payment attitudes to public services in the municipalities of …bigd.bracu.ac.bd/images/for...

BRAC Institute of Governance and DevelopmentBRAC University

Payment Attitudes to Public Services in the Municipalities of Bangladesh

BIGD Working PaperNo. 38 February 2017

Dr. Sultan Hafeez RahmanMd. Nahid Ferdous Pabon

BRAC Institute of Governance and DevelopmentBRAC University


Dr. Sultan Hafeez RahmanMd. Nahid Ferdous Pabon

BRAC Institute of Governance and Development (BIGD)BRAC University, Dhaka

BIGD Working Paper SeriesNo. 38, April 2017

Payment Attitudes to Public Services in the Municipalities of BangladeshByDr. Sultan Hafeez Rahman, Executive Director, BIGD.Md. Nahid Ferdous Pabon, Former Research Associate, BIGD; Currently Research Associate, Bangladesh Institute of Development Studies (BIDS), Dhaka, Bangladesh.

© 2017 BIGD, BRAC University

Cover: Md. Parvej

Published by

BRAC Institute of Governance and Development (BIGD)BRAC University

SK Centre (Basement, 3rd - 7th & 9th Floor), GP, JA-4, TB Gate Mohakhali, Dhaka 1212, Bangladesh

Tel : (+88 02) 5881 0306, 5881 0326, 5881 0320, 883 5303Fax : (+88 02) 883 2542Email : [email protected]: bigd.bracu.ac.bd

The views expressed in this paper are those of the authors and do not necessarily reflect the views of BRAC Institute of Governance and Development (BIGD), BRAC University. This volume is a product of the BIGD, BRAC University. The findings, interpretations and conclusions presented in this document do not necessarily reflect the views of BRAC University authorities.

All rights reserved. This book or any portion thereof may not be reproduced or used in any manner whatsoever without the expressed written permission of the publisher except for the use of brief quotations in a book review.

Terms of use

Disclaimer

BDT 50USD 5

Table of Contents

List of Tables iv

Acknowledgements v

Abstract vi

About Author vii

1. Introduction 1

2. Survey of the Literature 3

3. Operational Issues in The Municipalities and City Corporations 4

4. Methodology 10

5.Summary Statistics 13

6. Econometric Analysis 15

7. Test of Independence of Irrelevant Alternatives (IIA) Assumption 20

8. Summary and Conclusion 21

References 23

iii

List of Tables

Table 01: Number of sanctioned posts and staff working in municipal classes 5 Table 02: Number of sanctioned posts and staff working in administrative divisions 5Table 03: Number and percentage of municipalities with no staff 6Table 04: Persons per staff 6Table 05: Types of roads as a percentage of total road length 7Table 06: Infrastructures need repair or reconstruction as a percentage of total 7Table 07: Number and percentage of municipalities outside the coverage area 8Table 08: Percentage area covered within the covered by municipal services 8Table 09: Number of street lights, waste disposal rate and excess demand for water 8Table 10: Mean value of the explanatory variables across categories of the dependent variable 14Table 11: Mean value of the explanatory variables across categories of the dependent variable 14Table 12: Percentage distribution of payment categories 15Table 13: Class-wise percentage distribution of payment categories 15Table 14: Marginal effects estimated by MNL regression of willingness to pay for improved water supply service 16Table 15: Marginal effects estimated by the MNL regression of willingness to pay for improved waste management service 17Table 16: Marginal effects estimated by the MNL regression of willingness to pay for improved street lighting facility 18Table 17: Table 17: Average predicted probabilities for WTP 19Table 18: Table 18: Hausman (Chi-square) test statistic and P value 20Table 19: Chi-square test statistic and P value from seemingly unrelated estimation 20

iv

AcknowledgementAuthors are grateful to Mr. Ashok Madhab, Additional Secretary of the Local Government Division (LGD), Ministry of Local Government, Rural Development and Cooperatives of the Government of the People’s Republic of Bangladesh for facilitating data collection from the municipalities. Authors would like to thank many other staff of LGD and municipal offices, without whose cooperation, data collection from local government sources would not have been possible. Authors also thank Ms. Saika Nudrat Chowdhury of BIGD for editorial assistance and Mr. Mahmudul Alam Talukdar (Rasel) of BIGD for providing assistance from the publication department.

v

AbstractThe paper investigates the factors which influence households' willingness to pay for public services provided by municipal councils in Bangladesh. The analysis in the paper is based on a survey of all municipalities, with a sample size of over six thousand households. The survey data was complemented with administrative data obtained from offices of these local bodies. A multinomial logit model was used to investigate the factors (such as the rates paid by user households for public services, eg, water supply, etc. and socio-demographic characteristics of households, such as education, age etc.) which are significant in changing the probability of households' willingness to pay. The analysis was conducted with three different categories of households in the model, which differ in their attitudes to paying for the public services they use. The empirical results provide evidence that better public services significantly influence households' attitudes to paying more for the services, indicating that cost recovery-type of pricing policies may be used in the municipal areas, contrary to populist concerns. Thus, the municipalities may well be gradually 'weaned away' from dependence on grants from the government and instead, mobilize more resources locally to finance development.

vi

Municipality, Service coverage, Water supply, Solid waste management, Street lighting, Willingness to pay, Probability, Market orientation, Fiscal sustainability.

Key Words:

About AuthorDr. Sultan Hafeez Rahman

Dr. Sultan Hafeez Rahman is serving as the Executive Director of BRAC Institute of Governance and Development (BIGD), BRAC University. He started his professional career at the Bangladesh Institute of Development Studies (BIDS) in 1975 and worked at the Asian Development Bank (ADB) from 1992 to 2012. He retired from the ADB as the Director General and returned to Bangladesh in 2012. His work as a researcher and development practitioner includes the areas of macroeconomic policy, international trade, regional integration, development planning, transport, energy, financial sector, urbanization, education, environment and agriculture. He presented papers, participated in and organized many workshops, seminars and conferences around the world throughout his professional career spanning 40 years.

Dr. Rahman has worked in Cambodia, Thailand, Vietnam, Malaysia, Republic of Korea, Indonesia, Kazakhstan, Uzbekistan, India, Pakistan, Afghanistan, Nepal, Bhutan, Sri Lanka, Maldives and Bangladesh, Timore Leste as well as 14 Pacific Island countries.

He is a reputed economist and has published on a wide range of economic and policy issues. He was also the Editor of Bangladesh Development Studies (journal of BIDS). His current research interests are in the areas of economic growth and inequality, macroeconomic policy and sustainable development. Dr. Rahman undertook his post graduate studies in the USA.

Md. Nahid Ferdous Pabon

Mr. Nahid Ferdous Pabon is working as a Research Associate at the Bangladesh Institute of Development Studies (BIDS). Prior to joining BIDS, he worked with the BIGD, BRAC University as a Research Associate. He also worked at the Bureau of Economic Research (BER), University of Dhaka as a co-researcher.

Pabon obtained his Master’s and Bachelor’s degrees in Economics from the University of Dhaka, respectively in 2015 and 2014 with distinction.

His research interest relates broadly to the areas of economic growth, public policy and international trade.

vii

1. Introduction

Bangladesh is undergoing rapid urbanization. At independence in 1971, barely 8% of the country’s total population of 75 million lived in urban areas, while the remaining 92% lived in rural areas, i.e., Bangladesh was a predominantly rural economy. In 2016, the urban population increased to 35% of the total population of 160 million, and is forecast to exceed half the population by 2030. The government has thus prioritized the provision of public services to the municipal councils (Pourashabhas) and city corporations of the country, which currently number 332 including 11 city corporations. The government’s strategic agenda for the municipalities and city corporations are to achieve sustainable and planned urban development, improve the quality of life with emphasis on improving living standards of the poor, develop cities and towns to ensure better public services and infrastructure aimed at increasing productivity, employment and investment, as well as to enhance institutional and financial capability (7th Five Year Plan 2016-2021, p. 383).

The quality of public services, however, depends on adequate coverage and reliability of the services. Both of which are constrained by resources, management capacity, adequate staffing, and importantly, effective coordination between multiple government agencies and departments tasked with service delivery responsibilities, and importantly, by the degree of decision making authority (decentralization) at their disposal.

In addition to government budget transfers in the form of block grants, special grants and project grants, the municipalities finance the balance of their expenditures (Yunus & Rahman, 2015) from revenue collected locally through service or user charges, taxes, tolls, fees, etc. The revenue mobilization from local sources is low in most of the municipalities, and ranges from 30-60 percent of total revenue.

The conventional wisdom is that the low local revenue collection from charges for public services is due to poor willingness to pay (WTP) of the residents. This however, seems counter intuitive in light of the rise in incomes in urban, peri-urban and rural Bangladesh, especially over the past two decades. Since local revenue mobilization is important to the financial sustainability of the local government entities, to encourage participation in development processes and establishment of transparency and accountability at the local government levels, it is important to understand the factors that influence the households’ (service consumers) WTP in the municipal councils of Bangladesh.

Water supply, solid waste management (SWM) and street lighting are the most important public services delivered to the households by the municipalities, and charges levied on them are important sources of local revenue. The rates which municipal authorities may charge for these services are provided in the government’s model tax schedule first issued in 1976. The rate schedule which is currently in force was issued in 2003 and specifies the maximum rate which may be levied for each service. These are 10 percent, 7 percent and 3 percent, for water supply, SWM and street lighting facilities, respectively. However, the actual rates charged for the services vary substantially across municipalities.

Significantly, the municipalities cannot as yet, provide full coverage of the services to the entire municipality/city corporation. Twenty four out of 319 municipalities surveyed do not provide street lighting services, 165 do not provide water supply services and 40 municipalities do not provide SWM services. In percentage terms, these translate to 7.5 percent, 51 percent, 12.5

1BIGD Working Paper Series No. 38

percent of the municipalities which are unable to provide piped water supply, SWM and street lighting facilities, respectively to the households. The situation with respect to water supply is of particular concern. Even in the municipalities which provide these essential services, about 45-70% of their area is not covered by the three services. In other words a significant proportion of the population living in the municipal/city corporation areas is still deprived of essential public services.

The government has devised a 3-tier typology to classify the municipalities. They are grouped into three classes based on the annual revenues they collect during the preceding three years. Thus class A municipalities are those whose revenue collection exceeds Tk 6 million, class B municipalities collect revenue between Tk 2.5-6 million, and class C municipalities collect revenue between Tk 1-2.5 million (LGED website).

This study considers factors such as location of the household, i.e., whether the household is inside the utility service coverage area, household’s satisfaction with the service it receives, affordability of the rates charged, level of education, and age of the household head, household size to investigate their WTP for the 3 key public services provided by the municipality. The paper also examines whether there are differences in the WTP for the services due to gender (of the household head) or municipal class, i.e., A, B and C. The investigation involves evaluating household WTP for a hypothetically “better service” for successively higher charges/rates using the contingent valuation (CV) method, followed by extensive econometric analyses of the determinants of WTP.

Bangladesh has 332 municipalities, including 11 city corporations. The study uses data from the BIGD municipal survey (2014 and 2015) conducted on households in all municipalities and city corporations of Bangladesh. A second phase of data collection in 2015 was carried out to improve the quality of the data obtained from the municipal offices. The cluster sampling design used, yielded a sample size of 6634 (Yunus & Rahman, 2015). This involved interviewing 20 holding owners from each municipality in the household survey. As most municipalities comprise 9 administrative wards, the sub-sample of 20 households could be conveniently distributed across the wards, to do about 2 interviews in each ward. However, the number of wards in the city corporations is higher than 9 and therefore, at least one household from each ward was interviewed. This increased the sample size to 6752—the 2 municipalities which could not provide the required data were dropped from the sample. All samples for the household survey were drawn randomly.

The household survey collected detailed information on households’ sources of income, WTP for a hypothetical service of ‘better quality’, reasons for their WTP or lack of it, with both open and close ended questions. Survey conducted on the municipal offices comprised detailed data on revenue collection, expenditure, infrastructure (e.g. road, drain, bridge and culverts, etc.), actual and sanctioned staff in the municipal offices, services delivered, etc.

Payment Attitudes to Public Services in the Municipalities of Bangladesh2

2. Survey of the LiteratureDelivery of services such as SWM, water supply and street lighting are among the most important issues for city and municipal authorities throughout the world. These concerns have been researched extensively in both developed and developing countries. Determining households’ WTP for improved services using the CV approach constitutes a major part in this literature.

Major questions raised in the literature include ‘what factors influence households’ WTP for improved service delivery plans?’ and ‘what factors influence the maximum amount households are willing to pay for a hypothetical improved plan?”

In most of the studies, logit or probit models were used to model the probability of WTP for the proposed plans (e.g. Awunyo-Vitor et al. 2013, Hagos et al. 2012, Seth et al. 2014, Adepoju et al. 2010, Mary et al. 2014, Addai et al. 2014, Roy 2013, Mustafa et al. 2014, Rollins at el. 1997, Tussupova at el. 2015).

For example, a stratified random sample of five hundred respondents was used in a study on Islamabad, Pakistan by Anjum (2013) to measure the WTP of the respondents for improved SWM service and to find the determinants of WTP. Binary choices (i.e. yes, no) were offered for the payment of specific amount per month followed by an open ended question about the maximum willingness to pay (MWPT). Logit model was employed to determine the factors affecting WTP (i.e. the probability of paying or not) of the respondents. Then a multiple regression model was estimated to find the actual amount respondents would be willing to pay for the plan. Rollins et al. (1997) used a CV technique and a double-bounded logit model to identify the determinants of WTP for improved public water supply infrastructure.

Yunus & Rahman (2015) estimated the WTP for water supply, SWM, street lighting and property tax in the municipalities of Bangladesh using the BIGD municipal survey (2014 and 2015) data. Using a probit model, they found current tax rate and income of the household as significant determinants of WTP in all four cases.

Some studies have used the Tobit model to investigate the determinants of the maximum amount households would be willing to pay (MWTP) because the MWTP was not fully observed and it took zero values for a substantial part of the sample (i. e. censoring from below). For example, Hagos et al. (2012) used a Tobit model for SWM plan where the MWPT was censored from below. Awunyo-Vitoret et al. (2013) did the same. For water quality and supply enhancements, Brox et al. (1996) used a Tobit model which allowed the WTP to be non-negative. The income, age and number of children in the household were found to be statistically significant determinants of the WTP for better water quality. Willis et al. (2003) used a spike log-logistic model and the CV approach to estimate robust values of the WPT for the service and community benefits of replacing and/or upgrading street lighting.

Hazra et al. (2013) examined public perception towards attributes of SWM in terms of WTP values using a multinomial logit (MNL) model. However, in the presence of heterogeneity in consumption in terms of various socioeconomic characteristics, it is important to understand how these characteristics influence WTP values. From this motivation, Hazra et al. (2015) further analyzed the same data set used in 2013 by developing several random parameter logit (RPL) models and assessed the difference between WTP values from the two models and to capture the decomposition effect of socioeconomic characteristics of households on WTP values.


3. Operational Issues in The Municipalities and City Corporations

Beaumais et al. (2013) used a rank ordered logit model to estimate marginal WTP to decrease pollution (i.e. negative externalities) associated with SWM using data from a ranking choice experiment. They found that people were not satisfied with the status quo; homeowners and urban dwellers were found to be more willing to pay for the reduction of waste externalities. They also found that individuals who were aware of the existing fee (only 16 percent respondents stated that they knew how much they pay for SWM) showed a greater WTP for the reduction of waste externalities.

Ku et al. (2009) attempted to apply choice experiments with regard to the residential waste disposal system in Korea by considering various attributes that were related to the system. Multinomial logit model (MNL) and nested logit model (NL) were used to estimate the WTP for a clean food-waste collection facility, the collection of small items (such as obsolete mobile phones and add-ons for personal computers), and a more convenient large waste disposal system. Estimates of WTP from the empirical analysis based on multinomial logit models were quite similar to those of nested logit models.

This study aims to model the households’ WTP for improved water supply, SWM and street lighting services. The study uses a contingent valuation approach to assess individual household’s WTP for each of the three public services. According to their responses, households were categorized in three different response-groups: (1) Willing to Pay, (2) Reluctant and (3) Oppose. With this categorization of WTP, the study chooses to model the probability of WTP using a multinomial logit model (MNL). The model is estimated for each of the 3services (water supply, SWM and street lighting) provided by the municipalities rather than considering only one, as in most of the studies in the literature. We expect that such a study will yield a more precise understanding of the quality of municipal services. Thus the study not only extends and deepens the investigation by Yunus & Rahman (2015), but also is the first such study on municipal services in Bangladesh with significant policy implications in an area of high priority for the government, as well as a contribution to the literature in terms of its wider coverage of public services.


Effective management of the municipalities and city corporations requires that they are adequately staffed and positions appropriately filled. However, the reality on the ground seems to be quite different.

In total, the number of staff working is less than the posts sanctioned in all three classes of municipalities. Percentages of vacant posts are 20, 43 and 50 in A, B and C classes, respectively (Table 1). The number of staff working in water supply, tax collection, audit and health services is less than the number of posts sanctioned in each of the classes. For example, 21 percent of posts in water supply service are vacant in class A municipalities, whereas the vacancies are around 65 percent in the other two classes. The percentages of vacant posts in tax collection service are 37, 47 and 59 in class A, B and C municipalities. Similar distribution exists in the health services with 34, 37 and 67 percent of vacancies, respectively in A B and C class municipalities. Percentages of vacant posts in auditing varies from 55-58 in the three classes of municipalities. It is evident that the percentage of vacancies are higher in class C municipalities in each service; around 65, 59, 58, 67 percent posts are vacant in class C municipalities, respectively in water, tax, audit and health services. A large number of temporary and contractual staff works in SWM and this is the only sector which has more staff than the posts sanctioned in every municipal class. The number of SWM staff working in class A municipalities is 6 times higher than the posts sanctioned. The number goes down to 2.8 and 2.9 in class B and C municipalities, respectively.

The study surveys 319 municipalities across the country among which 109 are class ‘A’, 123 are class ‘B’ and 87 are class ‘C’.

The study surveys 25, 60, 62, 36, 26, 61, 30 and 19 number of municipalities from Barisal, Chittagong, Dhaka, Khulna, Mymens-ingh, Rajshahi, Rangpur and Sylhet divisions, respectively.

1

2

Table 2 presents the division-wise figures and reveals that as before, the number of working staff in SWM is several times higher than the posts sanctioned in each administrative division of the country. However, the scenarios in other services are totally the opposite. For example, every division, except Khulna, is producing the piped water facility with lower number of staff than the posts sanctioned. The working staff strength of water service in Khulna division is 10 percent higher than the posts sanctioned. In total, Khulna and Mymensingh divisions have the lowest percentage of vacant posts with 10 and 14 percent, respectively. The highest percentages of vacancies are in Sylhet, Barisal and Chittagong division with 50, 45 and 45 percent, respectively.


Table 1: Number of sanctioned posts and staff working in municipal classes

Table 2: Number of sanctioned posts and staff working in administrative divisions

Source: Authors’ estimate based on BIGD Municipal Survey

1

2


Average population density in class A, B and C are 5584, 4643 and 3708 per square kilometer (calculated from the survey data). 3

Table 3 provides a strong indication of the poor staff strength of the municipalities. It shows that 200 out of 319 (62 percent) of the municipalities do not have any audit staff. Percentage of municipalities with no staff in water supply service is 46 and the figures for SWM, tax and health are 13, 0.03 and 0.1, respectively.

Table 4 provides a measure of service delivery work load per staff, i.e. the number of service users per staff in each class of municipality. Earlier, in table 1 we noted that 20, 43 and 50 percent of the total sanctioned posts are vacant, respectively in A, B and C municipalities. This, coupled with a high population density, suggests a high work load per staff. Class A municipalities are the most densely populated3 and with 20 percent posts vacant, a single staff has to manage or work with around 1800 service users. Figures in B and C class are, respectively 916 and 1347 per staff which are also very high. Dealing with such large numbers of service users consequently weakens the quality and efficiency of service delivery. The scenarios in services such as water supply, SWM, audit, tax and health are worse. For example, service users being served per staff ranges from 13000 to 15000 in the three classes. Figures for SWM are 37000, 11000 and 13000 persons per staff in A, B and C classes, respectively. A tax staff in class A, B and C serves 19000, 10000 and 13000 persons, respectively, while a health staff serves 16000, 11000 and 13000 service users, respectively. Audit staff are the most over-stretched; persons per staff varies from 19000 to 68000 in the three classes. On the whole staff of class B municipalities, experience the lowest, while staff of class A municipalities experience the highest work load.

6

Table 3: Number and percentage of municipalities with no staff

Table 4: Persons per staff




Bituminous Carpeting Herring Bone-Bond Cement Concrete/ Roller Compacted Concrete Water Bound Macadam

45

67

On the one hand, many of the roads in municipalities are earthen, while on the other hand, a large proportion of the existing roads are damaged and, therefore, need repair or reconstruction (Table 6). A similar situation exists in bridge, culvert and drain infrastructure. The proportion of damaged roads in class C is the highest; 60 percent of total roads need repair, while the figures in class B and A are 55 and 32 percent, respectively. 69, 63 and 56 percent of drains, respectively in class A, B and C municipalities are damaged and need repair.

Full-fledged ‘municipalization’ of the existing municipal zones of the country remains an unfinished agenda of the government’s decentralization policy. A big number of the municipalities do not get any water supply, street lighting or SWM services. Table 7 presents that a total of 8, 52 and 13 percent of municipalities are outside the coverage area of street lighting, water supply and SWM, respectively. Number of municipalities without street lighting and SWM service is the highest in class C with 19 and 16 municipalities, respectively. 76 municipalities of class B have no water supply service coverage and the figures for class A and C are 28 and 61, respectively.

Table 5 describes the state of road infrastructure in the municipalities. Earthen and macadamized (BC4) roads constitute a large proportion of the total length of roads. In class A municipalities, the proportion of earthen (kancha) roads is approximately 60 percent whereas the proportion of macadamized (BC) roads is only about 22 percent. Proportion of other types of roads (e. g. HBB5, Brick Soling, CC/RCC6 and WBM7) are low with HBB, Brick Soling and CC/RCC roads constituting 4.6, 6 and 7.5 percent of total road length, respectively. The composition of road infrastructure in the other two classes is similar. However, in class B and C, the proportion of earthen roads is much lower than that in class A. Earthen, BC, Brick Soling, CC/RCC and HBB roads in class B are around 32, 36, 12, 10 and 6.5 percentage of total road length, respectively. In class C, about 40 percent of roads are earthen and BC roads constitute 28 percent of the total road length. Brick Soling, CC/RCC and HBB roads are 13, 8 and 7.5 percent, respectively of total road length. A large proportion of earthen roads show the poor state of road infrastructure in the municipalities.


Table 5: Types of roads as a percentage of total road length

Table 6: Infrastructures need repair or reconstruction as a percentage of total




Disposal rate =8

Even within the covered municipalities, a significant area still remains uncovered as regards provision of services. Percentage of the area covered within the covered municipalities is the highest in class A and the lowest is in class C. For example, only 41, 30 and 40 percent of the areas in covered class C municipalities are getting the services of street lighting, water supply and SWM, respectively. For covered class A municipalities, these figures are 56, 41 and 54, respectively for street lighting, water supply and SWM. In covered class B municipalities, 43, 34 and 46 percent of the areas have access to street light, water supply and SWM facilities (Table 8).

8

Table 9 shows the number of street lights per kilometer in each of the municipality classes. It can be seen that number of street lights is the lowest in class C and highest is class A municipality. Class A, B and C municipalities have 55, 39 and 38 street lights per kilometer. Disposal rate8 of solid waste, calculated in a daily basis, is 65-68 percent in all three classes. Municipal water supply does not meet the total demand for water in any class and average excess demand in class C is the highest. In class A, B and C municipalities, average excess demand for supplied water is 7124, 5554 and 8668 cubic meters, respectively.

Table7: Number and percentage of municipalities outside the coverage area


Table 8: Percentage area covered within the covered by municipal services


Table 9: Number of street lights, waste disposal rate and excess demand for water


4. Methodology


Households falling in the Reluctant category are unwilling to pay mainly because of their budgetary concerns. This study assumes that if they had not been constrained by their budget or if there had not been any cheaper alternative available, they would have agreed to accept the higher payment; because they do not raise issues such as corruption, uncertainty and lack of trust on the municipality. They are Reluctant to pay just because they cannot afford the higher payment or they have a less costly alternative.

Households falling in the Oppose category are not concerned about their budget rather they seemed more concerned about corruption, trust and uncertainty or service reliability issues. This study thus assumes that these households would not have paid the higher rates even if they had not been constrained by their budget.

10

9

10

The categorization:

In the first stage of the contingent valuation survey, a hypothetical description of a proposed plan was presented and a dichotomous question was asked: would the respondent be willing to pay an increased tax or service charge for the proposed plan or not? In the second stage, the respondents were asked about the reasons for their willingness or unwillingness to pay higher tax or service charges. The ‘unwillingness to pay’ respondents, were classified into two categories: Reluctant and Oppose. Respondents who stated (i) “The tax rates is too high” or (ii) “I have to pay a number of other taxes” or (iii) “I have other cheaper sources available” as reasons for their unwillingness were placed in the Reluctant9 category while the Oppose10 category comprised respondents who stated (i) “I do not trust the municipal authority” or (ii) “Officials of the municipality are corrupt” or (iii) “I have heard that the service quality is not good” as reasons for their unwillingness to pay. Individuals who showed WTP for the proposed plan, at the first stage, were placed in the ‘Willing to pay (WTP)’ category. The same 3 categories (WTP, Reluctant and Oppose) were used for each service (water supply, SWM and street lighting facility) in the empirical analysis.

The Model:

We fit a multinomial logit (MNL) model with three alternatives (e.g. WTP, Reluctant and Oppose) taking alternative invariant regressors, i.e. the same set of regressors are used for each three categories of the dependent variable.

In the general multinomial model, the utility of alternative j for individual i is specified as

where V_ij is the deterministic and e_ij is the random component of the utility. In order to maximize the utility, an individual chooses the alternative that gives him the highest satisfaction. For each category j, the outcome variable Y (i.e. the chosen alternative) is defined as

The probability of choosing alternative j is equal to the probability that the utility of alternative j is greater than the utilities of all other alternatives. Therefore,

Assuming that errors are identically and independently distributed (iid), the MNL specifies

First stage question: “Would you be willing to pay an increased tax or service charge for the proposed plan or not?” Second stage question (asked only to the respondents who showed unwilling to pay in the first stage): “What is (are) the reason (s) for your unwillingness?” Measured as (total tax liability-total tax payment)/total tax liability

11

12


where, Xi are the alternative invariant explanatory variables and βj are the coefficients for each alternative. The vector Xi' βj is expanded as

where, X1 is the current service charge (i.e. tax rate), X2 is the increase in service charge while facing the first stage question11. That is, X1+X2 equals the offered rate. X3, X4 and X5 represent each household’s annual income, age of the owner of holding and education measured as the total number of years of schooling of the owner of the holding, respectively. X6 is a measure of tax evasion12 and X7 represents household size (total member of the household). X8 is a dummy for class A which takes a value of 1 if the class is A, 0 otherwise. Similarly, X9 is 1, 0, otherwise in case of the dummy variable for class B The gender dummy X10 takes a value of 1 if the holding owner is male and 0 otherwise. The coverage dummy X11 takes the value of 1 if the household receives a particular service and 0 otherwise. X12, the high tax dummy, takes 1 if the offered rate exceeds 10, 7 and 3 for water, SWM and street lighting, respectively, and 0 otherwise.

Since Yj is 1 if alternative j is observed (i.e. the remaining Yk=0 ), for each observation on Y, exactly one of Y1, Y2…..,Ym will be nonzero. Therefore, the multinomial density for one observation can be written as:

and the likelihood function for N independent observations is

Therefore, log likelihood function in maximum likelihood estimation becomes

Category WTP is considered as the base category in the MNL regressions and therefore, marginal effects of regressors would in fact be the marginal probabilities of moving to either Reluctant or Oppose categories from the WTP category.

A limitation of the MNL model is that discrimination among the m (m equals 3 in this paper, namely WTP, Reluctant and Oppose) alternatives reduces to a series of pair wise comparisons (e.g. WTP vs Reluctant, WTP vs Oppose) that are unaffected by the characteristics of other alternatives. Since the conditional probability of any two alternatives does not depend on any other alternative, the MNL model reduces to a binary logit model between the pair of choices under consideration (Cameron &


H_0: difference in coefficients not systematic/ the IIA assumption satisfiedH_1: difference in coefficients systematic/ the IIA assumption violated

Hausman test statistic H=(βr ) -(βf )' [Var (βr ) - Var (βf )-1 ] (βr - βf ), where (βr ) and (βf ) are estimated coefficient matrices of the full and restricted models, respectively and Var represents respective estimates of variance matrices. His asymptotically distributed as chi-square with df = number of rows in (βr ) .

H becomes negative when Var (βr ) - Var (βf ) fails to meet the asymptotic assumption of Hausman test (i.e. not positive semidefinite). Such a case refers to an undefined Hausman test. However, Hausman& McFadden (1984) explains this as evidence that IIA assumption holds.

The variance (of the difference of the estimates) matrix estimated as Var (βr ) - Cov (βr ) ,(βf ) - Cov (βf ) ,(βr ) + Var (βf ) is always admissible.

13

14

15

16

Trivedi 2005). That is, the MNL model holds only under the assumption that a person’s choice between two alternative outcomes is independent of what other choices are available. This assumption is known as the Independence of Irrelevant Alternatives (IIA). In our case, we assume that all three categories are mutually independent. That is, we assume that the relative risk associated with categories WTP and Reluctant is not affected by the inclusion or exclusion of category Oppose, and the relative risk associated with categories WTP and Oppose is assumed to be unaffected by the inclusion or exclusion of category Reluctant.

A number of tests are available to test the validity of IIA assumption. Hausman & McFadden (1984) proposed a Hausman test which compares between the full MNL model (i.e. estimated with all outcomes/alternatives) and a restricted model that includes only some of the outcomes, where under the null hypothesis13, the coefficients of the full model are consistent and efficient and coefficients of the restricted model are consistent but inefficient. Statistical significance of the Hausman test statistic (H)14 indicates the violation of IIA assumption.

However, The Hausman test has at least two major limitations. First, the test may be undefined if the variance (of the difference of the estimates) matrix does not meet the asymptotic assumption of the test. In such a case, the test statistic H takes a negative value15. Second, the test relies on the availability of a fully efficient estimator, which may not be the case always. Another alternative test of the IIA assumption, which improves on the limitations of the Hausman test, is through seemingly unrelated estimation. The first advantage of this approach is that the test is always well defined16. Secondly, it allows estimation of the variance-covariance matrix of the multivariate normal distribution of the estimators of full and restricted models and test of equality of common coefficients. This study conducts both Hausman test and seemingly unrelated estimation to tests the IIA assumption.

12

( ( ( ( ( ( ( ( (

(

(

(

( ( ( (

( ( (

( ( ( ( ( (

5. Summary Statistics

Sources of local revenue collection, resources received from the central government, sectors of spending, holding tax paid by residents, overall infrastructure etc. of the city corporations are significantly different from the municipalities. As this study does not account for any individual-specific heterogeneity across the municipalities, we drop the city corporations in order to ensure a greater degree of homogeneity among the municipalities.

While offering the services, we increase the current rate by 0.2 percentage point for the first respondent and keep adding the same rate as we move to subsequent respondents until the current rate is increased by one percentage point. Once the increase in the current rate reaches one percentage point, the next offer becomes the current rate plus 0.2 and the process continues.

17

18

After dropping the city corporations17, we took 319 municipalities (109, 123 and 87 from class A, B and C, respectively) as the study sample. The number of households varies in each service with 4840, 4914 and 4031 households in water supply, SWM and street lighting, respectively. Table 10 presents mean of the independent variables of the MNL regression across different categories of the dependent variable for water supply and SWM.

Water Supply:

While the maximum amount of water rate the municipalities can impose is 10 percent, the actual average tax rate imposed is less than 2 percent in each category (Table 10). The average increase in tax rate for the proposed service is between 7 and 8 percentage points18 in each of the categories. The average annual income in WTP, Reluctant and Oppose categories is 440064, 372098 and 330002, respectively. Age distribution of the household head is roughly uniform across the categories at 50 years on average. Average years of schooling in WTP category is 8.37 whereas, for Reluctant and Oppose categories, the average years of schooling is 7.7 and 7.2 years, respectively. The average tax evasion is more than 80 percent of total tax liability in each category. It is higher in Oppose than the other two categories at88 percent. Household size is 4.5 members per household. On average, around 40 percent of households reside in class A and B in each three categories with the rest residing in class C. On average, around 94 percent of households are male-headed, 20 percent of households are receiving water coverage and 80 percent households are offered a high tax, according to our definition (see in the previous section).

Solid Waste Management (SWM):

Current service charge in SWM is around 4.5 percent on average in each category (Table 10). Increase in the service charge is around 1.5 percentage point on average. Average income is highest for the WTP category and the lowest for the Oppose category. Average age in each category is around 50 year. The average years of schooling is the highest for WTP and the lowest for Oppose category with 8.5 and 7.3 years, respectively. Average tax evasion is more than 85 percent of total tax liability in each category. Average household size is 4.5 members in each category. Average percentage of residents in class A and B are roughly 40 percent. Percentage of male headed households is 93 in each category. Percentage of households with SWM coverage is 30, 25 and 26, respectively in WTP, Reluctant and Oppose category. Households facing high tax are the lowest in Oppose category with a percentage of 56, while the percentages in WTP and Reluctant are 67 and 72, respectively.



‘Increase’ refers to offered rate minus the current rate. 19

14

Table 11 presents the mean values of the explanatory variables across the categories of the dependent variable for street lighting.

Street Lighting:

Average current service charge for street lighting is 2.2, 2.5 and 2.25 percent, respectively in WTP, Reluctant and Oppose category (Table 11). Average amount of increase is around 1 percentage point in each category. Income of the household is the lowest for Oppose category and the highest for WTP category. Average age of the household head is around 50 years and the member of household on average is around 4.5 in each category. Years of schooling of the household head on average is 7 years for Oppose category and 8.2 and 8 years for WTP and Reluctant category, respectively. Average tax evasion is around 85 percent of total tax liability. About 38, 53 and 33 percent of household in WTP, Reluctant and Oppose category, respectively are in class A while the numbers for class B are 43, 24 and 31, respectively. Around 95 percent of the households are male headed and the percentage of coverage in WTP, Reluctant and Oppose category are 54, 65 and 47, respectively. Households being offered high tax are 87, 89 and 75 percent in WTP, Reluctant and Oppose category, respectively.

Table 10: Mean value of the explanatory variables across categories of the dependent variable

Table 11: Mean value of the explanatory variables across categories of the dependent variable

n represents sample size, i.e. number of households in each categorySource: Authors’ estimate based on BIGD Municipal Survey

19

6. Econometric Analysis

Households falling in either Reluctant or Oppose category are called ‘Unwilling to pay (UTP)’20

Before going into the estimated results of MNL, we present in tables 12 and 13, the distribution of payment categories (i.e. WTP, Reluctant and Oppose) and classes of municipalities for the three services. According to our categorization, for the whole sample, in each section, a large number of household are willing to pay for the improved plan. More than three-fourths of the households show their WTP for better street lighting facility (Table 12). For water supply and SWM the percentage is around 60. Percentages of people opposing the payment are small in number; constituting less than 10 percent of the households in each section.

Table 13 combines the two categories Reluctant and Oppose and shows that in each three services, more households from class C municipalities are unwilling to pay for a better service compared to class A and B. In class A, 64 percent households are willing and other 36 percent are unwilling to pay for better water supply, while the percentages are 55 and 45, respectively in class C municipalities. Similar type of distribution exists in SWM. In street lighting, 73 percent of class A households are willing and 27 percent are unwilling to pay, while Percentage of households unwillingness to pay is slightly higher (28 percent).


Table 12: Percentage distribution of payment categories

Table 13: Class-wise percentage distribution of payment categories

n represents sample size, i.e. number of households in each categorySource: Authors’ estimate based on BIGD Municipal Survey


Source: Authors’ estimate based on BIGD Municipal SurveyTables 14, 15 and 16 present the estimated marginal effects of the MNL regressions.

20

20


Water Supply:

Table 14 shows the results of MNL regression for water supply service. The results indicate that a one percentage point increase in the current rate of water supply paid by households increases the probability of their moving from the WTP category to the Oppose category by 1.4 percent; the result being statistically significant at the one percent error probability. Difference between the offered rate and the current rate (termed as ‘increase’) has a similar effect on the WTP; the higher the amount of the ‘increase’ (i.e. the difference), the less is the WTP for the proposed service. A one percentage point rise in the amount of increase causes the probability of moving to category Reluctant to increase by 1.2 percent and the probability of moving to category Oppose by 1.4 percent. Age of the household head positively affects the probability of moving to category Reluctant but the probability of moving to category Oppose is affected negatively, i.e. households with older household heads are more likely to be Reluctant but they are less likely to Oppose. An additional year of schooling of the household head decreases the probability of being Reluctant by 0.6 percent and the probability of ‘Opposing’ by 0.2 percent. Households residing in classes A and B municipalities are less likely to show unwillingness to pay higher taxes. For example, being in class A and B reduces the probability of ‘Opposing’ by 1.5 percent and 1.7 percent, respectively compared to the probability that would be if they were in other classes (e.g. class A or C in case of class B and class B or C in case of class A). The coverage dummy is significant with positive sign; households within the coverage area are more likely to fall in the category Reluctant and vice versa. Households receiving the service of water supply are 6.7 percent more likely to be Reluctant than the households who are not receiving the service. The high tax dummy is significant with negative sign, which indicates that if the offered tax rate (current + increase) exceeds 10 percent then the households become less likely to Oppose, which seems counterintuitive. Household’s tax evasion behavior, gender of the household head and the size of the household do not significantly affect the WTP.

16

Table 14: Marginal effects estimated by MNL regression ofwillingness to pay for improved water supply service


Note: Robust standard errors are in parentheses and ***, **,* indicates statistical significance at 1, 5 and 10 percent levels, respectively


Table 15: Marginal effects estimated by the MNL regression ofwillingness to pay for improved waste management service



Solid Waste Management (SWM):

Marginal effects on current and increased rate are found statistically insignificant in both Reluctant and Oppose categories (Table 15). Although with a small margin, households having higher income are found to be less likely to be Reluctant. Effect of age on the WTP is similar to that in the case of water supply; households with older household heads are more likely to be Reluctant but less likely to Oppose. Education negatively affects the probability of unwillingness, e.g. one additional year of schooling decreases the probability of ‘Opposing’ by 0.23 percent. Bigger households (in terms of the total member of the household) are less unwilling to pay; household with a one member bigger family size is 0.8 percent less likely to fall in the Reluctant category compared to the probability of falling in the base category. Households in class A are found 2.7 percent less likely to Oppose the payment than households in other classes. Class B households are even more unwilling to pay; the probability of a class B household of falling in the Reluctant category is 8 percent lower than the households in other classes and probability of a class B household of falling in the Oppose category is 8 percent lower than the households in other classes. The coverage dummy negatively affects the probability of WTP. Households already receiving the service are 3.1 percent less likely to be Reluctant compared to the households outside the coverage area. As in the water supply, no significant gender differential is found and the high tax dummy turns out to be insignificant.


Table 16: Marginal effects estimated by the MNL regression ofwillingness to pay for improved street lighting facility



Street Lighting:

One percentage point increase in current rate increases the probability of being in Oppose category by 2 percent compared to the probability of being in WTP category, i.e. the current tax rate negatively affects the WTP (Table 16). Household heads with better education (in terms of the years of schooling) and higher income are less likely to be unwilling to pay. For example, one additional year of schooling decreases the probability of being in Reluctant category by 0.2 percent and the probability of being in Oppose category by 0.2 percent compared to the probability of falling in the WTP category. Marginal effect on tax evasion shows that households evading more taxes are less likely to show unwillingness to pay for improved plans. This finding is not counterintuitive. Tax evaders are motivated to deceive the authority. In any situation they would commit to pay but intend to do otherwise. Households with bigger family size are less likely to be unwilling to pay for the proposed plan; household with a one member bigger family size is 1.27 percent less likely to fall in the Reluctant category compared to the probability that the household falls in the WTP category. Also, households residing in class A and B municipalities are less likely to show unwillingness. For example, being in class A and B reduces the probability of falling in the Oppose category by 3.7 and 4 percent, respectively compared to the probability that would be if they were in other classes (e.g. class A or C in case of class B and class B or C in case of class A). The coverage dummy is highly significant with positive marginal effect for the Reluctant category, meaning people already having lights on their streets are more likely to be Reluctant to pay and vice versa. The high tax dummy, as in case of water supply, is negatively significant for the Oppose category which indicates that if the offered rate (current + increase) exceeds 3 percent then the households become less likely to Oppose the payment.



Table 17: Average predicted probabilities for WTP

Thus far the paper estimated the probability of moving from ‘Willing to Pay’ category to “Unwilling to Pay” category, while the latter category was broken down into other two categories, namely Reluctant and Oppose to facilitate a greater understanding of households’ attitude towards an offer of better service at a higher cost. Among the twelve regressors used, we found current tax rate, difference in the offered and current tax rate (the ‘increase’), income, age and education of household head, size and the class (e.g. A, B and C) of the household exhibit similar direction of probability (of moving from WTP to Reluctant and Oppose) across all three services.

Average Predicted Probability of willingness to pay (WTP):

Average predicted probability for the covered area of water supply is lower (0.65) than that for the not covered area (0.66), which indicates that the people getting the service are less willing to pay than those not getting the service (Table 17). For SWM, the average predicted probability in the covered area is higher (0.67) than that for the not covered area (0.62) which indicates that the people getting the service are more willing to pay than those not getting the service. In street lighting, average predicted probability for the not covered area is higher (0.8) than that for the covered area (0.76).Average predicted probability of WTP for street lighting is the highest in both covered and not covered areas with a probability of about 80 percent. This means people value street lighting the most. In the coverage area, people value water supply the least (with average probability of 65 percent) and in the non-coverage area, people value SWM the least (with average probability of 62 percent).


Table 18: Hausman (Chi-square) test statistic and P value

*** indicates that the IIA assumption is satisfiedSource: Authors’ estimate based on BIGD Municipal Survey

Table 19: Chi-square test statistic and P value from seemingly unrelated estimation

*** indicates that the IIA assumption is satisfiedSource: Authors’ estimate based on BIGD Municipal Survey

Hausman test results provide evidence (P value>0.05) of the non-rejection of null hypothesis that the IIA assumption is satisfied for both restricted equations (one excluding the Reluctant category and the other excluding the Oppose category) in case of water supply and street lighting services (Table 18). However, the test turned out to be undefined for the SWM service, which is not surprising; a negative chi-square value occurs when the variance matrix doesn’t meet the asymptotic assumption of Hausman test.

We now present the test results from the seemingly unrelated estimation, which is designed to overcome the shortcomings of Hausman test. In SWM service, the P values of the respective chi-square test statistics (P value>0.05) from the seemingly unrelated estimation of both restrictions, indicate that we cannot reject the null hypothesis that the IIA assumption is satisfied (Table 19). The same is true for street lighting; high P values (P value>0.05) associated with both restrictions confirm the validity of the IIA assumption. Although satisfied according to the Hausman test, the IIA assumption is violated in case of water supply when the Reluctant category is restricted from the model (P value<0.05). This result suggests that the probability of a household of falling in either WTP or Oppose category might be affected by the exclusion of the Reluctant category according to the seemingly unrelated estimation. However, the IIA assumption is satisfied when the Oppose category is restricted.

7. Test of Independence of Irrelevant Alternatives (IIA) Assumption


The paper set out to assess the state of public service delivery by municipalities in Bangladesh, the related issue of human resources available to manage and deliver the services to households and, finally, address the central concern of the paper, i.e., how different factors influence peoples’ payment attitude towards public services in the municipalities in Bangladesh.

The paper found a large number of vacant posts in all classes of municipalities, A, B and C, with the situation, shortage of staff (actual versus sanctioned posts), being most acute in municipalities of class C. In these municipalities, the percentage of vacant posts are 67%, 65%, 59% and 58% respectively, in health, water, tax and audit sections. Further, 200 out of 319 municipalities are operating without any audit staff, 148 do not have any water service staff, 42 lack staff in SWM services and 31 lack health services staff. Consequently, there are 1756 service users for every available staff in class A municipalities, 916 users per staff in class B municipalities and 1347 users per staff in class C municipalities. Although class A municipalities would be expected to have the best user: staff ratio, this is clearly not the case. This rather confusing staffing pattern mainly reflects the criteria used to classify the municipalities. A large number of municipalities still do not have any service coverage. In other words, they are municipalities in law but the necessary operating resources are not yet available to them. Even in the municipalities which provide public services, significant areas are still outside the reach of these services. For example, 61 out of 87 municipalities in class C are outside water supply service coverage area which translates to about 70 percent of the area of the municipality (which do provide services) is left out of water supply coverage. The road infrastructure in the municipalities is also in a poor state of repair, with a large proportion being unmetalled, non-all weather roads. Given poor state of public service delivery and road infrastructure, households are unwilling to pay the legally mandated charges for these services. In other words, households have an incentive to evade taxes and service charges and the municipal councils’ local revenue collection would be low. The survey data reveals that, on average, households evade more than 85 percent of the taxes that they are liable to pay. This, coupled with an inadequate budgetary allocations from the government, constrains both the quantity and quality of public services creating a vicious cycle which risks becoming endemic to the service delivery funding and delivery system in Bangladesh.

This paper’s analysis of households’ behavior in paying for better public services indicates that higher rates of service charge increase the probability of ‘Opposing’ the payment in case of water supply and street lighting. However, service charges were found to be statistically insignificant in explaining the willingness to pay in case of SWM. Households with higher income were found to be more willing to pay for improved SWM service, street lighting. Households with older household head had greater probability of being Reluctant to the higher payment for water supply service and SWM but were less likely to Oppose the higher payment. Higher level of education of the household head was found to reduce the probability of the household of being Reluctant to pay and ‘Opposing’ the payment. Residents in class A and B were less likely to be Reluctant and to Oppose the payment for all three public services.

The estimates above (Table 14-16) indicate a marked difference in the demand for public services between households living inside and outside the coverage area. Households in the covered area were found to be more unwilling (fall in Reluctant category) to pay for water supply and street lighting compared with their uncovered counterparts. However, the reverse is found in case of SWM; households in the covered area value the services more (less unwilling to pay) than the counterparts

8. Summary and Conclusion


in the uncovered. The average predicted probabilities for willingness to pay (table 17) reinforce this finding. The result probably reflects better SWM service relative to the other services. As shown in table 1 and table 2, SWM is the only service that employs more staff compared with the sanctioned posts for this service and almost 70 percent of the solid waste generated daily is disposed timely (table 9). Moreover, our analysis shows that for all the three services, more residents in class A and B municipalities are willing to pay for better service delivery compared to residents in class C (Table 13).

In light of the findings in this paper, we recommend extension to full coverage of services in all municipalities but, importantly, with improved quality of the public services. Apart from increasing the quality of life of households living in the municipalities which will become increasingly important in the future with rapidly growing urbanization, better quality of public services is vital to enhancing public trust in local government. Turning to financing of municipal services, our findings clearly show that households are willing to pay should better quality public services be provided. This calls into question the widely held view that residents of municipalities will not pay for public services and hence, that the government must bear the entire burden of providing for such services. While, it may be difficult to ‘roll back’ the current system of government grants in the short term, the results suggest that the government may do so in phases over the medium to long term (5-7 year period) but must ensure that the municipalities provide better services. Initially, the public service delivery policy might to cover operating costs through a modest service charge (price)but then, move to full cost recovery, i.e., towards self-financing of public services based on greater market orientation. With greater fiscal sustainability at the local level, the government can reallocate resources towards improving both the coverage and quality of road infrastructure in the municipalities which are in a poor state, as shown in section 3 of the paper.

Authors are grateful to Dr. Mohammad Yunus, Senior Research Fellow, Bangladesh Institute of Development Study (BIDS) and Dr. Wahid Abdallah, Research fellow, BRAC Institute of Governance and Development (BIGD), BRAC University for their suggestions regarding the model and other comments on the methodology. The authors also wish to thank an anonymous reviewer for his valuable comments on the paper.

i


ReferencesAddai, K. N., &Danso-Abbeam, G. (2014). Determinants of willingness to pay for improved solid waste management in Dunkwa-on-Offin, Ghana. Journal of Agriculture and Environmental Sciences, 3(1), 01-09.Adepoju, A. A., &Salimonu, K. K. (2010). Household willingness to pay for improved solid waste management in Osun State, Nigeria. Plenary Paper Session I: Water, 51.Anjum, R. (2013). Willingness to pay for solid waste management services: A case study of Islamabad (No. 2013: 03). Pakistan Institute of Development Economics.Awunyo-Vitor, D., Ishak, S., & SeiduJasaw, G. (2013). Urban households' willingness to pay for improved solid waste disposal services in Kumasi metropolis, Ghana. Urban Studies Research, 2013.Beaumais, O., Prunetti, D., Casabianca, A., & Pieri, X. (2015). Improving Solid Waste Management in the Island of Beauty (Corsica): a Latent-Class Rank-Ordered Logit Approach with Observed Heterogenous Ranking Capabilities. Revue d'économiepolitique, 125(2), 209-231.Brox, J. A., Kumar, R. C., & Stollery, K. R. (1996). Willingness to pay for water quality and supply enhancements in the Grand river watershed. Canadian Water Resources Journal, 21(3), 275-288.Cameron, A. C., & Trivedi, P. K. (2005). Microeconometrics: methods and applications. Cambridge university press.Hagos, D., Mekonnen, A., & Gebreegziabher, Z. (2012). Households’ willingness to pay for improved urban waste management in Mekelle City, Ethiopia (No. dp-12-06-efd).Hausman, J., & McFadden, D. (1984). Specification tests for the multinomial logit model. Econometrica: Journal of the Econometric Society, 1219-1240.Hazra, T., Goel, S., & Maitra, B. (2015). Willingness-to-pay and preference heterogeneity for service attributes of solid waste management: an experience in Kolkata, India. Global NEST Journal, 17(1), 82-92.Hazra, T., Goel, S., & Maitra, B. (2013). Willingness-to-pay for solid waste management service attributes: Kolkata Municipal Corporation area, India, as a case study. International Journal of Environment and Waste Management, 12(4), 406-421.http://www.lged.gov.bd/UnitAbout.aspx?UnitID=10Ku, S. J., Yoo, S. H., & Kwak, S. J. (2009). Willingness to pay for improving the residential waste disposal system in Korea: A choice experiment study. Environmental management, 44(2), 278-287Mary, O., & Adelayo, A. (2014). Household Willingness to pay for Improved Solid Waste Management in Akinyele Local Government Area. Journal of Biology, Agriculture and Healthcare , Vol.4, No.18.Mustafa, U., Ahmad, I., &ulHaq, M. (2014). Capturing Willingness to Pay and Its Determinants for Improved Solid Waste Management. Pakistan Institute of Development Economics.Rollins, K., Zachariah, O., Frehs, J., & Tate, D. (1997). Resource valuation and public policy: consumers’ willingness to pay for improving water servicing infrastructure. Canadian Water Resources Journal, 22(2), 185-195.Seth, K., Cobbina, S. J., Asare, W., & Duwiejuah, A. B. (2014). Household Demand and Willingness to Pay for Solid. American Journal of Environmental Protection , 74-78.Tussupova, K., Berndtsson, R., Bramryd, T., &Beisenova, R. (2015). Investigating willingness to pay to improve water supply services: Application of contingent valuation method. Water, 7(6), 3024-3039.Willis, K., Powe, N., &Garrod, G. (2003). The value of improved street lighting in rural areas. Final report for Department of Transport. University of Newcastle.Yunus, M., & Rahman, S. H. (2015). Municipal Finance in Bangladesh: Untapped Potential or White Elephant? Dhaka: BRAC Institute of Governance and Development (BIGD), BRAC University.

BRAC Institute of Governance and Development (BIGD)BRAC UniversitySK Centre (Basement, 3rd - 7th & 9th Floor), GP, JA-4, TB Gate Mohakhali, Dhaka 1212, Bangladesh

facebook.com/BIGDBRACUniversity twitter.com/BIGD_BRACU BIGD, BRAC Universityhttp://bigd.bracu.ac.bd

BRAC Institute of Governance and Development (BIGD), BRAC University is a centre of policy and academic excellence. BIGD is devoted to research on a range of governance and development concerns, as well as the nexus between the two. BIGD’s research is aimed at contributing to public policy as well as supporting its academic and training programs. It does not limit knowledge creation to being an end itself but rather to advance the pursuit of a just and prosperous society. It also plays an advocacy role to give voice to contemporary governance, political and economic issues. As an institute of BRAC University, BIGD works closely with the university motivated by its motto: ‘inspiring excellence’. BIGD also has the unique advantage of being associated with BRAC, the world’s largest NGO through its work. It is inspired by BRAC’s values in shaping its institutional work and professional standards.

payment attitudes to public services in the municipalities of …bigd.bracu.ac.bd/images/for...

Documents