The Economics of HOP

Paul F. Niehaus

February 2004

1

Contents

1 Introduction 4

2 Microfinance 5

2.1 Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2.2 Banking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2.3 Village Banking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2.4 Credit Alternatives for the Poor in the Developed World . . . . . . . . . . . . . . . . 7

2.5 Microfinance Opportunities in the Developed World . . . . . . . . . . . . . . . . . . 8

3 Income, Spending, and Credit 9

3.1 Consumption-Smoothing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

3.2 Moral Hazard . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

4 HOP and the Housing Market 11

4.1 The Housing Market: Substitutability . . . . . . . . . . . . . . . . . . . . . . . . . . 11

4.2 Public Policy and the Housing Market . . . . . . . . . . . . . . . . . . . . . . . . . . 13

4.3 Supersized HOP? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

4.4 Housing and the Law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

5 Optimal Repayment Rate Theory 18

5.1 Social Welfare . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

5.2 Probability of Repayment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

5.3 Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

5.4 Comparative Statics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

2

6 Group Liability and Individual Incentives 23

3

1 Introduction

This paper is meant to serve as an overview of topics that relate, more or less directly, to to theHousing Opportunities Program (HOP). It is intended primarily for people who are consideringvolunteering with HOP, and is strongly recommended to compers. More generally, the paper is foranyone who wants to understand the philosophy behind HOP. Since its founding in 1990, studentshave put a lot of critical thought into the program. They have identified premises on which theprogram rests, and tried to define and solve the decision problems that the program faces. Thispaper accumulates much of this thinking in formal language.

The paper begins with some comments on the microfinance movement, of which HOP is a part.Microfinance has created a lot of buzz in the past decade(s), and it is worth knowing what exactlymicrofinance is, what it isn’t, and where HOP fits into the spectrum of microfinance institutions(MFIs) that now dot the development landscape.

I continue with general considerations about housing markets, income and expenditure of low-income families, and credit constraints. These topics provide motivation for HOP, establishingthe existence of a social need, and rationale for HOP, identifying opportunities that may exist toaddress the need. Generally speaking they are meant to give you some intuition about why HOPexists and why HOP volunteers think their program is a good idea, compared to other possibleways of using time and money.

I move on to discuss issues that are specific to HOP. One section develops the concept of an “optimalrepayment rate” in mathematical detail. Readers who have not had intermediate microeconomicsmay want to skip the math, but should still try to get the intuition. The optimal rate conceptis basic to the way HOP evaluates loan applications, and you should fully grasp it in order toparticipate in group discussions about whom to lend to. The final section discusses group-lendingand village banking methods, and ways in which these models can be adapted and used by HOP.This section looks back to the earlier comments on the microfinance movement and draws on thehistorical experiences of other MFIs.

Comments, criticisms, questions, etc. are all welcome. This document is a collage of topics with atheme, and it is never meant to graduate from working-paper status. I can be reached before June2004 at niehaus@fas.harvard.edu.

4

2 Microfinance

[This section draws heavily on Jonathan Mor-duch’s paper ”The Microfinance Promise”1. Ihighly recommend Morduch’s paper to anyone in-terested in a more thorough overview of the his-tory and present state of the microfinance move-ment.]

2.1 Definitions

If finance is the borrowing and lending of money(abstracting away from the investment instru-ments and secondary markets that go along withborrowing and lending), then microfinance isabout “small” finance.

This does not mean that the institutions andorganizations that coordinate financial activityare small. Some microfinance initiatives arevery small — some are fairly autonomous villagecredit unions — but others are large conglom-erates like Grameen. Rather, “micro” usuallyrefers to the size of the average loan made bya lending institution. Technically this definitioncould get a bit hazy since, for example, a creditcard company may provide credit in small in-crements to many of its customers. What mostpeople mean by microfinance is something like,lending money in small increments to poor fami-lies who do not qualify for credit at conventionalfinancial institutions.

There is nothing about microfinance that makesit inherently a not-for-profit activity, or a for-profit activity, or a government program. All

1Journal of Economic Literature, v 37 n 4 (Dec. 1999),1569-1614.

these kinds of MFI exist, and there is a vigor-ous debate about whether the microfinance sec-tor will remain mixed or eventually become allone or all the other.

2.2 Banking

Regarding finance in general, perhaps the im-portant fact to keep in mind is that lending andborrowing as they are conducted in the UnitedStates and other developed nations today repre-sent the exception rather than the rule when onelooks across the globe and back through history.Liquid, functioning credit markets are rarities.The reason is simply that once someone else hasyour money, they have every incentive to findways to avoid having to ever return it. Withoutvery effective institutions to enforce debt con-tracts, lenders will lose most of their money. Al-ternatively, lenders will have to charge such highrates to cover default losses that no one will wantto borrow. In the developed countries this prob-lem has been solved in some cases with institu-tions that make lending possible.

One basic form of enforcement is contract lawthat allows lenders to secure their loans againstthe borrower’s collateral. If the borrower doesnot repay the loan, the courts allow the lenderto take possession of the collateral. If the jus-tice system functions well and without corrup-tion, this enforcement method allows people whoown physical property such as a house to borrowagainst it’s value.

In the absence of collateral, citizens of developedcountries are often still able to borrow using theircredit history as a guarantee that they will repay.In the United States, credit histories are main-

5

tained by a small group of companies that amal-gamate information from credit card companies,landlords, sellers who offer financing, and othersources into a record for each consumer. Bor-rowers have incentives to repay their loans be-cause non-repayment will hurt their credit his-tory, which will in turn hurt their ability to bor-row — or get a job or an apartment — in thefuture. Credit histories are a special case of areputational institution.

If enforcement is imperfect, a second problem inlending is distinguishing between potential bor-rowers who may be inherently better or worserisks. For example, suppose you are consideringinvesting either in a business run by entrepreneurA or by entrepreneur B. One may be significantlyless risky than the other, but both A and B haveincentives to underrepresent the riskiness of theirventure to you. Without some mechanism to dis-tinguish good risks from bad risks, lenders facea market for “lemons” and will again have tocharge a high rate just to break even. Again,reputational mechanisms can help mitigate theproblem by provide some record of the past per-formance of different individuals.

2.3 Village Banking

The above comments suggest that lending to thepoor, who have little or no collateral, would bedifficult. This is especially true if there are nofirms providing a reputational memory service.The facts fit this story. Microfinance has beenaround for a long time, but it was originally un-successful by any useful measure. Early microfi-nance programs from the 1950s on were generallypart of government strategies in less-developedcountries to eliminate poverty. The experiments

failed; repayment rates were low, the costs of theprograms high, and the amount of help actuallyreaching the poor — as opposed to the politicallyconnected — was low.

Recently, however, microfinance has experi-enced a change of fortune. New microlend-ing banks have achieved near-perfect repaymentrates while lending to clients without collateral.Millions of low-income families have gained ac-cess to credit that allows them to start smallbusinesses and work their way out of poverty.This recent change of fortune is due to the in-troduction of new lending models. Among otherthings, the new models help lenders (1) selectborrowers to lend to, and (2) enforce repay-ment from those borrowers. There is a greatdeal of heterogeneity among these models acrossdifferent MFIs, but collectively they have beendubbed “village banking”.

Village banking usually substitutes “social capi-tal” for the physical capital used to secure loansgiven to richer citizens. Social capital is a fairlyabstract concept with competing definitions, butfor our purposes it can be equated with relation-ships that a person has developed with his or herpeers. A relationship is valuable in a financialsense if the peer is able to observe characteris-tics of the person that a bank officer could not,and if they are able to put some pressure on theperson to take an action — such as repaying aloan — that the person might not want to do. Aperson with many strong relationships like thismay be able to credibly commit to repaying aloan, and this in turn will mean that they willbe able to get the loan in the first place.

How exactly is social capital used to secureloans? The technical details are best expressed

6

in game theory and I will omit them for thesake of brevity. Instead consider a simple group-lending model as an example. In a basic model,villagers who want to apply for a loan must forma group and apply jointly. After the loan is is-sued, if one member fails to repay their sharethe other members of the group become respon-sible for that share. Any member who does notpay what is expected of them, loses eligibility forfuture loans.

First, it can be shown that under these rulesapplicants will tend to form groups with otherapplicants who have similar “riskiness”. The in-tuition is that partnering with riskier types in-creases a “safe” type’s chance of having to shoul-der extra repayments.

Second, groups tend to enforce repayment in-ternally. Since default by one member hurtsthe others, they will use whatever peer pressurethey can exert to force that member to repay.If the relationships within the group are strongenough, i.e. if the value of the relationship tothe delinquent member is high enough, then re-payment will be enforced.

I leave the group-lending concept for now, butwill return to it in section 6 of this paper. Grouplending drives the recent and surprising successof microfinance banks worldwide; microfinanceprograms ignore the model at the risk of imitat-ing its far less successful predecessors.

2.4 Credit Alternatives for the Poor inthe Developed World

[This and the subsequent section combine expe-rience from HOP with comments from the excel-

lent overview “Replicating Microfinance in theUnited States: Opportunities and Challenges” byMark Shreiner and Jonathan Morduch.2]

Given the above discussion, one should expectmicrofinance in a developed country to be lesssuccessful precisely because more citizens inthese countries are already integrated into tra-ditional credit markets. However, there are stillcitizens of developed countries who do not havegood access to credit. Some have bad credithistories and are ineligible; others are unfamil-iar or uncomfortable with formal credit institu-tions. Neither reason necessarily makes them abad credit risk.

Low-income families in the U.S. take advantageof a variety of credit sources. If banks and creditcards are not options, they turn to pawn shops,check-cashing services, payday lenders, tax re-bate lending services, and some “fringe banks.”The cost of credit from these sources is typi-cally high. Another critical but often unnoticedsource of credit for poor families is borrowingfrom other family members or from friends. Thissort of borrowing is less visible but just as im-portant.

In public discussions about credit for the poorthere is apparent tension between acknowledg-ing need to charge higher rates to compensatefor higher risks, and accusations of extortion andusury. On the one hand, making small loans tothe poor inherently costs more than traditionallending because fixed costs are not as easily cov-ered and because the poor are less likely to ableto repay their creditors in adverse circumstances.

2Forthcoming as chapter 1 of Replicating Microfinancein the United States, ed. Jim Carr and Zhong Yi Tong,Washington D.C.: Fannie Mae Foundation, 2001.

7

On the other hand, some banks may use preda-tory practices towards poor and less-educatedclients. This often involves redlining, markingindividuals from poor or minority neighborhoodswith a red flag and then either rejecting their ap-plications or charging them higher rates. Otheroft-criticized practices are prepayment penaltiesand hidden balloon payments which make it dif-ficult to repay a loan, and flipping, offering refi-nancing with hefty fees.

Negative scrutiny of those who lend to the poorwill tend to discourage reputable people fromgetting into the business. Potential microlen-ders may not dare to charge cost-covering in-terest rates for fear of being labelled usurers. Incontrast, as a non-profit organization HOP doesnot have to cover costs; we seek to maximize so-cial impact rather than profit, and so we can loanmoney at low rates - currently 0%.

2.5 Microfinance Opportunities in theDeveloped World

Given the patchwork state of the credit indus-try serving low-income households, is there anopportunity for microfinance in the developedworld?

Microfinance for small businesses in the UnitedStates is a particularly difficult proposition.First, the economic challenges facing en-trepreneurs are different. In a developing coun-try, a startup business might expect to findless competition and less complementation in itsbusiness community. In the United States, sup-pliers and complementers are abundant but com-petition is also fierce. Whereas in Latin Amer-ica a small-goods store can find an immediate

niche, in the United States the same store hasto compete with hundreds of similar businesses,with huge supermarkets, and with online shop-ping alternatives. Many American markets aredominated by large corporations and chain busi-nesses which enjoy economies of scale and advan-tages of globalization. Generally speaking, thebarriers to entry to many industries, in termsof financial requirements which must be met tobe competitive, are high. There are also barri-ers in terms of human capital. Many industrieshave steep learning curves; on the level of themicroenterprise, business in the U.S. requires anoriginal product and marketing concept, educa-tion in accounting and negotiating, and a hostof other skills.

The U.S. economy also offers attractive alterna-tives to self-employment. There are more op-portunities to work for a wage as opposed to be-ing self-employed; there is also a more extensivepublic safety net (welfare and unemploymentbenefits) than exists in many poor countries.These conditions creates a minimum reservationwage for potential microentreprenuers. A highreservation wage means lower incentives for en-trepreneurism than in less developed countries.

In this context microloans to entrepreneurs have,unsurprisingly, proven relatively unsuccessful.Many of the ventures financed fail, so that repay-ment suffers and the clients fail to improve theirincomes. Further, evaluating business plans inthe United States tends to be more complicatedand costly than in rural villages. E.g. it is nothard to figure out if buying a milk cow is a goodinvestment, but it can be hard to predict theprofitability of a new barbershop downtown.

On the other hand, while HOP’s direct goal is

8

to prevent homelessness, it also functions indi-rectly as a method for sustaining productive em-ployment of individuals and of resources. Whentenants are evicted, they often are unable tokeep their jobs as they move, seek new hous-ing, or become homeless. It is true and remark-able that some homeless men and women go towork during the day, but holding down a family-supporting job is for the most part impracti-cable without housing or while seeking hous-ing. (There are also significant but less intan-gible efficiency costs of transition: loss of estab-lished community ties, the effort of learning anew neighborhood, and disruptions in childrenseducation, among others.) Meanwhile followingeviction the landlord loses the rent on his apart-ment during the ensuing transition period. Ifthe evicted client was chronically unable to paythen this is a necessary cost, but if the client wasevicted due to a temporary income fluctuationthen it is inefficient. By helping to prevent suchinefficiencies, HOP keeps people and resourcesproductively employed.

HOP can thus be seen as a demand-side lender tothe employed, as opposed to traditional supply-side microlending to the self-employed. HOPsupports its clients ability to pay for housing andthus indirectly supports their employment, edu-cation, etc. Since entrepreneurship is less em-powering in the U.S., less likely to lift the poorout of poverty, the HOP approach is a logicaladaptation.

The next section elaborates on the role that HOPlending plays in in household decision-makingand prospects. I will return in section 6 to thecritical question for microfinance in the UnitedStates: how to adapt village banking models tothe metropolis.

3 Income, Spending, and Credit

I offer a brief discussion of consumption subjectto credit constraints as it relates to the situationstypically faced by HOP clients. I also digress toconsider potential moral hazard problems asso-ciated with establishing a program committedto lending money to tenants in danger of beingevicted.

3.1 Consumption-Smoothing

The immediate rationale for HOPs operation isa response to the credit constraints which low-income households experience. Abstractly, theneed for HOP arises from the behavior of low-income tenants over time as they deal with finan-cial stress: medical expenses, job uncertainty,support of dependants, and so forth. One of themore stylized implications of the theory of utilitymaximization over time is that a person’s con-sumption during any given period depends noton income during that period, but on total ex-pected income over all periods. In strongest formthis principle is known as the lifetimes earningshypothesis, and it predicts a steady standard ofliving throughout life to the extent that such ispossible. Practically speaking, this means bor-rowing substantially during the years prior toone’s first job, repaying debt and saving duringemployment, and then spending again during re-tirement. Well and good, but we do not in factsee many ten-year-old’s buying cars and diningout!

Such discrepancies are accounted for by severalfactors, including uncertainty about the futureand, most significantly, the inability to borrow

9

T ->

$

->

Figure 1: Perfect consumption-smoothing: thelifetime earnings hypothesis.

money during periods of low income. This is re-ferred to as being liquidity constrained. In thecase of the ten-year-old it is not surprising thateven if the average citizen can expect a sizablesalary as an adult, the average child will notbe able to borrow based on that expectation.As a result we observe imperfect consumption-smoothing. This is why children remain econom-ically dependant on their parents.

T ->

$

->

Figure 2: Lifetime consumption-smoothing withcredit constraints.

Now HOP is also concerned with individuals fac-ing liquidity constraints, but at a mature timein life when a stable level of consumption mustbe independently sustained. This is also a time

when the uncertainties of childhood are replacedby the hard facts of adulthood: earning power,credit history, and unfortunately perhaps racialbackground. These factors determine each per-sons access to credit. Individual behavior stillexhibits consumption-smoothing, while liquidityconstraints and uncertainty cause a degree ofvariation. This is illustrated by the third figure,

T ->

$

->

Figure 3: Smoothing over income fluctuationswith credit constraints.

which represents consumption patterns during aperiod of low income (eg. job loss) or abstractlyduring any other exogenous financial stress. In-tuitively this means that, after being laid off orduring an expensive illness, the rational indi-vidual will not stop consuming altogether, butwill scale back consumption, especially of non-essential ‘luxury items.’

For HOP clients this response is not an option;there is a certain minimum level of consump-tion of essentials below which they cannot sim-ply ‘scale back.’ First, because most of ourclients earn just enough to pay for food, cloth-ing, and shelter, they do not have ‘luxury goods’in their monthly budgets which they can elim-inate in response to a loss of income. Second,they are usually consuming essentials at a sub-sistence level - enough food to stay alive, enough

10

clothing to keep warm in the winter, and thecheapest apartment they can find. Thus there islittle room to scale back consumption of essen-tials either. Graphically we can depict this con-straint with a subsistence line equal to or slightlybelow the level of average income. Economically,a lack of liquidity forces poor families below thisline. This means that our clients must look atfood, clothes, and housing, and choose two of thethree. The result is usually homelessness.

HOP attempts to alleviate this situation by re-moving the liquidity constraints which force ourclients’ level of consumption to dip below thesubsistence level. Our loans allow low-incomehouseholds to spend the amount of income whichthey will average over an extended period, ratherthan constraining them to what they are cur-rently earning.

3.2 Moral Hazard

It is worth noting that if HOP lending has po-tential benefits, it also has potential for abuse.Moral hazard is economics jargon for some sortof “bad” or “reckless” behavior that individualsare induced to take because of a deal they made.In this case, the “deal” is HOP’s implicit com-mitment to help people who fall behind on theirrent payments.

The hazard is that if individuals know HOP hasmade this commitment, they may not make asgreat an effort to avoid rent arrearages. For ex-ample, suppose that I wish to purchase an ex-pensive new TV. I cannot afford to pay out ofpocket, but the store offers 10% financing. I canactually do better as follows: the next time mylandlord comes to collect rent, I tell him I cannot

pay and apply to HOP for a loan. If my appli-cation is refused, I cough up the cash that I hadall along. If on the other hand my application isaccepted, I give my landlord a HOP check andspend my rent money on the TV. I repay HOPat 0% interest — or perhaps not at all — andend up with a TV for cheap.

This is of course an exaggeration. Eviction is afairly catastrophic event and any chance that theabove plan could fail might make it not worthtrying. But moral hazard is a real possibilitywhen many individuals know about HOP and areable to keep their financial information relativelyprivate from interviewing loan officers.

4 HOP and the Housing Market

To understand the role that HOP plays, wefirst considered its impact on individual decision-making subject to variable income and con-strained credit.

We now broaden our horizon to include the entirelocal housing market. HOP currently operateson a scale small enough that its impact on thehousing market is negligible. However, publicpolicies at the federal, state, and city level haveplayed major roles in shaping the housing mar-ket. We will discuss these policies and their ef-fects in brief to understand the environs in whichHOP operates. We will then consider the impactthat HOP would have if it were scaled up.

11

4.1 The Housing Market: Substitutabil-ity

The“housing market” is not a single market fora homogenous good, but a continuum of mar-kets for housing of different sizes and qualitieswhich are imperfect substitutes for each other.The market for studio apartments in Cambridgeis affected by the market for duplexes in Rox-bury and by the market for ranch-style suburbanhomes in Danvers. High prices in one of thesemarket niches will raise demand in the others,though not as much as if the various housingunits in question were identical.

We can imagine a simple two-good model of thisheterogenous market in which two kinds of hous-ing are sold, a high type and a low type. Thehigh type is characterized by more floor space,better quality of construction, more amenities,more desirable location, etc. all of which the lowtype lacks. It follows that in equilibrium andabsent any policy intervention the low type willcost less than the high type of housing. Whetheror not the equilibrium price of low-type housingwill be “affordable” according to any exogenousstandard of affordability, remains uncertain.

Supply of both types of housing depends onlyon the price that producers/landlords receive forhousing units, and their costs of producing andmanaging those units. Demand for each type ofhousing, however, depends both on the incomeprofile of the local resident population and on theprice of the other type of housing. Supply is rel-atively inelastic in the short run because the de-velopment of new housing units takes time (andbecause rental properties generally have fairlylow vacancy rates, so that changes in apartment-search behavior cannot do much to offset an in-

P'

QQ*Q'

P*

P supply

D(Phigh

)

D'(Phigh

)

High-Type Market

P'

QQ*Q'

P*

P supply

D(Phigh

, Plow

)

D'(P'high

, Plow

)

Low-Type Market

Figure 4: Substitutability between high- andlow-type housing.

crease in demand for housing.)

Now imagine that a boom in high-tech indus-tries increases the demand for high-type housingamong affluent young workers. This was the sit-uation in and around Boston through much ofthe 1990s. This significantly increases the priceof high-type housing (due to the short-run inelas-ticity of supply). This in turn raises demand forlow-type housing, which is an imperfect substi-tute. The result is an across-the-board increasein the prices of housing. Since the prices of differ-ent types of housing move together in this way, itmakes some sense to speak of a unified “housingmarket.”

Note that the economic causality describedabove is different from the phenomenon that so-ciologists term “gentrification.” This is a processwhereby a small pocket of the young noveau richechoose to move into an older, dilapidated inner-city neighborhood. Their increasing presencegradually restores the neighborhood to a newfashionability. Consumer stores, coffee shops,and of course more wealthy tenants arrive indroves, and the poorer former inhabitants of theneighborhood are priced out.

12

Gentrification could be seen as resulting fromchanges in tastes, not income (though of coursethe two changes may be connected). The de-mand for high-type housing falls while demandfor low-type housing rises due to the new fashion-ability of urban living. The result is a new equi-librium in which the price of high-type housingis lower than before, while the price of low-typehousing is higher.

Though the drivers of change are different, theend result for low-income families is similar: theprice of low-type housing rises.

4.2 Public Policy and the Housing Mar-ket

When the price of low-type housing rises, low-income families have to pay more for a place tostay. In current policy debates the term “afford-ability” is used to describe the situation wheretypical housing costs are below some fraction ofarea median income. This means that for house-holds far below area median income, housing willonly be affordable at very low prices. A typicalcutoff is 30% of median income, which defines“30% affordability”. “50% affordability” is de-fined analogously.

The housing market was a topic of political de-bate long before “affordability” was defined. Atthe turn of the 19th century, journalists andactivists publicized the living conditions of thepoor in urban centers such as New York andChicago. Some of these writers called for govern-ment intervention in the housing market. Otheractivists such as Jacob Riis opposed governmentintervention and called for the intervention ofprivate philanthropy (Riis urged landlords to

adopt a motto of “philanthropy and 5%,” mean-ing generosity to the poor with a below-marketrate of return on their properties.) Since thattime, government at different levels has used var-ious policy tools to intervene in the housing mar-ket: regulations, price controls, subsidies, anddirect public financing.

I will briefly discuss these policies in turn here.The discussion focuses explicitly on the impactsthat various government interventions have onthe housing market: prices, quantities, quality,shortages, etc. Some of these policies have ef-fects outside of the housing market. Taxes andsubsidies affect government balance sheets andmay indirectly lead to changes in funding forother programs. Regulations have both directand indirect effects. The purpose here is not toconsider the effects of policies in their entiretyand endorse or reject them, but only to under-stand their impact on the housing market.

Building Codes and Zoning

Building codes and zoning are the earliest formsof government intervention in the housing mar-ket. Building codes which require fire safetymeasures, limit occupancy, etc. date from theprogressive era. In part these laws were moti-vated by a desire to protect tenants (as well asemployees in workplaces) from the health andfire risks of their cramped living situations. Theywere also popular with the rich, who feared thatthe slums would become breeding grounds fordisease or start fires which would then spreadinto the better quarters of the city. Zoning alsohad strong support among the affluent classes,who wanted to preserve distance between them-selves and the slums.

Building codes had the unintended consequence

13

of raising the price of low-type housing. Simplyput, compliance with the codes cost money. De-velopers who had to build fire escapes chargedmore for the buildings. And when landlordswere required to limit the occupancy of theirrooms (to prevent unhealthy overcrowding) theyresponded by raising the rent per person. Build-ing codes have less impact on high-type housing,because those units are more likely to alreadycomply with the codes. Wealthy families willgenerally be willing to pay more for higher qual-ity, but poor families may be unable to do so.

The same remains true today. Building andhealth codes enforce safety standards but raisethe cost of building and maintaining housing.This is one cause of homelessness. When a fam-ily can afford half the cost of renting an apart-ment and is willing to squeeze into the space withanother family in a similar situation, they canboth be housed. But when that kind of crowd-ing is illegal, both families become homeless be-cause they cannot afford the minimum amount ofspace per person that the law requires. In someparts of Boston, for example in crowded immi-grant communities, this kind of informal sublet-ting is common. Similarly, some families may beable to afford an apartment that is built fromcheap but flamable materials. When the law re-quires that developers use more expensive, lessflamable materials, that family cannot buy legalhousing.

Zoning was also introduced in the early 1900s asthe first real policy tool at the disposal of ur-ban planners. Zoning has a specific economicfunction: it can be used to limit external ef-fects of property ownership. If a businessmanbuilds his factory next to your home, the noiseand smoke generated will undoubtedly devalue

your property and generally make your life mis-erable. Many zoning rules are intended to sep-arate such “incompatible uses” of land. Othersare not. For example, some municipalities limitthe fraction of a property that can be built-on toensure that open space is preserved; other ruleslimit the height of all buildings in a neighbor-hood. Such zoning codes operate like buildingcodes: they enforce a particular vision of “qual-ity” but in doing so raise the price of housing byconstraining its supply. Developers are unableto respond to increasing demand for housing bybuilding more if it is illegal to build either up-wards or outwards. Families that might be ableto afford an apartment built onto the 7th floorbecome homeless if a 7th floor is illegal. This isa major issue in Cambridge today, where build-ing heights are far less then what property valueswould predict because of the difficulty of gettingnew development approved by the city.

In the worst cases, some towns intentionally zoneout low-income families. For example, some sub-urbs require that land be sold and developedonly in increments of several acres, which pricesout the poor.

A related recent development is the “SmartGrowth” agenda, which resembles zoning at aregional level. “Smart Growth” encompassesmany elements, but is esentially a planning re-sponse to criticisms of urban“sprawl.” Thesecriticisms include use of vast quantities of land,loss of farmland, environmental harms, auto-dependency with heavy traffic, and other moreaesthetic criticisms such as the drabness of sub-urban subdivisions. Typical “smart growth”policies include urban growth bounds (whichprohibit new development beyond a perimeter)and set-asides of farmland and parkland. It

14

should be clear what the impact of such regu-lations is on the housing market. They reducethe quantity of land available for development,raising the cost of housing. Smart growth poli-cies are sometimes particularly biased againstlow-income housing. For example, rules banningmobile home parks, which are seen to epitomizethe ugliness of sprawl, force the poorest home-owners to return to the rental market or becomehomelessness.

Depression-Era Federal Programs

[Much of this information is drawn from Ken-neth Jackson’s “Crabgrass Frontier,”3 a worth-reading history of the suburbanization of theUnited States.]

A whole set of programs, inspired and fundedat the federal level, date back to the era of theGreat Depression and the Roosevelt administra-tion. The Depression wreaked havoc on the fi-nancial sector and on the home mortgage indus-try in particular, prompting federal intervention.Some of these programs remain today; all areimportant for understanding the current state ofthe housing market. Several have rather notori-ous track records.

Of some lasting importance was the Home Own-ers Loan Corporation, which refinanced manymortgages that were in danger of default. HOLCeffectively subsidized housing for eligible Amer-icans. It also introduced standards for ratingthe credit-worthiness of different neighborhoods,which unfortunately were in part racial drivenand introduced the practice of “red-lining” mi-nority neighborhoods (under HOLC’s system thepresence of black residents had a negative effect

3New York: Oxford University Press, 1985

on the credit rating of the neighborhood).

Perhaps the most important program was theFederal Housing Administration, which under-wrote loans from private banks to first-timehome purchasers. This intervention effectivelysubsidized home ownership (and thus home con-struction). This program is likely one of the rea-sons that ownership/rental ratios in the UnitedStates are much higher than in many other de-veloped countries.

Finally, the Depression saw the beginnings ofthe public housing and slum clearance movementunder the guidance of the United States Hous-ing Authority. Cities that declared low-incomeneighborhoods “blighted” could legally seize theland by eminent domain and then receive Federalsubsidies to redevelop it. In many cases theseprovisions were used by cities to try to restoretheir tax bases; in some worst cases low-income“slums” and minority neighborhoods were re-placed with fewer total units of more expensivehousing that the former residents could not af-ford. In other cases, public housing simply con-centrated the poorest tenants into small areaswith high crime rates and low social capital.

An attempt has recently been made to reformpublic housing through the HOPE VI program.Projects have been redeveloped to provide someprivacy, a mixture of residential and commercialneighborhoods, a mixture of income levels, andother aspects of more organic neighborhoods.

Public housing programs have also been criti-cized because they crowd out rather than sup-plement private development. Subsidy programssuch as the FHA or Section 8 vouchers (below)have the opposite effect, tending to stimulate pri-vate production.

15

Rent Control

Rent control is a common form of governmentintervention in the housing market. Rent controlis a price ceiling, a legal restriction preventingrents from rising about a given level.

Rent control has the same effect on the hous-ing market that price ceilings have on any mar-ket: it decreases supply and increases demand.[graph - both types] However, because the supplyof housing is relatively inelastic in the short run,the decrease in units supplied is not noticeable(no one tears down apartment buildings in re-sponse to a rent control ordinance). This makesrent control a politically attractive instrument,but does not change the long-run reality. Newhousing does not get built, and old housing doesnot get repaired, if the profits that owners ofhousing can earn are reduced or eliminated byrent control. The funds that would have investedin such projects are instead invested in anothermore profitable industry. In the long run, rentcontrol reduces the stock of available housing.

Rent control is politically attractive because ittransfers wealth from the owners of rental prop-erties to the tenants. Tenants represent a muchlarger voting block than landlords, so politicalsupport for rent control will often exist. Mean-while alternative policies, such as publicly fund-ing housing or subsidizing housing, impose costson taxpayers and are therefore less politicallyattractive (since taxpayers are a large votingblock).

Tenants as a class benefit from rent control, butnot all potential tenants benefit. Tenants in rent-controlled housing have strong incentives to stay,even if they would prefer to live elsewhere or nolonger need an an apartment of the given size.

When tenants do leave, landlords have many ap-plicants for each available apartment. But theyare forbidden to raise the price until only oneapplicant is willing and able to pay. In this sce-nario, landlords may use other factors to decidewho gets the apartment. They may rent only topeople of their race or gender, or only to fami-lies, or they may base their rationing decisionson any other factor that suits them.

Section 8 Vouchers

The federal program most relevant to HOP’swork is the Section 8 housing voucher program.

Currently, the voucher program receives lessfunding that would be necessary to providevouchers to every family that qualifies under theprogram’s means-testing. The resulting short-age of vouchers is handled by rationing vouchersthrough waiting lists. Many HOP applicants arecurrently on the Section 8 waiting list. Manyothers urgently need to use their vouchers be-fore a time limit expires and they return to thewaiting list.

The voucher program has advantages over otherforms of housing assistance. First, vouchers arefunded by federal tax revenues or by borrowing.This means that the costs of funding the voucherprogram are both widespread and apparent. Wehave seen that this is not the case with otherforms of housing intervention such as rent con-trol. Second, vouchers give greater freedom toindividual families to choose where to live. Ifthey wish to spend a little bit more for a biggerapartment, or a little bit less on a smaller one,they are able to do so. This is not the case withpublic housing (which tends to be fairly homoge-nous) or rent control (which encourages familiesto stay in any rent-controlled apartment they can

16

secure).. Third, vouchers stimulate private in-vestment in new units of low-type housing, whichis a sure way to increase the stock of “affordable”housing.

4.3 Supersized HOP?

Having considered the impact of public policieson the environment in which HOP operates, wenow turn to the effects of HOP lending itselfon the housing market. Currently these effectsare negligible because HOP operates on a smallscale. We consider a hypothetical scenario inwhich HOP expands and provides credit to a sig-nificant percentage of Boston’s low-income ten-ants.

Assume that landlords care only about rentalrevenues — that is, assume that they do not careabout the welfare of their tenants. This does notalways hold but it is a reasonable approximation.Then landlords decide whether or not to evict anindebted tenant by weighing the likely gains andlosses from doing so. The gains are the revenuethat a new, paying tenant would provide. Thechief loss is an opportunity cost, increasing inthe time the apartment sits empty before a newtenant is found and in the probability that theold tenant would have been able to resume pay-ing during that time. Another potential loss islegal fees if the current tenant sues to preventeviction.

Obviously, HOP exists to tip the cost-benefitscales and induce landlords to choose not to evictcurrent tenants. The primary purpose for inter-vening is to improve the situation of the tenant.But if the landlord chooses not to evict wherepreviously he would have evicted, then HOP’s

intervention has also made the landlord better-off. Generally speaking, HOP on a large scalewould serve as a kind of insurance policy for ten-ants at risk of falling behind in their rent — butthis would also amount to insurance for land-lords against tenant nonpayment. Both partieswould benefit.

In other words, HOP’s intervention has a pos-itive effect on the returns to being a landlord,or more generally the returns to owning rentalhousing. No one knows what the magnitude ofthis effect is. It is certain, however, that a larger-scale program would have a greater impact onthe profitability of building and owning low-typehousing. Whether the impact would be appre-ciable is an open question.

4.4 Housing and the Law

Earlier I mentioned the health and constructioncodes which regulate the construction and rentalof apartment housing. One significant aspectof these codes is that, under current law, ten-ants who are threatened with eviction for non-payment of rent can sue to forestall that evic-tion if they can prove that the apartment theyrented was in violation of the codes. For exam-ple, if there is a family of rats in the walls of thebathroom, the tenant may be legally entitled toremain in their apartment even if they have beendelinquent in paying their rent.

In the past HOP has been indirectly involvedwith legal proceedings of this nature. The SmallClaims Advisory Service (SCAS), with whichHOP shares an office, occasionally helps ten-ants win forestallment of eviction on groundsas above. Also, in 2002 a HOP member helped

17

the Tenants Rights Network of Boston developa website which informs tenants of their legalrights and helps them find resources and advice.

What the rights of tenants vis-a-vis their land-lords should be, is a contentious question. I willset aside the moral question to focus on positiveaspects. What are the effects on the housingmarket of the current allocation of rights, com-pared to alternatives? What is the impact ofprograms like SCAS or the Tenants Rights Net-work?

A first consideration for understanding law andeconomics is that legal activity per se is rent-seeking activity. Legal proceedings do not gen-erate more goods and services for the peoplewho participate in them. Instead they arrangefor some redistribution of values between theinvolved parties. Law is a essential to a well-functioning society in part because it protectsrights which should not be subject to commod-ification. Law is also important for economicoutcomes because it enforces rules that help peo-ple cooperate without fear of being cheated. Butthe courts function best when they achieve thosegoals simply by creating the threat of legal ac-tion, not with time-consuming and expensive lit-igation.

The laws governing the tenant-landlord relation-ship currently give expansive protection to thetenant, in some circumstances allowing him/herto stay in their apartment without paying rent.Apparently many evictions can and are legallyforestalled because tenants are made aware ofthis fact and take legal action. What is the im-pact of increased legal activity along these lines?Such actions are costly for landlords, taking uptime and money. If compliance with the codes

is cheaper than the costs of litigation (includ-ing possible fines) then the landlord will usuallycomply. But litigation also makes the overallprospect of owning and/or managing low-typehousing less attractive. Landlords find that theyare forced either to spend money improving theirapartments or to let tenants live for free afterunfavorable court rulings. Housing units thatwould be marginally profitable when the levelof legal activity is low become unprofitable tooperate when legal activity is frequent and willbe taken off the market. Investors will be waryof low-type housing when confronted with thecatch-22 of expensive code compliance or expen-sive litigation.

In the end, legal action against landlords benefitsthe the individual suitors (who often win per-mission to stay in their apartments) but hurtsother tenants who find that the housing stockshrinks while rents increase to cover legal ex-penses. Therefore, there are negative external-ities associated with legal action. These costsare not internalized by individual tenants ortheir lawyers/advocates because in each partic-ular case they have the welfare of one family inmind. But in aggregate, litigation may imposemore costs on low-income tenants as a groupthan it will capture benefits. From the pointof view of low-income families as a group, an in-crease in legal action may be harmful.

Legal action vs. lending and subsidy is notan academic question. Social workers, includ-ing HOP volunteers, face regular choices aboutwhether to pursue a remedial relationship with alandlord or an adversarial one. Both approachescan reap benefits for the individual tenant(s);the difference is in the effect on the landlord’sincentives to provide low-cost housing.

18

5 Optimal Repayment Rate Theory

One simple indicator of HOP’s performance isthe repayment rate which we get on our loans,or the amount of money that is repaid dividedby the total amount let. Historically our repay-ment rates have been poor, typically no morethan 10% - 20%. Higher repayment has alwaysbeen a program goal. This begs the question,how much higher? What repayment rate wouldbe “right?”

The quick answer is of course 100%. If we get allour money back every time, then a single dona-tion of $500 can be used over and over again anunbounded number of times. The total impactof the initial $500 would then be almost unlim-ited. (More accurately, it would be very largebut bounded and would depend on the rate atwhich housing prices increase. See below.) How-ever, I will argue here that this is an oversim-plification of the situation, and that the optimalrepayment rate may be less than 1.

To understand the reasoning behind an optimalrepayment rate you really only need to know twothings. First, HOP wants to lend to the neediestfamilies, which are also the riskiest. Second, lowrepayment reduces the impact we have over thelong term. There is an obvious tradeoff here be-tween helping the neediest families, and helpingfamilies that will repay us so that we can use ourmoney again. The purpose of this chapter is tocapture that tradeoff mathematically and thinkabout what factors might affect the optimal levelof repayment.

5.1 Social Welfare

We will start with a simple model of HOP’s pref-erence for helping poorer rather than richer fam-ilies. Imagine a world populated by n people;suppose that for each person 1 ≤ i ≤ n we de-note the income of that person by Yi. Supposefurther that we index persons in order of theirincomes, so that

Y1 ≤ Y2 ≤ . . . ≤ Yn

We would like to have some quantitative measureof the welfare of the group as a whole; we callsuch a measure a social welfare function. Per-haps the simplest choice of a social welfare func-tion is the sum of individual incomes,

W (Y1, Y2, . . . , Yn) = Y1 + Y2 + . . . + Yn

This function defines a preference orderingover all possible combinations of individual in-comes. However, this ordering lacks any income-egalitarian tendency, which can be seen from thefact that

W (0, 0, . . . , n) = W (1, 1, . . . , 1)

Even more relevant for HOP, a transfer t of in-come from a rich person to a poor person reducessocial welfare if there is any cost c > 0 associatedwith the transfer:

W (Y1, Y2, . . . , Yi + t, . . . , Yj − t− c, . . . , Yn)

= W (Y1, Y2, . . . , Yn)− c < W (Y1, Y2, . . . , Yn)

This would make any welfare-improving inter-vention by HOP impossible, since HOP loans al-ways involve some overhead costs. We need tointroduce some income-egalitarian feature to ourwelfare function. One possibility is to make the

19

welfare function concave in individual incomes(for example, by considering utilities instead ofincomes, or by choosing a totally different W (·).)An even simpler method that I will use is toweight each individual income Xi by a factor0 ≤ αi ≤ 1, and make our first normative as-sumption.

Assumption 1: αi < αj ⇔ Xi > Xj

This minimal condition places higher weight onchanges in the income of poorer people. If wethen define

W (Y1, Y2, . . . , Yn) = α1Y1 + α2Y2 + . . . + αnYn

we have a social welfare function that can in-crease with a costly transfer from rich to poor:

W (Y1, Y2, . . . , Yi + t, . . . , Yj − t− c, . . . , Yn)

= W (Y1, Y2, . . . , Yn) + αit− αj(t− c)

> W (Y1, Y2, . . . , Yn) iff αi > αj(1−c

t)

Finally, we will be a bit more specific about howwe choose α. Let α(Y ) be a strictly decreasingfunction of Y that is differentiable (hence contin-uous), and for any person i let αi = α(Yi). Forany finite collections {Yi}, {αi} we could choosea function α(·) that “fits” these points. So this isa reasonable extension from our discrete modelto a continuous analog. I will in particular re-quire that α : R → [0, 1] is onto, i.e. it spansits entire range; this requires that α(0) = 1 andα(1) = 0.

5.2 Probability of Repayment

Consider a household that borrows from HOP.To justify HOP’s lending, we have argued that

this household may be credit-constrained, whichmeans that they can only pay bills if they havecash on hand to do so. If a family is credit-constrained before they borrow from HOP, theywill still be credit-constrained after they borrowfrom HOP. This means that they will only beable to repay their loan if they have cash avail-able to do so. If we think that households havefinancial ups and downs — which are called pos-itive and negative shocks — then there is someprobability less than 1 that the household willdo well enough after the HOP loan to be able topay it back. Formally, if we think about house-hold wealth as a random walk, as in section 1,then we can associate some probability less than1 with the event, “the family is able to repay.”

However, a family that is able to repay does notnecessarily do so.

Suppose that HOP has no means of enforcing re-payment; we do not take clients to court, we donot put comments on their credit histories, andwe have no other leverage on our client. Tradi-tionally this has been the case with almost allHOP clients. Clients will only pay back if theyfeel morally or subconsciously obliged to do so,and if that impulse is stronger than the desire touse $500 for themselves. Different people will be-have more or less morally in this situation, butthinking about the population as a whole, call0 ≤ δ ≤ 1 the probability that a person willpay back a loan that they can afford to repay.Formally, δ = P (repay loan | able to repay). Itis here that we make our second ideological as-sumption:

Assumption 2: δ is independent of Y .

In English, this assumption says that there is nomoral difference between rich people and poor

20

people. Put in the same situation, and facingthe same external incentives to repay $500, richand poor people are equally likely “do the rightthing.” To explain repayment rates, then, wehave to think about other things: external in-centives to repay, and ability to repay in the firstplace.

We will return in section 6 to discuss ways ofenforcing repayment, but for now fix δ, and letvary the ability of a client to repay a loan. Ifr = P (able to repay) then

P (repay loan) = P (repay | able)P (able)

= δr

From the law of large numbers it follows thatover many loans the average repayment rate willbe close to this probability. We will thereforethink of the quantity δr as the average repay-ment rate on HOP loans.

What determines the probability r that a familywill be fiscally able to repay a loan? Roughlyspeaking, r depends on the possible combina-tions of incomes and expenses that the familymight have, and the probabilities of each. Wewill abstract away from all this detail and thinkabout r as an increasing function of Y, house-hold income. Households with higher incomesare more likely to be able to repay, householdswith lower incomes less likely. Let r = R(Y )with R′(Y ) > 0. R is strictly increasing andhence invertible, and we have

r = R(Y ) s.t. R′(Y ) > 0

αi = α(Yi) s.t. α′(Y ) < 0

⇒ define α(r) = α(R−1(r))

Then α′(r) = α′(R−1(r)) ddrR−1(r) ≤ 0, so α

is a decreasing function of r. Intuitively, peo-ple who are more likely to be able to repay arealso wealthier and thus do not receive as high aweighting in our social welfare function.

5.3 Optimization

We have now finished all the work we need to setup an objective function for HOP.

Suppose that on average HOP’s loans provide abenefit B to the recipient family, and that this Bdoes not depend on income levels. We can thinkabout B as the utility of not being evicted, withall its component parts: not losing a job, nothaving to take kids out of school, etc. A HOPloan also provides a benefit D to the client’s land-lord, who does not lose rent money that he/sheis owed and does not have to go through the pro-cess of screening potential replacement tenants.The total benefit from a single loan, weightedaccording to our social preferences, is therefore

α(rtenant)B + α(rlandlord)D

If each loan is repaid with probability δr, and ifwe start with enough capital to make one loan,then the total number of loans that can be madeover time is

1+1δr

+1

(δr)2+

1(δr)3

+. . . =∞∑

k=0

(1δr

)k

=1

1− δr

If we also consider that housing prices rise overtime, then we should adjust this figure downwardslightly. Let i be the percent increase in housingcosts during the time it takes to issue and collectrepayment on one loan. Then (assuming loans

21

increase in size to match housing prices) the totalnumber of loans made will be

∞∑k=0

(1

δr(1 + i)

)k

=1

1− δr(1 + i)

It is easy to see that if all we cared about wasthe number of loans issued, we would want to setr as high as possible. If we want to maximize thesocial impact of our lending, however, we have tomaximize the total impact of our initial capital:

I(rtenant) =α(rtenant)B + α(rlandlord)D

1− δrtenant(1 + i)

To simplify the math that follows, I will make theassumption that α(rtenant) ∝ α(rlandlord), whichseems reasonable since richer tenants are likelyto have richer landlords. This lets us express thenumerator as α(rtenant)B̃ where B̃ is some otherconstant. The problem is now

maxr

I(r) = maxr

(α(r)B̃

1− δr(1 + i)

)

where by r I mean rtenant. The first-order con-dition is

α′(r)B̃1− δr(1 + i)

+ δ(1 + i)α(r)B̃

(1− δr(1 + i))2= 0

⇒ (α′(r))(1− δr(1 + i)) + (δ(1 + i))(α(r)) = 0

⇒ α′(r)α(r)

=δ(1 + i)

δr(1 + i)− 1

Note that for an optimum to exist, it must holdthat δr(1 + i) < 1. If so, both sides are negativeand

−α′(r)α(r)

=δ(1 + i)

1− δr(1 + i)> 0

and thus

−α′(r)α(r)

= − d

drlog(1− δr(1 + i))

⇒ d

drlog(α(r)) =

d

drlog(1− δr(1 + i))

Substitute dummy variables and integrate bothsides:∫ r

0

d

dulog(α(u)) =

∫ r

0

d

dulog(1− δu(1 + i))

⇒ log(α(r))−log(α(0)) = log(1−δr(1+i))−log(1)

Since we assumed α(0) = 1, this implies

α(r) = 1− δr(1 + i)

orr∗ =

1− α(r∗)δ(1 + i)

Now recall that r is the probability that a clientis able to repay, and that the actual observed rateof repayment is the product δr, so that optimalobserved repayment is

δr∗ =1− α(r∗)

1 + i(1)

The numerator is less than 1, and the denomina-tor is greater than 1, so the entire expression isless than 1. In other words, the repayment ratethat maximizes HOP’s social impact is less than100%.

The intuition here should be clear. We have as-sumed that HOP can only raise it’s repaymentrate only by lending to people who are morelikely to be able to repay. In other words, wehave held constant people’s willingness to repayif they are able. We also suggested that peoplewho are more likely to be able to repay, are alsoless needy. Under these assumptions, the repay-ment rate that maximizes HOP’s social impactis less than 100%, reflecting a tradeoff betweenhigher repayment and targeting the neediest.

22

5.4 Comparative Statics

In this section we consider how the optimal re-payment rate changes with changes in δ and i.Because we defined r∗ implicitly, none of theseare obvious. In fact, without making some as-sumptions about α(·) these changes are of am-biguous sign.

Rewrite (1) as

(1 + i)δr∗ = 1− α(r∗)

Then differentiating with respect to δ gives

(1 + i)r∗ + (1 + i)δ(

∂r∗

∂δ

)= −α′(r∗)

(∂r∗

∂δ

)and therefore

∂r∗

∂δ=

−(1 + i)r∗

α′(r∗) + (1 + i)δ

The denominator could be of either sign, so theeffect of changing δ on r∗ is unclear. Similarlyfor i,

∂r∗

∂i=

−δr∗

α′(r∗) + (1 + i)δ

and the sign is ambiguous.

6 Group Liability and Individual In-centives

Our conclusions about what level of repaymentto aim for were based on a given probability pthat the client will be able to repay but willchoose to default. Minimizing this chance is aseparate but important program objective. It isacceptable for HOP to take risks on truly needyindividuals, and expected that they will at times

be unable to repay; but it is unacceptable forclients who could pay to instead default, andthus drain money out of the loan cycle. It istherefore critical for HOP to establish incentivesfor individuals to repay if they are able.

In the past few years HOP has sought to encour-age repayment by building closer relationshipswith clients, assigning a volunteer specifically toeach client and arranging follow-up meetings tocheck in on the clients progress. We have alsotried to play ”good cop, bad cop” with our clientsby emphasizing that their loan officer is advocat-ing for them to the rest of the group, which islikely to be more skeptical. This trust-buildingapproach to collecting repayment has producedweak results. One interpretation has been thatclients associated HOP with Harvard and thatthis association overwhelmed any sense of trustor obligation that might otherwise have devel-oped.

There have never been tangible incentives for in-dividuals to repay HOP. Our founders made anideological commitment to forego the enforce-ment methods used by commercial lenders, suchas legal suit or foreclosure on collateral. Real-istically, legal enforcement is probably too ex-pensive anyway: the costs of using courts areprobably fixed and high relative to the amountswe typically lend. Lawsuits would ever returnenough cash to justify the time and expense in-volved. Further, we choose as a matter of in-tegrity not to threaten our debtors with legal ac-tions which we will in reality never take - whetheror not such cheap talk would have an impact.In the past this has left our clients with littlereason to repay their loans other than personalethics, and as discussed the perceptions of HOPand Harvard have weakened that motive consid-

23

erably.

One alternative both to formal enforcement sys-tems and to case-by-case trust-building, is toleverage trust where it already exists. Thismeans working with pre-existing informal insti-tutions and social networks that exhibit an abil-ity to cooperate without the use of formal insti-tutions. In fact, microfinance programs in thedeveloping world work in this way. The proto-typical scheme is one of group liability. Typi-cally, a microfinance bank will issue a loan toa local group of anywhere from five to twentyor thirty people, who divide the funds amongstthemselves. The group as a whole is then respon-sible for repayment. Groups form spontaneously,and they tend to pool applicants with similar riskcharacteristics ex ante. This is because relatively”safe” individuals will not want to be pooled to-gether with ”risky” applicants for fear of endingup liable for their neighbor’s prodigality. Sim-ilarly, because the group is collectively respon-sible ex post, each member has an incentive toforce his neighbor to pay up; otherwise he is leftholding the bill. The resulting interdependencehas yielded impressive results and made microfi-nance the darling of developmental theorists.

Microfinance in the developing world dependson the promise of future loans as an incentive.Group members pressure each other to repaybecause they want to preserve their own futureaccess to credit. In the case of entrepreneurialloans, there is a high likelihood that the clientwill need additional capital in the future to fur-ther expand and develop his/her business. Re-paying each loan ensures the subsequent avail-ability of capital. This is analogous to the con-cept of a ”credit history” in the United States.

HOP can potentially combine the principles ofgroup lending and reputation-building to createa forceful incentive scheme. However, because ofthe urgent needs of our clients facing eviction, itis impractical to try to gather a group of clientstogether for a spontaneous group loan. And sucha group would have little internal consistency, lit-tle “social capital,” which members could use toforce others to repay. Instead, HOP should seekout an established group of tenants who are allpotential clients. This group should have someform of leadership or representation that can berelied upon to act in the best interests of thegroup. They might do so because of altruisticmotives, or alternatively because members of thegroup could provide them with compensation fortheir help. A leader must benefit in some wayfrom HOP loans to the group so that he or shehas an incentive to maintain a good relationshipwith us.

A leader should also be well acquainted with thesituations of individual group members. Theleader must help HOP overcome the informa-tional asymmetries inherent in lending to first-time applicants. This is particularly impor-tant because our groups are not formed spon-taneously, as in 3rd world microlending, and sogroups will not have homogenous risk character-istics.

In a prototypical partnership arrangement,HOP would provide a financial commitment, apromise to provide up to $X per year in loans tomembers of the group. In exchange, we wouldrequire that the group’s leaders 1) screen appli-cations and ensure accuracy of information, 2)use internal sanctions to force group memberswho receive loans to repay, and 3) help HOPmaintain contact with clients through the entire

24

loan and repayment process. If the leader failedto meet these conditions, HOP would scale backor terminate the partnership.

This threat to leave is critically. It implies thatfuture HOP loans to other members of the groupare contingent on the repayment of current loans.Even if the actual recipient of a loan deemed itunlikely that he or she would need another loan,other tenants would then have cause to pres-sure them to repay. And at the end of the yearHOP officials can sit down with group leaders,compare their performance with other partners,discuss successes and failures, and then decidewhether or not to renew the partnership. Thus,in order to make the incentive scheme work HOPmust be perfectly willing to break off a partner-ship if repayment is not rendered. If the lossof future credit is seen as a non-credible threat,the incentives fail. This is the rationale for part-nering with multiple tenant groups, giving HOPthe ability to sever any one particular tie with-out drastically reducing the number of clients weserve.

The group-lending model described here takesHOP’s most difficult tasks and sets the ’group’itself to work completing them. First, this modelhelps mitigate informational asymmetries be-tween HOP and applicants. HOP verifies ap-plication information whenever possible, but the’group’ can better observe the situation of theapplicant and supply HOP with more accurateinformation. Second, the ’group’ serves a screen-ing function. Because the entire group can onlyreceive a maximum of n loans per year, it willhave an interest in referring only the clients whoneed help most and are likely to receive a loan -otherwise it would ’waste’ its yearly allowance ofloans. Third, the group has incentives to force

its members to repay loans, since default by anyone member will cut off the remaining membersfrom HOP’s assistance.

What kind of “groups” might work well for sucha lending model? Some of the important fac-tors to consider are 1) the political power of theleadership and its ability to act in the best in-terest of the group; 2) administrative capabilitiesof the group, as they pertain to HOP’s obtainingaccurate application information and remainingin contact with current clients; 3) the risk/needtradeoff and likely annual need for credit that thegroup exhibits. The range of groups deservingconsideration is broad: tenant groups, churchesand temples, all the clients represented by a sin-gle caseworker, all the tenants of a single land-lord, etc.

25