioof their poverty incidence to the national incidence of poverty. This is indicative of ahigh degree of structural poverty among some of the poorest provinces in the country -especially those in the Northeast.

Poverty and Environmental Resource Constraints

120. The linkages between Box 7: Environmental Capitalpoverty and the environment arecomplex and not easily defined. ". .I think people claim that they are poorer because growth doesAccess to natural resources - i.e. destroy the environment. As people refer to me as a Buddhistwater resources, land and soil, the economist, I would like to quote P.A. Payutto (a revered monk).forest and fisheries - is critical to Economists see the merit of production; however, to Buddhism,

production does have a destruction aspect. Production is the usethe everyday functioning of rural of resources to make outputs for consumption or the next produc-households. For some poor tion process. In this light, resources have been converted fromhouseholds, limited or low quality the useable to non-useable resources. It is not enough to look atnatural resources endowments growth alone. Environment capital should be taken into ac-trap them in low productivity live- count..."lihoods. For others, their poverty Source: Participant comments at a TDRI Poverty Workshop, De-compels them to over-exploit the cember 22, 2000.natural resources they have, caus-ing a steady erosion in their natural capital base and in their incomes from it. And inmany instances, the poor are helpless bystanders as other more powerful interests expro-priate their natural resources, leaving them with degraded and less productive forests,fisheries and land. Thus in many ways, the deterioration and over-exploitation of naturalresources can have negative impacts on the livelihoods (opportunities) and security ofpoor people.

121. Extensive research has also illustrated that environmental problems can result inserious damage to the health of people in developing nations. Approximately 19 percentof the illness and death in the developing world can be attributed to environmental factors- access to safe water and sanitation, vector-borne diseases such as malaria, air pollution,etc. Furthermore, there is evidence that the poor tend to suffer disproportionately fromexposure to such environmental hazards.

122. Therefore, addressing the problems raised by environmental resource constraintscan help increase the incomes of the poor, improve their health outcomes, and reducetheir vulnerability to shocks.

123. In Thailand, rapid development over the last 30 years has been accompanied bysignificant pollution and degradation of natural resources in rural areas. Domestic sew-age, industrial effluents, and agricultural run-off have contributed to increasing pollutionof coastal, surface, and ground water. Forest cover has decreased drastically from 53percent in 1961 to 28 percent in 1989, when a logging ban was enacted, to 25 percent in1998. Deforestation is particularly serious in the Northeast, where forest cover has de-creased by nearly 60 percent between 1985 and 1995. Inappropriate agricultural prac-tices and encroachment, fueled by the need to provide income, food and shelter for an



increasing population, are among the main causes of land degradation. The latter in turnaffects agricultural productivity. The Government estimates that soil erosion affects be-tween 17 and 21 million hectares of land impacting both on and off-farm activities, andthis problem is more pronounced in the Northeast.

124. There is an increasing awareness and demand for environmental protection acrossa broad section of society. Civil society representatives have underscored the need to ad-dress environmental problems, given the clear links with poverty. Consensus is emergingamong the Government, private sector, and civil society on the need for increased atten-tion to preserving the environment through better management of natural resources,broader public participation, and effective enforcement of regulations for Thailand. In-deed, the Ninth National Economic and Social Development Plan (9 th Plan), for 2002-2006 explicitly emphasizes that sound natural resources conservation practices are impor-tant for building a strong social foundation as well as sustainable poverty reduction.

Box 8: The Nexus Between the Environment and Poverty

For many years, natural resources were used as inputs in national development to create higher overall income andimprove the living standards of Thai people. During these years, the Thai economy experienced rapid growth despitethe ineffective management of natural resources. However, with no line draw on the appropriate amount of natural re-sources that should be used, resources that once were rich and fertile started to decline....

The rise of natural resource and environmental problems has had a great impact on the level of poverty in Thai-land. A decline in the richness of natural resources and an unbalanced ecological environment has resulted in farmersfacing tremendous loss of profits - their outputs cannot cover their production costs, which means farmers do not haveenough income to spend on their everyday needs. Consequently, some farmers have moved to the urban area seekingemployment in industry - abandoning their work on the farms. An increasing number of labourers coming from thecountryside has resulted in the creation of slums in the urban areas, where the people are faced with a low standard ofliving and instability. The demand for limited natural resources has resulted in increasing debates about the right touse such resources and its allocation in Thai society - this conflict is waged between government and Thai citizens, aswell as among the citizen themselves. This is a structural problem that requires a comprehensive strategy in order toaddress the problems.

POTENTIAL STRA TEGIES TO RESOL VE THE PROBLEM

Preservation of natural resources and environment, and its management: This could be done by supporting communityinvolvement in solving problems; supporting a renewal program for areas that have suffered the most degradation ofnatural resources i.e. forests and marine resources; as well as providing opportunity and increasing local community'scapacity to be involved in resources preservation - ensuring a concrete way of managing the environment.

Improve management system at public level: This could be done by developing a mechanism and strategy to preventconflict in the society that involves the poor; supporting cooperation and agreement between government, private sec-tor, the people, and community institutions in order to solve problems peacefully; and supporting conflict-preventionmechanisms to provide participants with accurate information and establish feedback center to hear each voice.

Reconstruction of the natural resource and environment management system: This could lead to sustainable develop-ment and the reduction of violent conflicts on the use of natural resources. This could be done by improving resourcesand environment management mechanism in order to support the poor's livelihood; supporting the establishment of afund for preservation and restoration of natural resources and environment; supporting participation by local and com-munity in managing the resources and environment; as well as developing guidelines on the use of resources, and cityplan for environment and living standard.

Source: NESDB



125. This "nexus" of poverty and natural resource degradation is an important area forfurther analytical work and policy development. The Thailand Development ResearchInstitute has commissioned a study on poverty and natural resource conflicts in Thai-land's northern region to better understand causes and remedies. More work needs to bedone on how natural resource constraints affect the persistence of poverty in the northeastof Thailand, where both poverty and natural resource problems are concentrated.Whether restoration of natural resources can reduce poverty while increasing the stock ofenvironmental capital is an important empirical question. This "nexus" is a relativelynew domain for environmental science and for poverty analysis. It requires urgent atten-tion in Thailand.

Gender and Poverty

126. Gender issues have become an increasingly important aspect in the debate on de-velopment. Policy researchers and development practitioners have begun building a bodyof evidence and experience that links attention to gender in policies and projects to equi-table, efficient, and sustainable outcomes in development. While quantitative data maybe mixed on the link between poverty and gender, there is clear evidence that suggeststhat societies that discriminate on the basis of gender tend to experience more poverty,slower economic growth, and a lower quality of life than societies in which gender ine-quality is less pronounced. In all countries, but particularly in the poorest, giving womenand men the same rights - allowing them equal access to education, jobs, property andcredit, and fostering their participation in public life - produces positive outcomes.These outcomes include decreased child mortality, improved public health, and astrengthening of overall economic growth.

127. Attempts to estimate the number of women worldwide living in poverty has gen-erated a considerable amount of debate, and no consensus. The main stumbling block isthe lack of an acceptable indicator that can be compared for men and women. The litera-ture on gender has focused on non-consumption indicators such as vulnerability, empow-erment and access to health, education and land. Poverty analysis however tends to focuson differences in welfare between male and female-headed households. The argument isthat households headed by women, facing more constraints, will tend to be poorer thanthose headed by men. However, the evidence linking poverty with headship is mixed. Infact, in the case of Thailand, as earlier analysis has indicated, the incidence of povertyamong female-headed households is typically lower than in male-headed households, ex-cept when the female head is 60 years and older. The probable explanation for the lowerpoverty incidence among young female-headed households is that the latter are house-holds where the adult male has migrated to an urban area to seek employment and betterwages, and is sending remittances back home to a young wife who is temporarily takinghis place as family head.



128. Access to Resources and Opportunities. Research has shown that there is astrong relationship between poverty and gender disparities in basic education and healththat are critical to the ability of both men and women to take advantage of opportunitiesthat will enable them to attain a basic quality of life.

129. Over the last three decades, Thailand, has achieved remarkable progress in its so-cial indicators. The gender gaps have narrowed, especially in health and education. En-rollment of women has increased at all levels, and more women than men are enrolled inhigher education. At the primary level Thailand has close to full gross school enrollmentfor both girls (90%) and boys (93%). In private and public universities 51% of studentsare female and 49% male, and in vocational and technical educational programs 37% ofstudents are female and 63% male. Adult illiteracy has decreased drastically from 27.3%for women and 12.4% for men in 1970 to 7.2% for women and 3.3% for men in 1997.While life expectancy for women (72 years) is higher than the regional average, that ofThai men is lower (66 years), reflecting emerging health concerns specific to men. Infantmortality declined from 73 per 1,000 live births in 1970 to 26.1 in 1995-96. Maternalmortality decreased from 200 per 100,000 live births in 1990 to 44 by 1997. Thus, interms of key indicators related to well-being and better quality of life Thai women havewitnessed considerable improvements illustrating improved access to key social services.

130. While the disparities between men and women illustrate examples of discrimina-tion against women, they do not necessarily provide causality between poverty and thedeprivations in these areas. Nevertheless, there are areas for definite improvement.

131. The Social Monitor IV found that the adverse effects of the crisis were larger formales than for females.2 9 Employment, wages and aggregate wage labor earnings de-clined somewhat more for males than females, and both unemployment and underem-ployment increased more for males than females. However, a much larger proportion offemale workers than male workers earn less than the minimum wage. As in most othercountries, wages for men are 15-20% greater than those for women in Thailand. It is in-teresting to note that during the peak of the recent economic crisis, the wage gap betweenmen and women narrowed to about 14% during the crisis from approximately 18% in thepre-crisis period. However, by the third quarter of 1999, with recovery well under way,the male-female gap had increased to 19% -- illustrating the continued structural prob-lems in the labor market that prevent women from earning more.

132. The high concentration of women workers in semi-skilled positions in the manu-facturing industry, and the significant proportion of women employed in construction,mean that women are increasingly exposed to hazardous working conditions - withoutadequate occupational health and safety measures in place.

133. The National Statistical Office conducted a special survey of home-workers in1999, who are defined as individuals performing work outside the employer's work siteand who are paid by the employer for this work. The majority of home-workers arewomen - 80 percent of those surveyed. There is speculation that the number of home-

29 Thai Workers and the Crisis, July 2000. This section draws heavily from that report.



workers expanded during the crisis, and that they may continue to rise. Because they areoutside the workplace there is concern that these people, many of whom are women andare not protected by labor laws will be subject to substandard working conditions andpay. In fact late payments were a particularly widespread problem faced by home-workers. . There is however anecdotal evidence that women in such informal networksare becoming stronger and better able to voice their concerns. As a result, governmentagencies are beginning to pay more attention to the working conditions of women in theinformal sector.

134. In the formal credit market, women often face difficulties in applying for loans intheir own name, as banking regulations often require the head of household (assumed tobe the man) to make the loan agreement. There are some exceptions, credit programsadministered by the Community Organization Development Institute (see chapter 5 formore information on this Institute) and the Bank for the Poor administered by the Gov-ernment Savings Bank do not require borrowers to be the "heads of household." Thuspoor women can gain access to credit under these initiatives.

135. Managing Risk and Insecurity. Women are falling victim to AIDS at a faster ratethan men. From 1984 until 1998 they represented only one quarter of the total number ofAIDS patients. Today they constitute one-third of the total number of AIDS patients.HIV prevalence among pregnant women continues to rise, from 1.74% in 1997 to 2.02 in1999. In the absence of any intervention, an estimated 4,000-5,000 HIV-positive childrenwill be born annually to women with HIV/AIDs. There is some anecdotal evidence thatfemale and male AIDS patients tend to be concentrated in the lower income groups butthis has not been analyzed nationally.

136. The exact number of women working as commercial sex workers is unknown andestimates vary significantly. The Office of the National Commission on Women's Affairsestimates that about 200,000 women are commercial sex workers, of these up to 20% areunder the age of 18. The economic benefits gained from entering the commercial sextrade can be enticing to poor women and children. Villages in the North, in particularhill-tribe groups, and the Northeast have been susceptible to "agencies" willing to pay theequivalent of a year's salary in return for their children's services (UNICEF, 1997). An-ecdotal evidence suggests that the trafficking of women continues to be a problem in theNorth and Northeast.

137. Research has shown that domestic violence disables women, leads to a loss of in-come, and increases what society spends on healthcare. Since the onset of the economiccrisis over 2 years ago, when many Thais lost their jobs or had their wages severely re-duced as a result, there is anecdotal evidence that the number of women as well as chil-dren who are abused at home and in communities has increased.

138. More research is needed to understand the impact on poor women of AIDS, thecommercial sex trade and domestic violence.



139. Inequality in Voice and Measures to Empower Women. Representation of menand of women in both national and local politics is far from equal, and increasingwomen's political representation remains one of the greatest challenges to gender equalityin the country. Women hold just 8% of government positions-and only 1.5% at the lo-cal level. Few women are seated in the Parliament that consists of 367 (93.9%) male and24 (6.1%) female MPs.

140. The Prime Minister Office's and the National Commission of Women Affairshave issued a new regulation that requires the appointment of a Chief Gender ExecutiveOfficer (CGEO) in all government departments in an effort to ensure that gender issuesare taken into consideration in allocation of the government budget and preparation ofprojects. In September 2000, the Office of the Civil Service Commission issued guide-lines to promote gender equality in civil service administration. Currently women holdapproximately 22 percent of the higher level positions in the civil service.

141. The 1997 Constitution grants equal rights to both men and women. Building onthis framework, the empowernent of Thai women could be supported by increasing theirvoice at the national and community level. In some countries this has been done by re-serving a percentage of positions on party lists for women.

142. As in most areas of social development, the situation in Thailand yields mixedresults. Thailand has done a remarkable job in increasing access to education and healthservices to women, but more needs to be done to increase the voice of women and reducetheir sense of vulnerability and insecurity. The importance of bringing a gender and pov-erty perspective to policy analysis and design is still not widely understood in Govern-ment. While many are sympathetic, gender is still commonly viewed as a political issueand, something not central to country policy-making or to development effectiveness.More attention needs to be paid to achieving gender equity in order to advance the goal ofpoverty reduction.

Correlates and Causes of Poverty

143. As noted earlier, poverty can Box 9: "Causes" of Povertyhave many causes - economic, demo-graphic, social and cultural. But one In a lot of research there exists confusion between theof the single most important reasons characteristics of the poor and the roots of poverty. Weforfpovertyis the single m t i ancmea s say that the poor live in rural areas, mostly in thefor poverty iS the lack of income- Northeastern region. We indicate that the head of thegenerating assets. These include both family has low education and may be divorced - thus aphysical assets, such as land and capi- single parent. The majority of the poor suffer fromtal, and human capital, such as educa- health problems. Then, the conclusion is made thattion. Education is a particularly im- these characteristics are the causes of poverty. Actu-

ally, there is only a relationship or correlation betweenportant asset, the lack of which tends the two, but further study is needed to find out theto be self-perpetuating. The poor genuine causes of poverty. It is highly dangerous tohave low incomes and low productiv- conclude those characteristics are causes and then jumpity because they lack a good educa- into policy formulation.

tion, but they are also less likely to Source: Participant comments at a TDRI PovertyWorkshop, December 22, 2000.



invest in the schooling and education of their children because they cannot afford it. Thismeans that they are often trapped for generations in a vicious circle of poverty and loweducation.

144. Shocks (insecurity) are another important cause of poverty, especially transitorypoverty. Natural disasters (e.g., floods, cyclones, drought and earthquakes), macroeco-nomic shocks (e.g., financial crises, recessions, and a collapse in agricultural commodityprices), and catastrophic illness can easily push large numbers of people into poverty.The high cost of medical care and the large opportunity cost of illness (in terms of lostwages), combined with the general lack of medical and/or disability insurance, a catas-trophic illness in the family can wipe out years, if not a lifetime, of savings for a familyand impoverish it.

145. In addition to the lack of assets and income shocks, social and cultural factors alsoplay an important role in perpetuating poverty. The role of social and political exclusionis particularly important. Some groups are excluded from political participation becausethey are too few in numbers to attract government attention, too weak to lobby effectivelyfor their rights and interests, or too unimportant as voters or taxpayers in the eyes of gov-ernment. The poor may also face social exclusion, both in the form of social discrimina-tion (e.g., discrimination against women, ethnic or religious minorities, orphans, and sin-gle mothers) and in the form of not having adequate social capital - the network of in-formal norms and established relationships that enable people to act in concert for mutualbenefit.

146. Exclusion greatly increases the risk of poverty, as excluded individuals andgroups are less likely to find ways to benefit from economic growth opportunities andfrom formal safety-net programs. At the same time, severe poverty and deprivation canlead individuals or groups to exclude themselves from society out of a sense of shameand humiliation. In this sense, poverty and exclusion form a vicious circle, with exclu-sion increasing the risk of poverty and impoverishment in turn reinforcing exclusion. Forthis reason, exclusion is often identified as a facet of poverty and poverty as a facet ofexclusion. Indeed, within a distributive justice-based definition of poverty, the absenceof dignity and participation in social life are included as poverty indicators (Rawls 1971,Sen 1992).

147. An overwhelming share of Thailand's poverty is concentrated in the rural areas,especially in the rural Northeast. In the rural areas, land is the most important factor ofproduction. Hence, one would expect inequality in the distribution of land to be an im-portant factor in explaining poverty. 30

3OFor instance, one reason for the very low levels of rural poverty and income inequality in South Koreaand Taiwan is the egalitarian distribution of land in those countries. This did not evolve naturally, but in-stead was the result of deliberate large-scale agrarian reform in the late 1940s and early 1950s designed tostave off the threat of communism. In South Korea, the 1949 land reform redistributed about one-half of allagricultural land (Haggard 1990) while the Taiwanese reforms of 1949-1953 redistributed about a quarterof the land (Ho, 1978).



148. The Socioeconomic Survey data for 1998 are used here to analyze the determi-nants of household income per capita and of the probability of an individual being in-come-poor. 3 ' The results of the regression and probit analysis are shown in Table 3. Themain empirical findings are discussed below.

149. Size of land holdings is an important cause of poverty in the rural areas. The re-sults indicate that a one-percent increase in land ownership reduces the probability ofpoverty by about 0.1 percent. The effect of tenancy is much greater; controlling for otherfactors, the probability of being a tenant (as opposed to a land-holder) increases the riskof poverty by nearly 30 percent.

150. Demographic factors have an important influence on the risk of poverty in bothrural and urban areas. Larger families are at significantly greater risk of poverty thansmaller families. On the other hand, individuals residing in households that have a largerproportion of adults (as opposed to children) are less likely to be poor.

151. The effect of residing in a female-headed household on rurai poverty varies de-pending upon the age of the female head. Female-headedness increases the risk of pov-erty, but only when the female head is over the age of 55 years. Individuals residing inhouseholds headed by younger women actually have a lower risk of poverty than those inmale-headed households.3 2 This may reflect the fact households headed by youngwomen are likely to be those in which the adult male earner has temporarily migrated toan urban area for employment reasons. As such, these households may have access toremittances and transfers from the male migrant. However, households headed by olderwomen are likely to be those in which the adult male earner has died, and this may in-crease the risk of poverty for the widow and her relatives. Female-headedness has nosignificant effect on poverty in the urban areas.

152. Education has a powerful effect on reducing rural poverty. When the highest-educated adult in the household has primary or lower secondary education (as comparedto no education), the risk of poverty declines by about 66-74 percent. The correspondingfigure for tertiary (university) education is 90 percent. However, education has signifi-cantly smaller effects on reducing poverty in the urban areas, as education is more wide-spread there.

153. Finally, geographic location has a strong influence on poverty. Even after con-trolling for all of the above factors, residence in the rural areas of the Northeast increasesthe probability of being poor by 56 percent (in comparison to being resident in the ruralSouth). In the urban areas, residence in Bangkok and vicinity reduces the probability ofpoverty by as much as 80 percent.

3"Strictly speaking, the regression analysis does not identify the causes of poverty, but instead the associa-tions between poverty and certain observable characteristics of individuals and households.32 For instance, an individual residing in a household where the female head is 70 years old faces a 3.5 per-cent increase in the probability of being poor. But a person residing in a household with a female headaged 25 years old has a 6.5 percent lower probability of being poor.



154. Obviously, the empirical analysis discussed above is limited to variables that areobserved and measured in the SES data. As such, it cannot (and is not meant to) identifyall of the various determinants and correlates of poverty. The role of factors such as ex-clusion, social capital, and ethnicity in promoting poverty cannot be analyzed due to gapsin the available national data sets.

Poverty in the Northeast

155. Since such a large proportion of the poor in Thailand are concentrated in the ruralareas of the Northeast, it may be useful to see if the determinants of poverty in the ruralNortheast are different than those in the rest of the country. Table 4 shows the estimatesof separate probit equations for the probability of being poor for the rural Northeast.

156. The results suggest that the determinants of poverty in the rural Northeast differfrom those in the rest of the rural areas in the country in four ways. First, the amount ofland owned has a much stronger effect on reducing poverty in the rural Northeast than inother rural areas. For example, a one-percent increase in owned land reduces poverty byas much as 0.2 percent in the rural Northeast but only by 0.03 percent in other rural areas.Second, while being a tenant significantly increases the probability of poverty in the otherrural areas of the country, it has no such effect in the Northeast, perhaps because tenancyis not as widespread in the Northeast as in the other parts of the country.

157. Third, somewhat surprisingly, schooling (with the sole exception of tertiary oruniversity education) generally has a much smaller effect on reducing poverty in the ruralNortheast than in other rural areas of the country. In fact, none of the lower levels ofschooling (primary, secondary or vocational) have any significant effect on the probabil-ity of being poor in the rural Northeast. This is a most unusual result, since one wouldhave expected education to be an even more important factor in reducing poverty in theNortheast than in other regions of the country. However, the empirical result most likelyreflects the fact that the main value of education in the Northeast lies in moving out of theregion to other regions of the country where higher-wage employment opportunities canbe found. However, most such educated individuals who have moved out of the North-east are no longer captured in the SES survey. This is likely to substantially underesti-mate the estimated effects ofeducation on poverty using the Estimated effect of female-headedness on the probability of being

SES sample. 0.15 poor, by age of female head

158. Fourth, while residing in 0.10.

a household headed by a female 0

of any age reduces the risk of °poverty in the rural areas of the D ° 0.00T

rest of the country, the situation 0.05-E

is very different in the rural 9 C j

Northeast (Figure 39). There, - °'° X

the age of the female head 0.15- o,

makes a big difference to the --0.20- ,.E ,,

20 25 30 35 40 45 50 55 60 65 70Age of female head (years)

Figure 39Thailand Social Monitor 49

Pove-ty and Public Policy

risk of poverty. For example, residence in a household headed by a woman aged 20 yearsreduces the probability of being poor by 0.13, but residence in a household headed by a70-year old woman increases the risk of poverty by 0. 13. Again, this is likely to reflectmigration patterns. Younger women-headed households in the rural Northeast are typi-cally those where the main male bread-winner has moved to an urban area (typicallyBangkok) for higher-wage employment. Persons residing in such households are there-fore less likely to be poor. However, older females heading a household are most likelyto be widows, who typically have few sources of income, and persons residing in thesehouseholds are likely to be at higher risk of poverty.

159. Thus, poverty in the rural Northeast is not only greater than in other parts of thecountry, but is associated with different causal factors. Land ownership has a signifi-cantly larger effect on reducing the risk of poverty for individuals in the rural Northeastthan for individuals in other rural areas of the country. In addition, households headed byolder women (i.e., those aged 45 years or more) are at significantly greater risk of povertyin the rural Northeast, which is in stark contrast to the rest of the country, where femaleheadedness is actually associated with a lower risk of poverty. Thus, anti-poverty pro-grams in the rural Northeast need to target small and marginal farmers and householdsheaded by older women.

Economic Growth, Inequality, and Poverty Reduction Across Provinces

160. There is a large literature on the relationship between growth and poverty whichargues that economic growth is one of the important determinants of poverty reductionacross a cross-section of countries (e.g., see World Bank, 2000; Dollar and Kray, 2000).This point has been empirically established for a cross-section of developing countries aswell as for some individual countries over time. But little is known about the effect ofeconomic growth on poverty reduction in Thailand.

161. At an aggregate level, Growth rate in er ca ita GDP and chan es in incidence of' ~~~~~~~consumption-poverty, 1988-99

it is clear that poverty has 6

fallen rapidly in Thailand dur- 51996-984

ing periods of strong growth 3

in the economy, and that it has , 2

increased during periods of . 0 94

weak or negative growth (Fig- -I \ 1994-96

-2 -2t1909ure 40).Z -3

162. However, one problem 5\with analyzing the relation- 767 1988-90

ship between growth in per -8 199294

capitaGDPandchangesin -8-7-6-54-3-2-1O 1234567891011

poverty at an aggregate level % annual growth in per capita GDP

is that there are relatively few Figure 40data points to estimate the relation-ship precisely. In addition, most studies on the poverty-growth relationship at the cross-



country level have ignored the contemporaneous effect of changes in income inequalityon poverty. Changes in income inequality mediate the effect of economic growth onpoverty reduction. For instance, the impact of economic growth on poverty reductionwill be smaller if it is associated with a worsening distribution of income.

163. An attempt is made in this Social Monitor to use household survey data from theThai Socioeconomic Surveys, first, to calculate measures of growth, poverty, and distri-bution for each of the 76 Thai provinces, and to then analyze the relationship betweenpoverty changes, economic growth, and distributional changes. Data from the 1994,1996 and 1998 SESs are used for this exercise.

164. The regression analysis (shown in Table 5) suggests that poverty levels have astrong inverse association with mean per capita income in a province and a strong posi-tive association with income inequality. Surprisingly, these results continue to hold evenwhen changes in poverty are related to changes in mean per capita income (i.e., eco-nomic growth) and changes in income inequality.33 The empirical results indicate that anincrease of Baht 100 in real monthly income per capita is associated, on average, with apoverty reduction of 1.3 percentage points in Thailand's provinces. As would be ex-pected, the poverty-reducing impact of equivalent mean income growth is greater amongpoorer provinces than among better-off provinces. For a province with mean monthly percapita income of Baht 1,000 (in constant 1994 Bangkok prices), an increase of Baht 100in real per capita income would result in a reduction in poverty of 1.6 percentage points.However, for a muchricher province hav- Figure 41

ing mean income per Shares of each region in total GDP and population increase takingcapita of Baht 8,000, place in Thailand between 1987 and 1996

the reduction in pov- 100%

erty incidence from a - * South

Baht 100 increase in 90% N

income would only 80% 8

be 0.1 percentage 70% - North 32

points. Clearly then, 60% U Central

income growth at the 50% 0 Bangkok and

provincial level can 40% Vicinity 22

have a powerful im- 30%-

pact on poverty re- 20% 52

duction, especially 10%

among poor Thai 0%provinces. If this is % share in GDP increase % share in population increase

the case, why has

3 3Typically, the relationship between provincial poverty levels and mean per capita income levels becomesweak and insignificant when changes in poverty and changes in mean income are considered, because thelatter association effectively controls for unobserved province-specific factors, such as the geographicallocation of a province, the quality of its land, its history, the quality of its administration, and various cul-tural factors, that may be important determinants of its poverty level (as well as of its mean income level).Those factors are visible in the levels of poverty and per capita income, but are no longer present in thechanges in these indicators from one year to another.



poverty remained stubbornly high among poor provinces?

165. The regression results point to one reason why poverty reduction has been slug-gish in some of the poorer provinces in Thailand. The results indicate that increased in-come inequality is positively and strongly associated with increases in poverty, even aftercontrolling for the impact of economic growth. A one-point increase in the Gini index ofincome inequality is associated with a 0.8 percentage point increase in poverty incidence.Thus, a province having mean monthly income per capita of Baht 1,000 would experi-ence a poverty reduction of only 0.4 (instead of 1.6) percentage points with an increase ofBaht 100 in real per capita income if that increase in mean income was accompanied byan increase of 2 points in its Gini index of income inequality (e.g., its Gini index went upfrom 52% to 54%). Thus, worsening income inequality can substantially offset the bene-ficial effect of economic growth on poverty reduction, and appears to have done so overthe 1994-98 period. This will be analyzed for Thailand as a whole further below.

166. Another reason for the sluggish poverty performance in poorer provinces is thatgrowth of household income per capita has been far from equal across provinces.3 4 Asnoted earlier, many of the poorest provinces in the country have in fact experienced de-clines in real income per capita over the period 1994-99. Admittedly, this is the result ofthe uneven impact of the economic crisis on the poorest provinces. However, even in thepre-crisis period, economic growth was unevenly distributed across provinces and re-gions. Figure 41 shows that 'Bangkok and Vicinity' accounted for 52% of the increase innational GDP between 1987 and 1996, even though it accounted for only 11% of the na-tional increase in population over that period. In contrast, the Northeastern region, whichaccounted for nearly one-third of the population increase in Thailand during those nineyears, had a share of only 11% in the national increase in GDP over 1987-96. Thus,Bangkok residents enjoyed a share of the national income gain that was about five timesas large as their share of the population gain, while the corresponding ratio for the resi-dents of the Northeast was a mere one-third.

167. To better demonstrate the effects of changes in inequality on changes in poverty,35

an analysis similar to the one undertaken above can be done at the aggregate level for thecountry as a whole. In other words, one can decompose the total change in poverty inci-dence in Thailand between any two periods of time into the change that occurred due toeconomic growth (i.e., keeping constant the distribution of income) and the change thatoccurred due to changes in the income distribution (i.e., holding income constant). Ob-viously, the first effect will always be negative since economic growth in the absence ofany distributional changes will always reduce poverty. However, the latter effect can beeither negative or positive depending on whether growth is accompanied by improving orworsening inequality.

34Indeed, using the example in the previous paragraph, economic growth rates (i.e., per cent increase inmean real income per capita) have to be eight times as large in a province having mean income per capitaof Baht 1,000 as in a province having mean income per capita of Baht 8,000 for both provinces to experi-ence an equivalent (Baht 100) increase in mean income per capita.35A somewhat similar exercise has been done for Thailand, Korea and Laos by Kakwani and Pemia (2000).However, the figures shown in this section have been calculated independently and presented somewhatdifferently than in that paper. The overall conclusions remain the same.


Poverty and Pwblic Policy

168. As discussed earlier, Changes in the income-poverty headcount ratio and in the number of

the headcount ratio of pov- income-poor persons decomposed into changes due to growth and

erty in Thailand declined changes due to distributional shifts, 1992-94 to 1998-99

significantly from 1992 to Change in poverty headcount ratio (% points) Change in number of poor (millions)

1996 - from 23.2% to2 -11.4%. However, this de- -m

cline was reversed with the 0 _ 0.- ;

onset of the economic cri- o. H -02 3 -0.2

sis. The incidence of pov- °-2

erty increased to 13% in , -621998 and 15.9% in 1999. 4_

Figure 42 shows the results * Change due to distributional shif. . -6 - _ -Cl Change due to growth

of the decomposition analy- 6d

sis for the four time peri- _8 1992-94 1994-96 1996-98 1998-99 1992-94 1994-96 1996-98 1998-99

ods: 1992-94, 1994-96,1996-98 and 1998-99. What is Figure 42striking is the reversal in the con-tribution of inequality to poverty changes since the economic crisis. In other words, theincidence of poverty and the number of poor in the country in 1998 and 1999 would havebeen lower than they actually were, had there been no change in the distribution of in-come. This reflects the fact that the distribution of income worsened after the crisis.

169. In the beginning of the decade, rapid economic growth caused a large decline inpoverty incidence and the number of poor, with changes in distribution playing a rela-tively small role. For instance, between 1992 and 1994, the poverty headcount ratio de-clined by 6.9 percentage points (Figure 42). The contribution of growth to the total de-cline was 6.2 percentage points (or 90%), and that of a slightly improving distribution ofincome was 0.7 percentage points (or 10%). Had the income distribution not improvedduring this period, the reduction in poverty incidence would have been somewhat smaller(6.2 percentage points), implying that the improvement in income inequality reducedpoverty incidence by 0.7 percentage points more than what it might have been with nodistributional shifts. The size of the distributional effect relative to the growth effect wassmall because growth was so strong. Between 1994 and 1996, there was hardly anychange in the distribution of income, but since economic growth was strong, poverty de-clined by another 4.9 percentage points.

170. Just as income growth slowed down dramatically in the aftermath of the crisis,income inequality began increasing. This accentuated the increase in poverty. For in-stance, between 1998 and 1999, poverty incidence increased by 3 percentage points. Hadthe distribution of income remained neutral between 1998 and 1999, the incidence ofpoverty would have fallen by a modest 0.5 percentage points - not increased by 3 per-centage points, as it actually did. In terms of absolute numbers, the worsening inequalityresulted in an additional 2.1 million persons becoming poor between 1998 and 1999.



171. What this suggests is that income inequality can play a critical role in poverty re-duction, especially in a low-growth environment.

172. In a recent paper, Kakwani (2001) has developed a methodology to measure thetradeoff between inequality and growth in a country. The index, which he calls 'the ine-quality-growth tradeoff index or IGTI', indicates the amount of economic growth that isneeded to offset the adverse impact of an increase in income inequality on poverty. Asan example, an IGTI of 3 for a particular country would imply that a one-percent increasein the Gini index in that country would need to be offset by an economic growth rate of 3percent in order to keep poverty incidence in the country constant. Thus, the larger thevalue of IGTI, the greater is the potential benefit of inequality reduction.

173. Using household survey data for 1998, Kakwani then calculates the value of theIGTI for two measures of poverty - the poverty gap ratio and the poverty severity ratio -for four countries in Asia (Laos, Philippines, South Korea and Thailand). He finds that,for both the poverty gap and the poverty severity ratios, Thailand has the highest valuesof the IGTI among the four countries. For example, the IGTI for the poverty gap ratio is4.1 for Thailand, as compared to 0.9 for Laos, 1.2 for Korea, and 2.3 for the Philippines.This means that, for Thailand, a reduction of one percent in income inequality is equiva-lent to having a growth rate of 4 percent. Indeed, the benefits of inequality reduction aresignificantly greater in Thailand than in any of the other three Asian countries. Theseresults again imply that a strategy of inequality reduction can have very high pay-off inThailand relative to other countries.

174. In both the provincial and national analysis discussed in the two earlier sections,one maintained assumption is that economic growth and income inequality are independ-ent of each other. However, in a recent paper, Barro (2000) has shown that initial levelsof inequality have a negative effect on subsequent growth rates in developing countries,although he does not find a similar effect for more developed economies. Human capitalinvestments are one mechanism through which improvements in the distribution of in-come may foster growth. In developing countries, credit constraints may prevent thepoor from investing in human-capital investments (e.g., schooling) that offer high rates ofreturn. In such a situation, a redistribution of assets and incomes from the rich to thepoor will result in more human-capital investment, and this in turn will raise the averageproductivity of investments and thereby the growth rate of the economy.

175. The provincial data for Thailand strongly support Barro's findings (Table 5).Provinces that had a higher level of income inequality (as measured by the Gini index) in1994 were more likely to have lower economic growth rates in each of the subsequentperiods (viz., 1994-96, 1996-98, and 1998-99). The regression estimates indicate that aone-point increase in the Gini index is associated with a 0.4% decline in the growth ofreal income per capita. Thus, income inequality has two effects on poverty reduction:first, increased inequality is associated with increased poverty after controlling for eco-nomic growth; and, second, high levels of inequality can reduce future growth rates in theeconomy, thereby impeding the poverty reduction that would have taken place in thepresence of rapid growth.



176. The foregoing analysis strongly suggests that inequality in Thailand needs to betreated more seriously as a policy problem, on a par with poverty, and as a major deter-minant of poverty. Inequality is, however, difficult to change through standard policyinstruments, and tends to shift slowly over time. One challenge for Thailand policymakers will be to prevent further erosions in the distributional equity of the Thai econ-omy.


177. This chapter has explored the evidence on several policy issues that are central toaddressing poverty and inequality in Thailand. These more strategic issues will shape thechoices that policy makers will confront as they design programs to deal with poverty.

178. What are the findings on these key issues?

* Self Reliance and Safety Nets: Thai families provide a substantial and enduring safetynet to each other and within the family, and have thus moderated the severe impactsof the crisis. Through cash remittances and gifts, the poor in particular have benefitedon a net basis from the generosity of friends and family, and this pattern increasedduring the crisis. While the crisis itself added 4.5% to the level of poverty, thesetransfers were able to prevent up to a further 5 to 6% of Thai families from falling be-low the poverty line. Policy makers face a difficult dilemma in designing new secu-rity enhancing programs: they must avoid incentives that could serve to reduce volun-tary private assistance, but must take action to protect those who fall through thecracks. Insurance based mechanisms, rather than a purely transfer approach, may helpsustain self reliance and mutual assistance.

* Causes of Poverty: Land ownership, tenancy status, and education appear to be someof the strongest determinants of poverty in Thailand, especially in the rural areas. Inaddition, in the rural Northeast, households headed by females aged 45 years and overare at particularly great risk of poverty. This suggests that, in the short run, povertyprograms need to target small and marginal farmers, tenants, and households headedby older women. In the long run, land reform and expansion of education to the poorwould be effective poverty-reduction strategies.

* Structural or Transitory Poverty: Poverty in Thailand is becoming more structural orin another word more "chronic" which means that those households in poverty todayare likely to be poor tomorrow. The evidence for this conclusion comes from a num-ber of perspectives, such as the persistence of poverty among certain householdgroups, the stable ranking of regions in respect to poverty levels, and the growing di-vergence in mean income and poverty levels between the poorest regions and the bet-ter off. At the provincial level, the poorest provinces tend to remain in that statusyear after year. These findings point to the increasingly difficult challenge of liftingpeople out of a relatively small poverty group of 15% of the population. The chal-lenge for Thai policy makers is how to reach this group of increasingly chronic poor,and to do so with efficient targeting. There are a number of interventions that havebeen used in other countries to reach such groups, but Thai policy makers will need to



shape their choices to Thai realities. A critical question is whether economic growthalone can be expected to do this, once recovery in Thailand takes hold.

* Natural Resources and Poverty: The nexus of poverty and natural resource degrada-tion is an important area for further analytical work and policy development in Thai-land. More research needs to be done on how natural resource constraints affect thepersistence of poverty in the northeast of Thailand, where both poverty and naturalresource problems are concentrated. Whether restoration of natural resources can re-duce poverty while increasing the stock of environmental capital is an important em-pirical question.

* Gender and Poverty: The role of gender as a determinant of poverty and in the effortto reduce poverty presents a mixed picture in Thailand. While most social disparitieshave eased or disappeared (women outnumber men in Thai universities), others haveproved highly resistant to change, particularly economic disparities in wages and op-portunity. In the Northeast, female headed households show a distinctive probabilityof poverty. Some social problems affect women disproportionately, such as AIDSand domestic violence. And women remain heavily underrepresented in the politicalprocess. Gender issues need to be taken into account more systematically, particu-larly in respect to anti poverty interventions.

* Growth, Inequality, and Poverty: For most of its recent history Thailand has relied onrapid economic growth to lift people out of poverty, but this process has slowed, in-evitably, as the poverty sub group has steadily diminished in size. But is this the onlyfactor at work? Have changes in the distribution of income also played a role? Priorto the crisis, Thailand enjoyed a double boon for the poor: high rates of growth andimproving income distribution since 1992. After 1996, both trends have shifteddirection: the economy contracted and in the process, the poor endured the brunt ofthe decline, as income distribution worsened. From 1998 to 1999, poverty incidenceincreased by 3%; if inequality had remained unchanged, the increase in povertywould have been only 0.5%. At the provincial level, inequality also plays a strongrole in poverty outcomes, even after controlling for the impact of growth. Inequalityis much harder to deal with from a policy perspective, but, at already high levels, thegovernment will need to guard against further increases in income inequality thatwould exacerbate social tensions.



Table 3: Estimated maximum likelihood probit equations for the probability of being income-poor,by rural and urban areas, 1998

Villages (Rural areas) Municipal and Sanitarv Districts

Marginal effect Elasticity Robust Marginal effect Elasticity Robust

Independent Variable (dF/dX) at mean std. error (dF/dX) at mean std. error

Household size 0.0404 1.075 0.0041 0.0034 0.425 0.0008

Number of eamrlers -0.0203 -0.292 0.0051 -0.0005 -0.033 0.001C

No of members aged 25-44 yrs in household -0.0008 -0.210 0.0002 -0.0002 -0.318 0.0001

No of members aged 45-59 yrs in household -0.0014 -0.126 0.0003 -0.0003 -0.127 0.0001

No of members aged 60 yrs and older in household -0.0019 -0.125 0.0004 -0.0002 -0.070 0.0001

Whether highest-educated member of household has.-Primary schooling -0.1283 -0.740 0.0251 -0.0015 -0.043 0.0061

Lower secondary schooling -0.1147 -0.662 0.0148 -0.0086 -0.254 0.0040

Upper secondary schooling -0.1318 -0.761 0.0112 -0.0115 -0.339 0.0031

Vocational education -0.1415 -0.817 0.008 -0.0145 -0.426 0.0040

University education -0.1543 -0.890 0.0049 -0.0237 -0.698 0.0048

Age of household head 0.0005 0.154 0.0005 0.0001 0.145 0.0001

Whether household head female? -0.1210 -0.698 0.0233 -0.0068 -0.199 0.0070

Whether household head female X Age of head 0.0022 0.0 0 7 0.0001 0.0002

Dummy for second quarter -0.0029 -0.016 0.0120 -0.0003 -0.010 0.0025

Dummy for third quarter 0.0077 0.044 0.0121 0.0000 0.000 0.0024

Dummy for fourth quarter 0.0101 0.058 0.0124 0.0044 0.130 0.0028

Owned area (rai) -0.0018 -0.096 0.0004

Whether tenant cultivator? 0.0513 0.296 0.0170

Dummy for Bangkok _ _-0.0270 -0.795 0.0029

Dummy for Central region -0.0687 -0.396 0.0112 -0.0089 -0.261 0.0023

Dummy for North. -0.0360 -0.208 0.0121 -0.0037 -0.109 0.0023

Dummy for Northeast 0.0965 0.557 0.0132 0.0116 0.341 0.0040

Proportion of sample income-poor 0.1733 -226.1948 0.0340

Log likelihood ratio -4,226 -1,527

Weighted number of observations 16,189 7,359

Wald chi-square statistic 606.47 297.81

Notes: Coefficients in bold are significant at the 5 percent or lower level.



Table 4: Estimated maximum likelihood probit equations for the probability of being income-poor,Rural Northeast and Rural Other Regions, 1998

R ural Northeast Rural Other RegionsMarginal effect Elasticity Robust Marginal effect Elasticity Robust

Independent Variable (dF/dX) at mean std. error (dF/dX) at mean std. errorHousehold size 0.0577 1.062 0.0100 0.0305 1.242 0.003Number of eamers -0.0214 -0.221 0.0116 -0.0187 -0.398 0.0046No of members aged 25-44 yrs in household -0.0007 -0.118 0.0004 -0.0009 -0.398 0.0002No of members aged 45-59 yrs in household -0.0023 -0.141 0.0007 -0.0009 -0.127 0.0003No of members aged 60 yrs and older in household -0.0031 -0.119 0.0008 -0.0011 -0.132 0.0003Whether highest-educated member of household has:Primary schooling 0.0228 0.087 0.1080 -0.1074 -0.981 0.0186Lower secondary schooling 0.0115 0.044 0.1132 -0.0901 -0.823 0.0093Upper secondary schooling -0.0343 -0.131 0.1047 -0.0981 -0.897 0.0070Vocational education -0.1476 -0.563 0.0695 -0.1002 -0.915 0.0066University education -0.2486 -0.948 0.0146 -0.0981 -0.896 0.0048Age of household head 0.0013 0.243 0.0012 0.0000 -0.021 0.0005Whether household head female? -0.2323 -0.887 0.0477 -0.0568 -0.519 0.0241Whether household head female X Age of head 0.0051 0.01 1 5 0.0003 0.0007Dummy for second quarter 0.0119 0.045 0.0260 -0.0109 -0.100 0.0105Dummy for third quarter 0.0316 0.121 0.0263 -0.0065 -0.059 0.0104Dummy for fourth quarter 0.0564 0.215 0.0276 -0.0137 -0.125 0.0102Owned area (rai) -0.0045 -0.2111 0.0008 -0.000 -0.028 0.0003Whether tenant cultivator? -0.0150 -0.057 0.0344 0.0621 0.567 0.0156

Proportion of sample income-poor 0.2621 0.1094Log likelihood ratio -1,655 -2,261Weighted number of observations 6,181 4,441Wald chi-square statistic 176.38 341.08Notes: Coefficients in bold are significant at the 5 percent or lower level.



Table 5: Cross-province regressions of incidence of incom -poverty, 1994-99Le vels ((O_LS First difference (changes'

Independent Variable Parameter T-ratio Parameter T-ratio

Mean per capita income -0.01975 -20.13 -0.0182295 -9.02

Mean per capita income squared 1.56E-06 12.83 1.05E-06 7.86

Gini ratio (x 100) 0.694632 11.47 0.7756918 9.52

Dummy for 1996 -1.83657 -1.74

Dummy for 1998 -0.84451 -0.8 1.794845 1.86

Dummy for 1999 0.854989 0.81 3.443878 3.58

Intercept 20.58 896 6.09 -1.227703 -1.59

Number of observations 304 228

F ratio 148.73 30.55

R-squared 0.750 3 0.5571



Chapter 5. Empowering the Poor

"Ifyou don't have a strong community, anything you do will be set to fail. No matterhow much the government or authorities might try to help, nothing will be accomplished

But with a strong community, coupled with government support, you can fosterdevelopment which can stand on its own. "

Phra Subin Panito, the spiritual leader of Baan Pret, Bangkok Post, January 11, 1999.

179. Poverty is no longer seen as merely a lack of income and a lack of access to basichuman needs such as education and health. Poverty encompasses many more dimensions.To understand poverty in all its dimensions one must also examine issues such as thepoor's vulnerability to shocks be they from ill-health, economic or natural disasters andthe powerlessness of the poor in their ability to influence key decisions affecting theirlives due to lack of political freedom or voice. Thus to develop an effective strategy forpoverty alleviation one has to address these other dimensions. The success of anypoverty alleviation strategy will heavily depend on the ability and desire of any nation topromote opportunities, empower its people, and enhance their level of security.

180. The next chapter will assess the performance of programs that were introduced toexpand opportunity for the poor and to improve the security of the poor from unexpectedshocks. This chapter will focus on empowerment. Why is "empowerment" a key topoverty reduction? It means giving voice to the people and allowing them to have a sayin their own development. Many studies have demonstrated the positive effects ofparticipation and local decision making on the performance of projects and programs.What does this mean in the context of Thailand? It means decision-making by policymakers should be done in collaboration with civil society at all levels - NGOs,academics, community organizations, and communities themselves.

181. The concept of empowerment is not a new one for Thai society. Empowering thecommunity in Thailand means building social capital and promoting self-reliance. Itmeans crafting a development paradigm that is balanced and brings harmony back intothe society (as posited by the Ninth Development Plan) and moving away from the"excesses" of the high-growth era which led to commensurate high social costs. Effortsaimed at strengthening Thai society have been embodied in different ways: starting withthe lessons of H.M. the King calling for Thai people to be more self-reliant; to legalreforms mandated by the new Constitution; to the development agenda advanced by the8th and g9h National Development Plans; and finally to the growing grassroots movementsamong the people themselves seeking a larger role in policy formulation.

182. The political landscape in Thailand has been dominated by short-lived coalitionpolitics. Elections have encountered problems of vote-rigging and been subject to themanipulation by powerful business interests. The influence of the military shaped Thaipolitics for the last half century, however this influence has waned with the emergence ofcivil society. The failure of a military coup in 1992 marked the beginning of a newcivilian influence in politics, institutionalized with the passage of constitutional reform in1997.



Support From the Legal Environment

183. In the years leading up to the economic crisis, many currents of social reformwere gaining momentum in Thailand. Civil society organizations and leaders pushed forpolitical reform and government policies more responsive to the lower income andvulnerable groups. Their activities culminated in the drafting of the new Constitutionwhich was promulgated just months after the advent of the crisis in July 1997. Designedto end the inertia that has virtually paralyzed previous governments, it is widely regardedas the "People's Constitution" and the key to social reform in Thailand.

184. The new Constitution creates the framework for the restructuring of nationalrepresentative institutions and reform of the electoral process. It introduces a number ofnew measures to empower civil society and safeguard individual liberties. Along withthe promotion of greater politicalaccountability and transparency, Box 10: Giving the Power to the People -the Constitution emphasizes The Thai Constitution

gender equality, environmental Apreservation, local resource Article 45 - Grants individuals the right to form anmanagement, co mmunity association, a union, league, co-operative, farmer

management, co mmunity group, private organization, or any other group.empowerment, and basic rightsin education and health. It takes * Article 58 - Grants individuals the right to have accessunprecedented steps against to public information.abuses of power and conflict of o

interest. Infringements of Article 59 - Grants individuals the right to receivepersonalrigsis prevented by information and express their opinion from agencies

personal rights before permission is given for the operation of anyone of Asia's most liberal codes projects that may affect the quality of the environment,to support individual freedoms. health and sanitary conditions, the quality of life or any

other material interest concerning a person or a local185. The Constitution community.

provides the legal foundation for Article 60 - Grants individuals the right to participate incommunity em powermnent. decision-making process of government officials in

.Many of its provisions - from those areas which affects or may affect their rights orthe decentralization of power to liberties.the local level, to the right ofaccess to information, and the Article 76 - Duty of State to promote and encourageright to participate in the public participation in the decision-making process atdecision making process at all all levels.levels (see Box 10) - have * Article 78 - Duty of State to decentralize power to localbecome powerful tools for civil authorities.society in its efforts to reformthe political and socialenvironment of Thai society.



Institutional Environment

186. Complementing the Constitution, the Eighth Plan (1997-2001) prepared by theNational Economic and Social Development Board (NESDB), the Thai planning agency,emphasized an agenda of "people-centered development". A widely consultative processwas launched, seeking inputs from local communities and non-governmentalorganizations in the formulation of the Plan. The result was a document focusing on thepoor and under-privileged, and with initiatives designed to empower local communitiesthrough decentralization of government, public participation in the Government'sdecision-making process, increased transparency and improved governance, and thebuilding of local capacities and self-reliance.

187. Implementation of the 8th Plan has been launched in several ways. The Plan callsfor the coordination of government and non-governmental activities. Monthly learningforums held at the NESDB share local-level development experiences and newinitiatives. Provincial civic forums have been promoted to encourage all concernedparties including the public to participate in the planning and decision making in eachprovince. District and sub-district civic forums are also being encouraged.

188. Building on the 8th Plan and the reform agenda supported by the Constitution, theNESDB has solicited inputs from various segments of civil society for the Ninth NationalEconomic and Social Development Plan (2002-2006). The Ninth Plan espouses fourmain objectives (i) poverty alleviation (see box 11); (ii) support for good governance (iii)adherence to the new economy & self-sufficiency; and (iv) strengthening the socialfoundation through the enhancement of the quality of life and protection of naturalresources. The gth Plan embodies the growing sentiment in Thai society calling for amore balanced approach to development. It is believed having a more balanced approachwill improve the lives of all Thais and help to narrow the inequality gap as well as protectthe most vulnerable groups in society.

Progress in Implementing Reforms Mandated by the Constitution

189. Corruption can and will effect the economic performance of a country, not onlydoes it undermine opportunities, it makes efforts to reduce poverty that much moredifficult to achieve. Reforms emphasizing accountability and transparency are allmandated by the new Constitution. New institutions, such as the National Counter-Corruption Commission, the Electoral Commission, the Administrative Courts and theOffice of the Ombudsman have been established and are proving to be effective in theirefforts, thus gaining public confidence. These institutions are powerful mechanisms toincrease the voice and concerns of Thai communities, and in particular the poor.

190. The Constitution also contains several provisions aimed at combating endemicpolitical corruption and electoral fraud. The entire electoral process has been revamped -in the year 2000 Senate elections were introduced for the first time in Thai history. Thedemocratization of the election process, while faced with many problems, is an


Pover-ty and Public Policy

encouraging step towards making the state more accountable and responsive to itscitizens.

Box 11: The 9th National Economic and Social Development Plan and Poverty Alleviation

Poverty reduction has been recognized by the Thai Government as a high priority area in the NinthNational Economic and Social Development Plan. The Plan perceives poverty not only in terms of inadequacy ofincome, but also as a structural problem - lack of access to basic services, resources and information, and lack ofparticipation in development and decision-making. The Plan is based on the philosophy of sufficiency economy.This philosophy, in turn, is based on the principle of the middle path - moderation, balanced and rationaldevelopment, awareness of the changing world, and the quality of the people. One of the Plan's main objectives isto alleviate poverty and to improve the potential of and opportunity for Thai people to be self-reliant. The Plan'sgoals are to provide equitable access to education and social services to the poor and the disadvantaged. Amainstay of the Ninth Plan is pro-poor growth - viz., growth of economic sectors that the poor typically rely onand growth that provides employment opportunities and generates income for the poor. The Plan targets are toreduce poverty incidence to less than 10 percent of the population in 2006.

To achieve the above objectives, each of major strategies proposed by the Plan incorporate poverty-reduction measures as an integral element. The reduction of poverty is to be accomplished by three simultaneousmeans:

(i) Empowering civil society to recognize the importance of poverty and to participate in poverty alleviation.In particular, the Plan calls for all sections of society, including politicians, to have a proper understanding aboutpoverty and to have positive attitudes toward the poor.

(ii) Improving the capability of the poor and ensuring social security for the poor by urging formation ofoccupational and/or saving groups among the poor. Through these forums, the poor will be able to learn fromeach other's experiences, to make collective decision, and to establish networks of community welfare. Non-government organizations, local education institutions, and religious institutions will be encouraged to play anactive role in the poverty-reduction process. The ultra poor will be guaranteed basic necessities such as food,health insurance, and a living allowance. The state will also help improve productivity of the rural poor. Inaddition, the local economy will be enhanced, so that the poor can be self-reliant and have more job opportunitieswithin their own localities. For the urban poor, emphasis will be placed upon access to education and healthservices, securing a livelihood by linking community businesses with industrial businesses at various levels, andensuring a healthy living environment.

(iii) Adjusting the public administration system to correct structural impediments that cause poverty bycoordinating both public programs and the budget allocation process, so that both will be more responsive topoverty problems. The Plan sees the state as financing crucial programs, but local authorities as implementingthese programs. The Plan calls for legal reform and amendment of rules and regulations so as to grant the pooropportunities, rights, and equality in social welfare, labor protection, rights to information, and entitlement tocommercialize local wisdom. The Plan calls for taxation to be made more efficient and expanded in such a waythat it promotes social equity, while minimizing impact on the poor.

Source: National Economic and Social Development Board

Decentralization As A Tool For The Poor?

191. The Constitution requires the State to decentralize service delivery and financingto local authorities. Since 1999, 4 different decentralization laws have been enacted,including the Decentralization Act. In 2001, the share of local revenue relative to totalgovernment revenues will increase to 20%, and eventually to 35% of total government



revenues by 2006. In Fiscal Year 2001, central government subsidies were increased inorder to meet the 20 percent target.

192. Currently a plan for decentralization of functions has not been translated into localexpenditure assignment for different government functions. Ideally, decentralizationwould be driven by consideration of which level of government could most efficientlyand effectively deliver and finance public services. Decentralization experience in othercountries has shown that it is more effective to decentralize expenditure responsibilitybefore decentralizing resources. That the focus in Thailand remains on the revenue sideis in part because of the target set by the Constitution on increasing local revenues. Thereis currently no budget or plan for sub-national expenditures.

193. Local governments rely heavily on shared taxes and subsidies from the localGovernment. In Fiscal Year 2001, it is estimated that 43 percent of local revenues andgrants were from shared taxes, and 44 percent from grants leaving only 13 percent beinglocal collected revenue. Heavy reliance on shared taxes and subsidies reduces localaccountability and reduces the predictability of revenue flows. See Box 12 for adescription of the system of subsidies currently in use (Thailand Economic Monitor, July2001).

194. Efforts to decentralize central government functions have been underway in manydeveloping countries since the 1980s. This devolution of responsibility has occurred fora variety of reasons - as a result of increasing political pressure or the desire to delivermore efficient public services. It is argued that decentralization can result in increases inallocative efficiency because lower levels of government have better informationregarding local needs and preferences. Decentralization is also supposed to increaseproductive efficiency because it will result in greater accountability at the local level andincreased participation in the decision-making process by communities (Bahl 1999;PREM Notes 2001). Decentralization, however, if not done properly can also lead toincreased regional disparities and inequities. In fact, decentralization may not have aspositive an impact as desired if the constituents in local communities are not keptinformed or allowed to participate in the decision-making process. Therefore,empowering local communities to have a voice in the decisions that impact their dailylives can lead to more accountable local government and even better provision of publicservices.

195. Can decentralization strategies and mechanisms lead to better outcomes inpoverty reduction? Studies in both Vietnam and Argentina revealed that centralgovernments were more effective than subnational governments in targeting poorer areas(Ravillion, 1998 and Van De Walle 2001). Indeed the evidence has shown that centralgovernments are generally better at redistributing resources to poor provinces, thanprovinces are to poor areas.

196. Thus for decentralization to be an effective mechanism for making stateinstitutions more responsive to the concerns of the poor, civil society will need to beengaged in the local decision-making and monitoring process. Although, the process of



administrative decentralization has begun in Thailand, progress has been slow.Participation of civil society at the level of program and project design has visiblyincreased but their role in public policy formulation still needs strengthening. The latterwill be critical to ensure improved local accountability and transparency.

Box 12: Central Government Subsidies to Local Authorities

General grants: Grants distributed to local governments based on indicators. Indicators such as number ofstudents, elderly, villages, and population, and local revenue, are used as proxies for local expenditure needs andrevenue capacities. The grants are administered by the Department of Local Administration (DOLA) in theMinistry of Interior,

Specific grants: Also administered by DOLA, these grants are to be spent on specific purposes.

Grants with transfer of responsibility: Grants given only to Tambon Administrative Organizations (TAOs). Aninstrument used for FY01 and FY02 to help the transition from central to local service delivery. An amount is setaside for each local government for specific purposes. Before the funds are disbursed, local governments mustsubmit a plan and be able to procure the project. The grants are administered by various agencies.

Sectoral block grants: A grant given to local governments that specifies which sector to spend the money on, butgranting local governments much more autonomy than the two grants mentioned immediately above. Thisinstrument is not yet implemented, but planned to be used from FY03 and beyond.

The subsidy system is changing during the process of decentralization. Before FY01, the Central Governmentrelied on the general and specific transfers. During a transition period, "Grants with Transfer of Responsibility"were introduced to allow the Government to decentralize according to the pace of local capacity building. FromFY03, these grants will be replaced by sectoral block grants.

Source: Thailand Economic Monitor, World Bank Office Bangkok, July 2001.

Empowering Communities

197. The role of civil society in Thai politics and society has witnessed a remarkableevolution. Civil society organizations have developed and expanded in number since themid 1990s, when the new Constitution was being drafted and the 8th Economic andSocial Development Plan was in preparation. Both of those documents empowered civilsociety, and in the few years since the onset of the crisis in 1997, the voice of civilsociety, their overt role in the political process, and their participation in vigorousdevelopment debates have visibly grown and accelerated.

198. Approximately 18,500 non-governmental organizations (NGOs) are currentlyregistered, and many more are operating without legal status. These NGOs are formingthemselves into networks. The trends show an increasing rate of participation in a varietyof community organizations, but the most striking example is the high level of demandfor formation of organizations centered around the improvement of the economicsituation of group members. The exact numbers regarding geographical distribution of



such groups is not readily available, however there is evidence that a considerablepresence exists in the poorer regions of the country (i.e. the Northeast and the North).

199. Civil society organizations are playing an increasingly important role inproviding social protection at the community level. The Thai government as well asdonors have allocated an increasing proportion of their budgets to civil societyorganizations to implement programs including HIV/AIDs prevention and treatment,care of the elderly, small and medium enterprise development, and environmentalprotection. Data regarding the degree to which the poor are involved and participating insuch initiatives is not readily available, however the Social Investment Fund offers oneexample of a program that attempts to target the poor (see Box 14).

200. His Majesty The King's recently espoused theory of "self-sufficiency" at thefamily and community levels has encouraged the government to focus more attentionand resources on building strong communities. Thusin 1997, the Government viewed the economic crisis Box 13: The Remedy for Povertyas an opportunity to forge new developmentpartnerships with civil society and take steps per In my opinion, the remedy for

towards empowering communities to solve their poverty is empowering the poor "

own problems. This approach was aimed at Source: Participant comments at a TDRIrebuilding and consolidating social capital eroded Poverty Workshop, December 22, 2000.

by economic growth and to strengthen theunofficial, community-based "safety net." (see Chapter 4 for the empirical evidence oninformal safety net performance). Much support now goes directly to community-basedorganizations, such as savings and credit groups, women and children groups.

Programs Aimed At Supporting Community Initiatives

201. There were many government programs that focused on communityempowerment activities during the crisis. For example, the Government established theNational Social Policy Committee (NSPC) to address social issues in Thailand. One ofthe key approaches of the NSPC "is to empower communities to respond to the crisisimpact at the community level 'and to create an enabling environment for localcommunities to deal effectively and holistically with the impacts of the crisis on thecommunity" (United Nations 1999).

202. The NSPC set up a Community Empowerment for Response to Crisis Action Plan(CERCAP) which seeks to empower communities to identify and prioritize issues andproblems resulting from the crisis through participatory planning and action. UNDPfinanced this initiative. More than 200 staff of the Rajaphat Institute were trained to becommunity facilitators that would bring community groups and other partners to manageand monitor their own crisis-response initiatives. Moreover, the NSPC provided fundingin the amount of 40 million baht (approximately US$ lmillion) for a pilot project tocreate more opportunities for civil groups at the local level, both formal and non-formal,to participate in local policy formulation.



203. The National Social Policy Committee created the Pattana Thai Foundation tochannel over baht 40 million in government funds to: (i) conduct pilot projects to supportlocal communities to establish civic forums in every province; (ii) set up communitylearning centers for social services and sustainable development; and (iii) planning,monitoring and evaluation.

204. In September 1998, Government, with the assistance of the World Bank, UNDP,and AusAID, launched a new initiative in this area, the Social Investment Fund whichprovides grants to community-based organizations to undertake investments designedand implemented by the community (see Box 14).

205. Under the Thailand-United Nations Collaborative Action Plan (Thai-UNCAP),numerous United Nations agencies, including the UNDP, UNICEF, and the Food andAgriculture Organization have undertaken community-led activities in pilot areas withspecial attention to crisis impacts. Several bilateral donors have grant programs whichfund community-based activities, such as AusAid's Small Activities Scheme, theCanadian International Development Agency's Canada Fund and New Zealand's SmallProjects Fund.

206. All these initiatives have one common ingredient - strengthening the decision-making capacities of local communities; providing them with an opportunity to be heard;listening to their demands and understanding that in Thailand, the people demand anopportunity to help themselves.

Institutional Support

207. A number of actions taken by Government also recognized the need to establishinstitutional mechanisms of support. Providing financial support to communities hasbeen the initial focus, but the enabling environment for community organizations alsomust be strengthened. Below are some of the key initiatives.

208. The Ministry of Labor and Social Welfare has established 19 provincial NGOCoordinating Centers which coordinate activities of NGOs working on social welfareand safety nets in each province. In addition, the Ministry has supported legislation thatwould strengthen the ability of civil society organizations to provide support in the areasof social welfare. There is a draft bill which has passed Cabinet and awaiting revision bythe Council of State.

209. The Urban Community Development Office (UCDO) was established in 1992with an initial grant of 1.25 billion baht (US$ 34 million), as a special revolving fund tosupport urban community development activities and provide low-interest loans tocommunity organizations for emergencies, housing and income generation. Over half ofThailand's 2000 poor urban communities in 50 provinces are now UCDO members,linked together into 103 networks through a broad range of community developmentactivities.



210. A major step was the establishment of the Community Organization DevelopmentInstitute (CODI) which was approved by Cabinet on July 5, 2000. UCDO has beenmerged into CODI. The main objective of CODI is to help strengthen community-basedorganizations through financial and technical support. CODI for the first time helps todefine community based organizations which are legally eligible to receive directfinancial support from the government. A capacity building support arm will beestablished with a more comprehensive database of community based organizations inThailand. CODI will be a key institution bridging the gap between Government and civilsociety organizations - it will be an instrument of inclusion.

Measuring Results and Other Challenges

211. An important and still unfinished element of the empowerment agenda isadequate monitoring of the outcomes and impact of interventions. This is inherentlydifficult, particularly on a national scale. Empowerment investments are strong onprocess, that is the participatory decision making that must be introduced or strengthened,while measures of physical achievement, such as cubic meters of concrete or number ofgraduates or immunization rate, are sometimes seen as secondary objectives. Otherempowerment projects support unique community based activities without an obviouscommon output indicator. Many involve capacity building to make decisions, not thedecisions on investments themselves. Thus, with social capital building (various forms ofgroup formation and strengthening) as the objective, empowerment projects presentchallenges for conventional statistical treatment.

212. The Social Investment Fund and other projects include intensive feedbackmechanisms, transparency enhancing procedures, and objective monitoring of programsby independent consultants. These generate useful reports on how to improve the designand performance of the projects, usually from a community participation andtransparency point of view, but do not focus on maximizing outputs per se. Thechallenge for the future is to develop indicators of success that can be measured readily,aggregated across projects, and reported nationally. More daunting is the issue ofempowerment impact measurement. This will involve calculating the impact ofempowerment outputs on household incomes and expenditure, and on other key welfaremeasures such as health, education, family cohesion, and social gains in crime and drugabuse reduction. Those involved in community capacity building report these gains, andare confident that their interventions are key to the improved impacts in projectcommunities that they observe. Such feedback is a source of strength to the communityempowerment movement. For the future, the case needs to be made more systematically,so as to reach policy makers not yet familiar with empowerment and its advantages.

213. The next item for the empowerment agenda is to develop a more clear strategicview on targeting. Empowerment in Thailand, as practiced at the project level, hasdirected resources at rural and urban communities, usually ones that would be judgedpoor, but also others that are well organized, able to propose quality projects, and linkedto community networks. Such groups are not necessarily from the bottom 20% of theincome distribution, and in some projects, there is an overt attempt to build participation,



voice, or social capital in a broad cross section of Thai society. Like many other socialinvestments, building empowerment in some cases is best undertaken universally,without overt targeting. The SIF in Thailand has targeted poverty on a regional basis (seeBox 14), but more needs to be done however to measure SIF benefits at the householdlevel. This question needs further research, both on the benefit distribution of existingand past empowerment projects, and on the varying impacts of such investments byincome groups. Informal feedback suggests that the poorest benefit the most fromempowerment investments. Such findings will help better define the targeting strategiesappropriate for empowerment.

214. The third agenda item is empowerment and decentralization in Thailand, alreadynoted in the chapter. In this early phase of public support for empowerment, resourceshave flowed outside official bodies to informal and non governmental community groups.Such groups have begun to work on local planning activities, but the absence of stronglocal government bodies has not allowed such efforts to develop popular voice in localgovernment decision making. As decentralization efforts gather pace in Thailand, theempowerment agenda will need to shift attention to assuring popular voice in thedeliberations of official civic for a that have control over the allocation of publicresources.



BOX 14: THAILAND SOCIAL INVESTMENT FUND: SECURITY THROUGH EMPOWERMENT

The Social Investment Fund (SIF), launched in September 1998, marked the first time the Royal Thai Government channeledsignificant sums of money directly to civil society organizations. The general objective was to transform the crisis into an opportunityby supporting highly desirable reforms toward better governance, community empowerment, and the formation of broad developmentpartnerships among key elements of civil society.

Since the launch of its operations, SIF has utilized a Targeting Map to help assist in allocating funds across every region in thecountry. Each region was allocated a budget based on the following criteria: population, unemployment, seasonal unemployment andpoverty, with the respective weights of 25%, 25%, 10% and 40%. Based on this criteria, the allocation to the Northeast, is highest atapproximately 48.% of the total available budget.

During the early implementation of SIF, recognizing that certain groups had "slipped through the cracks," the Social FundOffice developed a new window of support in the early months of 1999 to finance social assistance to the needy. This new window orMenu 5 was designed to respond to the urgent needs of the poor and to increasing demands on organizations struggling to assist them,particularly those groups on the margins who cannot access regular services (i.e., hill tribe communities; unregistered families;communities in remote areas). The new window finances social assistance transfers to the most vulnerable through communityorganization networks. This new window of support has not only allowed the Social Fund to reach the vulnerable more quickly, but ithas also supported the strengthening of community networks and thus the core objective of SIF to build up the stock of social capital.

The funding support provided from this new menu is channeled directly to the beneficiaries through community organizationnetworks. Community organization networks are grouped according to geographical location, joint developmental activities, anddevelopment issues of local interest such as networks of small-scale fishing associations, or around social causes such as AIDSNetworks or networks of community environmental protection groups. Community organization networks are responsible for thecompilation of information gathered at the community organization level into subproject proposal packages, budget management,subproject implementation and the production of detailed reports of member organizations. The Social Fund enters into a contractualarrangement with these networks to implement Menu 5 sub-projects. The networks are then responsible for disbursing the grants to theintended beneficiaries within the communities. In practice, it has been found that the majority of the communities have decided to turnthese grants into a revolving fund in order to increase the number of potential beneficiaries over time. These community based fundssupport student fees and materials and start up cost for income generating schemes, while the support for indigent families and AIDSpatients often is in the form of an outright cash transfer.

The Menu 5 beneficiaries include (i) unemployed homeward migrants; (ii) the indigent; (iii) poor farmers; (iv) homelesschildren, disadvantaged children, neglected children, and school drop-outs; (v) battered women and women impacted by the socialcrisis; (vi) neglected AlDs/HIV patients; (vii) neglected senior citizens; (viii) the handicapped that need care providers; and (ix) patientswith chronic illness. Responding rapidly, it took the Social Investment Fund only three months to design Menu 5 procedures. In all, 457sub-projects worth about $50 million were approved between April 1999 and July 2000 and approximately $40.3 million has beendisbursed to date. More than 300 community organization networks have participated in Menu 5 activities, with approximately 538,414individuals benefiting from Menu 5 support.

The Social Fund Office has managed to mobilize a large number of volunteers all over the country, including individuals fromthe NGO community, Government civil servants, academics, physicians, teachers, and religious leaders to help prepare, review, screen,and help implement projects. The commitment of so many individuals from a wide range of backgrounds has been quite astonishing.The SIF has managed to attract close to 2,000 registered volunteers as well as hundreds of others who have not registered. All of themput in long hours without receiving compensation for their invaluable support. The SIF is widely recognized in Thailand as a majorelement of the Government's crisis alleviation strategy and a pathway toward government collaboration with CSOs.

In Thailand, there are altematives to Government for delivering social protection (security) programs to the poor. Civil societyorganizations are substantial and growing, and during the depths of the crisis they stepped forward with the SIF to target assistancewhere it was most needed. The costs of managing these efforts were contained by the large volunteer response. The success of thistemporary program contains lessons for the future management of Government social assistance in particular the benefits of workingwith civil society organizations in the communities to manage and deliver social assistance to those in need.



Chapter 6. Protecting and Investing in the Poor

215. With the rapid growth of the Thai economy during the 1980s and up until themid-1990s, poverty fell dramatically, and the government had little interest in focusingmuch attention on poverty-reduction programs. A limited number of programs were in-troduced both to protect Thai families from sudden shocks to their livelihoods and to in-vest in the capacities and assets of the poor. Taken together these programs, aimed atprotecting and investing in the poor, did not account for a large portion of governmentexpenditure.

216. In the domain of social protection, Thailand has no permanent income mainte-nance program, but does have various forms of social insurance. A compulsory socialsecurity system was introduced less than ten years ago when the government passed a lawrequiring large establishments (with 20 or more workers) to participate in the state-runsocial security system. The system, which covered maternity benefits, illness and deathbenefits in and out of the work place, and limited disability benefits, was expanded in theaftermath of the economic crisis (in late 1998) to include pension and child allowanceschemes for private-sector employees. Public employees are covered by the Civil Ser-vants' Pension Scheme and the Civil Servants' Medical Benefit Scheme. Many large pri-vate firms and establishments have also operated their own private pension systems.However, the number of Thai workers who are covered by these public and private socialsecurity systems is very small,3 6 which leaves the government-run safety-net programs asthe major source of official (as opposed to family and community) assistance to the thepoor and the indigent.

217. The government safety net system in Thailand can be divided into three catego-ries:3 7 cash transfer programs, in-kind transfer programs, and income-generation pro-grams. The Department of Public Welfare in the Ministry of Labor and Social Welfareoperates several cash assistance programs, such as the direct cash assistance to needyfamilies, the monthly cash assistance to the elderly without any means of support, andvillage community funds administered by village welfare committees to help poor villageresidents. The in-kind transfer programs include the School Lunch Program, adminis-tered by the Ministry of Education, and the Free Medical Service for Low-Income Fami-lies program, operated by the Ministry of Public Health. Finally, there are wage-employment creation programs that are open to the population at large within a givenarea, such as the Rural Job Creation program, The Green Earn Program, the Tambon De-velopment Program, and an array of job creation programs expanded during the crisis.The wage-employment schemes essentially finance rural work projects in the rural areas,such as construction and repair of rural roads, weirs, school buildings, and irrigation ca-

36For instance, the Social Monitor IV noted that only 15% of Thai workers (formal and informal) reportedbeing covered under the social security system, with the coverage strongly biased toward high-wage work-ers who are employed by large, formal-sector establishments. Indeed, as rules currently stand, only work-ers who are paid more than the minimum wage are eligible for social security coverage; so the lowest-paidworkers are automatically excluded from program participation (World Bank 2000).37This section draws from Krongkaew (1999).



nals. The objective of these projects is to provide short-term employment relief to thepoor.

218. Social safety net and social insurance schemes can be considered the mainstay ofthe government's protection programs. Programs to enhance opportunities for the poorinclude investments in health and education as well as various growth-oriented policies,such as fiscal and trade policies, that have over time yielded the high growth rates anddramatic reductions in poverty seen in Thailand over the past 30 years (see chapter 4).

219. At the level of operational projects and programs, the government has financedmany programs investing in the poor, most notably education and training programs thatare offered by the Ministry of Education, the Skills Development Department and SocialWelfare Department of the Ministry of Labor and Social Welfare, the Bangkok Munici-pal Administration and others. Although education programs are universal and not tar-geted to the poor, there has been a steady expansion of school enrollment and completionrates among the poor in Thailand. Micro-credit or small farmer credit programs constituteanother important anti-poverty intervention. These have been financed by the Bank forAgriculture and Agriculture Cooperatives, the Urban Community Development Organi-zation, the Rural Development Program (now under the Community Organization Devel-opment Institute), and, more recently, the Peoples' Loan program launched by the Gov-ernment Savings Bank. The Community Development Department of the Ministry ofInterior supports the Poverty Alleviation Program, also designed to make credit availableto the poor. The Ministry of Agriculture and Agriculture Cooperatives sponsors extensionand technical support programs targeted to poor farmers living in low-productivity areas.

220. Annex A contains a listing of major government programs that can be consideredtargeted to the poor, with the total expenditure on each of these programs during 1993-99.Most of these are targeted by design and objective, but others are included that success-fully target the poor even though that is not their stated objective. The following sectionanalyzes the public expenditure patterns on these programs, after which the chapter fo-cuses on the performance of several of the larger protection and investment programs onwhich data are available from the SES.

Government Expenditure on Anti-Poverty Programs

221. An important concern is the size of the government's anti-poverty effort, asmeasured by the total amount spent on public assistance programs in relation to variouseconomic indicators. 38 As Figure 43 shows, aggregate government expenditure on pro-grams primarily directed to the poor has increased significantly in real terms over time:

3"Total government expenditure on poverty programs is not routinely compiled or reported by governmentdocuments. The data reported here were compiled by the World Bank Office in Thailand from governmentexpenditure documents and the Bureau of the Budget. For the purposes of this calculation, poverty spend-ing includes expenditure on public programs that are primarily intended for the poor or which unintention-ally (and disproportionately) reach the poor (in the opinion of concerned Ministry officials). A total of 52such programs across all Ministries and Departments were selected as poverty programs. However, publicspending on universal programs such as basic education and primary health care (that may disproportion-ately benefit the poor) is not included in this measure of anti-poverty spending.



from Baht 4.5 billion in 1993 to Baht 21.1 billion in 1999 (in constant 1988 baht). Sincethe number of poor has generally fallen over this period, public spending on poverty pro-grams per poor person has also increased significantly - from a mere Baht 386 in 1993 toBaht 2,138 in 1999, implying an impressive annual growth rate of 33% over the six years(Figure 43).39 Since real GDP grew by 2.4% - and real total government expenditure at8.1% - annually over this period, public spending on poverty programs increased as aproportion of GDP and as a proportion of total government expenditure as well.

222. At the same time, it is Figure 43important to note that these Government expenditure on anti-poverty programs, 1993-99

- ____ - --_ _ _ _ _ _ _ _ -~~~~~ 2,546 10

large increases came about from 2,500- Real gov't exp on anti-poverty 2249 2 46

a very small base, and that, de- programs per poor person 138 9spite the growth in public 2,000 -As % of total govt expenditure - 8

spending on poverty alleviation, As%ofGDP 7overall public spending on pov- 1500 -,462 6

erty interventions was a mere 5

0.7% of GDP in 1999 (Figure ,000 97 40 4.2

43). Additionally, the data show 699 2.7 3

that, while government real ex- 210 l 2

penditure on poverty alleviation 386 a - 07 0.7

increased by 5.2% between 1.1 0.3 0.3 0.3 0.5 7 1

1998 and 1999, the population 0° - T --- 0of the poor swelled by 25.3% 1993 1994 1995 1996 1997 1998 1999

over the same period, resultingin a real decline of 16% in public spending per capita on poverty.

223. While 0.7% of GDP may seem like a small number, it is large in relation to theincome share of the poor in total national income. The 1999 SES indicates that the poor(who accounted for 16% of the national population) enjoyed a share of only 2.8% of ag-gregate national income. Thus, if all of the public spending on poverty programs were toreach the poor (and only the poor), it would represent about one-quarter of their income -not an inconsequential sum.

224. Needless to say, the quality of public spending on poverty is as important as thequantity of public spending. There is a wide perception in Thailand that a large portionof the government's expenditure on poverty programs is wasted owing to poor targeting,selection of the wrong types of poverty interventions, and inefficiency in execution ofpoverty-alleviation projects. Further, insufficient attention has been devoted tomonitoring and evaluating the effectiveness of the various government poverty programsin achieving their objectives, with the result that little is known about how much of thegovernment's spending on poverty reduction actually reaches the poor and how much, ifat all, it reduces poverty.

39Note that real government poverty expenditure per poorperson declined between 1998 and 1999 due to alarge increase in the number of poor persons - not because of a decline in aggregate public expenditure onanti -poverty programs.



225. The ultimate test of Total government expenditure and average household

success of an anti-poverty living standards across provinces, 2000

program is the extent to 10.0which it improves incomes of the poor and reduces -

9.6~poverty. Unfortunately, at , 9 *

9.4- this time, there are no 9

household survey data that v92can be used to evaluate the 9.0g *. .

Spoverty or income impact 8.8

of the government's spend- 0 8.6 *

ing on anti-poverty pro- - _

grams. However, in 1999, 9.8 10.0 10.2 10.4 10.6 10.8 11.0 11.2 11.4 11.6

the National Statistics Of- Log real household expenditure per capita (Baht)

fice added a special 'crisis' Figure 44module to the SocioeconomicSurvey that included a number of questions on any benefits received by the samplehouseholds from selected government welfare programs.4 0 These data offer one of thevery first opportunities in Thailand to examine the actual coverage of the poor by selectedanti-poverty programs as well as the benefit incidence of these programs. While theseissues are analyzed and discussed in the following sections, it is important to note at theoutset that coverage and targeting are merely two of many indicators of program impact.There are many other indicators - such as the volume and quality of actual benefits paidout by these programs to the poor and, more importantly, the extent to which these pro-grams affect income, health, nutritional and schooling outcomes of the poor - that cannotbe addressed in this Monitor with currently-available data. Hopefully, future surveys willcollect data that will permit analysis of these aspects of program impact.

226. Before analyzing the various targeted anti-poverty interventions, it may be usefulto examine the extent to which the inter-provincial allocation of overall public spendingresponds to provincial differences in living standards. Using data prepared by theNESDB Macro and Social Division, Figure 44 plots total government spending per capitain a province against average living standards per capita in that province. It shows thatbetter-off provinces tend to receive significantly larger allocations of per capita publicspending than poorer provinces. While this is not an unusual finding (and reflects the factthat local governments in better-off provinces are able to raise more revenues and thus

40The SES has been normally undertaken every two years since 1986, with data collection being spreadover a period of twelve months. The 1999 survey, done between the regular 1998 and 2000 surveys, was aspecial survey conducted only for a period of four months (June to September 1999) to assess the impact ofthe economic crisis on Thai households. The sample size for the SES 1999 was 7,789 households. A spe-cial module on who benefits from the most important government programs was included in the SES 1999.The respondent household was asked if it currently (or in the last year) received benefits from several gov-ernment programs, including the low-income medical card scheme, the health insurance card scheme, thefree school lunch program, the free supplementary food program, government scholarship and loans pro-grams, the social security program, and the social pension program for the elderly. For programs that aretargeted to individuals in the household (e.g., social pension for the elderly and educational scholarships),the identity of the respondent was also recorded.



finance their higher expenditures), it does suggest that there is substantial scope for cen-tral government spending in Thailand to be used more strategically toward poverty-reduction goals. The goal of public spending, especially by the central government,should be to compensate poorer provincial governments for their lower capacity to raiserevenues and finance public and social services.

Programs that Protect the Poor

227. In the following two sections, a number of poverty and other social program areanalyzed in terms of their uptake or utilization by the poor vs. other income groups. Us-ing the SES data base for 1999, the analysis allows an estimation of coverage of the pro-gram, nationally and by income group, and the relative share of the poor vs. other incomegroups in total use of the program services. Future studies with more complete data canextend this analysis to a full public expenditure incidence analysis of key social pro-grams.

228. The discussion in this chapter is not meant to suggest that universal (i.e., non-targeted) public programs and public policies that do not have the specific goal of reduc-ing poverty or helping the poor are of no relevance to the poor. On the contrary, someuniversal programs, such as provision of basic education, primary health care, rural roads,and communicable disease control, can have large impacts on poverty and the poor. Inaddition, fiscal and trade policies can have significant direct and indirect (via economicgrowth) effects on poverty and inequality. Policies that promote deconcentration of in-dustry - e.g., rural industrialization - can also significantly reduce income inequality andpoverty. However, it is beyond the scope of this report to undertake a detailed assessmentof such programs and policies.

229. Health Programs. Most Thai families participate in one of the many free orpartly subsidized public health coverage schemes, ranging from free health services to theuninsured to various insurance schemes that require some beneficiary co-payment orpremium. Four main schemes - the public assistance scheme, the civil servants' medicalbenefit scheme, the compulsory health insurance scheme for formal sector employees,and the voluntary health insurance program for the near-poor - receive public support.Administrative statistics indicate that, in 1998, 45 percent of the population was eligiblefor free medical care under the public assistance scheme (which includes 13.5 percent ofthe population covered by the low-income health card scheme), 11 percent under theHealth Welfare for State Employees' scheme (including the Civil Service Medical Bene-fit Scheme), 8.5 percent under the compulsory health insurance programs (viz., social se-curity and workmen compensation fund), and 16 percent under voluntary health insur-ance programs (viz., the voluntary health card and private insurance).

230. Low-Income Health Card. The public assistance scheme for health care coverslow-income adults, children under 12 years of age, the elderly (60 years and older), dis-abled, monks and war veterans. The low-income card program is a part of the public as-sistance scheme and is targeted to poor adults within the 13-59 years age group, with abudget of Baht 273 per capita. The card, which entitles poor adults to free health care at



hospitals and public health centers, is issued for a period of three years, and the incomeceiling for eligibility is Baht 2,000 per month for a single person and Baht 2,800 permonth for a family. Thecommunity leaders (e.g., Coverage and incidence of government health programs, by real per

the village head) identify capita consumption quintiles, 1999

who is poor, thus deterrmin- 60 1 * Poorestquintile USecondquintile [lThirdquintile C Fourthquintile *Richestquintile

ing eligibility for the card. 50 48

During the economic crisis, 3 843

the Government expanded 40 38 34

the coverage of this low- 30 27 27 25

income card to the unem- 20 19 18

ployed, provided they had h R12registered themselves as 10 4 1

unemployed with the Min- _ -istry of Labor and Social % of population % share among all % of population %shareamongall

receiving benefit beneficiaries of receiving benefit beneficiaries of

Welfare. l l program program

Low-Income Health Card Voluntaty Health Card

231. According to theSES 1999 data, only about 10 Figure45percent of the population reported having the low-income health card - somewhat lowerthan the coverage of 13.5 percent indicated by the administrative data. Households withthe low-income card had 2.7 adults in the 13-59 years age group. Just under a fifth of thepopulation in the bottom quintile, based on real consumption expenditures, had a low-income health card. Nominal per capita monthly income of recipient households wasBaht 1,368 - significantly less than one-half that of non-recipient households. The distri-bution of the low-income health card is progressive, with the poorest consumption quin-tile accounting for 38 percent of allbeneficiaries and the top quintile Box 15: Impact of the Crisis: Vulnerability of the Pooraccounting for only 3 percent (Fig- "...The crisis has happened so quickly it has left us con-ure 45). fused, puzzled and let down. We have lost our jobs but are

given no explanation," says a community leader from the232. The regional distribution of slums of Khon Kaen, Thailand. "It was the rich who bene-cards, as based on the household fited from the boom but we, the poor, pay the price of the

cards, as based on the household crisis." Within her community there is a feeling of uncer-data, appears biased toward the tainty, insecurity, and isolation. "Even our limited accesspoorest region - the Northeast - to schools and health is now beginning to disappear. Wewith over one-half of all recipients fear for our children's future," adds her husband. Poorbased there. Virtually no cards families report having to pull their children out of schoolwere reported in Bangkok, the to work and a mother says, "What is the justice in sendingour children to the garbage site every day to support therichest region. family?"- Thailand 1998

233. While the low-income Source: Narayan, Deepa, with Raj Patel, Kai Schaffl, Annehealth card program is progressive Rademacher, and Sarah Koch-Schulte, 2000, Voices of thein that the poorest quintile's share Poor: Can Anyone Hear Us? New York: Oxford Univer-

of beneficiaries is nearly two times sity Press.its population share, the fact that



the outlay on the program is small makes even better targeting to the poor necessary.41As it now stands, more than a third of the beneficiaries of the low-income card programare drawn from the top three quintiles. This effectively means that coverage for the poor-est quintile is low - less than 20 percent.

234. More important than the mere possession of the low-income health card is the ex-tent to which card-holders are actually able to obtain quality health services, whenneeded, without any payment. Unfortunately, the survey contains no data on householdutilization of health services or the payment made for health services, so it is not possibleto address these issues. Indeed, as noted below, there is a hint in the data that the benefitpackage associated with the low-income health card might be perceived as being inade-quate by the poor.

235. Voluntary Health Card. As noted earlier, there is also a voluntary health insur-ance card scheme in Thailand, under which a cardholder is entitled to free medical carefor an annual contribution of Baht 500. This program aims to serve the near-poor andthose in the population who do not have mandatory insurance, such as civil servants andformal-sector employees.4 2

236. Demand for the volun- Lorenz curves for beneficiaries of government health programs, 1999

tary health insurance card in- 100IT--

creased sharply during the cri- gosis period, even more so than 80

that for the low-income health 70

card (World Bank 1999). The 60

distribution of the voluntary 50/

health card is significantly less - roportionaltopopulation shae

progressive than that of the -Low-incomecard

low-income health card (Figure 20 -Voluntaryhealthcard

46). This is to be expected, as t0

the voluntary health insurance _ i

card is available, by design, to 0 t0 20 30 40 50 60 70 80 90 100

a wider population. Cumulative % of population, ranked by per capita consumption expenditure

Figure 46237. What is puzzling is that, at least according to the SES 1999 data, a significantlylarger proportion of the poorest quintile is covered by the 500-baht voluntary health cardthan by the free low-income health card (50% versus 19%). If correct, these numberswould suggest that the poor do not take advantage of the low-income card despite beingentitled to it, perhaps because they perceive the benefits package associated with the low-income health card to be much inferior to that obtained from the voluntary health card.Another explanation is that the per capita budget subsidy plus the co-payment to publichospitals for the volun'tary scheme exceeds the per capita hospital revenue of the low in-

4lFine targeting can, of course, entail higher administrative costs as well as undermine political support forthe intervention. In such cases, it could fail to achieve better poverty outcomes.42Civil servants are covered by the Civil Servants' Medical Benefit Scheme, while formal-sector employeesare covered by private insurance, social security, and workmen compensation fund.


Poverty and Public Policv

come scheme, thus presenting hospital management with an incentive to increase volun-tary card uptake. However, without knowing the actual utilization of health services bythe two types of cardholders, it is difficult to address these issues.

238. Finally, the data indicate that nearly a third of the poorest quintile does not reporthaving either the low-income or the voluntary health card. Since this group is unlikely tohave other forms of medical coverage, such as civil servants' medical benefits or privatehealth insurance, it would imply that a significant number of the Thai poor are unpro-tected from, and therefore vulnerable to, the adverse economic effects of catastrophic ill-ness.

239. The newly-elected Government has introduced a new health coverage scheme -the '30 Baht Health Scheme' - to provide health services universally at 30 baht per visit.This has been piloted in 21 provinces, and has recently been expanded to all provinces.Anecdotal reports indicate that the other insurance schemes are remaining in force, atleast for the present, while the 30 Baht Scheme "gold cards" are being distributed to thepreviously-uninsured. The aim of the 30 Baht Scheme is to absorb eventually other healthinsurance programs into a single program with consistent benefits and patient costs. Thefinancial implications, for Figure 47both the budget of the Min- Coverage and incidence of social security and social pension programs,both the budget of the Min- ~~~~by real per capita consumption quintiles, 1999

istry of Public Health and OPoorestquintile ESecondquintile [lThirdquintile fFourthquintile *Richestquintile

the personal health 40 3936

expenditures of Thai 35 3

beneficiaries, could be 30 - 30 28

substantial, and will need 25 - 21 22 2 21

careful monitoring and 20 r 17

analysis. IS 13 13 1 -

240. Programs to Protect '° F 5

the Elderly. Thailand has a 5- | . _ 63small program meant to % of elderly population % share among all % of population % share among all

assist the elderly poor. The (60+years)receiving beneficiariesofprograin receiving benefit 'beneficiaries ofprogramassist the elderly poor. The ~~~~~benefitDepartment of Public Wel- Social Pension for the Elderly Social Secunty Card

fare (DPW) of the Ministryof Labor and Social Welfare allocates a subsistence allowance to the elderly poor (thoseaged 60 and over) through its network of Village Welfare Assistance Centers (VWACs).In 1997, geographic coverage of the social pension program reached 56,538 villages, andby 1998 this had increased to 66,409 villages. The program is targeted to those over 60years of age who have no one to take care of them and who are economically inactive.These elderly poor must live in a village having a VWAC.

241. The program pays each elderly poor recipient a pension of Baht 300 per month,which represents an increase of Baht 100 during the crisis. The value of this income sup-port is also considerably less than the minimum subsistence requirements implied by theNESDB poverty line for the elderly, which averaged around Baht 700 per person in therural areas in 1997. The cash transfer expenditure (excluding administrative costs) on



this program during 1998 remained modest at Baht 763 million. The coverage in 1998was 318,000 elderly poor.

242. The SES 1999 data show that the program is progressive with nearly 40% ofbeneficiaries coming from the poorest quintile and only 5 percent from the top quintile(Figure 47). The main problem is that the program has low coverage, with only 8.1 per-cent of the population over the age of 60 years - and 13% of the elderly in the poorestquintile - being covered.43

243. A comparison of the characteristics of recipients to non-recipients among thesample of individuals aged 60 years and older indicates that the recipients of the programare more likely to be female, have lost their spouse, report their main occupation as beingretired, and their work status as economically inactive. On average, they are 73 years ofage. Among the recipients, 40 percent live in the Northeast and just another third in theCentral region (which is the second richest region after Bangkok). Recipients, on aver-age, received just under Baht 2,300 annually - which tallies quite closely with the stipu-lated program benefits. The average per capita real income of recipients is about Baht1,700 per month, in comparison to a real income per capita of about Baht 3,100 amongnon-recipients of the program.

Incidence of income-poverty among individuals residing in244. It is not immediately households headed by a person aged 60 years or older, by household

clear why the program targets 25- size,l998andl999 23only those over 60 years of age 20

who live by themselves (i.e., 20 1 01998 a 999

who have no one to take care of 16

them). Indeed, data from the ;SES 1998 and 1999 indicate , that the incidence of income- 10;

poverty among the elderly wholive by themselves is almost 5LInegligible (Figure 48). How- 2 3 4 5 6

ever, the incidence of poverty I 2 3 4 s rmore

among the elderly rises sharply Household size (members)

with their household size. This Figure 48suggests that the appropriate targetgroup for the program should be the elderly who live in large househoids, possibly con-taining many dependents, including (presumably) someone who can take care of the eld-erly poor person.

4 3 0f course, since the program targets only those elderly living by themselves . with anothe; elderly per-son, the intended target group is not all the elderly. It is not possible to calculate accurately with the SESthe actual size of the intended beneficiaries of this prograrr..



245. As noted earlier, the Figure 49social security card is avail-

Lorenz curves for beneficiaries of the social security and social pensionable to those working in the for the elderly poor programs, 1999

formal sector. Because - - -Proportional to population share -,

those working in es- - -Social pension for the elderly poor - .

tablishments that provide m0 _Social security - ,

social security are typically , 70 - ,

better-off than informal- a 60

sector employees or the 0 5

self-employed, the program = 40 /

ends up disproportionately E 30- /benefiting the affluent 20 -

(Figure 49). In 1999, 36% 10 /of all SSA beneficiaries _______

were drawn from the toj 0 10 20 30 40 50 60 70 80 90 10044 Cumulative % of population, ranked by per capita consumption expenditi

consumption quintile.The SES data indicate that, among households that report having the card, the householdhead either works as an employee in the private or public sector (53 percent) or is retired(25 percent). A third of the recipients live in Bangkok and another 36 percent in the Cen-tral Region.

246. It should be noted that the Thai government plans to expand the social securityprogram later this year to cover employees of companies with 10 or fewer employees.The expansion would add an additional 3.6 million workers to the program, boostingcoverage to 9.3 million workers. Workers set to benefit from the program's expansioninclude those working in garages, restaurants, beauty parlors, and all independent ven-tures. By some accounts, domestic helpers will also be covered by the social securityprogram. These moves will obviously go some way toward bringing low-wage employ-ees into the folds of the social security program.

247. Public Employment Programs. The wage-employment creation schemes werestepped up considerably in the aftermath of the economic crisis, and came to be known asthe Miyazawa fiscal stimulus package, which was put into operation in April 1999. Theprogram generated short-term employment for low-skill workers in the rural areas in re-habilitation of water supply, irrigation, and other public works. Some employment forskilled workers, mainly in computer-related fields, was also supported. In total, the Mi-yazawa package provided employment to 88,967 higher skill employees and to 3.5 mil-lion unskilled workers. These 3.5 million workers were employed an average of 18 days,which is equivalent to 319,182 person years. The public works program was meant topay wage rates designed to induce self-selection of the poor into the program. The objec-tive was to spend the money quickly to stimulate the economy through job creation and

44It should be noted, however, that this is not a failing of the social security program per se. The program isrelatively new in Thailand, and the history of social security programs in more developed countries sug-gests that initial coverage rates are typically low and are biased toward formal-sector employees who tendto be better off than informal-sector and own-account employees. However, over time, population cover-age of these programs expands and reaches into low-income and informal-sector groups.



productive investments, while at the same time cushioning the poor from the crisis andcreating a foundation for future competitiveness.

248. In addition to the Miyazawa package, the Social Investment Project (SIP), co-financed by the Thai government, the World Bank, JBIC-Japan, UNDP and AusAID, wasalso put in place in late 1998. One component of the SIP involved supporting existinggovernment programs aimed at providing jobs to the unemployed and the poor. Most ofthese programs involved creating jobs through the implementation of small-scale civilworks (weirs, foreground dredging, rural roads and school renovation). Additional publicemployment creation activities were also financed from the government's own fiscalbudget as well as from Asian Development Bank loans. In terms of scale, the Miyazawapackage was by far the largest of the job creation interventions.

249. The special crisis mod-ule of the SES 1999 did obtain

Coverage and incidence of public employment generation schemes,some infoirmation about house- by real per capita consumption quintiles, 1999

hold participation in any of l:Poorestquintile *Secondquintile [lThirdquintile 12Fourthquintile *Richestquintile

these public employment 42

schemes. Unfortunately, the 40 -

survey only asked if any mem- 35

ber of the household worked on 3025

a government- sponsored em- 20 1

ployment program since 1 April 15 . -

1999, and if so, how many peo- 10 ___

ple in the household worked. 5 2

Unfortunately, the survey did 0 _ _ _ _not ask who in the household % ofpop receivingbenefit T % shareamongall beneficiaries ofprogram

worked, what wages were paid, Public employment schemes

and how many days the indi- Figure 50vidual worked on the program.

250. Data from the crisis module of the SES 1999 indicate that 1.1 million householdsparticipated in the various public employment schemes. More than one member from thesame household may have participated in the schemes, and therefore this number may becloser to the administrative data on total wage employment created by the Miyazawa,SIP, and routine government programs. Another reason for the lower estimate is that im-plementation of these public works programs continued well into calendar year 2000,which was well after the 1999 survey was completed.

251. More importantly, the SES 1999 is the only source of data on the socioeconomiccharacteristics of persons participating in these wage employment programs. The datashow that, among those in the poorest consumption quintile, just under a fifth of thehouseholds participated in this program (Figure 50). The participation rate declined to 8percent for the next quintile and to just under 2 percent in the richest quintile. Looking atthe distribution of beneficiaries, the program was well-targeted to the poor, with 42



Figure 51percent of the beneficiaries Lorenz curves for beneficiaries of public employment schemes, 1999

being drawn from the poorestquintile and only 5 percent . 7100quintile and only5percet 90 -Proportional to population share

from the richest quintile (Fig- I ures 50 and 51). 80 -Publicemployment

.= 70d//

252. The level of targeting , 60

achieved by this program is 5

admirable, especially given > 401

that the targeting was E 30 /

achieved with minimal ad- 20 /

ministrative apparatus (i.e., 10through the payment of wage o

rates tat weredesigne to 0 0 20 30 4 0 60 70 80 90 100rate s that were designed to lCumulative % of population, -anked by per capita consumption expenditure

induce self-selection of thepoor into the program). This suggests that there may be considerable scope for usingpublic employment programs as a mainstay of an anti-poverty program during futuredownturns in the economy.

253. One problem often cited with public employment programs is that they need con-siderable start-up time to be put into place. However, once established, the employmentand income benefits of these programs can be immediate. This was the case during theThai crisis of 1997-98, with the public employment program interventions coming ratherlate in the crisis. For example, the Miyazawa package began to be implemented only inApril 1999 - more than a year and a half after the onset of the crisis and nearly a year af-ter the severest effects of the crisis were felt by the labor markets. But soon after its ini-tiation, it had already generated 63 million person days of work. This suggests that itmay be useful to have public employment schemes as a permanent feature of a poverty-alleviation and safety net program. The schemes could be designed so that they could beexpanded rapidly during times of economic crises and scaled back during times of strongeconomic growth. This would generally make for a swifter government response to eco-nomic downturns.

Programs Investing in the Capacities of the Poor

254. In addition to pro- Box 16: Disparity in Education Budget Allocationsgrams offering security, anumber of poverty programs ". .Though the state provides free education for everyone, whenin Thailand are intended to observing disparities of budget appropriated to schools in thecreate opportunities for the central and provinces (per head of students) - we find a reversion

of the Robin Hood concept in public policy. Schools in the cen-poor invasetsti inTheseirncapac tral enjoy handsome financial injections..."ties and assets. These includeeducational assistance pro- Comment of a participant during a TDRI Poverty Workshop,grams, various credit December 22, 2000.

schemes, and general publicpolicies to increase economic growth and redistribute incomes.



Coverage and incidence of government school nutrition programs,255. School Feeding by real per capita consumption quintiles, 1999

Programs. Thailand has c Poorest quintile * Second quintile E Third quintile 1 Fourth quintile * Richest quintile

two main school feeding 60- 54 592

programs - a free school 50 4 45

lunch program and a sup-plementary food (school 40 35

milk) program, both of 301 22 *

which are administered by 20.

the Ministry of Education. 1 8 10

Given their long term ob- Ljectives of encouraging en- o '--_ jectives of encouraging en- 5'%.of chitd-eged 6-ti1 % share among oil % of hitdren aged 3-8 % shame among all

rollment of poor students, years receiving benefit b heneficiaries of program years receiving benefit beneficiaries ofprogram

these programs are pre- School Lunch Program Supplementary Food Program

sented here under the "in-vestment" category, even Figure 52though they might also be categorized as safety net "protection" programs for the poor.The school lunch program applies to children in primary school, while the supplementaryfood program (largely a free milk program) is open to pre-primary students and primarystudents through the third grade. The original idea behind both programs was to assistundernourished children (as measured by weight-for-age) in village primary and pre-primary schools in obtaining adequate food. But, over time, the school lunch programwas expanded to cover all primary- and pre-primary school children from poor families,whether undernourished or not. The selection process of the target population in theseprograms is more arbitrary than that in the low-income health card program, as the teach-ers in each school decide which families are poor and which children are eligible to jointhe programs. For the school lunch program, the cost of each meal is pegged at aboutBaht 5, which is rather low for a nutritious lunch.

256. The SES 1999 data indicate that 29% of all primary school-aged children (aged 6to 11 years) benefit from the school lunch program, while 47% of children aged 3-8 yearsbenefit from the supplementary food (milk) program. Despite the arbitrary nature of thebeneficiary selection process, both programs are highly progressive, with as many as 45%of children in the bottom consumption quintile receiving free school lunch benefits andthe poorest quintile accounting for 54% of all beneficiaries of the free school lunch pro-gram (Figure 52). The supplementary food program is also progressive, although less sothan the school lunch program. The share of the poorest quintile among beneficiaries ofthis program was 45 percent in 1999.

257. Student Loan and Government Scholarship Programs. An Education Loan Fundfor the student loan scheme was established in 1996 prior to the crisis. The eligible bene-ficiaries include those studying at upper secondary and tertiary private and public educa-tional institutions with incomes below Baht 12,500 per month (World Bank, 1999). Ad-ministrative data show that the number of recipients increased more than four-fold be-tween 1996 and 1999 - from 148,444 to 675,614 beneficiaries (World Bank 1999).



258. The SES 1999 data corroborate that the loan program is largely for those currentlyin upper secondary school (26 percent), university (17 percent), and vocational and tech-nical education (50 percent). The average age of the recipients was 19 years, and 57 per-cent were females. The data indicate that there were over 300,000 beneficiaries of thestudent loan scheme in 1999. While this number may appear large, it represents a cover-age rate of merely 4.7% of upper- and post-secondary students. Given its target group,the student loan program is not oriented toward the poor, and the beneficiaries are muchmore likely to be from the higher quintiles (Figure 53).

Coverage and incidence of government educational assistance programs,259. The scholarship by real per capita consumption quintiles, 1999

program is less regressive 30 0 Poorest quintile * Second quintile [lThird quintile OFourthquintile * Richest quintile

than the student loan pro- 25 25 2524

gram, with 25% of all bene- 2 2 24

ficiaries belonging to the 20- 16 16 16

bottom consumption quin- 15 _ X1tile. However, the top three 1o -quintiles account for 52% 3 6 5 4 4 5 5 6

of all students on govern- 5 3iiL *_ *ment scholarships (Figure % of upper- and post- % share among all % of secondary and % share among al I49). It was noted in chapter secondary students beneficiariesof post-secondary beneficiaries of

receiving benefit program students receiving program3 that 90% of university benefit

students in Thailand are Government Scholarship Program Student Loan Program

drawn from the richest two Figure 53quintiles of the population.The relatively large leakage of student loan and scholarship benefits to the non-poor thusreinforces the highly unequal distribution of higher educational opportunities in the coun-try. More obviously needs to be done not only on expanding educational assistance pro-grams for the poor but in also targeting these programs much more narrowly to studentsfrom the poorest economic backgrounds.

260. Credit Programs. In many countries, micro credit schemes that enable thepoor to borrow for acquiring income-generating assets are an important element of thefight against poverty. In Thailand, the Poverty Alleviation Project administered by theDepartment of Community Development in the Ministry of Interior is such a program.Under this program, an amount of Baht 289,000 is provided to each village, to be used toprovide interest-free loans to poor households as seed money for investments in income-generating activities. The Government has begun implementation of a more ambitiousproject under which each village in the country will be provided a revolving fund of onemillion bath (Village Fund Scheme), also to be used for making loans to poor householdsfor income-generating investments.



261. Unfortunately, there isvirtually no information in the Share of each real per capita income quintile in total household (gross)

SES on howthe ifovration nlethe-borrowing from the Bank of Agriculture and Agricultural Cooperatives,SES on how the Poverty Allevia- 701996-99 (agricultural households only)

tion Project has performed and 59

how poor households have bene- ¢ 60 10 0l1996 019 :19

fited from this project. However, o50Q

the SES does routinely collect 3 |

information on household bor- 2°31

rowing from various sources, in- '5 30 27 2

cluding the Bank of Agriculture 19 2020 1

and Agricultural Cooperatives E

(BAAC). These data can be ana- v 0n

lyzed to examine how well tar- 0 i- ed -T - -- Rht

geted government agricultural Poorest Second Tird Fourth Richestcredit programs are to the poor . Real per capita income quintile (among agricultural households)

Figure 54

262. Relatively few households in the SES report borrowing from the BAAC. In 1999,for instance, only 2% of all agricultural households reported borrowing from the BAACduring the month preceding the survey. For those that borrowed, the average amountborrowed was large, amounting to approximately 4.4% of monthly household income.More importantly, better-off agricultural households appear to account for the bulk ofBAAC credit. This is true of 1996, 1998 and 1999 (Figure 54). Indeed, the data suggestthat the share of the richest quintile of agricultural households in total BAAC financingincreased sharply from 25% in 1996 to 59% in 1999. At the same time, the share of thepoorest quintile of agricultural households fell from 11% to 5%.45

263. Since only a small proportion of households in the SES report any borrowing dur-ing the reference survey period (past month), these results must be treated with caution.At the same time, the fact that all three SES surveys from 1996, 1998 and 1999 show thatBAAC credit is strongly biased toward the better-off agricultural households suggeststhat the results shown in Figure 54 are not entirely spurious, and that they may be broadlyrepresentative.

264. There is another important implication of these findings. There has been somerecent discussion in Thailand about granting debt (interest) relief to farmers. The resultspresented in this report suggest that uniform interest relief to all BAAC borrowers willbenefit the richest agricultural households much more than poor households. If the mainpurpose of the interest or debt relief is to assist the poor, the relief program will need tobe much better targeted to this group of borrowers.


265. While Thailand's record at reducing poverty during the last decade or two hasbeen impressive, the economic crisis has brought the issue of poverty back to the fore-

45The assumption being made here is that borrowing patterns during the reference month are broadly in-dicative of the total stock or portfolio of BAAC loans.



front. With economic growth faltering and the distribution of income worsening, the in-cidence of poverty increased sharply in the immediate post-crisis period and appears tohave stayed at that level. Given that the macro outlook for economic growth in Thailandin the near term is somewhat guarded, owing to global uncertainties, distributional issuesassume even greater importance in reducing poverty, as inequality mediates the relation-ship between growth and poverty. Viewed in this context, the sharp increase in inequal-ity between 1998 and 1999 is troubling, although it is not clear whether this was a one-time event due to the economic crisis or whether it is the start of a trend of worseninginequality over the coming years.

266. As Poverty and Public Policy is only a step toward a poverty strategy with manyanalytical issues still to be addressed, caution is called for in drawing policy conclusionsfrom results obtained so far. Some preliminary conclusions, however, are possible fromthis study:

Poverty is highly concentrated in a few provinces in Thailand, most ofwhich are in the Northeast (and a few in the South and the North regions).While the available data are not suitable for estimating poverty at the vil-lage level, early and tentative results suggest that poverty is also highlyconcentrated in a relatively small number of villages in the country. Thismeans that geographical targeting of poverty interventions, such as an ap-propriately-designed village fund, to the poorest provinces and villages islikely to be effective in reducing poverty.

* While poverty has generally responded well to economic growth in thepast, the problem has been that growth has been far too concentrated in afew regions, especially in and around Bangkok. This has reduced the rateof poverty reduction in the outlying regions, particularly in the Northeast.While migration of labor from the poorer provinces to Bangkok hasbrought about some diffusion of the benefits of concentrated growth, thereare significant costs of and limits to migration. Economic growth that ismore dispersed and less centered around Bangkok would probably resultin a more sustainable reduction in poverty in the future.

* There is also some evidence to suggest that some of the poverty, espe-cially in the Northeastern provinces, is structural in nature. Structuralpoverty cannot be eliminated with growth alone; more direct anti-povertypolicies, such as access to credit and education, may be needed in additionto economic growth.

* The Thai government has an impressive number of anti-poverty programsin place, and public spending on these programs has increased sharply dur-ing the 1990s. Further, the distribution of benefits of these programs ap-pear to be reasonably progressive, although overall public expenditure ismildly regressive on a provincial basis. The main issue with some of theseprograms is their low levels of coverage, with the result that a significantproportion of the poor are outside the purview of these programs.

* Public employment schemes, which were used widely during the eco-nomic crisis as a short-term anti-poverty tool, appear to be the most suc-



cessfully targeted poverty interventions. A strong case could be made formaking such schemes a permanent feature of a poverty-alleviation andsafety net program. The schemes could be designed so that they could beexpanded rapidly during times of economic crises and scaled back duringtimes of strong economic growth. This would generally make for aswifter government response to economic downturns in the future.

* In contrast, careful thought needs to be given to an important initiative ofthe new Government with obvious potential to help the poorest Thais -viz., debt relief from government loans. Household survey data suggestbetter-off agricultural households account for the bulk of credit from theBank of Agriculture and Agricultural Cooperatives (BAAC). Hence, uni-form (i.e., untargeted) interest relief to all BAAC borrowers will benefitthe richest agricultural households much more than poor households.

* Likewise, targeting the proposed Village Fund to the poorest villages inthe country would be significantly more efficient and equitable than thecurrent proposal of providing an identical amount of resources to everyvillage in the country.



ANNEX A

Government Expenditure on Anti-Poverty Prokrams (in millions of current baht), 1993- 991993 1994 1995 199 6 1997 1998 1999

All Income-Generation Programs 315.2 602.3 1,910.6 2,456.1 9,146.5 18,300.8 20,587.2Poverty Alleviation Program 0.0 0.0 922.8 1,342.0 563.5 0.8 587.2

Job Creation Program (road construction) 315.2 602.3 987.8 1,114.1 133.0 0.0 0.0Education Loans Program 0.0 0.0 0.0 0.0 8,450.0 18,300.0 20,000.0All In-Kind Transfer Programs 5,421.7 7,925.8 8,629.3 11,006.8 7,395.6 4,947.0 12,538.1School Lunch Program 819.9 1,832.7 2,570.4 2,633.3 3,764.8 2,425.4 2,364.6- for pre-primary school 99.9 332.6 523.6 574.6 699.4 687.9 809.1- for primary school 120.0 1,000.1 1,546.8 1,558.7 1,431.4 1,333.7 1,555.5- Fund for lunch programs 700.0 500.0 500.0 500.0 1,500.0 500.0 0.0

Housing Programs 1,131.6 1,648.4 1,831.0 3,387.6 1,484.2 502.6 352.7Urban Poor Development Program 500.0 500.0 0.0 266.1 0.0 0.0 0.0- Housing Assistance Program 0.0 0.0 0.0 500.0 500.8 0.4 0.4- Housing Development Program 330.8 704.7 828.8 552.8 300.9 171.6 2.0- Relief for Congested Communities 300.8 443.7 1 j002.2 2,068.7 682.5 330.6 350.3Health Program for the Needy 3,470.2 4,444.8 4,227.9 4,985.9 5,768.3 8,417.7 9,820.8

All Cash Transfer Programs 0.0 611.7 913.6 1,226.3 1,787.0 1,820.8 1,687.7All welfare programs 0.0 611.7 913.6 1,226.3 1,787.0 1,820.8 1,687.7

- for children and youth 0.0 277.5 305.0 356.4 548.8 452.1 462.6- for women 0.0 136.5 174.7 210.1 262.8 211.6 228.0- for the needy 0.0 73.9 75.8 104.4 102.0 298.5 108.6- for the elderly 0.0 123.8 358.1 555.4 873.4 858.6 888.5

All Anti-Poverty Programs 5,736.9 9,139.8 11,453.5 14,689.2 25,581.6 33,988.9 34,813.0All Anti-Poverty Programs (1988 baht) 4,476.8 6,778.3 8,062.1 9,943.5 16,637.6 20,113.3 21,164.8



References

Barro, Robert J. 2000. "Inequality and Growth in a Panel of Countries," Journal of Eco-nomic Growth 5: 5-32.

Constitution of the Kingdom of Thailand, B.E. 2540 (1997), Translation by the Office ofthe Council of State.

Dollar, David and Aart Kraay, 2000. "Growth is Good for the Poor," Development Re-search Group working paper, World Bank, Washington D.C., March.

Fields, Gary S., 2001. "Accounting for Income Inequality and Its Change: A NewMethod, with Application to the Distribution of Earnings in the United States," School ofIndustrial Relations, Cornell University, Ithaca, NY, July, mimeo.

Foster, J., J. Greer and E. Thorbecke, 1984. "A Class of Decomposable Poverty Meas-ures," Econometrica, May: 761-766.

Ho, Samuel P.S., 1978. Economic Development of Taiwan, 1860-1970, New Haven, Ct.:Yale University Press.

Kakwani, Nanak and Ernesto M. Pernia, 2001. "What is Pro-Poor Growth?" Asian De-velopment Review, forthcoming.

Kakwani, Nanak and Medhi Krongkaew, 1996. "Final Report: Establishment of Key In-dicators System for the 8th Plan Monitoring and Evaluation - Paper II: Specific Issues,"DED, NESDB, Bangkok, mimeo.

Kakwani, Nanak, 2001. "Pro-Poor Growth and Policies," paper presented at the ADBAnnual Meeting Seminar on "Pro-Poor Growth: The Renewed War on Poverty," Hono-lulu, Hawaii, 8th May.

Krongkaew, Medhi, 1999. "Thailand's Social Assistance to the Poor," paper presented atthe Manila Social Forum, Manila, November 9-12.

Lampietti, Julian A. and Linda Stalker, 2000. "Consumption Expenditure and FemalePoverty: A Review of the Evidence," Policy Research Report on Gender and Develop-ment, Working Paper Series No. 11, Poverty Reduction and Economic ManagementNetwork, World Bank, April.

Litvack, Jennie and Tripti Thomas, 2001. "Annex 8: Decentralization Strategiesfor Empowerment," draft note, PRMPS, World Bank.

Narayan, Deepa, with Raj Patel, Kai Schafft, Anne Rademacher, and Sarah Koch-Schulte, 2000. Voices of the Poor: Can Anyone Hear Us? New York: Oxford UniversityPress.



Ravallion, Martin, 1998. "Reaching Poor Areas in a Federal System," Policy ResearchWorking Paper 1901, World Bank, March.

Rawls, John, 1971. A Theory of Justice, Cambridge, MA.: Harvard University Press.

Sen, A.K., 1976. "Poverty - An Ordinal Approach to Measurement," Econometrica, Vol.44.

Sen, A.K., 1992. Inequality Re-examined, Cambridge, MA.: Harvard University Press.

Sl0k, Julie, 1999. "Thailand Country Gender Profile", East Asia and Pacific Region Gen-der Home Page, November.

Transcript of NESDB Workshop on Poverty, December 22-23, 2000.

UNICEF Office for Thailand, 1997. Children and Their Families in a Changing Thai So-ciety. Bangkok.

Van de Walle, Dominique, 2001. "The Static and Dynamic Incidence of Vietnam's Pub-lic Safety Nets," World Bank.

World Bank, 1999a. "Thailand Country Gender Strategy," Draft, East Asian and PacificRegion, September.

World Bank, 1999b. Thailand Social Monitor: Challenge for Social Reform, World BankThailand Office, Bangkok, January.

World Bank, 2000a. Beyond the Crisis: A Strategy for Renewing Rural Development inThailand, Rural Development and Natural Resources Sector Unit, East Asia and PacificRegion. World Bank Thailand Office.

World Bank, 2000b. Thailand Economic Monitor, World Bank Thailand Bangkok, De-cember.

World Bank, 2000c. Thailand Environment Monitor, World Bank Thailand Office,Bangkok, January.

World Bank, 2000d. Thailand Social Monitor: Social Capital and the Crisis, World BankThailand Office, Bangkok, January.

World Bank, 2000e. Thailand Social Monitor: Thailand's Response to AIDS - Buildingon Success, Confronting the Future, World Bank Thailand Office, Bangkok, July.



World Bank, 2000f. Thailand Social Monitor. Thai Workers and the Crisis: Social Moni-tor, World Bank, Thailand Office, Bangkok, July.

World Bank, 2001a. "Decentralization and Governance: Does DecentralizationImprove Public Service Delivery?" PREM Note #55, June.

World Bank, 2001b. Engendering Development: Through Gender Equality in Rights,Resources, and Voice. World Bank Policy Research Report. Washington, D.C.

World Bank, 2001c. Making Sustainable Commitments: An Environmental Strategy forthe World Bank. Washington, D.C.

World Bank, 2001 d. Thailand Economic Monitor, World Bank Thailand Office, Bang-kok, July.


WORLD BANK

OFFICE BANGKOKDietheimn Towers

14th Floor Tower APrinted By

93/1 Wireless Road DESIGN * COPY * PRINT

Bangkok 10330. Thailand

Tel. (662) 256-7792 alpho!uraphiosoPrintshops Of The Future

Fax (662) 256-7794-5 TEL.: 266-5335-8 Fax.: 266-5330

world bank document****n*i**r***** thailand social monitor poverty and public policy preface poverty...

Documents

world bank document****n*ir*** thailand social monitor poverty and public policy preface poverty...