Projects

Why are some entrepreneurs perpetually in debt? Around the world, individuals pay high interest rates to moneylenders or microfinance institutions in order to finance their working capital. In most cases, however, the micro-enterprises are not growing. Hence it is unclear in the long run why these businesses need financing. Why are these entrepreneurs unable to save sufficiently from their income to finance their businesses through accumulated prior earnings? Given the high rates of interest, such an approach would no doubt lead to higher consumption for the entrepreneurs and their families.

In this experiment, we will study the characteristics of vendors who repeatedly fall into debt. We will first conduct a baseline survey to collect detailed information on the vendor's borrowing behaviors as well as household and business characteristics; then we will randomly select 50% of the individuals from the sample and pay off their debts. After several months, we will observe which vendors have fallen back into debt. We will also conduct follow-up surveys to examine changes in behaviors after our intervention.

The borrowers here are vegetable market vendors; they typically borrow every morning and repay in the afternoon, paying an interest rate of about 5% overnight.



[close]

Why are some entrepreneurs perpetually in debt? Around the world, individuals pay high interest rates to moneylenders or microfinance institutions in order to finance their working capital. In most cases, however, the micro-enterprises are not growing. Hence it is unclear in the long run why these businesses need financing. Why are these entrepreneurs unable to save sufficiently from their income to finance their businesses through accumulated prior earnings? Given the high rates of interest, such an approach would no doubt lead to higher consumption for the entrepreneurs and their families.

In this experiment, we will study the characteristics of vendors who repeatedly fall into debt. We will first conduct a baseline survey to collect detailed information on the vendor's borrowing behaviors as well as household and business characteristics; then we will randomly select 50% of the individuals from the sample and pay off their debts. After several months, we will observe which vendors have fallen back into debt. We will also conduct follow-up surveys to examine changes in behaviors after our intervention.

Here the borrowers work in street markets selling a variety of products. They typically take month-long loans, paying interest of 20% per month.


[close]

Client selection is one of the most crucial determinants of the performance of microfinance programs. In group-lending, lenders rely on the better information among applicants/clients about others to screen out the 'bad type' of borrowers who are less trustworthy. In individual-liability lending programs, however, the client selection is entirely based on a credit investigation which captures observable characteristics of individuals and the subjective intuition of credit officers.

Aiming to help microfinance institutions make better selection decisions, we are conducting a field experiment in the Philippines to study the characteristics of microfinance clients who are more likely to default. In this experiment, all new and repeat clients are asked to take two surveys:

In the first survey, clients are interviewed by the bank staff at the time of loan disbursement in order to receive a certain amount of monetary compensation. The survey includes questions on the clients' business and household characteristics, time preference, and personalities. When the survey is completed, clients will receive greater compensation than what they were told they would receive. We record whether or not the client will return the excess amount of money (trustworthiness indicator).

Before clients leave the bank, they are given another short survey and asked to text back (by cell phone) the survey by a pre-specified date in order to receive monetary compensation. We observe who texts back and who does not mail back the survey (planning capability indicator).

The study attempts to predict two characteristics that are difficult for credit officers to observe; trustworthiness and planning capability of the clients. The study will benefit microfinance institutions in understanding the nature of its clients as well as provide insight into the necessary characteristics of a reliable bank patron.


[close]

Given the need for every MFI to set its own interest rates it is surprising how little empirical evidence exists to guide them in this process. MFI managers typically set their interest rates by surveying their closest competitors. From this starting point they can through trial and error adjust rates to a satisfactory level as their cost structure and competitive pressures change. But this leaves several important questions unexamined.

The first and most obvious question is whether MFIs are in fact maximizing sustainability. Could they increase profits either by raising interest rates and increasing revenue per borrower, or by lowering rates and increasing demand? The answer to this question depends on the price elasticity for credit among microfinance clients, that is, the extent to which demand increases with a decrease in interest rates and vice versa. It also depends on adverse selection and moral hazard: at higher interest rates does the MFI attract riskier clients (those who gamble on enterprises with high risk but high payoffs), and given higher interest rates are clients more likely to default? The second question relates to outreach: as MFIs change interest rates, how does this change the volume and composition of their borrowers?

In this study, interest rates are randomized at the branch level in order to determine the price elasticity of demand among village banking clients throughout Mexico. Half of these branches will offer a relatively higher interest rate to loan groups and the other half will offer a relatively lower interest rate. We will analyze outcomes on variables such as take-up rates, loan amounts, repayment rates, and client retention. In addition, socio-economic data will be collected to determine any differences among composition of borrowers.


[close]

This project works with several rural banks in the Philippines to study the viability of various monthly interest rates for the banks' core microfinance loan products. The study hits on traditional questions related to price elasticity of demand. How many additional (fewer) borrowers will be attracted to the lower (higher) rates? Will the changes in demand counter-balance the changes in per-borrower revenue? At the same time, we also hope to gain significant insight into the type of borrowers that are attracted to the various rates. Do better borrowers hold out for better prices? Conversely, are riskier borrowers less concerned with price?

Interest rates will be randomized for each marketing pitch made by the banks' account officers. To facilitate a clean randomization, every officer will be equipped with a personal digital assistant (or PDA). At the beginning of every marketing pitch, observational poverty measures of the microenterprise (e.g. materials used to construct the walls; presence of a cashiers machine) will be collected to help us learn more about which rates appeal to which types of micro-entrepreneurs.

The PDAs will then produce a random assignment for each potential borrower based on a pre-established range of rates set by the partner Bank. In addition to providing a clean randomization, the PDAs will also enable the account officers to better organize their marketing efforts. Full implementation of the study with First Macro Bank was scheduled for December 2006. We will expand the study to First Isabela Cooperative Bank, First Valley Bank, and Green Bank of Caraga in 2007.


[close]

There are two projects, one with ARARIWA and the other with Movimiento Manuela Ramos, a Peruvian female empowerment organization. The latter project goes beyond interest rate sensitivity to investigate effects on the types of clients that choose to participate in the community banks and thus gain access to credit.

The majority of MFIs aim to charge an interest rate that will allow them to achieve all of their financial and social goals. The interest rate should be high enough that it covers institutional costs, allowing for financial self sustainability. At the same time, however, it must be low enough to compete in the finance market and also allow them to reach their social mission of creating access to credit for the low income population. There is little information, however, to help MFIs determine the most suitable interest rate.

In order to study the impact of different interest rates, we partnered with the microfinance department of Movimiento Manuela Ramos, a Peruvian female empowerment organization. The aim of the project is two-fold. It will look at the effects of different interest rates and different incentives on institutional sustainability, but it will also look at how these factors effect the type of clients that chose to participate in the community banks and thus gain access to credit. In particular, we will focus on whether lowering interest rates attracts clients who have never received loans before or clients who previously received their credit from other institutions, banks, or NGOs.

We have randomly divided the banks into five different groups. One group has no incentive, and the remaining four groups receive a unique combination of high / low interest rate and positive / negative incentive. To measure impact, we will analyze institutional data on loan size, repayment rate and number of new clients. We will also survey all potential Manuela Ramos clients to determine characteristics of people who chose to take out loans at different interest rates and to see if they are women who have never before had access to credit.


[close]

Using data from 1438 households in six provinces in Indonesia, we aim to understand the profile of the 'unbanked.' Why do so many poor households lack access to finance? Are the unbanked creditworthy? Largely not interested in borrowing? Key findings (Summer 2007) are that about 40 percent of poor households were judged creditworthy, but only 14 percent had recently borrowed. Possessing collateral was a minor determinant of creditworthiness. Despite depictions of widespread pent-up demand for loans, about half of creditworthy poor households report being averse to taking on debt. Loans for small business were desired, but respondents often highlight broader household needs, including paying for school fees, medical treatment, and home repair. Note: This work does not use randomized control trial methods.


[close]

This is a project with Pro Mujer, a village banking institution along the lines of FINCA and Freedom from Hunger. The experiment involves dropping the group liability element of their program at the village center level (but keeping group liability at the group of 4 level). The research asks how it affects demand, repayment, and other features. The work complements work on group vs. individual lending contracts in the Philippines.

Several mechanisms contribute to the high repayment rates achieved in many group-lending microfinance programs around the world. These mechanisms include: group liability, progressive lending (dynamic incentives), mandatory savings, public payments, leveraging social capital through self-selection of group members, lending primarily to women, and frequent loan payments, among others. However, little is understood about the relative importance of each of these mechanisms. Some mechanisms may be highly effective at promoting timely loan repayment but at the expense of reducing the number of qualified participants in the program.

The objective of this experiment is to test two group-lending mechanisms. The first mechanism is group liability. We will reduce the number of fellow borrowers for which each Communal Association member is responsible from the current average of about 20 to around five to eight. We will then measure the impact that this change has on several outcomes that are important to Pro Mujer, such as repayment rate, loan size, dropout rate, and new member acquisition rate.

The second mechanism is group self-selection. Under this approach to group formation, members choose for themselves the other members that make up their solidarity group. We will study optimal group formation. When groups self-select members, it is likely that they overweight social ties relative to other characteristics such as business acumen. We will investigate other possible ways of organizing the groups, such as by loan size, by type of business, or by a reduced level of social connections.


[close]

The number of variables that can affect a farmer's ability to repay an agricultural loan is extensive; weather patterns and crop prices, for example, are entirely outside a farmer's control. Farmers are faced with taking a loan from a bank and being unable to repay due to lack of rainfall or low crop prices. Similarly, banks are faced with offering loans to farmers only to lose money (and potentially good loan customers) when the farmer is unable to repay for reasons out of his or her control.

Researchers are working with Mumuadu Rural Bank in Ghana to test out a loan product that could mitigate some of the risk, both to the bank and to the farmer, of offering and receiving agricultural loans. By analyzing historical crop price, rainfall, and crop yield patterns, researchers will develop a new loan product with a crop price insurance component for farmers that may encourage both take-up and repayment.

The repayment amount required by this new product would vary with crop price. If, at the time of harvest, the farm gate price for a particular crop is particularly low such that the farmer could not be expected to cover his or her loan payments, the repayment requirement is lowered. If the crop price is high such that the farmer could be expected to repay a higher amount, the repayment requirement is raised. Researchers will test the effect on take-up of marketing up front the potential for the farmer to be shielded from risk caused by fluctuating crop prices. Furthermore, the effect on overall repayment amounts attributable to the crop price insurance product will be tracked.


[close]

Microfinance institutions are in constant search of ways to increase their efficiency. One of their main objectives is thus to minimize default rates among their clients of credit products. These clients are usually too poor to offer any property as collateral, so micro-lenders need to use alternative methods to encourage repayment. The most common methods are: (1)the promise of new, larger loans, and (b) peer pressure from village banks, solidarity groups and/or loan sponsors. We have partnered with PRISMA, a microfinance organization in Peru, to identify ways to use these threats more effectively: We examine the use of gain/loss framing in a loan product and the extent to which peer pressure encourages repayment. Clients in default receive letters with either 'gain' or 'loss' frames, telling the client either that rectifying his credit standing will allow him access to credit in the future, or that his continued default will keep him from accessing loans in the future and also threaten legal action.

Additionally, late-payment letters prime either an individual identity, in which the threat is directed solely at the loan client, or a group identity, in which the client's sponsor also receives late-payment letters if the client is in default. To implement this study, all clients in PRISMA's individual-loans program have been randomly assigned to receive letters which prime either group or individual identities, or to receive no reminder letters at all. A second randomization assigns all individuals in letter-receiving groups to receive either gain- or loss-frame letters. The project began in March, 2006; data collection will continue until June, 2007.


[close]

Client selection is one of the most crucial determinants of the performance of microfinance programs. In group-lending, lenders rely on the better information among applicants/clients about others to screen out the 'bad type' of borrowers who are less trustworthy. In individual-liability lending programs, however, the client selection is entirely based on a credit investigation which captures observable characteristics of individuals and the subjective intuition of credit officers.

Aiming to help microfinance institutions make better selection decisions, we are conducting a field experiment in the Philippines to study the characteristics of microfinance clients who are more likely to default. In this experiment, all new and repeat clients are asked to take two surveys:

In the first survey, clients are interviewed by the bank staff at the time of loan disbursement in order to receive a certain amount of monetary compensation. The survey includes questions on the clients' business and household characteristics, time preference, and personalities. When the survey is completed, clients will receive greater compensation than what they were told they would receive. We record whether or not the client will return the excess amount of money (trustworthiness indicator).

Before clients leave the bank, they are given another short survey and asked to text back (by cell phone) the survey by a pre-specified date in order to receive monetary compensation. We observe who texts back and who does not mail back the survey (planning capability indicator).

The study attempts to predict two characteristics that are difficult for credit officers to observe; trustworthiness and planning capability of the clients. The study will benefit microfinance institutions in understanding the nature of its clients as well as provide insight into the necessary characteristics of a reliable bank patron.


[close]

This field experiment studies trust in social networks in poor communities in Peru. An initial pilot was conducted in urban shantytowns in Lima, and a second, modified, project is now being implemented in rural communities in Ancash, in the mountainous center of the country in conjunction with PRISMA, a Peruvian NGO with a large micro-lending program.

The current project aims to (1) measure trust in different communities and analyze the community determinants of the different valuations, and (2) test a novel microfinance model combining individual loans with screening and enforcement features originally developed for group lending schemes.

We will work with village banks in 30 poor communities and allow the existing bank members to co-sign small, individual loans for residents of the community who are not bank members. Social network surveys conducted in the communities before the implementation of the loan program will allow us to map the relationships between clients and sponsors. By varying interest rates between sponsors we will be able to examine the extent to which a lower interest rate is valued over (or under) a close relationship to the sponsor. After the loans are given, we can track how repayment rates vary by sponsor-client relationship. Additionally, because the program relies on the savings guarantee of the sponsor, it is a low-risk way for PRISMA to explore extending credit to people that lack the traditional pre-requisites. It also provides an opportunity to study the incorporation of individual lending practices into an existing group-lending structure.


[close]

At present, we are collecting data necessary to calculate the power of future statistical designs. This project will use a method developed by Karlan and Zinman where lenders using credit-scoring techniques provide credit access to a group of other-wise rejected loan applicants (marginal rejects). Their outcomes after borrowing are compared to a control group that was not given a second chance (and that are also on the margins of rejection).


[close]

Despite the ever-increasing global popularity of Microfinance, very little is known about the pricing and impact of loans for the poor. A number of important questions have been left mostly untouched by previous research: How sensitive is demand for Microcredit to changes in the interest rate? How do changes in the interest rate affect the type of clients that Microfinance Institutions (MFIs) attract? How does the presence of competition affect loan take-up at different interest rates? What is the impact of Microcredit in rural areas, and how does this differ across interest rates? What interest rates are financially feasible for MFIs? The rural expansion of Asociación ARARIWA, a Peruvian Microfinance Institution operating in the department of Cusco, offers a unique opportunity for us to investigate answers to these questions through randomized control trials and to share the results with Arariwa as well as global policy makers and MFIs.

We have randomized each of the 49 geographical zones covered by ARARIWA's expansion into one of three treatment groups: a control group that will receive no credit offer during the 18 months of the study, a low interest rate group that will receive a credit offer at 3.8% monthly interest (near what ARARIWA's competitors charge), and a high interest rate group that will receive a credit offer at 4.8% monthly interest. We are conducting a baseline survey collecting various measures of income, risk aversion, and past credit history in 25% of all homes in the expansion area. ARARIWA will then begin offering credit in the high and low interest rate areas and we will conduct additional surveys with each person who decides to take out a loan. We predict a sample size of 2000 new borrowers out of an eligible population of 50,000 in the expansion area. At the end of the 18 month study-period, the same 25% of homes in the expansion area will be resurveyed as well as all the additional people who took out loans.

The data analysis will compare the take-up rates in low interest rate zones with the take-up in high interest rate zones, the characteristics of the new clients at each interest rate, and take-up in zones with competing loan offers with the take-up in zones where there is currently no credit offered (around half the expansion area currently has alternative credit sources). Finally, we will measure the impact of microcredit by comparing how measures of income have changed for those people who took out a loan with how measures of income have changed for people in the control areas who we predict would have taken out a loan based on survey data. The impact of access to credit will also be measured by comparing changes in measures of income in non-control areas to changes in measures of income in control areas.


[close]

This study builds upon three earlier studies conducted by IPA related to finance among poor populations. In the earlier evaluation of the commitment savings product with Green Bank of Caraga in the Philippines, we found a significant impact of the savings product on overall savings as well as household decision-making power. Another earlier study about the psychology of finance, in which IPA partnered with a credit organization in South Africa to examine how design and content of letters offering credit can affect loan uptake, indicated that design characteristics such as presentation of loan options or photos in the promotion letter could affect take-up decision as much as the interest rate of the loan itself.

Given the success of the commitment savings product in Green Bank and the evidence from South Africa of the psychological impact on financial decision-makings, we designed the evaluation in which we examine various methods, beyond the financial incentive and benefit, of encouraging clients to complete their savings commitment with our partner banks in the Philippines. We partner with First Isabela Coop Bank (FICO) (Northern Luzon) and First Valley Bank (Western Mindanao) to test the impact of the commitment savings product which allows clients to set their own goals for the savings amount and purpose. We randomly assign different encouragement features for the savings product; offering different upfront interest rates; offering reward interest rate for reaching the goal; giving a piece of the puzzles with a image of savings goal at every deposit; and sending text message minders. We examine the impact of different product features on takeup decision as well as the savings behavior. The study aims to build an understanding of why the poor save and how to prime those goals to encourage savings.


[close]

Using a method developed by Karlan and Zinman, lenders using credit-scoring techniques provide credit access to a group of other-wise rejected loan applicants (marginal rejects). Their outcomes after borrowing are compared to a control group that was not given a second chance (and that are also on the margins of rejection).

[close]

Through this project we are also helping the two institutions design a credit scoring model and streamline their application evaluation procedure. Depending on their score, applicants are classified in three categories: applicants with a good or bad credit score are automatically approved or denied, respectively, and applicants whose score falls in a 'grey area' are randomly approved or denied. The randomization feature allows for clean studies of the impact of getting a loan and of the connection between credit score and repayment performance, benefiting the researchers and the partner banks. Using a score to rank applicants, banks can rigorously establish "safe" and "risky" levels of lending that maximize profit.

Baseline data is collected on applicants' personal characteristics and personality traits that may explain part of the impact of the loan on businesses. An extensive follow-up survey will focus on business performance and household characteristics of all applicants in the grey area, both approved and denied. We are currently working with First Macro Bank in full-implementation of the credit scoring system, and will launch the project with First Isabela Cooperative Bank and Green Bank of Caraga in 2007.


[close]

Currently, the use of savings services is low among remittance receivers at the Caja Nacional del Sureste. Increasing savings could mitigate the negative impacts of unforeseen circumstances, such as medical emergencies or economic hardship, as well as reduce dependence on the remittance schedules for those who rely on these funds for sustenance. To encourage savings, the Caja is offering a new emergency savings account, or 'Secure Future' as the account is named, to all remittance clients in order to allocate funds for future needs. After surveying these clients, a common theme of their savings needs emerged, with the majority citing future illness or medical emergency; future needs of their children; or future economic hardship.

This savings product is designed to serve the needs of the target population and incorporates ideas from economics and psychology. Mental accounting theory suggests that individuals naturally divide their assets into separate groups - current and future - and by labeling these assets their propensity to spend from each group differs (high for current assets, low for future), allowing for overall increases to savings levels. In addition, clients who open this emergency savings account will designate a certain portion of their remittance to be deposited as savings at every transaction. This self-commitment to save is not binding (they can change this amount at any time). However, if individuals stick with their prior decisions this pre-determined commitment could translate into a habit of saving.

Remittance clients will be randomly assigned to two groups - one will receive an offer to open an emergency savings account and the other will not. Savings levels at the Caja will be tracked over time to determine whether this account encourages remittance clients to increase their use of formal savings services. An initial survey will collect basic data on socio-economic indicators and savings practices, and a follow-up survey will explore welfare implications of this emergency savings account. This study will be implemented among Caja remittance clients in four distinct regions in Oaxaca.


[close]

Here, with FINCA-Peru, the study investigates the impact of integrating health and business training into their normal weekly credit meetings. The aim is to illuminate the debate on whether lenders should remain "business-like" and focus solely on the lending business, or whether they should take advantage of the frequent meetings to integrate various types of training. We randomly selected a subset of lending groups and integrated training into their group meetings. We then followed these groups, along with control groups that do not receive any training, to assess the impact of the training on institutional outcomes (such as dropout rates, repayment rates, loan size and savings), household outcomes (such as household consumption and health practices and outcomes) and business outcomes (such as business size and income). The project is in a final stage. Results coming out of the field show strong benefits for both the client and the microfinance institution. Clients significantly improved their sales and client retention improved 16%.


[close]

We are working with the Rural Green Bank of Caraga in the Philippines to evaluate health insurance as an add-on service to individual and group-lending microfinance clients.

This study aims to address several research, policy, and programmatic questions related to expanding access to health insurance, including:

1) The presence or absence of adverse selection: The impact of information asymmetries on insurance markets is important in theory but ambiguous in practice. Generations of studies have failed to produce a consensus on the presence, absence, or magnitude of adverse selection and moral hazard in most markets, especially with respect to the margins of expanded access that are critical in developing country settings.

2) The impact of access to insurance on household health and risk-taking behaviors/preferences: Evidence in existing literature suggests that the effects of being without insurance can be large. However, there is limited if any direct evidence from developing countries on this question.

3) Measurements of crowd-out of informal insurance arrangements: Formal insurance may displace community-based risk-sharing (informal crowd-out) but there is no empirical evidence yet of the magnitude of these kinds of spillovers.

4) The business strategy issues facing potential wholesalers of health insurance: Distribution channels are an important issue facing policymakers and program managers interested in expanding access to health insurance. In this study, the government health agency (PhilHealth) partners with a microfinance institution that offers PhilHealth policies on a wholesale basis. Our study will address the policy issue of whether such initiatives should have an element of compulsory participation where offered and how health insurance on a compulsory or voluntary basis affects client retention, repayment, and profits.

[close]