Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018

Neither a borrower nor a lender be; For loan oft loses both itself and friend. And borrowing dulls the edge of husbandry. This above all: to thine own self be true, And it must follow, as the night the day, Thou canst not then be false to any man. – Hamlet I.iii.79–841

In the last seven chapters, we have explored more perspectives on microfinance than have ever been explored before in one volume. We have looked at it from the points of view of the user at the metaphorical teller window and the manager behind it. We have placed it in the flow of history. We have surveyed its diversity. And we have taken seriously the strongest claims for its virtues, investigating each in turn: proven poverty reduction, freedom, as industry-building.

It is time now to sum up, draw lessons, and ponder what lies ahead.

You of course know the popular image of microfinance: It was invented by that guy in

India (or was it Bangladesh?) who won the Nobel Prize. It helps people lift themselves out of poverty through microenterpise. Part I of this book put the lie to that image. But it also teased out a story that is more credible, more complex, and impressive in its own way. Modern microfinance is not, as a cynic might have it, merely another foreign aid fad foisted upon the poor, doomed by its naiveté to fail. If it is a fad, then it must be the longest in the history of foreign aid. What explains its persistence is its remarkable success on the market test: poor people want formal financial services, and are willing to pay for them. Thus microfinance is best seen as arising organically from several sources: the real need of poor people for tools to manage tumultuous financial lives; a long historical process of experimentation with ways of delivering financial services to the masses; the creativity, vision, and commitment of the pioneers, including

Muhammad Yunus; and the business imperatives of mass producing small-scale financial services.

1 “Husbandry” might be “economy” in modern English.

1 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018

One of the most important lessons of part I is the one emphasized in chapter 5, that microfinance as we observe it today is the outcome of an evolutionary process. As we saw, this helps explain the emphases on credit, groups, and women. The evolutionary perspective also explains a trait little noted in chapter 5: the mythology that promoters have woven around the workaday business of disbursing and collecting loans. Just as it hardly matters from the evolutionary point of view whether joint liability was invented, discovered, or copied, it hardly matters whether microfinance promoters believe the mythology. What matters is that investors— again, broadly understood to include all who provide finance for microfinance—have often rewarded those who tell certain stories, creating a selective environment that favors the microfinance groups best at telling them. This should not surprise. All of us who believe in our work tell the best stories we can to illustrate our theories about how we make the world a better place. Pankaj Jain and Mick Moore put it well:

We are not suggesting here that the leaders of the big [microfinance institutions (MFIs)] perpetrated some kind of fraud….The picture is far more complex than that and notions of blame or of individual responsibility are irrelevant to our objective of obtaining practical understanding of why and how [MFIs] have been so successful. Our limited evidence suggests that the orthodox fallacy blossomed and spread in large part because that is what people in aid agencies wanted to hear, thought they had heard, or asked [MFI] leaders to talk about and publicise. To the extent that [MFI] leaders did foster a particular image, this could be seen simply as targeted product promotion in a “market” of aid abundance… …to justify the continuing flow of that money to their own particular organisations and to the microfinance sector as a whole, [MFI] leaders and spokespersons have gradually found themselves, through a combination of circumstances and pressures, purveying a misleading interpretation of the reasons for their success. They emphasise a few elements in a complex organisational system, and are silent on many key components.2

Ironically, microfinance succeeded in part by obscuring the businesslike nature of its success.

In Part II, I parted the curtains of the mythology to see what lay beyond. I tried, in the words of my colleague Ethan Kapstein, to be critical but not cynical, to investigate the evidentiary bases of the most serious arguments on behalf of microfinance. The lessons distill to:

2 Jain and Moore (2003), 28–29.

2 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018

 Credible evidence on microfinance’s success in development as poverty reduction

is scarce and not particularly encouraging. We have essentially two studies of microcredit

and one of microsavings. None says whether microfinance “works” in all places, times, and

forms because none could. The two of credit found no impact on indicators of household

welfare such as income, spending, and school attendance over 15–18 months. The one of

group credit in Hyderabad, India, did spot an increase in business starts and profits.

Meanwhile the small study of a savings account for market vendors in Kenya did find

positive impacts on income and spending, especially among women.

 The evidence on whether microcredit in particular spurs development as freedom,

which is essential to check given the uncertainty about the ultimate impacts on poverty, is

itself ambiguous. Researchers who have spent weeks or years with borrowers have collected

some happy stories of women of finding liberation in participating in financial business in

public spaces. Others have returned with disturbing stories—some mild, as of the women

made to sit in meeting till all dues are paid, some more serious, as of the women whose roofs

are taken by peers in order to pay off their debts. The contradictions within the evidence base

is not hard to understand, for credit is both a source of possibilities and a bond. It stands to

reason that poor people with volatile incomes need financial services more than the global

rich, in order to put aside money in good days and seasons and spend it in bad; and that

reliable loans, savings accounts, insurance, even money transfers, can help them do this.

Financial services inherently enhance agency. But credit inevitably entraps some people

through ill luck or judgment. Overall, it is hard to feel sanguine that success stories are the

whole story.

 The kind of success on which microfinance can stake its strongest claim is in

3 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018

industry building. With time, the microfinance industry is growing larger, more efficient,

generally more competitive, more diverse in its offerings, and increasingly creative in its

financing. More institutions are becoming national intermediaries, taking deposits and

lending domestically. (Because of inherent complexity, large-scale successes in insurance

have been rarer in the low-income market, which can only be served sustainably by cutting

costs to the bone.) Even the definition of success as industry building, however, leaves scope

for critique of the status quo, mostly relating to how enthusiastic, socially motivated credit

for microcredit is distorting the industry away from savings and toward loans. And an

important irony—and huge qualification relative to popular perception—is that microfinance

rarely turns clients into agents of economic transformation and growth. It does not fill the

role Joseph Schumpeter saw for finance.

Like all human institutions, the beast of microfinance has many warts. Leaving aside the inevitable imperfections, the hope that microfinance credibly offers lies in building institutions that give millions of poor people an increment of control over their lives, control they will use to put food on the table more regularly, invest in education, and, yes, start tiny businesses. Few lives will be transformed by microfinance; few will be lifted out of poverty. Yet because poor people are willing to pay for the services, microfinance institutions can huge numbers from a modest base of donor funds. Recently, Rich Rosenberg recalled his oversight while at the U.S.

Agency for International Development of “a few million dollars of donor subsidies in the mid-

1990s” for Bolivia’s Prodem (later, BancoSol). He reflected on the “value proposition” of microfinance, which he diagramed this way:

4 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018

Small one-time subsidies leverage large multiples of unsubsidized funds producing sustainable delivery year after year of highly valued services that help hundreds of millions of people keep their consumption stable, finance major expenses, and cope with shocks despite incomes that are low, irregular, and unreliable.3

You can’t have it all Economics is sometimes defined as the study of the optimal allocation of scarce resources. In truth, there is more to economics (resources are rarely allocated optimally anyway), but the definition is apt in that dismal scientists often think in trade-offs. Changing the allocation of labor and capital—rejiggering a factory—means more toasters but fewer microwaves. Having scored microfinance against various standards, it is helpful to introduce the idea of trade-offs.

Part II labored to think and gather evidence about each kind of success, one at a time. But that evaluation is only input to judgment, which is necessary for wise action.

Trade-offs await on at least two levels: in comparing microfinance to other charitable projects, and in comparing styles of microfinance, which score more or less well on the different dimensions of success. At the first, upper level, the notion brings us to the grand questions of this book: Does microfinance deserve all that praise and funding? Or are microfinance investors just chasing fantasies? Should they channel their charity elsewhere? Microfinance is not unusual in the degree of our ignorance about its impacts. Fragmentary evidence from half-believable studies are the norm in aid and philanthroopy. So the best use of funds is in any particular case is unknowable. I think that financial services for the poor do deserve a place in the world’s aid portfolio, for two reasons. First, microfinance has compiled impressive achievements in building institutions that enhance the freedom of millions. These achievements come with caveats, especially about the dangers of credit, but partly because microfinance is more than microcredit, the caveats are not fatal to microfinance generally. Second, a principle of diversification applies

3 Rosenberg (2010), 5.

5 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018 in charitable investing just as it does in conventional investing: given these achievements, as well as the importance of financial services in the lives of the poor and the inevitable uncertainties about the impacts of microfinance, school-building, road-building, or anything else, it is wise to invest in several strategies at once. Diversification reduces risk. That said, I will argue below that microfinance’s slice of the portfolio has grown too large.

As for trade-offs between different styles of microfinance, here some debate is longstanding, if in language different from mine. In the late 1990s microfinance specialists hotly debated the relative importance of serving the poorest—even if that required subsidizing interest payments—versus weaning MFIs off subsidies so that they could grow unconstrained by foreign aid budgets and serve more people. Economist Jonathan Morduch called the split the “poverty” and “sustainability” advocates the “microfinance schism.” Within this breach, however, a school of thought grew that questioned the inevitability of trade-offs—or at least submitted that business-like sustainability need not cost much in “depth of outreach” to the poorest. MFIs could sustainably serve legions of quite poor people while covering costs. True to his training,

Morduch doubted that the choices could be dodged so easily.4 With coauthors, for example, he demonstrated that increasing interest rates 1 percent (not 1 percentage point) in Dhaka, the capital of Bangladesh, reduced borrowing by slightly more than 1 percent on average. Within that average, poorer people cut their borrowing most. The implication: cutting interest subsidies at an MFI might make it more self-sufficient only by putting formal financial services beyond the reach of the poorest.5 With other coauthors, Morduch examined data on MFIs around the world, looking for relationships between profitability and the shares of clients that are poor and/or female. While hardly uniform, the overall correlation was negative. “[I]nvestors seeking

4 Morduch (2000).

5 Dehejia, Montgomery, and Morduch (2005).

6 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018 pure profits would have little interest in most of the institutions we see that are now serving poorer customers.”6

My own review of the evidence also hinted at trade-offs, especially between the development as institution building and development as freedom. Recall the end of chapter 7:

“There is a margin at which convenience for the institution and the needs of the client conflict.”

MFIs can do credit more easily than savings or insurance, yet it is credit that curtails freedom.

Layering non-financial services on top of financial, such as in Indian self-help groups and

Freedom from Hunger’s melding of lending and teaching, may enhance women’s agency but also takes subsidies. Higher interest rates may boost the profitability of MFIs and the dynamism of the industry—while bringing the whole business to a flirtation with “usury.” Likewise, the trade-off between development as industry building and development as poverty reduction is so mathematically direct as to almost escape mention: higher prices make clients poorer.

If it is easy to point out choices, it is harder to make them. Reality is complex, and so is morality. The consequences of, say, subsidizing microfinance (including through finance at submarket rates of return) vary over place and time in ways we cannot gauge any more than we can predict the precise consequences of a one-percent interest rate cut on a hundred different borrowing families. Even if we knew exact consequences, ethical imponderables would raise their heads. How are we to weigh the short-term benefits of cheap services against the long-term gains from growing, self-financing MFIs? That complexity, for anyone trying to help others, is life.

In the face of such unknowns and imponderables, I suggest two principles of action. First, don’t give up hope on dodging trade-offs. The dictatorship of hard choices is only absolute if

6 Cull, Demirgüç-Kunt, and Morduch (2009a), 169.

7 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018 microfinance institutions are squeezing every ounce of productivity from the capital and labor they consume. Such perfect firms reside only in textbooks. No real firm operates at what economists call the technological frontier, where every gain in one output must come at a sacrifice in another. Some of the inefficiencies in microfinance are created by investors—a regrettable fact which also represents an opportunity; Anyway, the most important economic developments occur not when people figure out how to get close to the frontier, but when they push it back, as Yunus once did. Thus the choices in microfinance today are not entirely dismal.

The chief opportunity I see is in savings, including through mobile phones. I will elaborate momentarily.

The second principle of action is that microfinance (or anything else) is mostly likely to achieve its potential when it follows its natural constructive tendencies. If your daughter were a piano prodigy, you would probably try to give her a balanced life and education, but not to the point of stunting her talent. Note the constructive though: you would probably not nurture her tendency, if any, to sociopathy. By this principle, microfinance is likely to do the most good when it plays to its strengths. Going by the review above, the microfinance project’s real talent is for building industries that serve millions with modest amounts of aid. Among charitable projects, it is in this respect truly prodigious. To echo the previous chapter: “There is no

Grameen Bank of vaccination. One does not hear of organizations sprouting like sunflowers in the world of clean water supply, hiring thousands and serving millions, turning a profit and wooing investors.” In contrast, client-for-client, microfinance does not stand head and shoulders above other forms of aid in eliminating poverty. With respect to Morduch’s “schism,” I therefore favor those who seek to reach the largest clientele.

8 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018

Seeking savings One opportunity for microfinance to excel on several definitions of success at once lies in taking deposits. The microfinance movement was born out of credit but should strain to move beyond it. Savings arguably performs better than credit in all three senses of success: a randomized trial found it reducing poverty; it does not impinge like debt on freedom (it is easier to imagine people getting in trouble by borrowing too much than by saving too much); and MFIs enrich the local financial fabric most when they interface with the poor bidirectionally, doing credit and savings. Experience and common sense say that the group of people willing to save is larger and poorer than the group willing to shoulder the risk of credit; recall from the last chapter that BRI in Indonesia has twelve times as many savers as borrowers below the poverty line.7 Meanwhile, as we saw in chapter 2, savings can do almost anything credit can. People can, for example, save up to start businesses. Where credit disciplines with mandatory weekly payments, commitment savings accounts can levy penalties to enforce agreed contribution plans and discourage early withdrawals. Nor is large-scale microsavings a pipe dream. As described in chapter 8, many leading MFIs do it: Bank Rakyat Indonesia (BRI), big MFIs in Bangladesh, the ProCredit banks, and others. Nevertheless, credit still dominates in many places, including India, Mexico, and the four nations whose bubbles popped in 2009 (Bosnia, Morocco, Nicaragua, and Pakistan).

A similarly principled case, but perhaps one less practical, can be made for microinsurance. After all, if the chief financial problem of the poor is managing unpredictability, insurance seems tailor-made to help. When I asked myself in chapter 2 which financial services I prize most, I chose life and health insurance because they blunt some of life’s worst traumas.

Efforts to bring such services to the poor should be supported and successes should be applauded and copied. However, insurance is inherently more complex than credit and savings, and with

7 Johnston and Morduch (2007), 29.

9 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018 complexity comes cost. Its nature cuts against the intense imperative to streamline in microfinance in order to keep expenses in line with the tiny sums at stake. Confirming insurable events as seemingly obvious as death sometimes takes work, in order to prevent fraud. Moral hazard (insurance encouraging irresponsible behavior) and adverse selection (only those mostly as risk taking the insurance) further complicate the economics. Many people will not buy insurance even when it is in their interest to do so, which is why insurance is so often bundled

(property insurance with a mortgage, credit-life insurance with credit) or mandated by law (car and health insurance in many countries). Why put your money into an insurance policy and risk never seeing it again, they think, when you could save the money and keep it yours? That thinking leads insurers to cover common events such as mild droughts and monthly prescriptions, so that everyone receives regular benefits from the insurance. And that drives up premia—and makes insurance more like savings. On balance, microsavings for the billions is a more practical ideal.

This call for microsavings needs one major qualification: where credit puts the onus of repayment on the client, savings places major responsibilities on the institution. And the ability of any institution to handle that responsibility cannot be assumed. Many MFIs have achieved scale first through laser-like focus on credit, then branched into savings. If credit is a pioneer species, then perhaps expecting MFIs to take savings in their earliest days would be like expecting oaks to root in bare rock. Or expecting caterpillars to sprout wings. Chapter 5 argued that credit has dominated microfinance for several reasons: fledgling non-profits are not in general trustworthy custodians of savings; savings is harder to mass-produce than credit, being more of a custom service. Fast, credit-led growth may continue to be a viable path for many microfinance institutions to reach the political and administrative heft needed to obtain licenses

10 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018 to take savings and to handle the responsibility with efficiency and propriety.

Still, while that path is not easy to gainsay categorically, neither is it inevitable. Human institutions are more malleable than the developmental stages of forests and butterflies. Among the exemplars of microsavings, the largest and smallest, BRI’s unit desa system and Stuart

Rutherford’s SafeSave in Dhaka, both took savings from the start. ProCredit banks in many countries took savings from birth, or moved into it quickly.

Accepting the belief in the potential of savings as an end, we should be ecumenical about the means. “Savings” is a simple word with diverse meanings in practice. There are liquid savings accounts, in which money can be moved in and out at any time. There are commitment accounts. Some savings accounts pay interest. Some don’t. Through SafeSave’s rural sister,

Shohoz Shonchoy, Stuart Rutherford has piloted a financial product, “P9,” that morphs see-saw- like from credit to savings: A client starts with an interest-free loan of 2,000 taka ($29), except that the bank disburses just two-thirds of it and puts the rest in an interest-free savings account; she pays back the full 2,000, then repeats cycles like these so that soon her savings exceed her credit balance. The loans give her the discipline to become a net saver.8 Diverse too are the institutions that take savings: informal village savings and loan associations organized with outside help; credit cooperatives and their more formal cousins, credit unions; private savings banks and postal savings banks; agricultural banks; and microfinance institutions. And modern technology is creating new possibilities, as I discuss below.

Too much credit for microcredit The practical question for various microfinance players is not whether savings is better than credit in the abstract, but whether the players are unnecessarily perturbing the plot in the

8 Stuart Rutherford, “Product rules,” sites.google.com/site/trackingp9/home/product-rules, viewed April 24, 2010.

11 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018 direction of credit. One important player is the investor. The financial innovations that are funneling billions a year into microfinance—mainly microcredit—are marvels. Yet as explained in chapter 8, ample finance has at least two downsides: it undermines political and administrative drives to take savings (as an alternative source of capital for lending); and it inflates bubbles.

Recognizing these risks forces a hard question: How should investors collectively define and enforce Aristotle’s golden mean in the realm of credit for microcredit? How do you legislate moderation?

Consider first the concern about undermining microsavings; even here defining the right balance is deceptively difficult. Seemingly, the risk justifies drastically curtailing outside money for microcredit, even halting it. Recall Dennis Whittle’s story from the chapter 8, of his repeated attempts while at the World Bank to lend money to BRI—and of BRI’s repeated and wise refusal. Perhaps all MFIs should copy BRI. Or perhaps they cannot: at the time, BRI was a century-old government bank. Many MFIs that now take savings started life not long ago as tiny non-profits doing just credit. Would it be wise to impede more MFIs from progressing through the same stages, from moss to oak, caterpillar to butterfly? If we accept the path as legitimate, should the amount of money made available to MFIs taper off as they grow, to wean them off pure credit? According to what formula?

Meanwhile, deciding when credit is so ample as to inflate bubbles is also difficult, notoriously so. To ground my thinking about this challenge, my research assistant Paolo Abarcar and I set out to answer an impertinent question about the microcredit bubbles that popped in

2009 (see chapter 8): Who inflated them? We combed through the annual reports of the largest microfinance institutions in the four affected countries. Where one institution, such as the U.S. government’s Overseas Private Investment Corporation, had guaranteed another’s loan—

12 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018 promising to pay it if the borrowing MFI did not—we attributed the amount to the guarantor. In

Bosnia, Nicaragua, and Pakistan, it turns out, foreigners supplied most of the air for the bubbles.

(See

One important message of One important message of Error! Not a valid bookmark self- reference. comes not from the data but the fact that they had never been compiled. Laboring to answer the question of who inflated the bubbles, I realized: almost no one knows. The data summed here are incomplete, uncertain in some respects, and at 1–2 years of age, ancient next to the tempo of 40-percent-per-annum hypergrowth. But they are the best that are publicly available. In years before the bubbles burst, hardly anyone saw the big picture because hardly anyone could. comes not from the data but the fact that they had never been compiled. Laboring to answer the question of who inflated the bubbles, I realized: almost no one knows. The data summed here are incomplete, uncertain in some respects, and at 1–2 years of age, ancient next to the tempo of 40-percent-per-annum hypergrowth. But they are the best that are publicly available. In years before the bubbles burst, hardly anyone saw the big picture because hardly anyone could..) Most of them are public institutions. Number one in Bosnia is the European

Fund for Southeast Europe, a conduit for European government donors. The Asian Development

Bank looms over the scene in Pakistan. Second there is the Pakistan Poverty Alleviation Fund, which has channeled a World Bank loan for microcredit. A few of the big creditors are private companies that manage funds from both public and private investors with social missions. Blue

Orchard, for example, is number one in Nicaragua and number two in Bosnia.

One important message of One important message of One important message of Error!

Not a valid bookmark self-reference. comes not from the data but the fact that they had never been compiled. Laboring to answer the question of who inflated the bubbles, I realized: almost

13 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018 no one knows. The data summed here are incomplete, uncertain in some respects, and at 1–2 years of age, ancient next to the tempo of 40-percent-per-annum hypergrowth. But they are the best that are publicly available. In years before the bubbles burst, hardly anyone saw the big picture because hardly anyone could. comes not from the data but the fact that they had never been compiled. Laboring to answer the question of who inflated the bubbles, I realized: almost no one knows. The data summed here are incomplete, uncertain in some respects, and at 1–2 years of age, ancient next to the tempo of 40-percent-per-annum hypergrowth. But they are the best that are publicly available. In years before the bubbles burst, hardly anyone saw the big picture because hardly anyone could. comes not from the data but the fact that they had never been compiled. Laboring to answer the question of who inflated the bubbles, I realized: almost no one knows. The data summed here are incomplete, uncertain in some respects, and at 1–2 years of age, ancient next to the tempo of 40-percent-per-annum hypergrowth. But they are the best that are publicly available. In years before the bubbles burst, hardly anyone saw the big picture because hardly anyone could.

14 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018

Table 1. Top five creditor/guarantors to top five microfinance institutions with data

This story should sound familiar: A set of borrowers, microcreditors in this case, are taking loans from many sources. Total borrowing is expanding rapidly. No one is tracking all

15 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018 this activity, much less whether the borrowers can reasonably be expected to handle all the debts they have contracted. The easy credit may hide and exacerbate the very problems it creates, since unpayable loans are quickly refinanced with new ones.

In other words, the cross-country microcredit financing scene resembles the within- country microcredit market in some places, with untracked multiple borrowing creating the risk of overborrowing and bubbles. Herman Daly one wrote about the need to move from an “empty world” mentality that treats natural resources as inexhaustible to a “full world” one that accepts limits.9 In remarkably short order, the world of microfinance finance has swelled from empty to full. Not that every poor person has microcredit who would want it; rather, the bottleneck is no longer wholesale finance. MFIs can only safely grow so fast. In the mid-1990s, Alex Silva struggled to raise a few million dollars for the first microfinance investment vehicle, Profund.

(See chapter 8.) Now microfinance investment managers are struggling to absorb millions per day. For the sake of the industry’s health, investors must adapt to the new reality. If they do not institutionalize collective limits, they may inflate more bubbles.

Together, the two concerns about easy money for microcredit—undermining microsavings and inflating microcredit bubbles—pose a complex problematique. How should the public and private social investors who dominate finance for microfinance decide how much credit is too much (or what price for their credit is too low)? And how should they regulate themselves to stay within such limits? Within nations, one standard corrective for over-eager lending is the credit bureau. By analogy, investors in microcredit need at a minimum to establish ways to share information at high-frequency on the financial obligations of MFIs. This could happen informally: in fact, managers of private microfinance investment vehicles are now

9 Daly (2005).

16 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018 sharing intelligence in the countries whose bubbles have popped. The investors recognize that they must, indeed, all hang together—collectively providing financial breathing room for MFIs on the brink, monitoring their moves closely—or most assuredly they shall all hang separately. If one investor calls in its loans, that might precipitate a bankruptcy that would damage the others.10

Partly in order to draw in the more dominant and less flexible public investors, the credit bureau analogy should also be taken more literally: investors should construct a formal body that would collect and publish high-frequency, high-quality data on the liabilities of microcreditors

(what they have borrowed) and assets (what they have lent). The body would professionalize, in other words, what Paolo and I did. It could also gather data relevant to the question of when credit for microcredit undermines the initiative to enter the savings business. It could analyze whether a given MFI could realistically obtain permission to take savings; study the cost of doing so; and compare that cost to that of external capital. It could also develop soft standards analogous to the rule of thumb that mortgage payments should not exceed a third of income. It might issue warnings of various severity levels based on these indicators.11 These external reference points could help microfinance investment managers dissuade higher-ups, politicians, customers, and citizens who are too eager for them to pour more money into microcredit.

External reference points would also help managers contain their eager inner demons.

Like ordinary credit bureaus, such a centralized brain for the microcredit investment business would contribute incrementally to the goal of making capital flows healthier—and perhaps prevent bubbles no more reliably than America’s three credit bureaus recently have.

Investors as diverse as the World Bank, Blue Orchard, TIAA-CREF, and Kiva might resist external regulation, however light. They might not find common ground on how to do it. If they

10 Daniel Rozas, microfinance consultant, Brussels, e-mail to author, March 29, 2010.

11 Liliana Rojas-Suarez, Senior Fellow, Center for Global Development, conversation with author, April 30, 2010.

17 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018 did agree, the guidance might be so muddy as to be ineffectual; the institutional imperative to keep investing is often strong. The deepest problem might be accepting the restraint implied by a serious commitment to savings: would microfinance investors support a body that advised them to slash their operations, to stop picking the plum MFIs?

An alternative to regulating the flows is to regulate who can emit them, favoring those constitutionally more apt to act with care. In a 2007 report called Role Reversal, which inspired

One important message of One important message of Error! Not a valid bookmark self- reference. comes not from the data but the fact that they had never been compiled. Laboring to answer the question of who inflated the bubbles, I realized: almost no one knows. The data summed here are incomplete, uncertain in some respects, and at 1–2 years of age, ancient next to the tempo of 40-percent-per-annum hypergrowth. But they are the best that are publicly available. In years before the bubbles burst, hardly anyone saw the big picture because hardly anyone could. comes not from the data but the fact that they had never been compiled. Laboring to answer the question of who inflated the bubbles, I realized: almost no one knows. The data summed here are incomplete, uncertain in some respects, and at 1–2 years of age, ancient next to the tempo of 40-percent-per-annum hypergrowth. But they are the best that are publicly available. In years before the bubbles burst, hardly anyone saw the big picture because hardly anyone could., Julie Abrams and Damian von Stauffenberg argued that public investors ought to exit MFIs when private ones enter. The job of public investors (which they call International

Financial Institutions, or IFIs) is to “go where the private sector does not yet dare to tread; to assume risks that private capital would find unacceptable”:

Whether top decision-makers are aware of it or not, there are powerful incentives for IFIs to maximize their microfinance exposure, and to do so by concentrating on the largest and safest borrowers. Microfinance has acquired such a positive image, that a sizeable exposure in this sector has become a sign of a IFIs commitment to development. This is reinforced by an IFI’s need to disburse its microfinance budget each year. Since IFIs are not primarily profit-driven their

18 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018

success is often defined by the amounts that have been lent. If a budget has been allocated to microfinance, that budget must be spent—and spending it on a few large loans to top MFIs is far quicker, cheaper, and less risky than lending to, and nurturing immature institutions.12

“Public should exit when private enters” is a blunt rule with a fuzzy rationale. It is not obvious why socially motivated, below-market-price public capital would behave better than socially motivated, below-market-price private capital. But given the current surfeit of investment in microcredit, if the rule were followed, the world might well be a better place.

Perhaps closer to the heart of the matter than the public-private distinction is that between generalist institutions such as the World Bank and CARE on the one hand and, on the other, specialists such as microfinance investment vehicles and dedicated microfinance network groups. (The latter include the Grameen Foundation, Women’s World Banking, and others.)

Among generalists, microfinance is one line of business among a hundred, and staffers often rotate to another country or department before consequences of their decisions in the previous one arrive. Notably, CARE withdrew from microcredit in 2005, recognizing that it lacked the competency and focus to do it well.13 (CARE still supports village savings and loan associations.) It would probably improve on Role Reversal’s rule if all generalist institutions withdrew from directly financing microcredit—or were removed by funders such as the U.S.

Congress. Surgery should be left to surgeons.

Of course, some surgeries would be better left undone. Surgeons are not always the best judges of when to stay their own hands. If the microcredit investment industry cannot enforce credible, collective self-discipline that minimizes harm to the goal of taking savings, then perhaps it should be shut down altogether, save for a modest catalytic role in developing new

MFIs through seed capital and training grants. A fundamental problem here is that the evidence

12 Abrams and von Stauffenberg (2007), 1.

13 Wilson (2007).

19 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018 on the impact of microcredit on poverty and freedom is ambiguous—more so than for microsavings. If the sign on microcredit’s impact was more clearly positive, we would not need to engage in such mental contortions to define a healthy role for investing in microcredit. If it is this hard to assure that such investment does more good than harm, that suggests we are barking up the wrong tree.

Building trustworthy institutions Banking is a perplexing business: essential, but prone to fraud, manipulation, and manias. This is why it is regulated almost everywhere. However, regulation is generally imperfect and sometimes thin on the ground, especially in poor countries and especially in microfinance.

Ideally, government officials in each country where microfinance is done would shoulder the technical and moral responsibilities ensuring stability and protecting consumers. That would arguably absolve even social investors of worrying about whether, say, the microcredit they finance is arguably usurious. Investors could then morally confine themselves to deciding which

MFIs to invest in and on what terms. In the real world, investors can influence the operations of

MFIs they support, and with that power comes responsibility. They must accept it or tread cautiously, lest they fuel unhealthy growth. Fortunately, many MFIs have proved by example that is possible to win customers’ trust in an environment of incomplete or imperfect regulation.

Some have greater reputations for integrity than their own governments.

Making financial institutions trustworthy has many aspects.14 Here are some:

 Preventing abusive lending practices such as aggressive marketing, harsh

collection tactics, and opaque explanations of interest rates and fees.

 Auditing for fraud and corruption.

14 Christen, Lyman, and Rosenberg (2003).

20 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018

 Assuring that lending is prudent and growing appropriately, and that banks have

adequate capital to absorb losses.

 Improving the portfolio of services available to the poor, such as by adding a no-

frills savings account.15

 Assuring that deposits are safe and available to customers on stipulated terms.

One project aimed at the first bullet point is the Smart Campaign, a joint production of

CGAP and ACCION’s Center for Financial Inclusion. The campaign has signed up more than

250 MFIs to endorse six principles of responsible lending: avoidance of over-indebtedness; transparent and responsible pricing; collection practices; staff behavior; mechanisms for redress of grievances; and privacy of client data.16 Nice words may do little in themselves; but they give investors a benchmark to which they can hold MFIs accountable. Recently, for example, an MFI that had endorsed the Smart Campaign quietly increased the forced-savings percentage on its loans. Continuing to charge interest on the full loan amount while reducing the portion that clients could take out increased the effective interest rate in a way that violated the Campaign’s principle on transparent and responsible pricing. Two dedicated microfinance investment funds that financed the MFI wrote a pointed letter to its management expressing displeasure with the change.17

The second and third bullet points are naturally of great interest to investors, who require external audits and reviews by ratings firms (MicroRate, PlaNet Finance, or M-CRIL) to keep tabs on management. Here, the interests of investors and customers tend to align, and investors therefore do good simply by investing.

15 For example, the Mzansi account in South Africa; see Bankable Frontier Associates (2009).

16 smartcampaign.org/about-the-campaign/campaign-mission-a-goals.

17 Rozas, op. cit. note 10.

21 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018

There may also be an opportunity for investors to assist with the last two bullet points, encouraging savings by assuring that they are safe. Earlier, I argued that it is hard to imagine how people can get in trouble by saving too much. In fact, it is not hard. Lots of people have lost savings they put in banks. Why? For one, propriety does not always prevail: a World Bank report on postal savings banks delicately explained that, “In some countries, mainly in Africa,...deposits have not been managed with transparency and are transformed into substantial unfunded liabilities.”18 Sometimes the cancer of corruption grows on the lending side instead, sending deposits into the pockets of cronies who don’t pay back. Economic crises turn good loans bad and bring banks to their knees. Or the government pays its debts by printing money, causing inflation that erode the value of savings. Then there is the problem of bank runs, like those that fed the Great Depression. Doubts about the soundness of a bank become self-fulfilling as depositors line up to withdraw their money—and the bank cannot call in its loans as fast.

One institution that has been devised to keep savings safe is deposit insurance. U.S.

President Franklin Roosevelt introduced first, in 1933, as an emergency response to bank runs.

The Federal Deposit Insurance Corporation would insure each bank account up to a certain amount. In exchange, the FDIC could levy a fee on banks equal to a small percentage of insured deposits, as well as regulate and supervise going banks and take over failed ones. By 2007, 79 countries had deposit insurance programs. However, most were weaker on paper or in practice than the FDIC: for example, only about a dozen had the authority to take over failed banks.19

And how many of these insurance plans extend to microfinance institutions is unclear.

Could social investors set up a microdeposit insurance scheme to cover MFIs in countries where such insurance is not available domestically? One worry is moral hazard: freed of the

18 World Bank and ING Bank (2006), 7.

19 Barth, Caprio, and Levine (2008) and associated database, j.mp/8YSfQz.

22 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018 responsibility to keep savers whole, MFIs could promise unsustainably high interest rates to attract funds and on-lend the money carelessly. This hazard could be reduced through co- insurance—requiring the MFIs to share the risk—and modest ceilings on insured amounts per account.20 Another challenge is the sand castle nature of microcredit. The binding agent of the credit portfolio is as transient as water: the longstanding relationships between borrower and credit officer, and, within them, the prospect of continuing access to credit. The water can evaporate quickly if a failing MFI halts new loans or transfers collection of old ones to a healthier MFI with a staff of strangers. Contrast this with the way that rich-country banks can pass mortgages among them, knowing that the collateral is always substantially recoverable.

Microfinance consultant Daniel Rozas explains:

For property to be considered an asset, it must generally retain value irrespective of its owner. However, in rare cases, assets have value only when associated with a given entity. Like David Beckham’s foot, such assets can generate cashflow to their owners, but they cannot be independently transferred or sold….Microcredit appears to be just this type of asset.… This is a world apart from the relatively liquid asset model of traditional financial institutions. Take as an example the worst of the worst—the subprime [mortgage-backed securities]. To this day they are still very much passing aggregate loan payments (much reduced, of course) through to investors, despite the fact that nearly all of the original subprime lenders have long ago been marched off to their infamous grave. As bad as a loss of 20% or even 50% might be, it is still considerably better than losing 100%, which unfortunately, has been more common than not in the world of microfinance investment.21

Thus, MFIs usually don’t fail partway. If savers at an MFI get wind of trouble on the lending side, panic may well be the right response. The MFI may be not just illiquid but insolvent—not merely unable to return all deposits as quickly as depositors demand, but unable to return any of them ever. To this extent, the rationale that deposit insurance would prevent bank failures does not hold. And the likelihood of a complete loss might force the insurance premiums up from, say, 0.5 percent of covered balances each year to five or ten times as much. If MFIs passed that

20 See proposal by Muhammad Yunus in Counts and Meriweather (2008), 11.

21 Rozas (2009), 4, 10.

23 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018 cost on to savers, and savers did not feel like paying that much for insurance, they might switch to uninsured MFIs. The scheme, after all, would not be mandated by law; and people often don’t buy insurance even when it would be wise to.

Still, there may be hope for the idea. If you had a choice between lending $100 to

Esmeralda in Bolivia through Kiva and contributing $100 to an insurance fund that would protect

$100 of her savings, which would you do? The main risk to your money (the failure of

Esmeralda’s MFI) and the operating costs would be similar. Yet with the second option,

Esmerelda could live freer of debt. The symmetry of this comparison suggests that if one arrangement can work, the other ought to as well. Perhaps the insurance fund could even leverage your money, insuring $200 or $300 of hers. The higher the leverage, the less the insurance would cost per dollar insured. On the other hand, the greater the need would be for backing from a deep-pocketed institution to reinsure the fund in the event of simultaneous MFI failures that caused losses of more than $100 for each $100 in the fund. Given the surplus of money and financial engineering swirling in microfinance, this puzzle does not seem unsolvable.

And if social investors moved en masse from extending credit for microcredit to using their savings to back microsavings, MFIs will have little choice but to adapt.

The technological frontier Up till here, this book has been entirely backward looking. It has examined history ancient and modern, asked what has been achieved, what has been proved. Even the forward-looking recommendations in this chapter have largely assumed that the institutional forms developed over the last few decades will continue. But they may not. They may be displaced by quite different approaches made possible by new technologies—happily, ones that could also greatly reduce the currently unhealthy dominance of credit by making other, safer services more

24 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018 economical for the poor.

The flavors of microcredit that powered the microfinance revolution—solidarity and village banking, individual microcredit (see chapter 5)—are what economists call technological advances: particular ways of producing a service that was once thought impossible, or at least impossibly expensive. By producing more useful outputs with the available inputs, they pushed back the technological frontier. But in lay person’s terms, the popular methods of microcredit are anything but high-tech. As designed in the 1970s and early 1980s (or perhaps we should say

1720s and 1850s), they involved long meetings under trees, exchanging bits of wrinkled paper we call cash, and recording each transaction on several pieces of paper. More and more it appears that the future of microfinance is not merely the past with a high-tech add-on–group meetings with Palm Pilots—but things radically different. Advances in computing and communications will trigger a revolution here as in so many spheres. If so, then the next generation of microfinance will deserve a book of its own.

The single fact that makes talk of a high-tech microfinance revolution credible is that some 5 billion people, including half of all people in developing countries, now tote mobile phones. And the community of the connected is growing fast even in the poorest countries.22 Put otherwise, there are now 5 billion people carrying around globally networked computers. What can they do with those besides talk and text? Accessing financial services seems like a leading candidate. In the Philippines, the country with the highest use of text messaging per capita, mobile phone users can pay for milk at a store or send a friend money by pulling out a phone.23

In South Africa, the upstart WIZZIT Bank is expanding rapidly to bring financial services to the

22ITU (2010), ix.

23 [Ibid.]

25 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018 poor by phone, and has provoked all the large banks into following suit.24

Other technologies, it should be said, are advancing. In Brazil, correspondent “banks” at post offices and corner stores have handled trillions of dollars in utility bill and other payments.

The technological package behind these services includes bar code readers and communications uplinks to central computers. Every one of Brazil’s 5,561 municipalities now has at least one correspondent bank, including 2,300 that have no regular bank branch and thus until recently had no access to formal banking.25 In South Africa and Namibia, a company called Net1 delivers financial services to nearly 4 million low-income people, mostly government welfare beneficiaries, through smart cards. Unlike ordinary debit and credit cards, these contain computer chips and memory, so that they can operate even when the card reader they are inserted into is disconnected from the financial institution’s central computers. They can, for example, extend a loan on the fly, based on on-card data about the card holder’s history of welfare payments and servicing of past loans.26

[write about M-PESA after Kenya trip, linking to savings] [Jack, Pulver, and Suri, The

Performance and Impact of M-PESA--Preliminary Evidence from a Household Survey.pdf—nice stats and graphs on M-PESA. before-after pie charts of ways to move moneys; reminder that users not the poorest of the poor, but moving downmarket

http://technology.cgap.org/2010/03/08/mobile-money-takes-off-where-is-the-innovation- in-product-design/ --nice stats on saving through mobile, though I think Kenya ones wrong

“when I talk to them, I hear little effort from operators to really think freshly about their

24 Brian Richardson, Chief Executive Officer, WIZZIT Bank, presentation at “Expanding Financial Services to the Poor: The Role of ITC” conference, International Finance Corporation, June 9, 2006.

25 Terence Gallagher, Consultant, presentation at “Expanding Financial Services to the Poor: The Role of ITC” conference, International Finance Corporation, June 9, 2006.

26 David Schwarzbach, Vice President, Business Development, Net1 UEPS Technologies, Palo Alto, California, presentation at Center for Global Development, June 14, 2006, cgdev.org/content/calendar/detail/8069.

26 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018 products. Maybe it’s the rush to get to market.” “not all mobile money launches will get the response M-PESA did. In fact, it’s clear that was an outlier. M-PESA-like services in other countries may well have to do more, and try different things, if they want to hit for six (to use a cricket term, or hit a homerun, for my fellow Americans).I’m surprised more operators are not launching with a multi-product push to try and pick up several kinds of consumers – those who will pay to send money, those who want to save, etc.”

“processes more transactions domestically than Western Union does globally.” http://financialaccess.org/node/2916]

High technology has had its share of hype and financial bubbles; but its revolutionary potential is inarguable. Certainly, challenges remain. Importantly, phones may remain out of the reach of the poorest. Management of the interface between the cash and electronic economies will be a hindrance until the day (if it ever comes) when electronic money entirely replaces paper money. Tellers, automated or human, are needed to convert between the two forms of currency, and they cost. And while electronic money may seem to travel at the speed of electrons, it must often drag paper money along with it. When the net flow of e-money is from Nairobi to the country side, armored trucks with cash must be dispatched, and they cost too. Nevetheless, M-

PESA has surmounted these barriers. Ignacio Mas, who spearheads the Gates Foundation’s work in the area argues that the real challenge at this point is not the technology, but building businesses models that link together the banks, mobile phone companies, and agents in ways that give proper incentives (profit) to each.

What is intriguing and promising about high technology is its capacity to solve several problems that have bedeviled microfinance. It makes possible cheap, reliable identification, through fingerprint or retina recognition, paving the way for credit bureaus for the poor. Those in

27 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018 turn will lower the cost of credit for reliable borrowers. By replacing dedicated MFI offices and employees and with generalist corner stores and embedded software, the high-tech approach may reduce the cost of small, ad hoc transactions needed for voluntary savings. By linking the poor en masse to big banks, it gives them safer places to put their money and safer ways to transfer it to others. Technology can even helpfully nudge people into saving through automated reminders.27

The more viable these services, the less need for traditional microcredit based on peer pressure and time-consuming meetings.

Coming full circle I began this book with two opposing stories, one of Murhsida who climbed out of poverty on a ladder of microcredit, one of Eva Yanet Hernández Caballero who, if anything, slipped down a rung. This I did to expose how storytelling forms the public image of microfinance, and to make the case for serious research. We need good research not to move beyond narrative knowledge, but to test it, to inform us about which stories are most representative. That is as close as we can come to the truth about something as diverse as the microfinance experiences of 100 million people.

Though this book examines services other than credit and notions of success other than proven poverty reduction, there is no denying that the grain of sand that seeded this imperfect pearl is the impression that microcredit cuts poverty. As a child of bitterly divorced parents, it goes against my nature to choose sides. I see the world in grays and it those who see it in black and white, whether they agree with me or not, who most stir my ire. I cannot dismiss traditional microcredit. But it is hard for me to defend it as a strategy for helping poor people. Consider:

 Credible studies have so far shown no average impact on poverty. More

27 Karlan et al. (2010).

28 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018

high-quality research is needed and is underway, including a three-year follow-up on

the randomized trial in Hyderabad, India.

 Common sense says that microcredit gives people a new option to manage

their complex and unpredictable financial lives and helps some build businesses. But

common sense also says that it leaves some worse off and has an addictive character,

with the need to pay off one loan feeding the need for the next. Overborrowing is

more likely where several creditors are competing and growing fast.

 Careful qualitative studies, done by people who immersed themselves in a

village for a month or a year, corroborate this ambivalent reading. Some women find

liberation in doing financial business in public. Others find entrapment in the peer

pressure.

 Perhaps the best that can be said for microcredit is that good things can be

built on top of it. Pro Mujer (“For Woman”) in Peru uses the convening power of

credit groups to provide basic healthcare and business training. BancoSol in Bolivia

began with credit and expanded to savings.

If the best that can be said for microcredit, pending further study, is that it makes other, good things possible, is it right to build these things on such a questionable foundation? Uncertain evidence cannot support a certain answer. But choices today must be made on the evidence available today. To the practical question of whether social investors ought to keep pouring billions of dollars per year into microfinance—mostly microcredit—I say no. Finance for microfinance ought to go down, not up. The priority should not be building giant machines for indebting the poor.

At the end of day, I cannot dismiss the story of Eva, the one Compartamos featured on its

29 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018 web site until her business unraveled and she began missing payments on her 100-percent- interest loan.28 I cannot dismiss the story of families in northwest Bangladesh who “starve themselves or just take rice mixed with water” in the two days before each weekly meeting in order to scrimp for the installment.29 I cannot dismiss the story of Jahanara, the microcredit borrower and moneylender who boasted “that she had broken many houses when members could not pay.”30 And I cannot dismiss the report that Sufiyah Begum, Muhammad Yunus’s first borrower, died a beggar, as poor as Yunus found her.31 I cannot dismiss these stories as so atypical as to be immaterial to the morality of pushing credit.

But neither can I dismiss the manifest hunger of poor people for reliable tools to manage their money; nor the extraordinary success of some microfinance institutions in creating and serving this market over the last third of a century. The best way forward is to celebrate this achievement and build on it. The success of the microfinance movement to date has proven the viability of businesslike provision of financial services to the poor. The need now is to diversify more aggressively beyond microcredit—indeed to deemphasize it as a line of business to the extent that it dulls the appetite for deposits. Microfinance institutions such as the ProCredit banks have shown how to balance credit and savings in countries as poor as the Democratic Republic of Congo. But if savings, money transfers, and insurance can also be done through institutional forms less associated with traditional microcredit—through member-run village savings and loan associations (VSLAs) or their formal cousins the credit unions, or national post offices, or commercial banks, or mobile phone operators—let them be done so. At this writing the VSLAs

28 Epstein and Smith (2007). See chapter 1.

29 Gillingham and Islam (2005b), 26. See chapter 7.

30 Karim (2008), 23. See chapter 7.

31 Al Amin (2010). See chapter 4.

30 Roodman microfinance book. Chapter 8. DRAFT. Not for citation or quotation. 4/9/2018 appear to be particularly promising for assisting the poorest of the poor, such as the millet farmers in the dry lands of Niger.32 For people wealthy enough to own mobile phones—half the population of the developing world and counting—high technology may forge the link to the formal financial system. Among supporters of financial services for the poor, none possesses more financial muscle and autonomy than the Gates Foundation; not by chance is it working along these lines.33

Over the next third of a century, a global industry could arise to deliver to a billion or more poor people the tools they need to gain a modicum of mastery over the vicissitudes of their financial lives. Better banking will no more end the poverty than more clinics a more schools or more roads ever have. Most poverty reduction has arisen from profound processes of economic transformation nearly impossible to push from the outside. But the poor rightly value financial services, enough that they are often willing to pay the costs of delivery. There is good reason to hope, then that, just as a touch of funding did from the U.K. government did in Kenya, modest outside support can catalyze the growth of a giant global industry serving the poor. If this vision is made real, that will be a mighty achievement. And it will cast the last third of a century as an essential chapter in humanity’s ongoing discovery of ways to help the poor manage their wealth.

32 See chapter 4.

33 Christen and Mas (2009).

31