31
Jan

All’s well that repays well? Not necessarily.

By Vaishnavi Prathap, IFMR Finance Foundation

The past year has seen many commentaries on the rapid expansion of microfinance in India warning of the imminent consequences of unbalanced growth. The most striking statistic in this context — that the average client’s dues more than doubled in just four years (between 2012 and 2016), far outpacing only moderate growth in numbers of branches, employees or clients, and surely clients’ incomes — was estimated from data that, at best, captures only a large proportion of the microfinance market. While this has triggered ruminations of an emergent repayment crisis, these fears have been tempered on two grounds. First, the enforcement of new regulations since 2012 limit the risk of client over-indebtedness. Second, delinquencies have consistently remained low over the expansionary period, and wherever reports of distress have surfaced, they seem mostly uncorrelated with the sector’s growth rate.

But do these arguments show us the full picture? Considered against primary evidence from the financial diaries of low-income households in India, we find that often they do not. The data, collected by IFMR Finance Foundation during a study supported by the CGAP Customers at the Center Financial Inclusion Research Fund, provides rich detail on the financial lives of borrowers in a competitive and mature microfinance market. It reveals that the indicators cited by both arguments above are poorly correlated with the incidence of over-indebtedness and with the ways in which borrowers experience and cope with repayment distress.

Timely repayments and borrower distress are not mutually exclusive

Aggregated data from lenders’ administrative records, such as delinquency estimates published by credit bureaus, have traditionally served to indicate portfolio quality. When delinquencies are low, it is interpreted as a signal of positive borrower outcomes. However, the repayment record may not fairly represent borrowers’ experiences if lending practices emphasize timely collection above all else.

The financial diaries of over-indebted borrowers illustrate this fact. Of the 400 households we studied, nearly one of every five borrowers reported repayment obligations higher than they could reasonably afford, given their incomes and minimum living expenses. Yet, as many as 85 percent of those over-indebted borrowers never missed a repayment on formal loans, arguably since they had strong incentives (both institutional and social) to do so.

Further, the records submitted to bureaus seldom distinguish between repayments made by the client and payments made by group members on her behalf. Thus, perhaps as an unintended consequence of the design of joint liability, the administrative data reveal few meaningful insights at a sector-level on borrowers’ distress or well-being.

Not consumption smoothing but repayment smoothing

How is it possible for so many borrowers to consistently avoid delinquency while carrying multiple, unaffordable loans? The data suggest they are using several coping mechanisms, such as lowering consumption or postponing essential expenses; raising resources from friends and social networks; and using large formal loans when available to settle old debts, including smaller informal ones accrued in past months. This use of coping mechanisms in the face of shocks is not unlike previously documented evidence. Low-income households use a variety of strategies to insulate their consumption and standard of living from the risk of volatile incomes; alternatively, they try to minimize the impact of income volatility by diversifying their occupations and resources. These strategies have a limited ability to protect households from poverty, and it has been shown that severe or persistent shocks are a major cause for chronic poverty.

The use of coping mechanisms by over-indebted borrowers differed from these practices in one regard — the incidence of coping behavior was highly correlated with the unaffordability of household debt and appeared to revolve around insulating repayments rather than consumption. Borrowers with more unaffordable debt used negative coping mechanisms more often and to a greater degree than others. Their financial behavior was not unlike the expected response to an income or health shock. But in this case, the shock came in the form of multiple non-negotiable loan repayments. Unlike a random occurrence, these “repayment shocks” persisted month after month.

Additionally, not only was the level of debt correlated with distress, but also with certain product features embedded in the loan contract. For example, a subset of borrowers experiencing highly volatile cash flows might have more trouble meeting repayments at certain times of the year. We found that these borrowers, for whom repayments were occasionally unaffordable (when calculated against a given month’s income instead of the average), experienced higher levels of distress, almost on par with those for whom the repayments were almost always unaffordable.

The implication for microlending is that poorly matched repayment schedules and other product features could be just as harmful as too much debt — and more harmful if combined with high levels of debt. This is a critical dimension of the experience of over-indebtedness, yet it is often overlooked.

New lending rules to prevent unsuitable loans

It is evident from the observed level of financial distress that current practices to prevent over-indebtedness are not effective. In fact, critical fault lines in their implementation have created an environment where unsuitable credit remains the primary coping mechanism even for over-indebted borrowers. It is also concerning that they focus heavily on limiting the amount of client debt while ignoring other aspects of borrowers’ cash flows that are significant in mediating financial distress. These include large seasonal or cyclical effects for specific lines of income or, even more generally, total income volatility (the median household in our sample experienced monthly income swings as high as plus or minus 45 percent) as well as large and significant uninsured (but insurable) risks.

Beyond better repayment assessments

Microfinance in India is no longer dominated by monopolistic or mono-product markets. With the licensing of several large lenders as small finance banks, the expectation is now that low-income households will have access to better financial services. This means not only easier access to a wider set of services, but services provided by institutions that are better equipped to respond to low-income households’ primary needs and vulnerabilities.

Many have argued that the continued success of lending to low-income households will require the evolution of robust mechanisms to assess clients’ capacity to deploy credit and manage repayments. By itself, this will not be enough to prevent borrower distress since repayments alone do not signal that a loan is suitable.

Lenders must also adequately detect clients’ cash-flow vulnerabilities and respond to them with appropriate design and service improvements, complementary savings and insurance products, flexible repayment schedules where appropriate, and best practices for delinquency management.

Read the latest version of the paper here. This article first appeared on the CGAP blog.

19
Jan

Guidelines for Suitability in Lending to Low-Income Households

By Vaishnavi Prathap, IFMR Finance Foundation

Img_1In December 2014, the Reserve Bank of India published the Charter of Customer Rights as a commitment to protecting the interests of consumers of financial services. The charter includes the Right to Suitability, defined as the principle that “products offered should be appropriate to the needs of the customer and based on an assessment of the customer’s financial circumstances and understanding”. The MFIN and Sa-Dhan, in a joint Code of Conduct, also enshrine a similar principle but specific to the context of lending: “We, as part of the Microfinance Industry promise the customers that we will […] conduct proper due diligence to assess the need and repayment capacity of customer before making a loan and must only make loans commensurate with the client’s ability to repay”. Both the RBI and the SROs directed financial institutions to understand the parameters of suitability within the context of their respective product offerings and to formalize policies to prevent unsuitable sales to customers.

In a new research paper published as part of our Working Paper series, we focus on biggest barrier that financial institutions might face in complying with this directive – a lack of clarity on what may be deemed suitable and how this is to be determined for each client. Towards this end, our new research investigates the nature and incidence of unsuitability in a competitive lending market and the variety of ways in which low-income borrowers may experience or cope with loan-related financial distress. Our findings are both a reality-check on the effectiveness of the current approach to customer protection (particularly the efforts to prevent borrower over-indebtedness) as well as a guide to how, going forward, lenders can institute formal processes to prevent unsuitability.

In our previous writing on this topic, we have acknowledged that successful suitability practices must be iterative and that even at-best, they can in no way guarantee positive outcomes for clients. The focus of both compliance and supervisory efforts must rest instead on understanding patterns in product-client interactions – especially when such interactions result in substantial hardship to clients – and meaningfully improving sales processes to prevent unsuitable sales.

Key Findings

The primary data for this study was collected in a year-long panel survey of 400 low-income households in Krishnagiri district, Tamil Nadu; the full sample included clients of 7+ MFIs and 20+ formal financial institutions. The survey adopted a financial diaries approach and a detailed socioeconomic survey was administered every 4-6 weeks to capture the dynamics of households’ cashflows. The resulting dataset has a primary focus on the details of borrowing and loan servicing but also rich detail on the volatility of occupational income, the frequent incidence of small and large shocks to household budgets and the use of other income sources, resources from social networks and a variety of financial instruments to smoothen consumption, repayment obligations and other expenses.

From the survey, we were able to create a full picture of borrower indebtedness across multiple institution types – perhaps more completely than the credit bureaus for microfinance clients. Comparing the sum of monthly repayment obligations to borrowers’ average monthly incomes, we found that one of every five borrower households in the sample held an unaffordable level of formal debt. If we included informal loans or factored in the volatility of incomes, an even higher proportion held unaffordable levels of debt for a few or all months of the loan tenure. However, the incidence of repayment delays or the proportion of delinquent borrowers was much lower, and only weakly correlated with households’ debt levels. Even at very high levels of unaffordability, borrowers were prioritizing repayments on formal loans over essential expenses, and willing to take on even further unmanageable debt to get through a difficult period.

Further, the average households’ incomes varied month-on-month by as much as 45% and as a result, even borrowers with sufficient year-end surplus were observed experiencing periods of distress and using harmful coping mechanisms comparable to those whose incomes were much lower.

Implications for Suitability Practices

These patterns in borrower behaviour are perhaps not new to experienced practitioners of microfinance and further, may only be a reflection of practices designed to achieve repayment discipline. What is alarming however, is the relative ease with which some over-extended borrowers remained undetected, and were able to continuously receive new formal loans on the same terms as others.

NBFC-MFIs are subject to regulatory directives that restrict the level of indebtedness per client and additionally, a large part of the non-NBFC microfinance lending is also required to be reported to credit bureaus so that it may be available at the time of loan appraisal. Notwithstanding, we find that critical faultlines in the preparation and use of credit reports placed as many as 33% MFI clients in the sample at risk of being mis-sold an unaffordable loan.

More critically, the types of client assessments that inform loan-making are largely unregulated and often do not triangulate borrowers’ actual repayment capacity (relying instead on unverified or indicative measures, peer selection and group enforcement). Our results show that in a mature and competitive market, clients with similar incomes and livelihoods may in fact have very different borrowing portfolios and vice versa. In this scenario, universal lending limits— such as those currently in effect— poorly safeguard customers’ interests.

Instead, determined efforts should be directed towards building market capacity to conduct thorough client assessments and to respond meaningfully to clients’ financial situation. Credit reports urgently need to be strengthened to reflect a comprehensive view of all formal borrowing, without exception. On the lenders’ side, the use of comprehensive “combo” credit reports will still fall short if not also complemented by a robust understanding of what portion of household income can be made available for repayments. Further, clients with unique liquidity constraints or cashflow risks must receive adequate insurance either through appropriate products or through modified terms of service.

This research highlights not only how critical these measures are for borrower well-being, but also the challenges involved in suitably serving low-income households’ financial needs. As a step in this direction, this research outlines two minimum components for suitability assessments –

  1. All lenders should ensure that loan amounts are appropriate and the agreed repayment terms are affordable for every borrower given their income flows, outstanding loan repayments (including self-reported informal loans) and critical payment obligations.
  2. Lenders must also evaluate the harms of selling standardized products to those borrowers with highly volatile cashflows or those with unique liquidity or flexibility constraints. Uninsured cashflow risks must be provisioned for in the assessment, product design or terms of repayment.

The paper also outlines recommendations for coordinated regulatory, practitioner and research effort that can enable successful implementation. 

The working paper is available online here and we welcome both questions and comments.

18
Sep

Aggregate Risk, Saving and Malnutrition in Agricultural Households

Guest post by Dr. Anjini Kochar, Stanford University, C. Nagabhushana and N. Raghunathan, Catalyst Management

Why is malnutrition in India’s central belt, which includes the state of Madhya Pradesh and Bihar, so high and so persistent despite relatively high rates of income growth? High rates of malnutrition reflect diets that predominantly feature cereals at the expense of more nutritional foods including pulses, vegetables and fruit. In turn, high cereal intensity is believed to be a consequence of agricultural land use patterns that favour wheat and rice. In recent research, we show that the link between cropping choices and nutrition exists only because of households’ savings choices. Faced with high and variable prices for pulses, households in this region primarily save in the form of stocks of wheat, using home stocks as an inflation hedge to substitute for the consumption of high-priced pulses. The attractiveness of wheat stocks as a buffer against high prices arises not just because stocks insulate households from inflation, but also because of the high costs of transacting with formal financial institutions in this region, despite the government’s recent attempts to ensure that all households have savings accounts in formal institutions.

Recent data from the National Family Heath Survey (2015-16) estimates that 44% of children under the age of 5 in the state of Madhya Pradesh are stunted, and 45% are underweight.[1] Though detailed data from this report are not yet available, earlier rounds (2005-06) confirm the limited role of income in explaining these rates: The percentage of malnourished children remains relatively stable over the wealth distribution, falling off only for the richest quintile of households. The percentage of stunted children in the lowest four quintiles of the wealth distribution was found to be 53%, 54%, 54% and 52%, respectively, falling to 42% for the richest quintile of households. Similarly, the percentage of under-weight children was 67%, 67%, 62% and 63% amongst the bottom four quintiles and 50% for children in the richest quintile.

Poor nutrition, undoubtedly, plays a major role in explaining the pervasiveness of stunting and low weight. Diets in this belt heavily favor cereals (wheat and rice) for all households, regardless of wealth or occupation. Expenditure on cereals amounts to 26% of total food expenditure in Madhya Pradesh and 28% of expenditure in the neighboring states of Uttar Pradesh, Bihar, Chattisgarh and Jharkhand. In Madhya Pradesh, as in neighboring states, this percentage shows almost no variation across households distinguished by principal occupation; it is as high (26%) amongst agricultural households who derive their income primarily from the cultivation of their own land as it is amongst households who are primarily dependent on causal wage work in unskilled labor markets (28%).[2]

While the dominant role of cereals in the diet of Indian households has often been noted, less attention has been paid to the striking importance of consumption out of home stocks for farming households. In Madhya Pradesh, a predominantly wheat growing state, data from the NSS (2011, round 68) reveal that consumption out of home stocks of wheat amounts to 42% of total wheat consumption of all households, but as much as 81% of the consumption of agricultural (farming) households. Similarly 62% of the wheat consumption and 67% of the rice consumption of agricultural households in neighboring states is also from home stocks.[3]

The hypothesis that poor nutrition reflects households’ decisions to hold large stocks of wheat as a precautionary response to price uncertainty may be surprising, only because prices of wheat and rice are far less volatile than that of other crops. This in turn is a consequence of government intervention in grain markets, both on the consumption side (through welfare programs such as India’s Public Distribution System that distribute food grains at highly subsidized prices) and on the production side (through minimum support prices). Relatively low and stable prices compared to other food crops reduces the likelihood that households will sell stocks of wheat to purchase more expensive food items, thereby lowering their value as an inflation hedge. Their value to households lies, instead, in their substitutability for other foods. Though such substitution would occur, even if households did not maintain grain stocks, these savings allow households to transfer consumption across seasons allowing more consumption in seasons characterized by poor rainfall and low incomes than would otherwise be possible. Substitution of wheat for other foods, such as pulses, does, however, come at a cost: It reduces the nutritional content of households’ diets. Additionally, the retention of wheat by farming households for consumption purposes reduces market supply and raises market prices. This adversely affects landless households. It also increases the cost of government purchases of wheat for the public distribution system.

We examined the relationship between nutritional status and savings using rich household data from a sample of approximately 2800 households from rural areas of Madhya Pradesh, collected in January 2016, matched to monthly data on market (mandi) prices for wheat and the main pulse consumed in this area, tur or pigeon pea, for the 2010-2015 period. Using this data, we tested three hypotheses. First, we assessed whether stocks of wheat are held as a precautionary response against variability in the price not just of wheat, but also tur. Second, we examined whether households’ stocks of wheat affect wheat consumption in the household in regressions that control for the effect of total savings. That is, we tested the hypothesis that the composition of a household’s portfolio of assets affects nutrition. In a final set of regressions, we considered the effect of the share of wheat in household diets on child health, as measured by their height and weight relative to WHO standards for children of the same age and gender.

Our findings support all three hypotheses. This has important implications. First, it links poor nutritional outcomes to the methods utilized by households to save against price and income uncertainty. Second, the insurance value of wheat stocks suggests that improved access to financial institutions will increase financial savings only if they offer a significant risk premium. This helps explain why the significant improvement in financial sector access in the country has not generated commensurate increases in financial savings. Finally, our research also helps reconcile two conflicting literatures. The first examines seasonality in consumption expenditures and generally finds that households are able to smooth consumption relative to income. In contrast, a second set of studies finds that children born in the monsoon months, and particularly those born in periods of low rainfall, have poorer health outcomes. We suggest that households are able to protect total food intake in the face of poor rainfall and other income shocks, but that this is achieved by increasing the share of stored grains of lower nutritional value. Thus, while they are able to maintain consumption levels and stave off hunger when incomes are low, nutrition suffers.

The findings of this paper suggest the importance of policies that help reduce price volatility, including the integration of agricultural markets. It also suggests that policies that increase the relative return to financial savings, such as flexible delivery options, lower transaction costs and financial literacy programs may also help improve nutrition. Finally, since the insurance value of wheat comes from its substitution for other crops, educating households on the value of a balanced diet may also affect household’s willingness to save in the form of stocks of wheat.

As part of our latest series of knowledge management sessions in our office, we had the pleasure of hosting Dr. Anjini who presented on this research. Here is the PPT from her session and below is the video of her talk.


[1] A child is considered stunted or underweight if his or her height for age or weight for age, respectively, is less than 2 standard deviations below that of children of the same gender and age in the reference population. The NFHS surveys calculate a household asset index that is the basis for comparisons across the wealth distribution.

[2] In neighboring states, this percentage varies from 27% amongst agricultural households to 30% amongst casual wage households.

[3] In contrast, in the southern states of Maharashtra, Tamil Nadu, Andhra Pradesh and Karnataka, characterized by lower levels of child malnutrition, while the share of expenditure on cereals (23%) is also high, only 6% of wheat consumption and 11% of rice consumption is from home stocks. Amongst agricultural households in these states, these percentages are 16% and 29%, respectively.

20
Jul

Reorienting Financial Well-being through FWR 2.0

fwr_img_1

By Dhivya S, IFMR Rural Finance

For an institution focussed on delivering high-quality and customised financial services to low-income households, the Wealth Management approach has been the one of the key underlying layers that is core to the KGFS Model. The sole objective of the approach is to maximise the financial well-being of households by offering tailor-made and suitable recommendations to them. Financial well-being, in this case, is to help customer households achieve financial goals, as per their priorities, in a secured and sustainable manner.

We have always believed that this approach of understanding the inherent composition and risks of rural households to help create their financial road-map requires a deep level of expertise. This approach of wealth management proves to be very different from the traditional approach of providing a full suite of products to customers and ask them to choose based on their financial needs. In this context, IFMR Rural Finance (IRF) first designed a Financial Well-being Report (FWR) five years ago to meaningfully engage with the bottom of the pyramid clientele.

The FWR is an automated customer-centric financial planning tool that uses customer data and back-end algorithms to make specific and actionable financial recommendations to enrolled rural households. The concept of devising wealth management conversations with customers takes us to the basic premise that there is a customer touch point that is primarily driven by the front-end staff called Wealth Managers. Wealth Managers across all KGFS branches rely on the FWR report as a guide to provide suitable financial advice aimed at enabling customers to realize their financial goals. To better understand the household, comprehensive data collection process by way of enrolments is undertaken, capturing data on the household’s members, demographics, cash-flows and financial goals. Post on-boarding the customer, a complete cash-flow analysis and risk assessment of the household is done to determine the customer’s financial position. While the data collection process forms the building block; the crux of wealth management however rests with capturing the financial goals of households. This is a continuous process of engagement with the customers that gets refined with subsequent conversations and financial transactions at the KGFS.

FWR in its current version has evolved over the years through a process of continuous improvement. However, there are several substantial improvements that need to undergo in order to make it even more customer-centric. Through anecdotal experiences and client interactions, we realised that wealth management isn’t simply about making financial plans for one’s future but is a means of realising one’s priorities. It is in this process, we realised that the process of customer engagement needs to be improved across various components as a precursor to having quality wealth management conversations. These conversations with customer focus on helping them achieve their financial goals by identifying and prioritizing what is most important to them. This then becomes imperative for us to make the entire process simple, intuitive and easy for both the Wealth Managers as well as the customers to achieve its full potential.

This post explains the perspective on the approach, methodology and design of FWR 2.0 and seeks feedback on improving the tool.

In the latest version of the FWR 2.0, the report aims to build on the legacy of the earlier system and is intended to be a significant leap towards delving much deeper into the financial lives of households. FWR 2.0 is engineered with concepts of Human Centered Design (HCD) to offer more practical and actionable insights keeping the customer interests at its core. The key areas of development that we plan to include in its redesign are:

1) Strengthening the data collection process to ensure high quality inputs from customers to have an in-depth understanding of their financial lives – Data collection is a fundamental process whose quality in turn determines the quality of financial advice given to the customer. For instance, if the Wealth Managers at KGFS are unaware of the customers’ high social expenses or if they have borrowed through informal channels or they are just a few thousands short of cash to achieve their goal; the Wealth Managers would not be able to provide appropriate recommendations which may have an adverse effect on the household’s financial lifecycle.

The scope of FWR 2.0 seeks to bridge the lacunae created through the development of various prototypes of a smart tool and associated processes. The aim is to identify the best technique of asking questions to customers during enrolment in a way that they are able to relate the most. We also plan to design a data quality score to explicitly measure the quality of data captured. This would be one of the key constituents to make sure that the entire wealth management process is based on sound first-principles.

2) Process redesign for achieving customer goals – Greater emphasis would be laid upon the quality of engagement with customers to enable them to reflect on their financial situation, identify and prioritize their individual & household goals. Assuming the data collected is of good quality, there are other important factors impacting the conversation that are to be rethought of. Some of the immediate alterations thought of are related to articulation of goals and logistics of organising wealth management conversations – for instance, should we have these conversations at home or at the branch; should we use laptops or just record customer stories and so on.

3) Better customer connect and usability through intuitive services – In regards to redesigning inclusive and progressive wealth management process, the aim is also to enhance the interface for mobiles and tablets through responsive web design and effective visualisation. The revamp would entail interface and visual improvements that are intuitive enough for the Wealth Managers to have meaningful conversations with customers.

We plan to finalise the above stated areas by creating various contending prototypes that aim to fulfil the stated objectives of FWR 2.0. These sets of prototypes would be tested in KGFS branches with existing and potential customers.  The revamp would entail conceptual, process and system related modifications that would be intuitive enough for both the staff and customers to equally participate in wealth management conversations. We are also scoping through the feasibility of creating a customer version of Financial Well-being Report that can be offered to the customer at the end of every conversation.

FWR 2.0 would not only aid in augmenting KGFS business performance and minimising business related risks due to improper or erroneous recommendation, but most importantly, would lead to an even more improved and meaningful customer engagement and retention.

14
Jul

Thinking about Micro-insurance Penetration and Entrenchment

Guest post by Renuka Sane

Background

Insurance contracts to lower income households (micro-insurance) are typically for one year. This implies that when the contract expires, the household needs to renew the purchase for the next year. It is intuitively appealing to consider that micro-insurance is truly an effective means of smoothing consumption when households continuously renew their contract. The question arises: do households choose to repurchase micro-insurance and enjoy continued cover? In Sane and Thomas (2016), From participation to repurchase: Low income households and micro-insurance, we evaluate this question for life and accident micro-insurance, along with what drives the repurchase. We also ask how long it takes customers to repurchase once the policy has expired. We especially focus on two such drivers: access to credit and wealth.

Data

We use data from the IFMR Rural Channels and Services Private Limited (IRCS) which implements the Kshetriya Gramin Financial Services (KGFS) branch-based model of distributing financial products across India. KGFS branches distribute two insurance products: the term life (TLI) which covers mortality risk, and the personal accident insurance (PAI) which covers mortality risk or permanent disability risk of customers arising due to accident. The data includes demographic and wealth information for 132,000 micro-insurance customers whose first policy expired between March 2011 and March 2014. The data also includes information about micro-credit contracts between KGFS and these customers prior to the purchase of the micro-insurance. To this dataset we add rainfall data gathered for the relevant districts and time periods, to indicate if the policy expired in a period when rainfall was scanty, versus when rainfall was normal.

Findings

We find that 65 percent of the sample renewed their insurance policy at least once, after their first policy expire. Five characteristics stand out:

First, there is a large difference in re-purchase probability (almost 33 percent) between the group with a micro-finance (Joint Liability Group) loan before the original purchase of the insurance policy, compared to those without a JLG loan. What could be the reasons?

When we examined the date of insurance repurchase and the take-up of a JLG loan, we find that 17 percent of those who renew insurance have taken a new JLG loan within 7 days of the insurance purchase, and another 18 percent have taken a new loan within 14 days of the insurance purchase. This suggests that while some part of the loan may be used to pay the insurance premium, it does not appear to be an over-whelming driver for the purchase, at least for two-thirds of those who renewed insurance.

A popular voiced perception is that life or accident insurance acts to protect the credit payments in case the borrower dies or suffers a debilitating injury. In this case however, most lenders would waive repayment of loans in the event of death of the debtor, giving customers little reason to purchase insurance to ensure repayment. Further, the insurance producer has nothing to gain from the point of view of repayment. There is little incentive for either intermediary to push the insurance product only to loan clients.

However, a common financial intermediary for credit and insurance may be important in other ways. Since credit and insurance are offered in the same branch, a higher demand for credit may translate into higher repurchase of insurance as customers visit the branch more frequently, and get more exposed to other financial products, and are perhaps able to build trust about the financial service provider.

Finally, there could be unobserved differences between those who have chosen to take a JLG loan and those who have not. It could be these differences that are driving the result, except that we are unable to test for this in the present data-set.

A second feature is that when the policy expires in months with scanty rainfall, the repurchase probability reduces by almost 7 percent. This is statistically significant at 1 percent. It suggests that collecting premiums during a lean period (caused by poor rainfall) restricts the ability to pay premiums.

The third feature is that repurchase probability rises with assets, but falls for those in the highest asset quartiles. This suggests that individuals only consider the purchase of insurance when they do not have enough buffer stock wealth. Households primarily demand life insurance when they lack accumulated reserves, or wealth, for self-insurance.

A fourth feature is that the largest number of repurchases occur within the first one to two months of expiry. Repurchases then continue to fall further after 12 months. This implies that if an insurance customer does not repurchase her policy within 12 months of expiry, she is unlikely to do so after. This helps to guide policy on improving insurance uptake: the first few months are the right time for an intervention to improve repurchases.

The fifth feature is that only 28 percent of those who repurchase the policy, increase the amount of cover purchased. We also find that 47 percent of those who increased their cover had gone from having one policy (accident cover, for example) to purchasing both policies (accident and term life cover).

Conclusion

Improving insurance participation of low-income households has become an important objective in the access to finance movement. The market for micro-insurance products will mature once people continuously purchase these products, and also make decisions on the sum assured purchased. Our research on understanding repurchases can provide inputs to the design of government programs as well as private sector initiatives. This is also the start of what we hope is an exciting research agenda on the drivers of sustained participation in micro-insurance.

References:

Sane and Thomas (2016), From participation to repurchase: Low income households and micro-insurance, FRG WP. http://ifrogs.org/releases/SaneThomas2016_microInsurance.html