do financial services build disaster resilience? · do financial services build disaster...

DO FINANCIAL SERVICES BUILD DISASTER RESILIENCE? Examining the Determinants of Recovery from Typhoon Yolanda in the Philippines

ABSTRACT

In November 2013, the Philippines was devastated by Typhoon Yolanda. In response, Mercy Corps initiated programming aimed at speeding affected households’ economic recovery and promoting their resilience to future natural disasters. The interventions were designed with the assumption that access to formal financial products, such as savings accounts, as well as the provision of cash assistance would increase households’ resilience to future shocks. This study analyzed survey data from households in Western Leyte, Philippines to test this and other common assumptions regarding what characteristics contribute to disaster resilience. Specifically, it tests the extent to which formal financial products, diversity of income sources, social capital, and other sources of support are linked to households’ recovery from effects of the typhoon and their perceived ability to cope with similar disasters in the future.

The analysis found that use of savings among households is positively associated with greater recovery from Yolanda, while both savings and loans are positively related to families’ perceived ability to manage future shocks. The results suggest that informal financial tools are as effective as formal ones in supporting disaster resilience. Both informal assistance from neighbors and formal government aid both are also positively linked to households’ recovery. Livelihood diversification is not shown to be linked to greater resilience or recovery, but this may reflect that diversification often occurs due to economic necessity rather than as intentional risk management in preparation for disasters. The findings shed new light on the potential effectiveness of increasing access to formal financial services, supporting livelihood diversification, and enhancing bonding social capital in contributing to household-level resilience to natural disasters.

Do Financial Services Build Disaster Resilience? | MERCY CORPS i

Mercy Corps Working Paper | Dan Hudner and Jon Kurtz

Do Financial Services Build Disaster Resilience? | MERCY CORPS ii

TABLE OF CONTENTS

ABSTRACT i

1. INTRODUCTION 1

2. RESEARCH QUESTIONS 2

Financial Services 2

Livelihood Diversification and Independence 2

Social Capital and other Forms of Assistance 2

3. METHODOLOGY 3

Sample Selection 3

Dependent Variables: Resilience Outcomes 4

Independent Variables 4

Analytical Methods 5

4. DESCRIPTIVE STATISTICS 5

Dependent Variables 5

Independent Variables 7

5. REGRESSION RESULTS 9

Determinants of Financial Instrument Use 9

Determinants of Recovery and Resilience 11

Financial Services 12

Income Diversification 13

Social Support and Other Sources of Aid 13

Other Factors 14

6. CONCLUSIONS 15

Implications for Programming 15

Sources of Bias 15

Study Limitations 16

Future Research 16

Do Financial Services Build Disaster Resilience? | MERCY CORPS iii

APPENDIX

APPENDIX A 18 Formulation of the Perceived Economic Resilience Index 18

APPENDIX B 19 Formulation of the Coping Strategies Index (CSI) 19

APPENDIX C 20 Definitions of Formal and Informal Financial Instruments 20

APPENDIX D 21 Full Regression Models 21

LIST OF FIGURES

Figure 1: Conceptual Framework 3

Figure 2: Self-reported Recovery from Yolanda 5

Figure 3: Self-reported Ability to Cope with Future Disasters 5

Figure 4: Perceived Economic Resilience Index 6

Figure 5: Coping Strategies Index 6

Figure 6: Use of Financial Instruments Prior to Yolanda 7

Figure 7: Sources of Income by Economic Activity 7

Figure 8: Sources of Relief and Recovery Support Post-Yolanda 8

Figure 9: Type of Aid Received Post-Yolanda by Source 8

Figure 10: Severity of Storm Damage by Municipality 9

Figure 11: Logistic Regressions Results on Predictors of Use of Financial Instruments 10

Figure 12: Regression Results on Relationships between Households Characteristics 11

and Resilience Outcomes

Do Financial Services Build Disaster Resilience? | MERCY CORPS 1

1. INTRODUCTION

Located on the typhoon belt of the Pacific Ocean, the Philippines is struck by approximately 20 typhoons every year. As of 2012, more than 30% of the population was engaged in the agriculture sector, making their livelihoods particularly vulnerable to the effects of these frequent typhoons12. In Eastern Visayas, the region where Leyte is located, and where this study was conducted, 70% of cities and municipalities had no commercial banking presence at the end of 2011.3 However, mobile penetration is extensive throughout the country, with an estimated 100 million SIM cards for the population of 105 million.4

Within this context, in November 2013 Typhoon Yolanda (also known as Typhoon Haiyan locally) struck the Visayas region of the Philippines. The deadliest recorded storm in Philippine history, Yolanda killed 6,500 and affected 14 million, displacing more than 4 million. Farmers in the Leyte region were struck especially hard.

In response, Mercy Corps initiated programming aimed at speeding affected households’ economic recovery and promoting their resilience to future natural disasters. For the purposes of the program and this study, resilience is defined as “the capacity that ensures adverse stressors and shocks do not have long-lasting adverse development consequences.”5 Using this definition, resilient households may suffer property damage and other losses to natural disasters, but in the aftermath will be able to meet their needs of food and shelter without sacrificing investments in productive assets, education, or their own health. As a result, they will be able to recover assets and restore their livelihoods more quickly, without extensive humanitarian intervention.

As part of its recovery programming, Mercy Corps has leveraged its partnership with BPI Globe BanKO, the only mobile-based microfinance bank in the Philippines to deliver initial unconditional electronic cash transfers and access to formal savings accounts and other financial services to the most severely affected families. BanKO also has plans to offer a suite of other financial services, including loans and insurance products, to this population in 2015.

This research is intended to examine the main assumptions underlying Mercy Corps’ programs – and similar interventions conducted by actors – aimed at strengthening households’ and communities’ resilience to recurrent natural disasters. The analysis uses household data collected during the program baseline to estimate the relationships between resilience and use of financial instruments, livelihood diversity, social capital, and sources of aid. The study takes advantage of pre-existing variation in these factors prior to the beginning of Mercy Corps’ program activities. Resilience is demonstrated by recovery from the effects of Yolanda and perceived ability to handle future shocks and disasters. The findings shed new light on the potential effectiveness of increasing access to financial services, supporting livelihood diversification, and enhancing bonding social capital in contributing to household-level resilience to natural disasters. This initial analysis will be augmented through additional studies, including a randomized control trial evaluating the effects of Mercy Corps’ program on households’ recovery.

1 Philippine Statistics Authority, 20122 Bangko Sentral Ng Pilipinas, “Report on the State of Financial Inclusion in the Philippines”, 20113 Mobile & Online, http://mobileandonline.com/mobile-marketing/smartphone-users-in-the-philippines/4 UNOCHA, 20145 Constas et al., Resilience Measurement Principles: Toward an Agenda for Measurement Design, 2014


II. RESEARCH QUESTIONSFinancial Services

The connection between access to vehicles for savings and credit and household income has been well-established.6 Studies have also shown that increased rates of saving allow small business-owners to maintain their asset stocks and consumption levels in the face of small-scale shocks such as illness.7 Intuitively, it seems clear that access to cash, whether savings or credit, would help a household quickly repair or replace their belongings damaged during a natural disaster. However, the source of funds may matter – predatory moneylenders may exploit their borrowers, and informal savings may not be as secure as bank accounts. The disorder in the period following a disaster would exacerbate these hazards, impacting families’ ability to meet their basic needs during the aftermath. Alternatively, in situations when a whole community is affected, the terms of informal credit may be loosened to provide respite for families which suffered the most harm.8

This study examines the relationship between households’ use of financial products and resilience to or recovery from shocks, focusing on the roles of savings, loans, and insurance. The driving questions are:

• Are households with greater financial capability more likely to use formal financial services?• Does use of financial services bolster household resilience to natural disasters?• Which financial products and services are linked to more successful recovery? Is the benefit imited to formal services?

Livelihood Diversification and Independence

Among poor and vulnerable populations, it is common to engage in several income-generating activities rather than specializing in a single trade. This practice is often a result of limited working capital and local market constraints, and tends to be associated with lower levels of income. However, in the case of severe natural catastrophes and other major shocks, diversification may protect households’ livelihoods if one of the income sources is impaired or lost. In particular, households with more independent income sources – which draw income from different economic sectors, such as both salaried work and farming – may be most likely to have at least one source which remains stable in the wake of a disaster. This study seeks to test this assumption by examining the question:

• Does diversifying sources of income across economic sectors protect livelihoods from severe natural disasters?

Social Capital and other Forms of Assistance

In the wake of Yolanda, affected areas received relief and recovery assistance from the Government of the Philippines, foreign governments, and international non-governmental organizations (INGOs). Many households also received material support from their neighbors, friends, and families – an important form of bonding social capital. The contribution of social capital – the level of cohesion and mutual assistance among a group or community – to disaster recovery has been well-documented. In many disaster situations, neighbors and peers are faster to respond than emergency services9, and continue to provide support during the recovery period. However, the exact nature and effect of this informal support is not fully understood.

6 Dupas and Robinson, Savings Constraints and Microenterprise Development: Evidence from a Field Experiment in Kenya, January 2009 7 Burgess and Pande, Do Rural Banks Matter? Evidence from the Indian Social Banking Experiment, June 20058 Udry, Risk and Insurance in a Rural Credit Market: An Empirical Investigation in Northern Nigeria, July 19949 Aldrich and Meyer, Social Capital and Community Resilience, October 2014


To address this knowledge gap, this study seeks to address the questions:

• How does bonding social capital contribute to resilience to and recovery from natural disasters?• How do informal sources of assistance compare to formal aid in supporting disaster recovery?

III. METHODOLOGYSample Selection

Data used in this study were collected via a household survey administered to randomly selected households targeted by Mercy Corps’ economic recovery program. The survey was conducted during between May-June 2014, immediately after the registration process and before project implementation began, covering 3,184 households in Northern Cebu, Eastern Leyte, and Western Leyte. A more comprehensive version of the survey was used in West Leyte, and only data from that region are examined in this study. The sample size for the baseline survey in Western Leyte is 1,751.

LIVELIHOOD DIVERSITY• Number of

income sources • Independence of

economic activities

SHOCK EXPOSURE

• Severity of storm damage

OUTCOMES• Food security

• Self-reported recovery from shock

• Preceived ability to cope with

future shocks

FINANCIAL SERVICES

• Use of savings (formal and informal)

• Use of loans (formal and informal)

• Use of insurance

• Financial literacy

SOCIAL SUPPORT AND OTHER AID

Reciept of assistance from: • Other households

• Govermnent of Philippines • Foreign sources (including INGOs)

FIGURE 1 presents the conceptual framework of the factors examined in this study and their presumed effects on mitigating the impact of the typhoon on affected households.


The key criterion for inclusion in the Mercy Corps program was the extent of damage a household suffered from Typhoon Yolanda. Only households whose homes were considered to have been “totally damaged” by the Department of Social Welfare and Development (DSWD) were eligible to enroll in the program. As such, the data reflect the population most heavily affected by the typhoon. Within this group, there was still variation in their exposure to the effects of the storm, which was included in the analysis. The estimates of storm severity used were obtained from www.mapaction.org, which drew on “a multivariate formula incorporating physical factors, (estimated storm surge, proximity to storm path, etc.) reported affected population stats and baseline vulnerability indicators.”10

Dependent Variables: Resilience Outcomes

The survey included questions capturing data on households’ levels of resilience, to be included in the study as dependent variables. As resilience is a multi-faceted concept, the study employed multiple measures to estimate households’ resilience. The outcomes measures for resilience fall into two categories: predictive and demonstrated. Predictive resilience is based upon the respondents’ anticipated ability to cope with shocks in the future.11 Demonstrated resilience reflects households’ actual and perceived recovery from Yolanda.

The first measure of predictive resilience is an index consisting of questions on how well households feel they could cope with a range of future household level shocks, such as illness or death of the primary income-earner. The index ranges from 10 to 50 with higher values indicating more expected ability to cope without compromising economic status or living conditions. A full list of the questions included in the perceived economic resilience index is found in Appendix A. The second predictive indicator is a straightforward inquiry on whether respondents believed they could successfully cope with a future natural disaster similar to Yolanda. Only 1% of households believed they could cope without any difficulty, so this study focuses on the difference between households who reported that they could cope with some changes to economic status and living conditions, versus those who said they could not cope at all with a natural disaster of the magnitude of Yolanda.

The first indicator for demonstrated resilience is a standard measure of food insecurity, the Coping Strategies Index12 (CSI). CSI is a weighted measurement of food consumption-related strategies used by the household to respond to income and assets lost to the typhoon, such as skipping meals, reducing portions, and scavenging. For this study, the CSI was scaled from 0, indicating daily use of many distressful coping strategies, to 108, indicating no use of distressful strategies. A full list of the questions used to calculate the CSI, including their weights, is found in Appendix B. The second measure for demonstrated resilience is subjective. Respondents were asked the extent to which they had recovered from the effects of Yolanda. As very few households reported full recovery, the analysis compares households which reporting having recovered somewhat or totally against those who said they had not recovered at all.

Independent Variables

To assess levels of access to financial services, respondents were asked about their use of financial instruments prior to Typhoon Yolanda, including informal and formal savings, loans, insurance, and bank accounts. A four-question quiz of financial knowledge was used to construct a financial capability score for each respondent. Other questions addressed respondent household’s primary income-earning activities prior to Yolanda, and the sources and types of assistance they received after the storm. The survey also included questions on demographic characteristics of households, to be used as controls. The independent variables are described in more detail Descriptive Statistics section below.

10 http://www.mapaction.org/?option=com_mapcat&view=mapdetail&id=316611 The authors recognize that other studies have treated these factors as explanatory variables rather than outcomes. For example, Smith, Frankenberger, et al.

Baseline Results Report of the Resilience Impact Evaluation of the Pastoralist Areas Resilience Improvement through Market Expansion (PRIME) Project. USAID Feed the Future FEEDBACK, Washington, DC., 2014

12 Maxwell, et al, The Coping Strategies Index: Field Methods Manual, Second Edition,2008


Analytical Methods

To identify important determinants of resilience to Typhoon Yolanda, the explanatory factors were combined into a single model. This model controlled for the strength of the typhoon by municipality. as well as household demographics such as family size, income, and education. This was essential, as households who were better off or less severely hit by the storm would otherwise appear more resilient. Holding these variables constant in the model allowed the analysis to estimate the unique contribution to resilience of the use of financial services and the other factors examined among households that were similar prior to the typhoon in terms of poverty status and other socio-economic factors.

A second model allowed the effects of each factor to vary between more and less severely-struck areas, in the belief that some characteristics might be more beneficial in areas which suffered more damage. For example, a household which was not heavily affected by Typhoon Yolanda may only benefit slightly from property insurance, but if it were in the most severely affected areas insurance may greatly contribute to their recovery.

For the binary (1/0) outcome variables, logistic regression models were used. For the continuous indices, the logs of indices were taken to normalize the distribution, since the CSI was positively skewed and the index of economic resilience was negatively skewed. Robust OLS regressions were then used to assess correlation between the independent variables and the log-indices.

Additionally, propensity score matching (PSM) was used to estimate the causal impact of financial instruments on a household’s resilience. This method predicted the probability of using each financial instrument – savings, loans, and insurance – based on financial literacy, education, and other household characteristics. A comparison was made between households which did and did not use the instruments, but which were equally likely to have used them based on the prediction. The difference between the two on the resilience outcomes is therefore considered attributable to the use of the financial tool.

IV. DESCRIPTIVE STATISTICSDependent Variables

The data (Figure 2) paints a picture of communities which are on the road to recovery, but still suffering from the effects of Yolanda. Most respondents reported having recovered to some extent, while just over a third had not yet recovered at all. However, less than 4% reported having recovered to pre-typhoon levels.

FIGURE 2 shows the variation in self-reported, recovery from Yolanda

Percentage of Households

RECOVERED AND IMPROVED

RECOVERED TO PRE-YOLANDALEVEL

SOME RECOVERY

NO RECOVERY34.4%

61.9%

2.5%1.1%

FIGURE 3 shows the variation in self-reported ability to cope with future disasters

Percentage of Households

UNABLE TO COPE

ABLE TO COPEWITHOUTDIFFICULTY

ABLE TO COPE WITH CHANGES TO INCOME AND FOOD

46.7%

50.5%

2.3%


Just over half of households (Figure 3) felt that they could cope with future disasters with some changes to their income and food sources, while under half felt that they would be unable to cope with future disasters. Just over 1% believed they could cope without any difficulty. For the analysis, this small minority was combined with those who felt able to cope with some changes.

The index of perceived economic resilience (Figure 4) is a continuous variable, with a range from 10 to 50 and a mean of 19.8. As lower values indicate less perceived economic resilience, this downward skew shows that most respondents predicted that their economic status and living conditions would suffer significantly from future household-level shocks, such as the death or illness of the primary income earner. This is unsurprising, but the variation in responses is interesting. About 20% of households felt that they would suffer substantially from all of the listed shocks, but a few with scores of 50 felt that they would not suffer at all from any of the shocks.

The Coping Strategies Index (Figure 5) is also a continuous variable, with a range among the sample from 27 to 108 and a mean of 95.0. The scores are skewed towards higher possible values, indicating that most households were fairly food secure at the time of the survey. One in eight households had a full score of 108, meaning that they did not make use of any distressful coping strategies such as begging or skipping meals in the weeks preceding the survey.

0%

5%

10%

15%

20%

25%

% o

f Hou

seho

lds

with

Sco

re

10 15 20 25 30 35 40 45 50

Score on Perceived Economic Resilience Index

FIGURE 4 shows the percentage of respondents who scored at each value on the index of perceived economic resilience

0%

2%

4%

6%

8%

10%

12%

14%

% o

f Hou

seho

lds

with

Sco

re

27 50 55 60 65 70 75 80 85 90 95 100 105 108

Score on Coping Strategies Index

FIGURE 5 provides the percentage of respondents who scored at each value on the Coping Strategies Index


Independent Variables

The survey (Figure 6) reveals some use of informal financial tools, but a low level of engagement with the formal financial sector. The criteria for determining formal and informal tools are given in Appendix C. This lack of naturally-occurring variation may cause difficulty in assessing the effect of formal services on resilience and recovery. However, sizeable minorities of the affected population did make use of both informal loans and insurance.

Over a quarter of households had diversified income sources (Figure 7), meaning they earned income through engagement in two or more independent economic sectors. Engaging in multiple income-generating activities within the same sector was not considered to be diverse, as all activities within the sector would likely be affected in similar ways by the typhoon. More than half of all households earned income through agriculture, the most common income-generating activity. One in three earned income through daily labor, while other sectors were less prevalent.

0

10

20

30

40

50

60

% o

f Hou

seho

lds

That

Use

d In

stru

men

ts

Insurance Loans (Informal)

Savings (Informal)

Loans (Formal)

BankAccount

Savings(Formal)

FIGURE 6 displays the percentage of households which made use of various financial instruments prior to Typhoon Yolanda

0

10

20

30

40

50

60

% o

f Pop

ulat

ion

Agriculture Labor Commerce Salary Driver Remittances

FIGURE 7 shows the percentage of households earning income from different economic activities


Almost 40% of households received in-formal social support from their neighbors – including shelter, food, and cash – (Figure 8), while more than 90% received aid from the Government of the Philippines and over half received aid from a foreign source such as international NGOs and other governments.

The nature of aid received differed substantially between sources (Figure 9). Assistance from neighbors and the local community consisting largely of meals, temporary lodging, and rebuilding. Households which received aid from the national government were very likely to receive food, and some were also given non-food items (NFIs) such as tarps and blankets. Assistance from foreign sources also primarily included food and NFIs, but fewer households received foreign assistance than received help from the national government.

Severity was divided into ten categories, indicated by the colors on the map. (Figure 10). This study collapsed the categories into a binary variable, with municipalities in the four most heavily affected of the ten groups considered to have suffered severe damage. 52.59% of all respondent households lived within severely damaged municipalities.

0

20

40

60

80

100

% o

f H

ou

seh

old

s Th

at R

ecei

ved

Su

pp

ort

GovernmentAid

ForeignAid

SocialSupport

FIGURE 8 gives the proportion of households which received relief and recovery assistance from different sources in the 6 months after Typhoon Yolanda

0

20

40

60

80

100

% o

f Hou

seho

lds

Rec

eivi

ng A

id

Foods/Meals

Housing/Hosting

Non-FoodItems

Rebuilding Money

SOCIALSUPPORT

NATIONALAID

FOREIGNAID

FIGURE 9 shows the type of aid received by households from each source


V. REGRESSION RESULTS Determinants of Financial Instrument Use

The table below (Figure 11) shows the effects of financial capability and other household characteristics, in the use of financial instruments. Only relationships which are statistically significant are reported. The coefficients are in the form of odds-ratios, meaning that a one-unit change in the independent variable is associated with an “X times” likelihood of using the given financial instrument. These coefficients can also be expressed as a percentage change in the likelihood of using each instrument. For example, a respondent who scored one point higher on the financial capability test was 1.46 times as likely, or 46% more likely, to use formal savings compared to a similar respondent who scored one point lower. Coefficients less than 1 indicate that the respondent was less likely to use the financial instrument, meaning the relationship is negative. Male financial decision makers, for example were only 0.51 times as likely to use informal savings as female financial decision makers, holding all else equal. Dark blue cells show positive relationships, while light orange cells show negative relationships.

FIGURE 10 shows severity of storm damage by municipality, as calculated by www.mapaction.org

Severity score was generated from amultivariate formula incorporatingphysical factors (est storm surge,proximity to storm path, etc) reportedaffected population stats and baselinevulnerability indicators. Data originallycompiled by Miguel Antonio Garciafrom the Barcelona Graduate Schoolof Economics and then geo-referenced by UN-OCHA, MSF-UK,and WFP. The operational presenceoverlay uses the proxy of number oforganistions reported as active,through the 3W matrix as at 30November.

The depiction and use of boundaries,names and associated data shownhere do not imply endorsement oracceptance by MapAction.

!(

!(

!(

!(!(

!(

!(

!(

!( !(

!(

!(

!(

!(

!( !(

!(

!(

!(

!(!(

!(

!(

!(!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(

!(!(

Hinunangan

Anahawan

Bontoc

HinundayanLibagon

LiloanMaasin

City

Tabogon

Tudela

Almeria

BiliranCabucgayan

Caibiran

Culaba

Kawayan

Maripipi

Naval

BienUnido

Jetafe

Pres.Carlos P.GarciaTalibon

BogoCity

Borbon

Carmen

Catmon

Compostela

Consolacion

Cordoba

Daanbantayan

DanaoCity

DanaoLake

Lapu-LapuCity

Medellin

Pilar

Poro

SanFrancisco

SantaFe

Libjo

Loreto

Tubajon

Balangiga

Balangkayan

BoronganCity

GeneralMacarthur

Giporlos

Guiuan

Hernani

Lawaan

Llorente

Maydolong

Mercedes

Quinapondan Salcedo

SanJulian

SulatTaft

Abuyog

Alangalang

Albuera

Babatngon

Barugo

Bato

BaybayCity

Burauen

Calubian

Capoocan Carigara

Dagami

Dulag

Hilongos

Hindang

Inopacan

Isabel

Jaro

Javier

Julita

Kananga

La Paz

Leyte

Macarthur

Mahaplag

Matag-Ob

Matalom

MayorgaMerida

OrmocCity

Palo

Palompon

Pastrana

SanIsidro

SanMiguel

Tabango

Tabontabon

TaclobanCity

Tanauan

Tolosa

Tunga

Villaba

EsperanzaPio V.

Corpuz

Basey

Calbiga

CatbaloganCity

Daram

Hinabangan

Jiabong

Marabut

Motiong Paranas

Pinabacdao

SanSebastian

SantaRita

TalaloraVillareal

Zumarraga

Malitbog

SaintBernard

San Juan

Silago

Sogod

TomasOppus

´MA1003_V01

Produced byMapActionwww.mapaction.org

Data sources

0 5 10 15 20 25

kilometres

Created

Map Document

Projection / Datum

Glide Number

02 Dec 2013 / 16:00

MA1003_Severity+3WEVisayas

WGS 1984 UTM Zone 51N

TC-2013-000139-PHL

UTC+08:00

Philippines:Typhoon Yolanda -

Severity and OperationalPresence,

Eastern Visayas(as at 30 Nov 2013)

Severity scores: see above3W data: Clusters/UNOCHABoundaries: GADM

Philippine Sea

Supported by

OSOCC

Sub OSOCC

Storm track

Number of orgs(3W matrix 30 Nov)

!( 10

!( 100

Ranked Severity

Most severe

Least severe

Not ranked

Data for Guiuan includes island of Homonhon


FIGURE 11 Logistic Regressions Results on Predictors of Use of Financial Instruments

Use of Savings Use of Loans Have BankAccount

HaveInsuranceFormal Informal Formal Informal

Higher Financial Capability 1.46 1.23 1.14

Higher Income (Log) 1.65 1.31 1.24 1.11 1.33 1.20

Increased Poverty Likelihood 0.12 0.24 0.08 0.41

More Productive Assets 1.01 1.01 1.01

Additional Family Member 1.14 0.88 1.15 1.16 1.13

Male Financial Decision Maker 0.51

More Years of Education 1.06 1.17

Constant 0.0003 0.18 0.01 0.11 0.0007 0.08

Sample Size 1566 1566 1580 1580 1585 1580

Pseudo R-Squared 0.16 0.05 0.08 0.01 0.15 0.04

Financial capability was measured through a series of questions on arithmetic, calculating interest earned, and business practices. When the household financial decision maker scored one point higher on these questions (out of four points), they were 1.46 times as likely to have used formal savings, 1.23 times as likely to have taken loans from formal sources, and 1.14 times as likely to have taken loans from informal sources. However, such households were no more likely to have opened a bank account, purchased insurance, or used informal sources to save.

Other personal and household characteristics were also related to higher use of financial tools. The strongest relationship was between income and formal savings: a 100% increase in a household’s income was associated with a 65% increase in the likelihood of using formal savings, a 31% increase in the chance of using informal savings, and 33% increase in the chance of having a bank account. The magnitude of the relationship was smaller for use of formal loans (24%), informal loans (11%), and insurance (20%). These findings likely reflect that households with higher incomes will also have more money available to save, and more incentive to insure their savings are protected in formal financial institutions. More productive assets were also associated with the use of savings and insurance. Poorer households were less likely to use formal instruments, but were not significantly less likely to have informal savings or loans.

Interestingly, larger families were more likely to use formal instruments, but were less likely to save informally. Each additional family member was associated with a higher probability that the household used formal savings (1.14 times), formal loans (1.15 times), a bank account (1.16 times) and insurance (1.13 times), but also a 12% decrease in the probability of having informal savings. An additional year of education on the part of the financial decision maker was surprisingly only associated with a 6% higher chance to have taken formal loans and a 17% higher chance to have a bank account, but was not related to use of savings or insurance. Male financial decision makers were 49% less likely to have used informal savings than female decision makers, suggesting significantly different saving and spending patterns between men and women.

Overall, the results from this model show support for the premise that building financial capability may be a viable way of increasing the use of formal financial instruments. However, wealth and income appear to be greater determinants than financial literacy, so this may not hold true for the poorest segments of the population.


Determinants of Recovery and Resilience

Figure 12 below shows the results of correlations between resilience and household characteristics including use of financial instruments, income diversification, sources of aid, and demographics. Only relationships which are statistically significant are reported here; the full results of the model can be found in Appendix D. The first two columns address predictive resilience, with Column 1 displaying the determinants’ effect on the index of perceived economic resilience, and Column 2 showing their effect on households’ perceived ability to cope with future natural disasters. The latter two columns cover demonstrated resilience, with Column 3 showing the determinants’ effect on the Coping Strategy Index (CSI) and Column 4 showing the effect on households’ reported recovery from Yolanda. Dark blue cells show positive correlations, meaning the factor in column A is correlated with greater resilience or recovery; light orange cells show negative correlations, indicating the opposite.

FIGURE 12 Regression Results on Relationships between Households Characteristics and Resilience Outcomes

Predictive Resilience Demonstrated Resilience

Perceived Economic

Resilience Index (Log)

Perceived Ability to Cope

with a Major Natural Disaster

Coping Strategies

Index (Log)

Reported Recovery

from Yolanda

Savings, Formal 3.3%P*

Savings, Informal 8.1%P 3.13P 1.9%P 2.12P

Loans, Formal 7.4% 1.70

Loans, Informal 9.3%P 1.31P*

Bank Account

Insurance -4.9% 0.73P

Financial Capability Score -1.1% 0.86

Diverse Income Sources -9.5% -2.4%

Relied on Community Support 2.08 -3.8%

Aid from Philippine Government 13.6% 3.9%

Aid from Foreign Source

Storm Severity

Total 2013 Income (Log) 2.2% 1.9% 1.18

Productive Assets 0.1%

Family Members -0.6%

Male Financial Decision Maker (FDM)

6.8%

FDM Years of Education 1.05 0.3% 1.05

FDM Literate

Household Owns Land 0.67

Poverty Likelihood -3.2%

Constant 2.60 0.41 4.37 0.40

Sample Size 1194 1353 1286 1359

Adjusted R-Squared 0.160 0.079 0.091 0.058

P Indicates that the results are supported by the propensity-score matching estimate, which was only used to estimate the effect of financial instruments.*This result was significant using propensity-score matching, but not using the OLS regression.


The coefficients for the log indices, Columns 1 and 3, are the percentage change in the index associated with a one-unit change in the independent variable. For example, households with informal savings scored, holding all else equal, 8.1% higher on the index of perceived economic resilience. The coefficients for the binary outcomes, Columns 2 and 4, are odds-ratios. Following the example used for Figure 11 above, households which used informal savings were 3.13 times as likely to say that they felt able to cope with future natural disasters than households without informal savings.

Financial Services

Use of certain financial services prior to Yolanda is positively linked with both predictive and demonstrated indicators of resilience. Households that saved informally – such as in their home – scored 8.1% higher on the index of perceived resilience to small-scale shocks compared to households without informal savings, and were 3.13 times as likely to feel that they could cope with natural disasters. Additionally, these households scored slightly higher (1.9%) on the index of food security, and were 2.12 times as likely report some recovery since the typhoon. The analysis controlled for pre-Yolanda poverty status, meaning that these results were not simply because the households were better-off prior to the storm. These effects are confirmed through the PSM analysis, and are intuitive: a store of readily available funds allows families to smooth over the damage of natural disasters and other shocks. These positive relationships also suggest that, in general, informal savings were resistant to the physical damage wrought by Yolanda, and households were still able to make use of them after the storm passed.

More surprisingly, households with formal savings in banks or credit unions did not have higher predictive or demonstrated resilience. This is most likely due to the relatively small proportion of families which made use of these types of savings, only about 5% of the sampled population. It may also reflect difficulty withdrawing money in the chaotic aftermath of the storm, or other pitfalls of relying on the formal banking sector. Affected households reported that long distances to commercial bank branches were cited as one of the major obstacles to their use of formal financial services.

Use of any loans was positively correlated with some measures of resilience, although the relationship was not as clear as that of informal savings. Families that took out loans from formal sources, like financial institutions, prior to Yolanda were 1.7 times as likely to feel able to cope with future disasters, and scored 7.4% higher on the index of perceived ability to handle household-level shocks. However, they were not significantly more likely to be more food secure or report greater recovery from Yolanda. Households with informal loans from sources such as employers, shops, and local moneylenders also had greater predictive resilience. These households scored 9.3% higher on the index of predicted ability to cope with shocks. The positive connection between informal loans and predicted coping was confirmed by the PSM analysis, and PSM showed a significant correlation with reported recovery from Yolanda as well.

The results on the use of loans are also intuitive: households with access to credit of any type felt better prepared to meet their emergency needs after a disaster, and felt more security knowing that they could obtain funds if needed.

Overall, having savings and loans through formal means, including bank accounts, did not appear superior to informal means in terms of supporting households’ resilience to Yolanda. In particular, there was no evidence of a benefit from formal savings accounts, whereas savings through informal means had strong beneficial effects on both predictive and demonstrated resilience.

Having insurance prior to Yolanda had mixed effects on households’ recovery and resilience. Households that reporting having any type of insurance scored 4.9% lower on the index measuring perceived ability to coping with household shocks. They were also 27% less likely to feel able to cope with future natural


disasters than households without any form of insurance. This finding was confirmed by the PSM analysis. However, in the most severely affected areas families which had insurance scored no lower on the perceived economic resilience index, and had slightly higher food security than families which were comparably affected and did not have insurance.

These mixed results may reflect the fact that in the Philippines, the sale of insurance is targeted to particularly vulnerable households who are likely to suffer the most from disasters and other shocks. These individual vulnerabilities, such as physical impairments, lack of family network, and personal health status, may not be accounted for within the survey data. The existing insurance market may also be limited to catastrophic loss, which would mean that households which suffered extensive damages benefit but those which did not lose many assets or property would gain little from insurance.

One puzzling result is the negative relationship between financial capability and demonstrated resilience. Holding use of financial instruments constant, a higher score on the financial literacy test is correlated with less likelihood to report some recovery since Yolanda and worse food security status. This bears further exploration in future studies. It may be that respondents with more financial knowledge are also better able to assess their status before and after Yolanda, and are more acutely aware of their assets lost in the storm.

Income Diversification

Households were considered to have diverse income sources if they earned money through activities in more than one economic sector, such as both agriculture and commerce. Households with diverse income sources appear to have been less resilient to Typhoon Yolanda than those which only drew income from a single sector. They scored 9.5% lower on the perceived economic resilience index, meaning they felt more susceptible to household-level shocks. They were also less food secure than other households, scoring 2.4% lower on the CSI.

These counterintuitive results may reflect that income diversification often occurs due to economic necessity rather than as intentional risk management in preparation for disasters. Poorer, more vulnerable households are known to diversify their income when the returns from each individual source are insufficient to provide for their needs, and often require all their sources to be intact in order to make ends meet. Even if one of the income sources is left intact after the typhoon, the household may be unable to meet their living costs with the reduced money. It is also possible Yolanda was so devastating that all economic sectors were severely affected, from direct damage to fields and fishing boats to reduced demand for services. An alternate model which tested the effect of salaried income sources – which should have remained stable following the typhoon – against all other forms of livelihoods found no difference between the two groups, reinforcing the theory that the effects of the typhoon rippled throughout the economy.

Social Support and Other Sources of Aid

Relief and recovery assistance provided after Yolanda was divided into three categories: support from other community members, aid provided by the Government of the Philippines, and aid through foreign sources such as international NGOs and other governments. Households which relied on others in their community for assistance were considered to have greater bonding social capital, and were 2.08 times as likely to feel able to cope with future disasters than households that did not receive this type of social support. However, these same households also scored 3.8% lower on the index of food security – meaning that they were more likely to use distressful strategies to obtain or preserve their food supplies following Yolanda.

These results suggest that community support increases households’ confidence in their ability to manage large shocks, but does not necessarily increase their prospects for actual recovery, at least in the short term. Based


on Aldrich’s research,1413help from other community members may have addressed the most urgent needs, such as rescue and shelter in the immediate aftermath of the typhoon. Respondents who benefited from this type of assistance would therefore feel less concerned about the worst consequences of a storm, including physical danger, starvation, and exposure. However, by the same token, these respondents may have been the hardest hit within a community, thus making them less food-secure than their neighbors. The analysis attempted to control for the severity of damage to households, which would reduce this bias. However, the measures used for this may not have been sufficiently sensitive to pick up on the variance within broad categories such as having “completely damaged” houses.

Households that received aid from the Philippines government were more food secure, scoring 3.9% higher on the index of food security than comparable families. This follows the finding that a large portion of government support was in the form of food aid. They also felt more confident in their ability to cope with small shocks, scoring 13.6% higher on the index of perceived economic resilience. This may be because their experience seeing the government support system in action increased their confidence in their chances of receiving emergency assistance in the future.

Unlike for aid from the Philippine government, having received assistance from international sources was not associated with greater household recovery or resilience. Speculatively, it may be that international assistance was not delivered in a timely manner or distributed effectively. The type of aid may also be relevant: Households mainly received non-food items from international sources, which may not have been as important for recipients’ recovery as other forms of assistance.

The correlations between sources of aid and resilience are expectedly mixed, given that aid is targeted at the families which have suffered most from the typhoon and are considered least able to recover on their own. Receiving aid would therefore be beneficial, but would only bring these severely-affected households up to the level of those around them. An accurate estimate of the results of each type of aid would require disentangling types of assistance, selection of beneficiaries, and speed of delivery. A cross-sectional analysis, like the one used in this study, cannot effectively break apart these different possibilities. Further research is needed before drawing stronger conclusions regarding the relative contributions of different sources of support to households’ recovery.

Other Factors

Certain socio-economic characteristics of households are clearly linked with resilience and recovery. Most notably, each additional year of education for the family’s financial decision maker is associated with 5% increase in the chance of feeling able to cope with future natural disasters, and an additional 5% chance to report some recovery from Yolanda. Each year of schooling is also correlated with a 0.3% increase on the CSI. The exact mechanism through which education affects resilience is not well understood, but it may reflect factors such as planning for the future, knowledge of disaster risk, and greater ability to access assistance.

Unsurprisingly, income is also a contributor to demonstrated resilience. Households which earned more were also more likely to have partially recovered from Yolanda, be more food-secure, and be more confident in their ability to handle future household-level shocks. Intuitively, more wealth helps a family to meet their food needs and rebuild their lost assets. For this reason, the analysis controlled for wealth status, education levels, and other socio-economic factors in the model. This enabled the results to pinpoint the unique contributions of use of financial products and services, livelihood diversity, and sources of aid on households’ recovery and resilience.

14 Aldrich and Meyer, Social Capital and Community Resilience, October 2014


VI. CONCLUSIONSImplications for Programming

The findings of this study reinforce the importance of financial instruments as mechanisms by which households can build their resilience to natural disasters. In particular, families with informal savings had higher predictive and demonstrated resilience than others, and those which had taken out loans before Yolanda were both more likely to feel that they could cope with a similar disaster in the future. However, there is not clear evidence for the benefits of formal financial tools over informal ones; informal savings appear more effective than formal, while the advantage of formal over informal credit is ambiguous. Development actors engaging in programming aimed at enhancing resilience should use the expansion of formal financial services as a means of increasing financial inclusion, but would be wise not to view informal financial outlets as inherently inferior. Programming aimed at boosting financial capability also appears to have high efficacy in encouraging a diverse range of formal and informal financial services.

Livelihood diversification has not been proven to be an effective means of building resilience to the effects of natural disaster as severe as Yolanda. However, since the study addresses only naturally occurring diversity, it may be that planned and intentional expansion of income sources for the purposes of spreading risk may be more effective. Based on the current evidence, development programs designed to support livelihood diversification should be cautious about assuming this will result in contribute to greater resilience to major natural disasters.

Greater social capital – as demonstrated by a households’ ability to rely on neighbors or others in their community during a time of need – has been shown to be positively correlated with predictive resilience, meaning households have increased confidence in their coping abilities. Agencies and INGO’s should consider programs which build the capacity of community support networks and social capital among vulnerable groups. These appear to serve valuable roles in protecting families from the most damaging effects of disasters, particularly in the short run. Humanitarian aid in the immediate aftermath of a major disaster might also take into account existing informal support systems and seek to complement them or use them as a vehicle to deliver assistance. For example, networks within the community may hold more information on the damage suffered, the immediate priorities in response and rebuilding, and the families which do not possess the resources to manage on their own. These networks could be well-placed to distribute aid or to direct government and international aid to the most vulnerable recipients. Community-based targeting has also been shown to be more cost-effective, and lead to greater satisfaction among beneficiary communities compared to more traditional targeting methodologies15. 14

Sources of Bias

This study has drawn upon data collected at the baseline of the Mercy Corps’ economic recovery project, and reflects pre-existing factors which may influence household resilience. As the data is cross-sectional, the results of the analysis show correlation rather than causation. One primary source of bias in this type of analysis is risk of reverse-causality, meaning the outcome variable may actually be causing a change in the explanatory variable. However, the survey did collect information on key socio-demographic factors prior to Yolanda, such as use of savings and loans before the storm, and time-invariant characteristics like gender and education. The flow of time from pre-storm status to Yolanda to post-storm recovery limits concerns about reverse causality, meaning the level of recovery almost certainly did not directly affect households’ actions before Yolanda.

15 Atlas, V, et al. Targeting the poor: evidence from a field experiment in Indonesia, 3ie Impact Evaluation Report 12. New Delhi: International Initiative for Impact Evaluation (3ie), 2014.


Omitted variable bias is another concern. This comes about when unobserved characteristics affect both the explanatory and the outcome variables. When unaccounted for in the analysis, omitted variables create the illusion of a relationship where none exists. In the context of this study, many personal characteristics such as risk preferences, optimism, and foresight, were not captured. All of these may affect households’ decisions to protect themselves before the storm by saving or purchasing insurance, and could also influence their speed of recovery after Yolanda. Similarly, community-level factors such as a well-functioning local government or market might allow households living in the vicinity to achieve faster recovery after a disaster.

Study Limitations

Some characteristics of interest, particularly use of formal financial tools, were not prevalent in the population studied, and their effects may be understated as a result. The marginal benefits could vary when promoted at a large scale through development interventions. This analysis should therefore be combined with program impact evaluations and qualitative research to fully understand how changes in use of financial services affects communities’ abilities to cope with shocks and stressors.

The timing of the baseline survey several months after Typhoon Yolanda presents both benefits and limitations to the measurement of recovery resilience. By being conducted 6 months after the typhoon struck, it was possible to assess patterns and determinants of recovery. However, it was potentially too long after the height of the disaster to capture the immediate effects. For example, the Coping Strategy Index was used to indicate recovery after the storm, but it is most sensitive in the days and weeks immediately after a disaster. By the time of the survey, the daily behavior of many households had begun to return to normal, and many had repaired the damage done to their homes and physical property. Since the data is cross-sectional, long-term indicators of recovery such as recuperation of assets and income are not yet available. As a result, the predictive markers of resilience are presumed to be more informative than the demonstrated indicators of resilience, i.e. households’ feelings about coping with future shocks better reflect their resilience than their current progress recovering from Yolanda. In spite of this limitation, the recent experience of Yolanda means that households were well-positioned give accurate estimates of their predictive resilience, being able to anticipate how well they would respond to similar shocks in the future.

As discussed in Section IV, the sample of respondents is drawn from beneficiaries of Mercy Corps’ program in Western Leyte. The results are those of a population heavily-affected by Yolanda living within a heavily-affected region, and may not be indicative of households which suffered only minor losses in the typhoon or other regions of the country.

Future Research

Future studies, particularly those associated with Mercy Corps-supported projects in the Philippines, should further explore whether the formal and informal financial products and services have different effects on resilience. Concerns among communities regarding formal services will be particularly important for programs aimed at promoting their use – for example, formal financial institutions may be viewed as overly rigid or inaccessible. The results also challenge the assumption that informal financial services, such as those provided by moneylenders, are exploitative. A more thorough understanding of the operations and possible benefits of informal financial tools would be of value to future program design.

Given that the data do not support livelihood diversification as a means of building resilience, both researchers and program implementers should seek examples of households which were more resilient through specific livelihood choices, and carefully assess why most families did not benefit from diversification. Such examination may help reveal the specific forms of livelihood diversification hold the greatest – and least – potential to strengthen resilience.


Additionally, social capital appears to be a crucial element for households coping with the aftermath of a disaster. Studies may seek to delve further into systems of support, and clarify the types of social connections which determine who benefits from this support. Future project design could tap into these networks as a means of bolstering community resilience, but there may be marginalized groups which do not benefit from the support of their neighbors. If so, identifying them would help direct humanitarian aid and development assistance to the most at-risk groups.

Mercy Corps and its partners are engaged in additional research around these and other outstanding questions. At the time of writing, Mercy Corps is conducting an impact evaluation of its economic recovery program in Western Leyte, Philippines. This evaluation will provide credible evidence on the causal effects of providing households cash transfers, financial literacy, and access to formal bank accounts on their recovery from Yolanda. Further data collection and analysis in Western Leyte is also planned in early 2015 to more robustly test if and how the factors examined in this study are important for households during the later recovery period.


APPENDIX A: Formulation of the Perceived Economic Resilience Index

The index of perceived economic resilience is created by combining responses on how severely a negative shock would affect the family’s financial status and living conditions. The shocks are below:

ShockIllness of primary income earnerDeath of primary income earner

Business/livelihood/crop performs poorlyLoss of formal/informal employment

Food price inflationLoss of remittance income

Loss of/damage to primary productive asset (fishing boat, vehicle, sewing machine, livestock, etc.)

Loss of/damage to household assets (appliances, furniture, etc.)Loss of/damage to dwelling

Natural disaster

Respondents chose from the same set of responses for each shock, which were scored from 1 to 5, with 1 indicating the most severe impact and 5 the least severe:

Score Level of Impact1 Severe (very large) impact2 Large impact3 Moderate impact 4 Small impact

5 Little to no impact


APPENDIX B: Formulation of the Coping Strategies Index (CSI)

The CSI is created through combining responses to the following questions regarding respondents’ actions taken to access food, which are weighted to reflect the level of distress associated with them:

In the past 2 weeks, how frequently did your household use one or more of the following strategies in order to have access to food?:

Activity WeightRely on less expensive or less preferred foods 1Limit/reduce meal portion sizes subsequent 1

Reduce number of meals eaten per day 2Skip entire day without eating 4

Reduce adult consumption so children can eat more 2Borrow food or rely on help from friends or relatives 1

Rely on begging for food 2Gather unusual types or amounts of wild food/hunt 2Consume seed stock to be saved for next season 2

Take children out of school to work 4

Respondents chose from the same set of responses for each activity, which were scored from 0 to 4, with 0 indicating most and 4 indicating least distressful coping:

Score Response0 Daily1 3 or more times per week2 1-2 times per week3 Less than once per week

4 Never


APPENDIX C: Definitions of Formal and Informal Financial Instruments

Savings tools are classified according to the following criteria:

Savings Instrument

Formal InformalBank Community Welfare Scheme

Credit Union Informal Savings ClubFormal Savings Association Inside Home

Sources of loans are classified according to the following criteria:

Loan Source

Formal InformalFormal Institution Employer

Microfinance Institution PawnshopDaily Bank

Community Welfare

Neighborhood Community

Store Credit

Local Moneylender


APPENDIX D: Full Regression Models

The table below contains the full regression outputs for the relationships between household characteristics and resilience. For each measure of resilience, Column (1) displays the results of the basic model, and Column (2) displays the results of the model which allowed for different effects of characteristics between more and less affected areas. T-statistics are given in parentheses below each coefficient, and significance is indicated by * for P<0.05, ** for P<0.01, and *** for P<0.001.

Predictive Resilience Demonstrated Resilience

Perceived Economic Resilience

Index (Log)

Perceived Ability to Cope with a Major Natural Disaster

Coping Strategies Index (Log)

Reported Recovery

from Yolanda

(1) (2) (1) (2) (1) (2) (1) (2)

Savings, Formal 0.0208 0.0162 1.277 1.085 0.330 0.0138 1.316 1.189

(0.50) (0.24) (0.77) (0.14) (1.89) (0.48) (0.73) (0.30)

Severe X Formal Savings 0.0397 1.203 0.0281 1.189

(0.46) (0.27) (0.77) (0.22)

Savings, Informal 0.0808*** 0.0418 3.128*** 3.116*** 0.0192* 0.0192* 2.118*** 2,382***

(3.79) (1.57) (6.94) (5.44) (2.23) (1.67) (4.07) (3.64)

Severe X Informal Savings 0.0913* 0.993 -0.000393 0.737

(2.12) (-0.02) (-0.02) (-0.82)

Loans, Formal 0.0736* 0.0524 1.703* 2.289* -0.300 -0.0162 1.530*** 3.144***

(2.17) (1.05) (2.45) (2.29) (-1.93) (-0.60) (1.72) (2.67)

Severe X Formal Loans 0.0382 0.619 -0.0216 0.313*

(0.57) (-1.06) (-0.66) (-2.22)

Loans, Informal 0.0930*** 0.0806*** 1.072 1.136 -0.00424 0.00527 1.309 1.727**

(4.94) (3.35) (0.53) (0.68) (-0.54) (.47) (1.85) (2.67)

Severe X Informal Loans 0.0223 0.926 -0.0153 0.604

(0.60) (-0.29) (-0.96) (-1.75)

Bank Account 0.0109 -0.0984 1.458 2.274 0.0121 0.0257 1.175 1.033

(0.26) (-1.37) (1.21) (1.43) (0.68) (0.98) (0.46) (0.06)

Severe X Bank Account 0.214* 0.444 -0.0195 1.289

(2.55) (-1.19) (-0.55) (0.36)

Insurance -0.0493* -0.0222 0.728* 0.662* 0.00243 -0.0283* 1.109 0.855

(-2.54) (-0.88) (-2.46) (-2.14) (0.30) (-2.43) (0.73) (-0.78)

Severe X Insurance -0.0507 1.197 0.0564*** 1.644

(-1.33) (0.69) (3.51) (1.76)

Financial Capability Score -0.00481 -0.00744 1.001 1.017 -0.0113** -0.0115** 0.859* 0.866*

(-0.51) (-0.78) (0.01) (0.27) (-2.88) (-2.95) (-2.28) (-2.14)

Diverse Income Sources -0.0946*** -0.0862** 0.982 0.680 -0.0243** -0.0181 1.011 0.771

(-4.62) (-3.13) (-0.14) (-1.95) (-2.87) (-1.42) (0.07) (-1.21)

Severe X Diverse Income 0.0189 1.992** -0.00932 1.822*

(-0.47) (2.59) (-0.55) (1.97)

Community Support -0.0321 -0.0150 2.082*** 1.650* -0.0382*** -0.0498*** 0.824 0.525**

(-1.21) (-0.43) (4.94) (2.36) (-3.83) (-3.62) (-1.21) (-2.86)


Severe X Community Support

-0.0287 1.644 0.0203 2.534**

(-0.55) (1.66) (1.04) (2.86)

Aid from Philippine Gov’t

0.136*** 0.186*** 0.676 0.546* 0.0390* 0.0389* 1.061 1.071

(4.00) (4.44) (-1.64) (-2.11) (2.33) (2.11) (0.23) (0.22)

Severe X Government Aid -0.168* 1.938 0.000819 0.794

(-2.36) (1.21) (0.02) (-0.39)

Aid from Foreign Source 0.0238 0.00882 0.832 0.663* -0.00702 -0.000179 1.012 1.013

(1.28) (0.34) (-1.44) (-2.22) (-0.86) (-0.02) (0.08) (0.07)

Severe X Foreign Aid 0.0295 1.588 -0.0142 1.004

(0.80) (1.79) (-0.87) (0.02)

Severe -0.000688 0.120 1.036 0.336 0.000553 -0.0109 1.198 1.085

(-0.02) (1.52) (0.16) (-1.75) (0.04) (-0.25) (0.71) (0.12)

Total 2013 Income (Log) 0.0221** 0.0217** 1.043 1.060 0.0190*** 0.0184*** 1.180** 1.176**

(2.75) (2.70) (0.83) (1.13) (6.05) (5.67) (3.21) (3.02)

Productive Assets 0.00122* 0.00118* 1.002 1.002 -0.000320 -0.000292 1.004 1.004

(2.24) (2.08) (0.36) (0.45) (-1.16) (-1.04) (0.73) (0.74)

Family Members -0.00899 -0.00781 1.021 1.020 -0.00648*** -0.00632** 1.010 1.019

(-1.62) (-1.42) (0.59) (0.56) (-3.34) (-3.26) (0.27) (0.51)

FDM Male 0.0680*** 0.0713*** 0.956 0.945 0.00545 0.00536 1.060 1.058

(3.72) (3.85) (-0.37) (-0.46) (0.72) (0.70) (0.44) (0.41)

FDM Years of Education -0.00476 -0.00416 1.048* 1.047* 0.00319* 0.00323* 1.048* 1.043

(-1.54) (-1.35) (2.18) (2.13) (2.31) (2.36) (2.00) (1.78)

FDM Literate 0.0389 0.0371 1.250 1.237 -0.00574 -0.00676 0.822 0.806

(1.84) (1.75) (1.55) (1.47) (-0.59) (-0.69) (-1.25) (-1.35)

Household Owns Land 0.0380 0.0344 0.866 0.853 -0.00915 -0.00798 0.665* 0.655*

(1.58) (1.45) (-0.93) (-1.02) (-0.91) (-0.80) (-2.46) (-2.54)

Poverty Likely 0.0592 0.0543 1.448 1.443 -0.0323* -0.0320* 1.086 1.016

(1.67) (1.53) (1.60) (1.57) (-2.29) (-2.26) (0.33) (0.06)

Municipality Constant (7) 0 0 1 1 0 0 1 1

(.) (.) (.) (.) (.) (.) (.) (.)

Municipality Constant (8) -0.193*** -0.205*** 1.038 0.996 0.00748 0.00407 1.131 1.047

(-4.43) (-4.54) (0.12) (-0.01) (0.36) (0.19) (0.41) (0.15)

Municipality Constant (9) 0.0130 0.0152 1.334 1.311 0.0407*** 0.0441*** 0.838 0.857

(0.51) (0.59) (1.50) (1.38) (3.84) (4.16) (-0.87) (-0.73)

Municipality Constant (10) -0.128*** -0.113*** 1.699* 1.734* 0.0253 0.0247 2.009* 2.106*

(-3.87) (-3.34) (2.22) (2.24) (1.71) (1.62) (2.40) (2.55)

Municipality Constant (11) 0 0 1 1 0 0 1 1

(.) (.) (.) (.) (.) (.) (.) (.)

Municipality Constant (12) -0.168*** -0.159*** 1.259 1.320 0.00990 0.0133 0.824 0.879

(-5.29) (-4.74) (1.02) (1.19) (0.67) (0.88) (-0.75) (-0.50)

Constant 2.600*** 2.575*** 0.4061426 0.5326863 4.365*** 4.375*** 0.401972 0.4788146

(28.37) (27.73) (-1.58) (-1.08) (115.90) (115.55) (-1.51) (-1.16)

N 1194 1194 1353 1353 1286 1286 1359 1359

adj. R-sq 0.160 0.168 0.079 0.088 0.091 0.095 0.058 0.072

Do Financial Services Build Disaster Resilience? | MERCY CORPS

ABOUT MERCY CORPSMercy Corps is a leading global humanitarian agency saving and improving lives in the world’s toughest places.Poverty. Conflict. Disaster. In more than 40 countries, we partner with local people to put bold ideas into action, help them overcome adversity and build stronger communities. Now, and for the future.

45 SW Ankeny StreetPortland, Oregon 97204

888.842.0842mercycorps.org

CONTACT

JON KURTZMercy Corps Director of Research and [email protected]

ERYNN CARTERMercy Corps Philippines Country Representative [email protected]

40 SciennesEdinburgh, EH9 1NJ, UK

44 131 662 5160mercycorps.org

EUROPEAN HEADQUARTERS

do financial services build disaster resilience? · do financial services build disaster...

Documents