The Urban Unbanked in Mexico and the United States

This paper examines the ways in which lower-income households obtain basic financial services in urban communities in Mexico and the United States. And it discusses the efforts that private sector and government organizations are making to lower the cost or improve the quality of those services. The paper summarizes available information on these issues and assesses the rationale and challenges facing the strategies that both countries are using to improve the financial services available to lower-income households, giving particular attention to unbanked households, meaning households that do not have deposit accounts with any regulated deposit-taking institution, and also to lower-income households in large urban areas. In comparing the experiences of the two countries, the paper reviews the extent to which lower-income households are unbanked, their use of non-bank financial services, and strategies for improving financial services to the unbanked. The underlying differences between the countries' typical household incomes - national income per capita in Mexico in 2002 was US $8,540, compared with $35,060 in the United States (World Bank 2003) - may also influence the difference in percentage of unbanked - 9.1 percent of families in the United States compared with 76.4 percent found in a recent study in Mexico City.

Abstract: This paper examines the ways in which lower-income households obtain basic financial services in urban communities in the United States and in Mexico. In addition, the paper discusses the efforts that private sector and government organizations are making to lower the cost or improve the quality of those services. It summarizes available information on these issues and assesses the rationale and challenges facing the strategies that both countries are employing to improve the financial services available to lower-income households, giving particular attention to "unbanked" households, meaning households that do not have deposit accounts with any regulated deposit-taking institution, and also to lower-income households in large urban areas.
In comparing the experiences of the two countries the paper reviews the extent to which lower-income households are unbanked, their use of non-bank financial services, and strategies for improving financial services to the unbanked. The underlying differences between the countries' typical household incomes -national income per capita in Mexico in 2002 was U.S. $8,540, compared with $35,060 in the United States (World Bank, 2003) -may also influence the difference in percentage of unbanked -9.1 percent of families in the U.S. compared to 76.4 found in a recent study in Mexico City.
The paper surveys data on the urban unbanked in the U.S. It asks five questions. Who are the unbanked? Where do the unbanked get basic financial services? Why don't the unbanked use banks? What is the problem with being unbanked? And, finally, what private-sector and government initiatives are underway to improve financial services for the unbanked? It then addresses the same questions for Mexico. The answers to these questions are offered in a summary comparative form in the introduction. In both countries unbanked households are similarly characterized by low income and education levels although in Mexico the unbanked include persons well above the median income. The unbanked in both countries tend to rely on cash transactions and on services provided by commercial outlets, although in Mexico the unbanked rely on informal forms of saving and borrowing not present in the U.S. At the same time, the unbanked in Mexico show a high rate of home ownership, suggesting an alternative form of saving not present in the U.S. As to why the unbanked do not use banks, in both countries the issue of trust appears to a minor degree, but while the unbanked in Mexico perceive barriers in the costs and requirements of financial institutions, in the U.S. the unbanked point to their own financial situation and lack of savings as a reason for not using banks. In both countries the unbanked pay a significant cost in terms of additional transaction fees, time, and insecurity, in not using formal sector financial services. Concerning initiatives to improve access to financial services, the U.S. has a longer history of both regulatory and private sector practices encouraging retail banking practices. The Mexican government has recently begun to address the situation by focusing on the "popular" banking sector, and assisting credit unions and cooperatives to develop broader services. In its final section, the paper analyzes the rationale and challenges facing the strategies that both countries are employing to improve the financial services available to lower-income households.

Preface
This paper was written at the request of a World Bank team examining access to financial services in Mexico and Colombia. At the time, 2002, the unbanked, and the financial service needs of households represented a new issue for the World Bank. Previous Bank work on access to financial services had focused on access to credit-specifically for small businesses, rural and agri-business and housing. Indeed, the Bank has a long history of supporting access to credit for small and micro business, for rural development and for housing in the interest of furthering economic development and increased productivity in client countries.
Until recently, however, other financial services such as savings, payments and transfers at the household level have not figured in the World Bank research or project agenda. While the existence of high percentages of "unbanked" households in developing countries and the contrasting high percentages of "banked" households in developed countries has long been acknowledged, links between access to these financial services and increased economic growth have not been well documented. The high percentages of unbanked households and low access to formal savings, payment and transfer instruments was generally viewed as a symptom of underdevelopment, but not a factor in development.
Beginning, however, in 2002 World Bank Department of Finance, Private Sector and Infrastructure for the Latin American and Caribbean Region began to look at issues of access to financial services more broadly, and to enquire from the household point of view what services might be needed or in demand and what costs were associated with lacking access to formal sector financing institutions. The work in Latin America, beginning with studies in Brazil, Colombia Mexico has since led to the creation of a new Bank-wide working group on issues of access to financial services in general and on the unbanked in specific terms.
This new work began with a review of the literature which relates access to financial services to economic development (that is the broad coverage of formal sector financial institutions, measured by holding of deposit accounts), as well as the experiences of developed countries which have put in place programs to help increase access to banking services.
The literature on access to financial services and economic growth begins with Walter Bagehot who noted in 1873 the important role of the financial sector in England's economic growth when financial markets were able to mobilize savings to finance the implementation of new technologies in England. Walter Rostow linked financial intermediation to economic development in his studies of economic growth in the 1950's. More recent empirical studies have demonstrated not only that the development of financial sectors accompanies economic development (Goldsmith, 1969, King andLevine, 1993), but also that financial development is necessary for economic development. Others have convincingly demonstrated that deeper financial markets can help reduce aggregate volatility (Caprio and Honohan, 2001.) Certainly a formal banking system with widespread coverage is a consistent characteristic of developed economies, while widespread financial exclusion is associated with lesser-developed countries.
Links between access to financial services and growth are several. Rostow's argument that countries achieving an aggregate savings level above 5% of GDP will reach a level of reinvestment leading to economic "take-off" implies a certain link between financial intermediation and growth, insofar as it is access to financial services which convert aggregate savings to aggregate investment. Savings kept at home (under the mattress or in a tin can) cannot contribute to economic growth. By the same token, unbanked entrepreneurs who cannot obtain credit also represent an opportunity cost to a national economy. Hernando de Soto has argued that access to capital and to financial services is the key to economic growth both in advanced economies and in the developing world ( de Soto, 2000.) Access to financial services can help promote equity, and can, in this context, be linked to better economic equilibrium and to faster economic growth (Stegman, 1999.) Recent research has centered on the impact of access to credit on micro-economic development, suggesting that improving the access of micro-enterprises to financial services could have an important positive impact on a country's income distribution (Westley, 2001.) Other literature links better income distribution to macro-economic growth. Surveys of the literature on financial intermediation and poverty reduction conclude that development of the financial sector contributes to economic growth and thereby to poverty alleviation (Holden and Prokopenko, 2001.) Finally on-going research at the World Bank looking at cross-cutting evidence substantiates the hypothesis that "countries with better-developed financial intermediaries experience faster declines in measures of both poverty and income inequality." (Beck,Demirguc-Kunt, and Levine, 2004.) A number of recent studies have looked at the positive results stemming from access to a broad range of financial services and payment system services (Christen, 2000;Wenner and Campos, 1998). Reports from the development banking community and studies of international NGOs (Dichter, 1997), strongly suggest that lower income families need a wide range of complementary financial services both for everyday life and for asset building purposes (Caskey, 2001;Robinson, 2001;Rutherford, 1999).
In the USA, the responsibility of government to "bring its broad-based experience in capital markets and financial services to bear on the inner city" was underlined in 1997 by the then Secretary of the Treasury, Robert Rubin. Referring to an initiative that gave Treasury the responsibility of bringing unbanked welfare recipients into the mainstream banking system, Rubin declared. "If this works… it will not only give them a more efficient way to cash checks and access to other financial services, but it may also encourage people to save, to plan financially and, therefore, to improve their economic life over time." A further argument to link broader coverage of financial services to economic growth comes from the private sector. Experience from high-income countries demonstrates that niche "community development" financial institutions catering to low-income clients can be good businesses. Recent studies by the US Federal Reserve and the Office of Controller of the Currency have favorably compared the profitability of investments under the Community Reinvestment Act credits to commercial investments, and the profitability of community development financing institutions to other start-up banks. Indeed the fact that the banking crisis experienced in the US in the 1980's -the results of deregulated banks' over-investing in high risk venture -did not in fact cause a major sector collapse -as occurred in Mexico and in Colombia under similar circumstances, has been attributed to the fact that the US banks had spread their business over a wider gamut and depend on a variety of population sectors, not just the wealthy. Anecdotal evidence from banks such as Banco Popular of Puerto Rico, Union Bank of California, Pa-cific Bank of San Francisco, First Community Bank of Boston 1 , Harris Bank of Chicago, Bank-America and CitiBank Corp, all major banks with strong business lines in low-income, immigrant communities, all suggest that working with the poor and the unbanked is good business.
Regarding the second new direction the World Bank team took in approaching the issue of the unbanked in Latin America, learning from and working with experts in developing countries, particularly the United States, has considerably enriched the Bank's work. The USA has a long history of promoting access to financial services and on working with the commercial banking sector to help the poor build assets. Inputs from current and former staff of the Office of Controller of the Currency, from Fannie Mae, from FDIC have been invaluable, as have the collaborative efforts of teams from Ford Foundation, Annie E. Casey Foundation, Woodstock and Brookings Institutions. Moreover, recounts of experiences from private banks have been extremely illuminating. Finally, the work of scholars who have studied the issues in the US has served as both model and benchmark. In the same vein, the World Bank team has drawn on knowledge and experience of policy makers from Germany and Spain. Such links represent a break from much past work in access to financial services, which tended to emphasize experiences mainly cited from developing countries, on the grounds that the experiences of developed countries depend on too advanced an economic level to be replicable outside of the OECD.
Partnering or "twinning" as sharing experiences between developed and developing countries has been referred to, has been successful in other sectors. In this case, the collaboration of experts from the US, Brazil, Colombia and Mexico in the World Bank initiative has led to a new partnering in the area of access to financial services for the unbanked. This paper is an example of a product of the partnership in the course of the World Bank's work on the unbanked in Latin America. The discussions of John Caskey and Clemente Ruiz, two scholars and leading experts in the field of access to financial services, one from the United States and one from Mexico led to comparisons between the two countries. In this paper Caskey and Ruiz put to rest any lingering reluctance to make comparisons between developed and developing countries in favor of greater mutual understanding. They also demonstrate that, contrary to the some previous assumptions, there are important lessons to be learned from both sides, and there are replicable experiences. Indeed, as Caskey and Ruiz point out, private banks are already adapting mechanisms, and strategies for reaching the unbanked across borders. Equally important, they have set the stage for continued dialogue based on a serious analysis of issues, and consideration of solutions, in other countries concerned with the problems of the unbanked.

I. Introduction and Summary Findings
This paper examines the ways in which lower-income households obtain basic financial services in urban communities in the United States and in Mexico. In addition, the paper discusses the efforts that private sector and government organizations are making to lower the cost or improve the quality of those services. The goal of the paper is to summarize available information on these issues and to assess the rationale and challenges facing the strategies that both countries are employing to improve the financial services available to lower-income households.
In the paper, we give particular attention to "unbanked" households, by which we strictly mean households that do not have deposit accounts with any regulated deposit-taking institution. But we also use the term more broadly to include lower-income households that use high-cost informal-sector non-bank lenders when they seek consumer loans. In many cases, these are unbanked households, but they may also include households with deposit accounts.
As noted above, our emphasis in the paper is on lower-income households living in large urban areas. Two considerations explain this focus. First, in both countries, the best survey data on the use of financial services among lower-income households applies only to urban areas. Second, while there are undoubtedly many similarities among the unbanked in rural and urban areas, there are also differences in the financial institutions with which they interact and differences the reasons that these households are unbanked. To try to cover both the rural and urban unbanked in two countries in one paper, would make this rather long paper much longer and less focused.
Before plunging into the details, it is useful to review some distinguishing features of the two countries that explain some major differences in the extent to which lower-income households are unbanked, their use of non-bank financial services, and strategies for improving financial services to the unbanked. Most importantly, there is a substantial difference in typical household incomes across the two countries. On a purchasing power parity basis, national income per capita in Mexico in 2002 was U.S. $8,540 p.a., while it was $35,060 p.a. in the United States (World Bank, 2003). This difference means that what we consider to be a lower-income household in the U.S. would often qualify as a middle-income household in Mexico. The difference in typical household incomes between Mexico and the U.S. also creates a substantial difference in the percentage of households who are unbanked. Based on 1992 data, the research division of the US Federal Reserve concluded in 1997 that 151 percent of families in the U.S. do not have any type of transaction accounts. In Mexico City, a recent survey found that 76.4 percent of households in the same situation.
Numerous other distinguishing features of the two countries contribute to differences in the percentage of unbanked households and ways that they obtain financial services. Here we mention three. First, banking in Mexico is much more concentrated than in the U.S. In 2004, the year of the World Bank study, Mexico had fewer than 20 commercial banks and a similar number of registered cooperatives and credit unions in Mexico City 2 . The five largest banks hold 82 percent of all deposits. The United States, in contrast, has over 10,000 banks, without counting cooperatives and credit unions. In New York City, one of the more highly concentrated urban markets in the U.S, the five largest banks account for 54 percent of all deposits. Competition for deposits is also more varied in the United States. Large commercial banks compete with community banks, thrifts, and credit unions, all of which are regulated and offer deposit insurance. In Mexico, non-bank savings institutions are much smaller and many have traditionally been largely unregulated. Second, computerized credit histories for consumers and automated credit risk assessments are highly developed in the U.S. and cover nearly all working adults. In Mexico, such institutions are far less comprehensive and sufficiently less developed that there can be little meaningfully predictive automated credit risk assessments for most lower-income Mexican households. Third, labor costs are significantly less in Mexico compared to the U.S. This means that labor-displacing financial service technologies employed in the U.S. may not be appropriate in the Mexican context. Despite these differences, however, this paper identifies many parallels in the issues that the two countries face in trying to improve financial services for lower-income households, even if the scale and context of the problem differs.
Who are the unbanked?
In both countries, the unbanked are characterized by lower incomes and lower education levels than the population at large, and they also tend to be marginalized in socio economic terms. In Mexico this is inferred by the high percentage of unbanked dependent on the informal sector, while in the U.S. it is defined by the disproportionate representation of immigrant and minority groups. (According to US Federal Reserve Bank reports based on surveys of consumer finance from 1997, 54% of the unbanked are Latin American migrants, compared with 15.1% over the population as a whole 3 .) However, although in line with income and education levels overall, the percentage of unbanked is markedly greater in Mexico than in the U.S. Surprisingly, home ownership among the unbanked is also greater in Mexico. Where do the unbanked get basic financial services?
The unbanked in both countries tend to rely on cash transactions and on services provided by commercial outlets, although in México the unbanked rely on informal forms of saving and borrowing not present in the U.S. At the same time, the unbanked in México show a high rate of home ownership, suggesting an alternative form of saving not present in the U.S.
In both countries the majority of unbanked are paid by check or by deposit to a third party account. Less than a fifth of the U.S. group and a tenth of the Mexican group are paid in cash. This suggests an important cost to the unbanked in transaction costs. At the least it involves a trip to a bank, at most, a commission paid to cash a check or to access a third party account.
The unbanked in both countries face additional costs in making payments. In the U.S. unbanked persons generally rely on payment services associated with the "check cashing outlets", a service industry which has grown up in response to the unbanked' needs. Among the "CCOs" those associated with "payday lending 4 " have been under investigation recently for predatory lending. In Mexico the unbanked carry out their transactions in cash at a cost of travel and time, although focus groups also report that the unbanked frequently circulate signed checks as a form of cash or cash checks in stores where they also (by obligation) make purchases. In terms of borrowing, surveys looked at what credit sources the unbanked have used in the previous three years. Interestingly, family, friends and third parties make up the principal sources of loans, in both countries suggesting the importance of informal support network for marginalized groups in the United States. (Recall that a high percentage of the unbanked are recent immigrants and minority groups.) In Mexico, friends cover about the same percentage for credit needs as in the US (between 8-9%) but the retail stores lead in importance as lenders to about half of the unbanked population. Government programs -absent in the US -also have a small but significant presence providing credit to about 6% of the unbanked in Mexico. 4 Payday lending actually refers to loans made between paychecks, guaranteed by the turn over of the paycheck to the lender on payday.

Payments recei ved US A Mexico
B y C h e c k 5 0 . 5 % 8 5 . 6 % (Check cashing operations charges 2-3%) I n C a sh 1 8 . 8 % 7 . 8 % D i r e c t b a n k d e p o si t 1 6 . 6 % 6 . 3 % (to third party account)

Where do the Unbanked borrow?
In both countries the majority of the unbanked households report that they have no financial savings. Nonetheless, a significant percentage of the unbanked do save in cash, especie or other informal means. In Mexico this percentage translates into approximately one-fifth of the country's population, while in the U.S. it represents around 2%. (The unbanked in Mexico represent close to 75% of the population compared to less than 10% of the U.S. population.) The high rate of home ownership among Mexico's unbanked adds further evidence to their capacity to save.
What is the problem with being unbanked?
From the forgoing, it is possible to make some estimates of what the unbanked in both countries pay for financial transactions. One telling comparison, for example, in Mexico arises from the use of credit from retail stores, at monthly interest rates of between 15 -30%, or with 6 -10% per month from the credit unions. These rates are well over the interest charged by credit card companies, let alone bank lending rates. But costs of being unbanked depend, of course on the household situation. We assume that all unbanked households must pay utility bills, both in the U.S.A. and in Mexico. Additionally, for those who save, there are costs associated with not having access to a bank deposit account. Finally, costs associated with receiving payments, such as check-cashing are calculated. The accumulated costs can run up to 4% of a median income in the US, or to 15% of median income in Mexico. Credit is not factored into this estimate, but for U.S. families using pay day lender services, or for Mexicans buying goods on a lay-away plan, the percentage of income devoted to financial services affecting the unbanked would be higher. Why don't the unbanked use banks?

Sa v i n g s M e c ha n i sm s U SA M e x i c o
As to why the unbanked do not use banks, in both countries the issue of trust appears to a minor degree, but while the unbanked in Mexico perceive barriers in the costs and requirements of financial institutions, in the U.S. the unbanked point to their own financial situation and lack of savings as a reason for not using banks. In both countries the unbanked pay a significant cost in terms of additional transaction fees, time, and insecurity, in not using formal sector financial services. Concerning initiatives to improve access to financial services, the U.S. has a longer history of both regulatory and private sector practices encouraging retail banking practices. The Mexican government has recently begun to address the situation by focusing on the "popular" banking sector, and assisting credit unions and cooperatives to develop broader services.
Here a serious difference emerges. While a similarly serious percentage of each group (close to one-fifth) underline their discomfort with or lack of trust in banks, the unbanked in the United States do not consider that they have enough money to justify opening an account, whereas the unbanked in Mexico City point to the high costs of opening an account in Mexico City. Actually further analysis from CONDUSEF showed that the main commercial bank savings programs in 2003 were not affordable to 75% of the population, based on an initial and minimum deposit equal half of monthly earnings and maintenance fees of 1% of monthly earnings.

Reasons
And, finally, what private-sector and government initiatives are underway to improve financial services for the unbanked?
In both countries governments and the banking industry have developed strategies to help the unbanked gain access to financial services. Both countries have developed programs to support microfinance institutions through grants (USA and Mexico) and subsidized credit programs. The US supports community development institutions, while Mexico's focus is on lenders to small businesses, car retail and new home developers. In addition, the Mexican government has long managed a postal savings program and is currently developing a major regulatory capability to stimulate growth and modernization in the "social financial sector" (cooperatives and credit unions.) The US has experimented with regulatory policies and incentive programs, aimed at both the banks, the informal lenders and at the unbanked themselves.

Mexico
Public policy focus on Non-Banking Instituti ons BANSEFI -Regulating and setting up electronic base for cooperative and social sector SOFOLES -Lending to finance corporations for housing, car and business loans PRONAFIM -Grants to micro finance companies

Innovati on by Mexican B anks
Banco Azteca -New and outside the banking sector -virtually without competition Banamex and Co mpartamos has introduced a debit card option USA Public policy i nitiati ves: Co mmunity Reinvestment Act (CRA) applies mild pressure on banks to maintain offices in lo w-income communit ies and offer low-cost bank accounts. Support for Commun ity Development Financia l Institutions (CDFIs) that specialize in serving low-income communit ies. Some states use price controls for non-banks that provide payment or credit services to the unbanked.

Public and pri vate sector ini tiati ves foc using on new products:
Pay cards, Electronic Transfer Accounts, ID Cards for non-registered aliens

Educating the Unbanked:
Savings and lending education intended to help lower-income households to build savings and improve credit histories, which are the main barriers to accessing mainstream financial institutions. Individual Develop ment Accounts (IDAs) increase incentive to save.

II: The Urban Unbanked in the United States
This section uses a survey of households conducted in lower-income communities in New York City and Los Angeles to profile the unbanked and to explain why they are unbanked. The Office of the Comptroller of the Currency (OCC), the agency that charters and regulates national banks, conducted this high-quality survey of the use of financial services by residents in low-and moderate-income census tracts in both cities in 1998. The survey sampled 2,000 adults asking, among other things, numerous questions about how they receive and make payments. The surveys were conducted by telephone and in face-to-face interviews. The interviews were conducted in either English or Spanish depending on the respondent's choice. Because of careful design and persistent efforts, the OCC obtained a remarkable 70 percent response rate. For people using the survey data, the OCC supplies a set of weights to convert the sample responses into responses representative of the 2.6 million adults living in the low-and moderate-income census tracts of the two cities. All of the tables in this paper based on the OCC survey use the weighted data.
For our purposes, there are two weaknesses to the survey. First, it focused on only the residents of two large cities who may not be representative of urbanized lower-income households generally. As a result the percentage of unbanked in this survey is higher than that given by the Federal Reserve's 1997 percentage for the country at large 5 . Both of these cities, for example, have much larger Hispanic and immigrant populations than is typical of American cities. And New York has a lower percentage of homeowners than is typical of American cities. Second, the OCC with little discernable rational basis decided to omit a number of survey responses from the publicly released data base. Consider one example. The OCC survey asked people without bank accounts why they did not have an account and read a list of reasons that they could agree with, including such reasons as "don't have enough money," "bank fees are too high," etc. The public data set includes peoples' responses to this question. The OCC survey followed this question with an identical question with a second set of possible reasons, such as "you need a Social Security number to open an account," "bank won't let you open an account," etc. The public data set does not include peoples' responses to this question. Despite these weaknesses, the OCC data provides a number of insights into the use of financial services by lower-income urban households. Table 1 presents an overview of the socioeconomic characteristics of the adults in the survey. As indicated in the table, almost 60 percent of the adults work and almost an equal percentage have household incomes of $30,000 or less. We will refer to these households as lowerincome households. 6 More than half (53%) of the surveyed population identified themselves as Latino and 33 percent as black non-Latinos. Slightly more than one-third (37%) of the adults did not have a deposit account of any type. In contrasting the data between the two cities, the most striking differences is the much greater representation of Latinos (63%) in Los Angeles, and the greater presence of younger individuals and individuals in the lowest income category in this city.  Table 2 contrasts the characteristics of the unbanked individuals with the banked. Recall that "banked" individuals need not literally use a bank. They can have a deposit account at a bank, credit union, or thrift.
As shown in the table, the unbanked tend to be younger than the banked, they have less education and are much more likely to rent than to own their homes. They are less likely to be working and they are much more likely to have low incomes. They are more likely to identify themselves as Latino. Finally, they are much less likely to have maintained any financial savings over the past year. Presumably, individuals who had bank accounts but who answered that they had no financial savings drew their account balances down to near zero at the end of each pay period. As we discuss later, the strong correlation between having financial savings and using a bank has led many policy analysts to focus on helping lower-income households to build savings as a way to bring them into the banking system. Individuals who do not use banks still need to obtain financial services. If they have any financial savings, they need a means to safeguard their savings. Even if they receive income in the form of a check, they need a means to convert the check into cash. Finally, they many need to borrow periodically. This subsection draws on the OCC survey data to examine where and how the unbanked obtain these financial services. Table 3 presents data on the forms in which the unbanked keep financial savings and contrasts this with the banked. As noted in Table 2 above, 86 percent of the unbanked said that they had no formal sector financial savings. The OCC survey also asked about the use of informal means of financial savings, such as holding cash, money orders, uncashed checks, jewelry that can be sold, etc. As shown in Table 3, 21 percent of the unbanked respondents said that they use such means and 16 percent of the banked did. Allowing for both formal and informal means of financial savings, 69 percent of the unbanked had no financial savings and 22 percent of the banked had none.  Table 4 examines how people receive their incomes and convert that income into cash. Among banked individuals, 38 percent received an electronic deposit to their account, 49 percent received a check, and 6 percent were paid in cash. Among the unbanked, 51 percent were paid by check, 19 percent were paid in cash, and 17 percent receive an electronic payment through a non-bank. Almost all of the people in this latter category live in New York City where many check-cashing outlets (CCOs) participate in a network that allows them to distribute cash to people who are paid electronically but who do not have bank accounts. People receiving a check must deposit it or cash it in order to convert it into a spendable form. Among the banked, 78 percent generally deposit their checks or a part of the value of their checks and 21 percent generally just cash their checks. Among the unbanked, 95 percent cash their checks. There is also a striking difference in where people cashing their checks go to do so. Among the banked, 79 percent go to a bank and 16 percent to a check-cashing outlet (CCO). 7 There are a variety of reasons that individuals with bank accounts would cash their checks at CCOs. If a customer's bank account does not have sufficient funds to cover the check in case it is returned unpaid, many banks will not cash it. They will instead insist that the individual deposit it and wait two to three days for it to clear before the banks will provide access to the funds. In addition, many individuals may wish to cash their checks, buy money orders, pay utility bills, and buy stamps and envelopes in which to mail payments ---and do this all in one location. A typical CCO will handle all of these transactions. A typical bank does not handle utility bill payments or sell stamps and envelopes.
Among unbanked individuals cashing checks, 21 percent mainly use banks, 68 percent use CCOs and 5 percent use grocery stores. Undoubtedly, one of the explanations for the heavy reliance on CCOs is that many banks will not cash paychecks for non-depositors. In addition, an individual cashing his or her check at a CCO can buy money orders, stamps, envelopes, and pay utility bills at that same location. As indicated in Table 6, another payment that many lower-income urban households make is to transfer funds to family members living outside the U.S. In the OCC survey, 20 percent of the respondents indicated that they had made at least one such remittance over the past year. Of those who did so, 45 percent wired the funds and 24 percent mailed a money order. No other method accounted for over 5 percent of typical means of transferring the money. The households in the OCC survey reported using a variety of forms of consumer credit. As shown in Table 7, 56 percent of individuals with bank accounts had a major credit card but only 8 percent of those without banks accounts did. When banked individuals looked for a loan, the largest percentage reported applying to a bank. In the case of the unbanked, the largest percentage turned to a family member. In terms of their use of nonbank forms of credit, 16 percent of the banked reported that they had obtained a credit card cash advance at some time in the previous 3 years, 10 percent had used a layaway plan to purchase a consumer good, and 5 percent had obtained a loan against their expected tax refund. In the case of the unbanked, 10 percent reported that they had used a layaway plan and 4 percent had obtained a tax refund loan. It is likely that the unbanked make much less use of bank loans and credit cards because they would not be able to pass traditional risk screening procedures. Interestingly, when the OCC survey asked if the respondents could borrow $500 for 3 months if the need arose, 68 percent of the banked were confident that they could do so as were 51 percent of the unbanked. This suggests that about a third of the banked and half of the unbanked in the survey are severely limited in their access to credit of any type. A striking finding in Table 7 is the low percentages of individuals who reported that they used a pawnshop or payday lender. 8 This is likely a misleading indication of the use of these credit sources in lower-income urban areas generally. New York sets a severely binding interest rate ceiling on its pawnshops, so pawnshops in New York City are scarce and generally take only jewelry as collateral. In addition, New York regulations do not permit payday lending to be profitable. Although payday lenders were reasonably well-represented in Los Angeles in 1998, they have grown explosively in California and many other states since that time. Undoubtedly, a contemporary survey would find much heavier use of payday lenders than is suggested by the OCC survey.
Payday loans are only available to banked individuals since, in a traditional payday loan, a customer writes a personal check made out to the lender. The lender agrees to hold the check for about two weeks before depositing it. In exchange, the payday lender advances a cash payment to the customer that is somewhat less than the amount of the check. The difference, which is the "finance charge," in combination with the maturity of the loan determines the annualized interest rate. In the states where payday lending thrives, lenders typically charge $15 to $25 for each $100 that they advance. That is, in a typical transaction, a borrower might write a check for $235 that the lender agrees to hold for two weeks and the lender would provide the borrower with a $200 cash advance. The annualized interest rate on such a loan is 455 percent.
Prior to the maturity of the loan, the borrower can pay the lender the face value of the check in cash, extinguishing the debt and concluding the transaction. If the borrower does not repay the loan by its maturity, the lender may deposit the check. Assuming that the check clears, the loan is fully repaid and the transaction is complete. If a borrower does not want to repay a loan at maturity, or cannot, a lender will frequently allow the borrower to renew the loan by "rolling it over". In a rollover, the borrower pays the lender the finance charge due at maturity and the lender agrees to hold the check for another specified period of time.
Survey data (Caskey 2002) indicate that most payday loan customers are not officially poor. Rather they belong to moderate-income households, with incomes between $18,000 and $50,000, but they struggle to pay their bills on time. The vast majority of payday loan customers do not have access to convenient lower-cost credit from mainstream lenders because they have severely impaired credit histories or because they have reached the limit of the credit lines these lenders are willing to extend.

C. Why don't the unbanked use banks?
A number of surveys have sought to answer the question: Why don't the unbanked use banks? In almost all of these surveys, unbanked individuals respond most frequently that they don't need a bank account because they have no month-to-month savings to safeguard. This response is also consistent with the OCC findings, discussed above, regarding the financial savings of the unbanked. Generally, the second and third most frequently cited explanations are bank minimum balance requirements and bank fees. But these too are related to a household's level of savings since minimum balance requirements will generally not be binding on individuals able to maintain about $100 of month-to-month financial savings. In addition, low balances in an account frequently trigger many bank fees, such as monthly account fees and bounced check fees.
Other reasons that people give to explain why they do not have a deposit account include concerns about privacy, a lack of comfort interacting with banks, language barriers and banks not letting them open accounts. People who banks do not permit to open accounts likely have histories of writing bad checks, have severely impaired credit records, or do not have required identification. People who say that they do not open an account out of a desire for privacy may have a number of concerns, including: • fear that a creditor might seize the savings of a delinquent debtor; • fear that a former spouse might seize the savings of an individual behind on his childsupport payments; • fear that welfare eligibility would be threatened by a substantial account balance or by a history of deposits from under-the-table earnings; • a desire to hide earnings from the tax authorities; or • fear that a bank might report suspected illegal immigrants to the Immigration and Naturalization Service or that the INS could use bank records to discover their presence. • Interestingly, the least important reasons people give for being unbanked are the hours and locations of bank branches.
The OCC survey was typical of other surveys in the way that it investigated why people did not to have bank accounts. In the OCC survey, if a person did not have an account, the surveyor read a list of possible reasons to explain the lack of an account and asked the respondent to choose the main reasons that applied to him or her. For our purposes, however, there are two problems with the data. First, the most common reasons found in other surveys -the person did not want an account because he or she had no savings -was not included as a possible choice. Second, the public data set excluded the responses of people who selected such reasons as: the bank would not let them open an account, their bank account could be frozen by a creditor, or they thought that they would need a Social Security card to open an account.
Despite these shortcomings, the OCC data are still broadly consistent with the findings of other surveys. As shown in Table 8, the two most common reasons that people cited from this list of possible reasons are the lack of money necessary to open an account and bank fees. But 58 percent of the unbanked indicated that none of the listed reasons were important to them. This is consistent with the finding from other surveys where the most common reason people give for not having an account is that they do not have any savings, a response that was not possible in the OCC survey. Interestingly, in the OCC survey 77 percent of the unbanked said that they were aware that some banks had "basic" checking accounts with low minimum balance requirements, low fees, and a small number of monthly free checks. And 63 percent of the unbanked indicated that they had never had a bank account. This is much higher than is commonly found in other surveys, and may reflect the much higher percentage of Hispanic individuals in the sample areas compared to the sample areas of other surveys. Because of the shortcomings of the OCC survey regarding the reasons that people do not have a bank account, we include the results from one other survey that addressed the same issue. In a 1996 survey of 900 lower-income urban households, Caskey (1997) asked households without deposit accounts, why they do not have an account. He provided respondents with a list of possible reasons from which they could select one or more. They could also provide a reason that was not on the list. As shown in Table 6, 53 percent of the respondents cited "don't need account because we have no savings" as a reason, making this the most frequent reason cited; another 45 percent cited bank fees or minimum balance requirements. No bank has convenient hours or location 8.5 Source: Caskey (1997) As discussed earlier, large percentages of lower-income urban residents see themselves as having severely limited access to credit or turn to non-bank lenders, such as finance companies or payday loan shops. Although the OCC survey did not examine why people lack access to bank credit, other data support the conclusion that the main reason that people are excluded from bank credit is that they have impaired credit histories or, in some cases, no credit histories. In addition, many households have reached the limit of the credit that banks are willing to extend them.
A comparatively large percentage of lower-income and minority households report a history of failing to fulfill payment obligations on time, heavy debt-payment burdens, bankruptcy, or liens placed on their property. For example, Fair Isaac and Company Inc., one of the largest U.S. credit scoring bureaus, reported that it examined loan application data from "tens of thousands" of individuals applying for installment loans between July 1992 and December 1994 (Martell et al). Using its proprietary data base, Fair Isaac assigned scores for each applicant ranging from 0 to 240, with a higher score indicating a lower credit risk. The report (p. 14) notes that, "Many lenders set their cutoff score -the score…below which they decline applicants -around 200." The study found that that 54 percent of low-and moderate-income individuals (defined in the study as individuals with annual incomes under $21,000 a year) had scores below 200, but only 33 percent of individuals with incomes over this amount had scores below this level. Similarly, the Freddie Mac Corporation (1999), a large government-sponsored housing enterprise, conducted a survey of 20,000 households with incomes under $75,000. The survey focused on the households' credit histories and financial behaviors. It classified a household as having a "bad" credit record if the household reported that: • it had been at least 90 days late on a payment in the previous two years, • it had been 30 days late on a payment more than once in the previous two years, or if • it had a record of bankruptcy or liens files on its property due to payment delinquencies.
As shown in Table 10, by these criteria a substantially higher percentage of lower-income households have bad credit records that would likely exclude them from prime loans than do higher-income households. In addition, substantially higher percentages of African-American and Hispanic respondents had bad credit records than did white respondents.  (1999) In a national household survey conducted in 2001, the Federal Reserve found a similar pattern linking household incomes to indicators of credit risk. As shown in Table 11, lower-income households were much more likely to have such indicators than were higher income families. The strong link between household income and indicators of credit risk should not be surprising. After all, since many lower-income households have no financial savings, any economic disruption, such as a health crisis in the family, a job termination, etc, can force a family to put off non-essential expenditures, including debt service obligations. In addition, there is a strong correlation between household education and income. If education is linked to money management skills or to an awareness of the benefits of a good credit record, this may also partly explain the correlation between household income and indicators of credit risk.
An impaired credit record does not necessarily shut one off from credit. Mainstream lenders will extend loans to people with somewhat impaired credit histories or with fairly heavy debt burdens, but they commonly increase the interest rate on the loans to reflect the increased risk. If an individual has a more seriously impaired credit history or heavier debt burden than mainstream lenders will accept, the individual can often borrow from "subprime" lenders, many of which are subsidiaries of mainstream prime lenders. Subprime lending is generally a separate business from prime lending because the business practices of the subprime lenders can differ substantially from those of the prime. Subprime lending frequently involves much more person-to-person contact and much faster responses when a borrower falls behind on a scheduled payment. Individuals with seriously impaired credit histories or very heavy debt burdens can even lose access to the subprime market. At this point, if they need credit, they can turn to pawnshops, payday lenders, or rent-toown operations that do not pull traditional credit reports. 9 A pawnshop will lend to anyone since the loan is based only on the value of the collateral that the customer leaves in the possession of the pawnbroker. A payday lender will lend to almost anyone who has a checking account in good standing, recent steady employment, and who has not failed to honor a previous payday loan. Rent-to-own stores will also implicitly provide credit for the purchase of furniture and other consumer goods because the stores retain legal ownership of the goods, making repossession in the case of missed payments much simpler. D. What is the problem with being unbanked?
One major problem with being not having a bank account or not having access to bank credit is that the alternatives tend to be significantly more expensive. Most check-cashing outlets, for example, charge two to three percent of the face value of a paycheck to cash it. 10 They also levy fees for money orders, stamps and envelopes, and for handling utility bill payments. This means that a household with $20,000 in take-home income that regularly uses a CCO for all of its payment services can easily spend $600 year on just payment services. Were the same household consistently able to meet a bank's minimum balance requirements and not bounce any checks, it would almost always pay less than $100 a year in bank fees, and often substantially less. Such a difference can make a noticeable impact in the standard of living of low-and moderate-income households whose budgets are already stretched to pay for household necessities.
Similar cost differences arise in the case of bank credit versus the credit of lenders that serve mainly high-risk borrowers. Pawnshop interest rates are regulated by states, not the federal government. In states where pawnshops thrive, they generally charge annualized interest rates of 150 percent or more on typical loans of around $100. Payday lenders, which have become more numerous than pawnshops, are also regulated by the states. In the states where payday lenders are common, the lenders charge annualized interest rates of 300 to 500 percent on typical loans. Small loan finance companies, which make somewhat larger loans and refuse to accept very high risk customers, typically change annualized interest rates of 50 to 100 percent. Annualized interest rates on bank credit cards are have generally hovered between 18 and 25 percent in recent years. Households that must borrow in the alternative financial sector because of impaired credit histories or heavy debt burdens, pay a substantial penalty for their status. When these are lowerincome households who already struggle economically, this compounds their problems.
Interestingly, in an ethnographic study conducted by Caskey et al (1997) among lowerincome households in a large city and in a small town, the researchers found that the unbanked did not complain about the cost of payment or credit services. Rather, they complained most about the personal stress of living paycheck to paycheck without easy access to credit. They fully expected that minor or major personal financial setbacks, such as a required automobile repair, a large health care expense, or an employment interruption, were coming their way as such setbacks had in the past. They said that they worried constantly about this because their credit histories and complete lack of savings turned every such set back into a very stressful personal financial crisis. The stress associated with their situation may, in fact, be the major problem of being unbanked and of lacking access to bank credit.

E. Policy initiatives aimed at helping the unbanked
Agencies and legislators of the federal and state governments have long expressed concern that millions of lower-income households are unbanked and millions are excluded from bank credit. They have pursued and are pursuing a wide range of initiatives to lower the cost of financial services for the poor and to bring more of the poor into the banking system. Some private sector institutions, both for-profit and not-for-profit, have also launched initiatives of their own. This subsection of the paper reviews these initiatives. In doing so, it divides them into four categories. First are initiatives that focus on making banking services more convenient and affordable for a larger segment of lower-income households. Second are initiatives that seek to lower the cost or improve the quality of the non-bank financial services often used by lowerincome households. Third are initiatives that seek to change those characteristics of the unbanked that may leave them dependent on higher-cost non-bank financial service providers. Finally, there are initiatives intended to develop new financial service products that may better meet the needs of unbanked households or that may lower the fees that these households pay for basic financial services.
Before reviewing the initiatives aimed specifically at broadening the coverage of banking services for the poor, it is worthwhile noting that banking in the United States has historically been widely dispersed in geographical terms. Branch banking was heavily regulated through the first half of the 2oth century and interstate banks were illegal through the its first three quarters. As a result, the U.S. currently counts on a far greater banking presence -with some 10,000 banking institutions versus the forty counted in Mexico. Furthermore, given the difficulties of spreading out geographically, U.S.banks have traditionally drawn intensely on their local community for clients.

Initiatives targeting deposit-taking financial institutions
In 1977, the federal government enacted the Community Reinvestment Act (CRA). The main intent of the CRA was to ensure that banks provide home mortgages in all of the communities from which they gather deposits. The CRA was passed in response to concerns that many banks, in their lending decisions, consciously or unconsciously discriminated against lowerincome communities and communities with high percentages of racial and ethnic minorities. In addition, there were concerns that bank branches were largely absent from many lower-income urban communities. Bankers, in some cases, agreed with this latter point, but emphasized that it was difficult to serve a lower-income population profitably. They argued that deposit balances are low, transactions are numerous, and loan opportunities are limited. Some community activists countered that the absence of the banks reflected preconceived notions rather than truly limited profit opportunities.
Over time, the CRA has been revised substantially. As of this writing, banks undergo an annual, or less frequent, CRA examination. The examiners give the banks one of four possible CRA ratings: outstanding, satisfactory, needs improvement, and substantial noncompliance. These ratings are based on a bank's performance in three areas: its record of lending (primarily home mortgages and small business lending) in the communities from which it gathers deposits, its investments in community development projects, and its delivery of retail banking services in its market area. The last category is the one that concerns us here. A bank's service rating depends on its geographic distribution of branches and its record of opening and closing branches, especially branches in low-and moderate-income (LMI) communities. The rating also depends on other community development services or service delivery mechanisms that the bank may offer that benefit low-and moderate-income individuals.
The point of the service test is to pressure banks and thrifts to open or maintain branches in LMI communities and to offer products, such as low-cost basic checking accounts, or services, such as consumer education efforts, that benefit LMI individuals. That said, it must be admitted that the pressure is very light. For one, there is no immediate consequence for a bank that receives a CRA rating below "satisfactory." Regulators simply take into consideration a bank's CRA rating when the bank must seek regulatory approval for future actions, such as a bank merger. Even then, the bank's CRA record is only one of several factors that regulators consider in granting approvals. Beyond this, a bank's service record is one of three criteria that examiners use in determining a bank's overall CRA record, and it gets a weighting of 25 percent, not 33 percent. Finally, as Michael Stegman and Robert Faris (2001) point out, the standards for the service test are vague, so almost all banks receive a satisfactory rating or better in this area.
In addition to the CRA, there have been periodic legislative efforts to mandate that banks offer low-cost basic checking accounts. A few states have passed such laws, but the federal government has not. While the bank trade associations have opposed these laws, they have called on their members to offer such accounts voluntarily, and a majority of banks claim that they do. Periodically, members of Congress have proposed that credit unions, which are not subject to the CRA, also be evaluated based on their service to LMI communities. The main credit union trade association has vigorously opposed this, but at the same time it and the federal regulator of credit unions have encouraged credit unions to reach out to LMI communities. In 1998, the regulatory agency for federal credit unions made it relatively easy for credit unions to make residents of LMI communities eligible for membership. But one recent study found that the mainstream credit unions in Chicago serve relatively few lower-income households (Jacob et al, 2002).
In 1999, the U.S. Treasury Department launched a major effort to pay all federal benefit payments, such as social security benefits, electronically. One impediment to this initiative was the large number of benefit recipients without bank accounts. As a result, the Treasury urged banks to offer Electronic Transfer Accounts (ETAs). The Treasury offered to pay banks $12.60 for each ETA account they established for benefit recipients, and the Treasury specified a minimum set of characteristics that these accounts must meet. The accounts could not cost account owners more than $3 a month, they could levy no fee for electronic deposits coming from the federal government, they could have no minimum balance requirement, and they had to permit four free cash withdrawals per month. Although hundreds of banks, thrifts, and credit unions agreed to offer the accounts, usage rates are very low. Most recipients, who currently receive their benefits by check and cash the checks at a check-cashing outlet or grocery store, probably see no benefit to the account. Even if they had an ETA account, they would still need to purchase money orders, stamps, and envelopes to pay bills, and they can do this in one stop at a check-cashing outlet and at many grocery stores.
Finally, some government agencies and philanthropic organizations have provided financial support to credit unions and banks that have, as a main goal, promoting the economic development of LMI communities. Such banks, credit unions, and venture capital firms are generically known as community development financial institutions (CDFIs). Some are organized as not-for-profit institutions and others as for-profit, but to be classified as a CDFI a for-profit organization must be willing to limit its profits in order to achieve community development goals.
CDFIs have been around for decades, but most started in the 1980s and 1990s. During this time, many policy makers and community activists promoted the idea that access to financing was a major barrier to the economic development of LMI communities. Philanthropic organizations, such as the Ford Foundation, backed financial institutions that committed to make special efforts to provide financial services in LMI communities. In addition, in 1994 the federal government created the CDFI Fund in the U.S. Treasury Department. Between its founding and late-2003, it has provided $534 million in financial support to certified CDFI financial institutions. The vast majority of these funds have gone to CDFIs that provide mainly financing for housing and business, but some of it has subsidized or capitalized banks and credit unions that focus on providing basic consumer financial services. It would be fair to say, however, that CDFIs that emphasize the provision of basic consumer financial services to LMI communities have had a very small impact nationally. The institutions are relatively few in number and most are small and serve hundreds or a few thousand households in an urban area, not tens of thousands.
In addition, the performance of CDFIs has not supported the views of those community activists who argued that traditional banking institutions could be profitable while serving a predominantly LMI community. Some CDFIs are profitable, but most require ongoing explicit or implicit subsidies. There are no formal studies of the banks and credit unions that manage to earn profits while serving predominantly LMI communities, but anecdotal evidence suggests that they deviate from traditional banks in a number of ways. First, they work hard to keep their operating costs low. Their branches have low-cost furnishings and are often located in low-rent buildings, frequently the former branch offices of banks that withdrew from deteriorating communities. Staff salaries, especially for top management, are well below those found in traditional banks. Second, the front-line staff, such as tellers and even the top managers, dress and communicate in ways that make the LMI customers comfortable. In other words, they either belong to the community they serve or make an effort to belong. Third, the institutions raise as many large deposits from outside of the community as they can. This is an implicit subsidy. That is, the community development banks and credit unions ask churches, foundations, local governments, mainstream banks, and others to deposit up to $100,000 (the maximum covered by federal deposit insurance). The CDFIs offer a below-market interest rate on the deposits and invest the funds in market-yielding instruments. The earnings support the operations of the CDFIs. These institutions make the deposits because they support the community development goals of the CDFIs and, in the case of banks, because they receive CRA credit for doing so.

Initiatives targeting non-bank alternatives
In addition to efforts to push banks to serve lower-income households or to subsidize and capitalize those that make special efforts to do so, many policy initiatives have focused on the non-bank financial institutions that serve LMI households. The initiatives have sought to reduce the costs associated with using the non-bank institutions and improve the quality of their services. Specifically, numerous state governments have imposed price controls on the fees or interest rates that check-cashing outlets, pawnshops, payday lenders, and others can levy for their services. In addition, many state governments require these alternative financial institutions to clearly post their fees, and have made it easy for their customers to file service complaints with state agencies.
These state regulations on the fees and interest rates charged the non-bank financial institutions are highly controversial. On one side, the advocates claim that there is some market failure, such as the monopoly power of the financial institutions or an inability of their customers to understand the fees or interest rates, which justifies the price controls. Opponents argue that there is no evidence of general monopoly power. Indeed, entry barriers for these non-bank financial service providers are relatively low. In addition, the opponents of fee regulations acknowledge that many LMI customers may not understand annualized interest rates, but they argue that the customers do understand the dollar costs of their transactions and all other relevant aspects of the transactions. Unlike home mortgages, payday loans, pawn loans, or check cashing transactions have simple-to-understand terms. Finally the opponents to regulation emphasize that low ceilings on permissible fees or interest rates will simply drive marginal non-bank financial institutions out of business. There are, for example, no payday lenders operating in states where they are not able to charge annualized interest rates over 100 percent. The opponents of regulation argue that the absence of the non-bank financial institutions that serve LMI households pushes those households to less-desirable alternatives.
International remittances between U.S. residents and residents in Mexico and a few other countries is one area in which there has undoubtedly been progress in lowering the cost of nonbank financial services. In the late 1990s, most U.S. residents who sent typical remittances of $200 to $300, paid $20 to $25 for each transfer. In addition, the transfer was often made using an exchange rate that deviated substantially from market rates and the recipient frequently had to pay an additional fee upon receipt of the funds. The Mexican government brought attention to these high costs and, in the U.S., lawsuits were filed against the largest money transmitters over their hidden and unfavorable exchange rates. At the same time, increasing numbers of banks and credit unions in the U.S. established partnerships with Mexican counterparts to handle the remittance business. The increased competition and the pressures brought upon the traditional transmitters to cut their fees and make their exchange rates more transparent have reduced U.S./Mexico remittance costs substantially. In 1999, Western Union, for example, charged $20 for transferring $200 to Mexico. In 2001, it charged $15 and many banks and credit unions have even lower charges (Orozco, 2002).

Initiatives targeting the unbanked
The third group of public-and private-sector initiatives focuses on the unbanked households themselves. In broad terms, these initiatives provide consumer financial education, seek to alter consumer spending behaviors, or provide increased incentives for LMI households to build financial savings.
Many non-profit organizations and government agencies sponsor consumer education programs targeting LMI adults (Braustein and Welch, 2002). These sponsors often argue that one of the reasons that large numbers of LMI households use high-cost payment or credit services is because they are not aware of lower-cost alternatives or are not comfortable using the lower-cost alternatives. The educational programs seek to correct these deficiencies. Typical curricula teach adult students to compare, for example, the relative cost of payday loans versus bank loans, the cost of a rent-to-own purchase versus a simple purchase financed with a credit card, and the cost of using a check-cashing outlet versus maintaining a checking account. The courses frequently teach the students how to balance a check book, how to open a bank account, and even take them to visit a bank and practice the skills in order to make them feel comfortable using the banking system. Often, banks help support such educational efforts, and receive CRA credit for doing so.
Although such courses are common, there are no high-quality studies of their effectiveness (Hilgert and Hogarth, 2003). Most reports are based on simple before-and-after studies, meaning that the program sponsor reports how many of the course participants had banks accounts, to cite one example, prior to the course and how many did so afterwards. Standard impact assessments do not examine the persistence over time in the new behaviors. In addition, since these studies lack a control group, it is hard to discern the effect of the education from the self-motivation of the course participants. In any case, the data reviewed above suggest that the impact of such educational efforts on rates of account ownership or use of high-cost lenders is likely to be modest. Most people using check-cashing outlets, payday lenders, etc., are aware that these are more costly alternatives, but they do so because of the convenience of the payment services and because their credit histories and debt burdens prevent them from patronizing lower-cost sources of credit.
A second, and sometimes overlapping, set of initiatives tries to teach LMI households methods that they can use to build savings. Motivating these efforts is the belief that inadequate savings are a major barrier to account ownership, are highly correlated with an inability to pay bills on time, and are a source of substantial personal stress. Educational efforts that try to help LMI households build savings generally involve a detailed analysis of their spending patterns intended to identify unnecessary expenses and a discussion of behavioral "tricks" that many people use to limit their spending. These tricks can include immediately setting aside income from a paycheck into a separate savings fund ("pay yourself first"), never using a credit or debit card for purchases, and limiting the amount of cash that one carries for impulse purchases. As in the case of educational efforts focused on conveying knowledge about alternative financial services, there are no fully satisfactory studies documenting that these behavioral modification efforts significantly affect savings, credit histories, or the use of banks among the LMI households to complete the courses. One quasi-experimental study (Staten et al, 2002) did find that one-on-one credit counseling had a positive and sustained impact on indicators of the recipients' credit risk profiles. Such results are encouraging, because the data reviewed above gives strong support to the notion that a lack of savings and impaired credit histories are the critical barriers preventing many LMI households from lowering the cost of their payment and credit services.
A related third initiative supported by foundations and government agencies has sought to build the savings of LMI households by increasing their incentives to save. Since the late-1990s, several non-profit organizations partnered with banks and credit unions to create "Individual Development Accounts," known as IDAs. IDAs are special savings accounts open only to qualifying LMI individuals. Using funds supplied by a philanthropic foundation or government agency, the non-profit group offered to match the qualifying savings put into the IDA account by an IDA participant. Match rates vary across IDA programs but typically range from $1 for each dollar saved to as high as $3 for each dollar saved. To be eligible for matching, generally the funds must be withdrawn only to pay for education, the down payment on a home, to start a business, or to support oneself in retirement.
Several ideas motivated the creation of IDAs (Sherraden, 1991). One was the observation that most incentives to save in the U.S. come from federal and state income tax exemptions which bring little benefit to LMI households since they face low marginal income tax rates. Another motivation was the observation that people tend to build savings by putting them into forms that are costly to liquidate quickly, such a pension funds or equity in a house. A third motivation was the claim that anti-poverty programs in the U.S. give too much emphasis to peoples' incomes and not enough emphasis to their wealth. Proponents of IDAs argue that ownership of assets makes one more forward-looking and increases the likelihood that one will become civically engaged.
A number of foundations sponsored an evaluation of 13 IDA programs (Schreiner et al, 2002), but the results are inconclusive. Of the LMI individuals who volunteered to participate in IDA programs, 56% saved a net of $100 or more over an average participation period of two years. The average net savings among all participants was $528. Unfortunately, since the report does not include a random assignment study, one cannot separate the incentive effect of the IDA from the self-motivation effect. Presumably, people who are motivated to save sign up for IDAs. In addition, the administrative costs of the IDA programs in the study exceeded the net amount of money that the participants saved (Sherraden, 2000). These costs could undoubtedly be lowered by cutting staffing devoted to the programs but there is no indication that low-cost IDA programs would result in similar levels of saving.
Philanthropic foundations and government agencies have also supported the efforts of some non-profit groups to build the wealth of LMI households through microenterprise development. But, in the U.S., these programs have remained very small scale and have been plagued by very high administrative costs. In addition, most studies find that they have done little to raise participants' incomes. While microenterprise may have had impressive results in lowincome countries, there are serious doubts that it can be effective in the U.S. context (Schreiner and Woller, 2003).

Initiatives to develop new financial institutions or products tailored to the unbanked
The final set of outreach initiatives underway in the U.S. focuses on altering the traditional products or delivery channels for basic financial services to better meet the needs of many LMI households. These initiatives also look to new technologies to lower the costs of delivering traditional payment and credit services to unbanked households. These initiatives are largely private-sector efforts. Some have been modestly subsidized by foundation grants or other means, but many have received no subsidies at all. 11 Several initiatives are based on banks or credit unions offering the traditional services of a check-cashing outlet along with traditional banking services. In some cases, the banks form partnerships with check-cashing outlets to deliver the combined services. There are several rationales for this approach. First is the recognition that many unbanked individuals see little need for banks' savings services, but they do need payment services. Banks can earn additional income by selling the payment services to unbanked individuals. This can help support a branch in a LMI community that might otherwise be uneconomical. Equally importantly, it will bring the unbanked into the branch and make them comfortable interacting with a bank and give the bank an opportunity to market savings products to those who might be interested. If the bank forms a partnership with a check-cashing outlet (CCO) and uses the CCO to deliver consumer banking services, this to can lower its operating costs in the LMI community and provide more convenient banking services to many LMI households.
The best known of the bank/CCO hybrid models is the Cash & Save division of Union Bank of California. This large bank opened 13 non-traditional bank branches displaying the name Cash & Save in a variety of settings where high numbers of LMI households walk or drive by. These branches are small and look more like a CCO than a bank. They offer the full range of CCO services as well as traditional consumer banking services. The branches have been financially successful and Union Bank claims that about 40 percent of its check-cashing customers have begun to use one or more traditional bank products. 12 Another such model that has received substantial attention is the partnership between a check-cashing firm, RiteCheck, in New York City and a credit union, Bethex Federal Credit Union. RiteCheck operates 11 CCOs in the City and Bethex has five full-service branches. Through the partnership, Bethex depositors can obtain all of the traditional payment services at a RiteCheck outlet as well as make deposits or obtain cash withdrawals from their accounts. This enables Bethex to provide basic banking services though the CCOs at a fraction of the cost of maintaining a full-service branch. RiteCheck earns fees from handling the deposits and withdrawals on behalf of Bethex. 13 In addition to the bank/CCO hybrid model, several banks, including some very large banks, have redesigned the account that they suggest unbanked individuals first begin to use. As 11 The Treasury Department in the last year of the Clinton Administration launched the "First Acounts Initiative." Funded with $10 million, it provided financial support to depository institutions that were making innovative efforts to bring the unbanked into the banking system. Under the Bush Administration this initiative stopped, and there were no reports issued on the success that the institutions that were funded had with their efforts. 12 For more information on the Union Bank model, see the report by Richter and Tan (2002). 13 The report by Richter and Tan (2002) has more details on this partnership. discussed above, advocates for the unbanked urged banks to offer basic checking accounts, a low-cost checking account with low minimum balance requirements and a specified minimum number of free checks per month. While this account was suitable for some of the unbanked who wished to open a bank account, it was not suitable for all. For one, many of the unbanked who had previously had a checking account were forced to close those accounts because they overdrew their accounts and did not promptly return the accounts to a positive balance. Banks not only closed their accounts, but they also commonly reported them to "ChexSystem," a network owned by member banks and credit unions that contribute information on mishandled checking and savings accounts to a centralized database. Most other banks will refuse to open checking accounts for people whose names are in ChexSystem. In addition, even among those unbanked individuals who can open a checking account, many may not want to for fear of the fees that they will incur if they overdraw their accounts. Most banks charge $20 to $30 for each "non-sufficient funds" (NSF) check, and the merchants that receive the checks traditionally impose a $10 to $15 returned check charge. For an individual living from paycheck to paycheck, it is easy to overdraw a checking account at the end of a pay period. This can make a "free" checking account quite expensive.
In recent years, some banks have encouraged unbanked individuals who express an interest in opening an account to open transaction accounts without checking privileges. These are low-cost low-minimum-balance accounts. Individuals can withdraw money from their accounts by using an ATM or visiting a teller. They can generally make debit card purchases using the ATM card, but the ATM card is usually an "on-line" card that does not permit the account holder to overdraw the account. The account holder pays bills by purchasing money orders or by making electronic transfers from the account. There is no data yet on the success of these accounts, but they do appear to be well designed to meet the needs of many of the unbanked who wish to open starter accounts.
Banks are also making efforts to create cost-effective means for LMI individuals to obtain emergency loans ranging from $200 to $500. Traditionally banks do not make such small loans except through credit card advances or some other prearranged line of credit, which excludes people unable to pass a traditional credit-risk screening procedure. Many credit unions and banks offer deposit-secured credit cards to customers unable to meet traditional credit risk standards. In addition, many credit unions will make deposit-secured small-value loans. For example, if a depositor has $500 in his account, he can borrow $500 rather than withdraw the money. The deposit balance serves as the collateral. Many customers apparently prefer to borrow the money rather than withdraw it because they like the budget discipline that repaying the loan imposes upon them. Some credit unions have also begun to offer rapidly-disbursing unsecured small-value loans to customers who might not meet traditional credit risk standards. Typically they require that the recipient has used direct deposit for several months and the loans are repaid out of future direct deposits. Early reports from credit unions offering this service indicate that it is popular and profitable for the credit unions (Richter and Tan, 2002).
The development and marketing of automated "payroll cards" is another recent privatesector innovation that could benefit the unbanked. 14 Until recently employers had to pay their unbanked employees using a check, or in some cases cash. In recent years a number of firms have developed payroll cards that enable employers to pay unbanked employees electronically. A payroll card is an ATM-type card that employees can use at an ATM machine to withdraw their pay in cash. In most cases, it can also be used as an on-line debit card. The card is linked to an electronic account that keeps track of the balance available to the cardholder. Although the number of payroll cards issued by mid-2003 was small compared to the unbanked population, the growth rate in the number of cards in use is reportedly high. Whether this rapid growth continues will depend largely on the fees associated with using the card and on people's willingness to change their payment habits. It should be emphasized that individuals who use payroll cards do not have actual bank accounts in which they can make deposits, not can they have face-toface interactions with bank tellers. Nevertheless, payroll cards have the potential to lower the costs of payment services to unbanked households. In addition, over time they could be enriched to permit electronic bill payment and to offer an easy transition to non-checkable deposit accounts.
Finally, just as consumer banking is becoming increasingly automated, there are signs that the same may be true of check-cashing and bill-paying for the unbanked. Several companies have developed and deployed automated check-cashing machines. The large chain of retail convenience stores, 7-Eleven Incorporated, first installed automated check-cashing kiosks in several of its stores in 1998. Since that time, it has worked to refine the technology. Its current automated kiosks, known as "Vcom" kiosks, can cash paychecks, handle money order purchases and money transfers, pay bills through Western Union's "Quick Collect" payment service and other services, and originate orders for many products sold over the internet. As of December 2003, 7-Eleven had placed Vcom kiosks in nearly 1,000 of its approximately 5,800 stores located across the U.S. (7-Eleven news release, December 11, 2003). If this technology succeeds, it could lower the cost of delivering payment services to the unbanked and lower the fees that they pay.
Summarizing the effectiveness of these initiatives As implied in the discussion above, we cannot know the aggregate impact that the outreach initiatives outlined above have had on the percentage of U.S. households without deposit accounts. It is quite likely, however, that the aggregate effect has been quite small. This is true for three reasons. First, almost all of the initiatives have been small scale, pilot initiatives that would have small aggregate impacts even if they were effective. Second, we argued that the major reason that people do not have bank accounts is because they do not have any month-tomonth savings. Assuming that successful methods can be found to help lower-income households build financial savings, the savings are likely to accumulate slowly. Finally, available data do not indicate any major changes in the percentages of unbanked households over the past 25 years. In 1977, the Federal Reserve's Survey of Consumer Finances found that 9.5 percent of all households did not have deposit accounts (Caskey and Peterson, 1994). As noted in the introduction to this paper, the same survey in 2001 found that the percentage was 9.1 percent.

III. The Urban Unbanked in Mexico
This section uses a survey of households in the Metropolitan Area of Mexico City to profile the unbanked and to explain why they are unbanked. In Mexico there has been no systematic national analysis of the extent to which people are inside and outside of the banking system. In 2002, the World Bank commissioned a survey in Mexico City Metropolitan Area. The survey, known as the Encuesta Nacional de Servicios Financieros (ENDSFI) was conducted by Instituto Nacional de Estadística, Geografía, e Informática (INEGI) as an addendum to the Encuesta Nacional de Empleo Urbano. The survey sampled 1,500 households. The surveys were conducted in face-to-face interviews with an 80 percent response rate. For people using the survey data, INEGI supplies a set of weights to convert the sample responses into responses representative of 11.4 million adults living in the Metropolitan Area of Mexico City. All of the tables in this paper are based on the ENDSFI survey and use the weighted data.
One basic weakness of the survey is that it focuses only on the largest city of the country. The population in this city differs from that in most other urban environments and differs dramatically from that in the rural areas, where almost 30 percent of the population lives. Data from Mexico City, for example, could be misleading because it has the highest average income in the country, $14,180 U.S. dollars per capita in 2003 compared to a national average of $5,450. 15 Undoubtedly, due to this difference in average income and other factors, a national survey would find a larger percentage of the population to be unbanked compared to the percentage found in the capital city. In terms of comparison, the Mexican survey drew on the city with highest per capita earnings, while the two cities surveyed in the U.S. were considered below average for that country. Table 12 presents an overview of the socioeconomic characteristics of the survey population. As remarked in the introduction, income per capita in Mexico is about one fourth what it is in the United States. Indeed, the surveys show less than 45% of the U.S. households with annual income below US$15,000 compared with close to 80% in Mexico City. As noted in the table, in Mexico City 23.6 percent of the population reported that they had savings in a bank (21.3% had savings but did not borrow from a bank and 1.8% had savings and credit from a bank). As expected, this implies a much higher percentage of the population in Mexico is unbanked compared to the U.S. While noteworthy, the differences in household income across the two countries do not appear to account for the differences in the coverage of banking services.  Table 13 contrasts the characteristics of the unbanked individuals with the banked. Here a banked person is defined as someone who has a deposit account in a formal-sector deposit institution or who has credit from such an institution. The unbanked thus include persons with "afore" account s-pension accounts opened by their employers on their behalf. This accounts for the approximately 25% of unbanked persons who reported formal savings. As shown in the table, the unbanked population in Mexico is somewhat less educated than the banked population, is less likely to be working, and has a lower household income. These patterns are broadly consistent with those found in the U.S. survey.

Characteristics of the Survey Population % of the survey population
B. Where do the unbanked get financial services?
Individuals who do not use banks still need financial services. If they have any financial savings, they need a means to safeguard their savings. If they receive income in the form of a check, they need a means to convert the check into cash. They need to pay bills. Finally, they many need to borrow periodically. This subsection draws on the survey data to examine where and how the unbanked obtain these financial services. Table 14 presents data on the forms in which the unbanked keep financial savings and contrasts this with the banked. As noted in Table 13 above, 74.9 % of the unbanked said that they had no formal sector financial savings. The survey also asked about the use of informal means of financial savings, such as holding cash, money orders, uncashed checks, jewelry that can be sold, etc. As shown in Table 14, 18 percent of the unbanked respondents said that they use such means; only 1.1 % of the banked did. Allowing for both formal and informal means of financial savings, 56% of the unbanked had no financial savings and 0.7% of the banked had none.  Table 15 examines how people receive incomes and convert their incomes into cash. Among banked individuals, 54% received a deposit into either their savings or checking account, 32 percent received a check, and 11.5% were paid in cash. Among the unbanked, 85.6% were paid by check, 7.8% were paid in cash, and 6.3% received a deposit to a savings or checking account that is controlled by an employer.  In Mexico, unlike in the U.S, there are few practical alternatives for people to cash checks other than to go to a bank. Banks in Mexico commonly cash checks for people who do not have deposit accounts and they do not charge a fee for this service. As shown in Table 15, 80% of the unbanked cash their income checks in a bank, or make a deposit into a relative's or friend's account. Table 16 presents data on the means that people use to pay bills. As indicated in the table, both unbanked and banked individuals pay the overwhelming proportion of their bills in cash. The only notable exceptions are for payments of purchases of appliances and furniture and for purchases from department stores. For such purchases, about 7.8% to 10.3% of the banked use credit cards. As shown in table 17, a large share of the individuals in the survey reported that they had requested credit from a department store. This credit was undoubtedly used to finance purchases. Among banked individuals, 38% requested a loan from a bank. By definition, none of the unbanked reported borrowing from a bank, however almost half of them reported requesting credit from a department store. Significant percentages of banked and unbanked individuals borrowed from the "other" category, which refers to a variety of informal sources. A striking result in table 17 is the low percentage (0.8%) of individuals who reported that they used a money lender. Even more striking is that only banked individuals reported using a money lender as a source of credit. We suspect that this pattern would not hold in a larger survey.  Individuals in the survey who reported that they did not have a bank account were asked why they do not. Most (72%) responded that they had not tried to open an account. As shown in table 18, when these individuals were asked why they had not tried to open an account, 49% replied that they didn't have enough money and another 11% stated that the minimum balance requirement was too high. Although a recent study has demonstrated that banks in Mexico City tend to locate in censal tracts with high income levels and up market commercial activity, are disproportionately represented in low income areas and practically absent from the "informal" neighborhoods 16 , only 3% of the respondents reported that bank location was a barrier. This is similar to what was found the U.S. Although bank offices have a far wider presence in the U.S. In both countries, financial access to banks was perceived as a much larger barrier than physical access. Unlike in the U.S. where a large percentage of the unbanked say that they do not need a bank account because they don't have any savings, only 7% of the unbanked in the Mexico City survey stated that they did not need an account. This difference could be due to the wording of the survey or could reflect the absence in Mexico City of institutions, such as check-cashing outlets, that provide convenient payment services for unbanked individuals. Another difference between the responses in Mexico and the U.S. is that almost 9% of Mexicans in the capital city said that they did not trust the banks, a response that was well under 1% in the U.S. Undoubtedly, this reflects the turbulent history of Mexican banks over the past 20 years.  A small percentage of the unbanked (2.7%) reported that they tried to open a deposit account but were not able to do so. As shown in table 19, two-thirds of them could not do so because they did not have the required minimum initial deposit. Another 22% did not have the required documents or personal references. It is likely that many of these individuals work in the informal sector where written documents are rare and where formalities are often avoided to prevent governmental authorities from tracking business operations. In the survey, 13.5 % of the respondents reported that they had applied for a loan from a bank. As shown in Table 20, of this group 74 % received the credit, 20% were rejected and 6% were still in the approval process at the time of the survey. As indicated in table 21, most of the people whose applications for a bank loan were denied believed that they were denied due to bank concerns about the risks associated with the loan. These risks were caused by a lack of a credit history, insufficient or unsteady earnings, and the lack of collateral or an outside guarantee.  Among those individual who reported that they did not apply for bank credit, 71% said that they did not need any bank loans or that they preferred to finance their purchases out of their income flows (Table 22). Much smaller percentages cited factors that suggested a fear of being rejected. This suggests that many people never or rarely borrow or have other sources of credit so that a lack of access to bank credit does not cause a hardship. Finally, as shown in table 23, of the individuals who reported that they had taken out a bank loan, 65% were still making payments on the loans. The survey did not ask whether or not they had experienced periods of delinquencies with their payments.  There are no formal studies of the costs paid by Mexicans as a result of having such a large percentage of the population outside of the banking system, but there are good reasons to believe that these costs could be significant. They must, for example, pay extra fees to pay their utility bills through a bank. The 83% who are paid by check must pay, direct commissions or through favors and time to cash salary checks. Focus group interviews and cost surveys carried out as part of a World Bank survey of Mexico City's unbanked suggest that households earning below median income (about U.S. $2,500) can spend up to $100 -or 0.5% per year on such extra fees, not counting the time lost -a heavy burden for persons dependent on daily wages. In addition, the lack of a deposit account can make it difficult for households to build assets. The households may lack a safe place to guard any financial savings that they may have. More importantly, one's access to formal-sector credit is linked to the maintenance of a bank account. And credit can be used to acquire such non-financial assets as a home or an automobile, to smooth consumer purchases over time, or to finance small business activities. The cost of credit outside the formal banking system is high, as the data in Box 1, below. Finally, all Mexicans may pay a price as a result of such a large percentage of the population being outside of the banking system, as informal savings are often not channeled to productive activities, lowering the potential rate of growth of the economy. Box 2, on the following page, estimates the costs of being unbanked to a median income couple in Mexico City at 15% of total income. The individuals who are unbanked bear most of these costs. Source : Solo et. al., 2003 E. Bank initiatives to bring the unbanked into the banking system By necessity, much of financial sector policy over the past decade has focused on restoring the financial health of the banking system. In the 1980s and again in 1994-95, the banking system was subject to severe crises that left it deeply undercapitalized and preoccupied with managing troubled portfolios. At the time of this writing, however, the banking system has made substantial progress toward recovery. In addition, it has changed as many weak domestic banks were been acquired by foreign banks. Nevertheless, as shown in table 24, levels of intermediation have continued to fall. Undoubtedly, the weak state of the banking system over much of the past decade constrained its growth and limited its ability to think about new initiatives. In recent years, however, banks have introduced new products that could benefit a significant share of the unbanked. Most importantly, 12 of the 45 commercial banks in Mexico City offer payroll debit cards. These cards enable a worker who does not have a traditional deposit account to be paid electronically. The worker can use the card to withdraw funds from an ATM or to purchase goods at a store that accepts debit cards. The card is linked to an electronic account that keeps track of the remaining balance. The account is not a true bank account, however, since the worker cannot make independent deposits into the account or write checks on the account. Bank began to offer these accounts partly in response to pressures from unions, which demanded the cards as a way to reduce insecurity in payments. Reportedly, payroll debit cards have been quite successful among workers, white and blue collar.

Box 1 Non-bank credit terms are expensive
The second, and even more interesting initiative, came from a non-bank. A large department store chain, Elektra, with a largely working-class clientele, applied to obtain a banking charter, which the authorities granted. Elektra named its bank "Banco Azteca," and located many of its offices within its department stores. The bank structured its products, both deposit accounts and loans, to meet the needs of moderate-and middle-income households. It makes, for example, small-valued unsecured personal loans, although Banco Azteca levies relatively high interest rates on these loans to make them profitable. Banco Azteca appears to be very successful which could encourage other banks to start serving this market. Banco Azteca opened its first office in 2002. By mid-2003, it had over two million deposit accounts and 836 branches. Undoubtedly, part of the basis for such rapid growth was that the bank could market to households that had been credit customers of Elektra. And, by opening small branches within existing department stores, the bank could keep its facilities costs low.
F. Non-bank initiatives to attend the unbanked Some finance companies, institutions that make loans but do not take deposit accounts, have sought to serve lower-income households. Formally, these non-bank financial institutions are known as Sociedades Financieras de Objeto Limitado (SOFOLES). They are also called "specialized" banks since they are exclusively dedicated to one sector (for example: construction, automobile, etc) or activity (for example, credit cards). They finance their assets by selling debt securities or by obtaining financing from other financial institutions. Their main activity consists in granting loans for the acquisition of specific assets such as cars or houses, or issuing credit cards. SOFOLES has become the main source of financing for new auto purchases, making it possible for many people to buy cars who might not be able to obtain bank financing for this purpose. The housing SOFOLES do not generally provide housing financing to low-income families. Their market niche is middle-to high-income households.

G. Popular banks and financial services for lower-income households
In addition to traditional commercial banks and finance companies, the Mexican financial system includes a variety of semi-formal and informal small-scale "popular" banks. Traditionally, these popular banks served moderate-income households, providing both credit and deposit facilities. They were largely unregulated, and many were severely undercapitalized. Most of the popular banks are organized as cooperatives.
In the 1990s, the government recognized that the popular banks could play an important role in serving lower-income households, but it also wanted to limit the risk that these banks created for their depositors and for the financial system. Consequently, the government initiated a series of measures intended to clarify their rules of operations and to bring them into regulatory oversight. Its first step was to broaden financial legislation to bring in the Cajas de Ahorro Popular; transforming them into regulated Sociedades de Ahorro y Prestámo (savings and loan associations, SAP law 1991). Many savings and credit cooperatives remain ouside that legal framework and promote a new law in 1994 which further segments the sector. The process culminated in 2001, when the Congress enacted the "Ley de Ahorro y Crédito Popular" designed to encompass the entire sector and bring it under the supervision of the Comisión Nacional Bancaria y de Valores (CNBV); simultaneously the old National Savings Board (Patronato del Ahorro Nacional) 17 was transformed into "Banco del Ahorro Nacional y Servicios Financieros" (BANSEFI) in order to coordinate the reorganization of all popular financial intermediaries and to serve as development agency for these institutions by implementing (non-credit) support programs intended to ensure that they follow sound banking practices. The main goal of the new legislation is to lend credibility to the popular institutions and to promote healthy financial intermediation among the low-and middle-income groups.
BANSEFI is a descendent of the "Patronato del Ahorro Nacional", commonly identified by its main product the "Bonos del Ahorro Nacional" which collected savings but did not make private-sector loans. BANSEFI continues to provide this savings service, but as noted above it also functions as a development bank, or "second floor" bank, to assist and support programs that could help popular banks to increase their income, reduce costs, improve their management, and broaden the services that popular intermediaries could provide to their members. One of the first tasks of BANSEFI was to develop a database on all popular banking institutions. For this purpose in 2002 BANSEFI conducted a census of the entities (Entidades de Ahorro y Crédito Popular), 18 including those financial institutions that were not recognized officially by the financial authorities. Table 26 presents an overview of the results from this census. As shown in the table, the cooperatives with the largest membership are the savings and loan associations (Sociedades de Ahorro y Préstamo), which are formally regulated, and the non-regulated savings and loan cooperatives (Cooperativas de Ahorro y Prestámo). The third largest membership is that of the Cajas Solidarias. These were originally rural savings institutions which were instituted in a number of communities using funds provided by the government and technical support from Développement Iternational Desjardins of Canada. The average deposit in a Cajas Solidarias is around U.S. $700 and the average loan is about $600. Credit unions have larger average deposits of about $2,000 and average loans of about $4,000. Recent estimates of BANSEFI show that in 2003, membership in popular banking institutions reached 2.9 million persons. If we add the direct clients of BANSEFI, a grand total of 5,378,094 persons benefit from the services of the "Popular Banking Network" or Red de la Gente. BANSEFI's direct clients had an average deposit balance in September 2003 of $120 U.S. dollars. 19 As noted earlier, BANSEFI does not make loans to its direct clients. The average deposit balance suggests that BANSEFI reaches a population with lower average incomes than that of Popular Banking Network.
One of the reasons that BANSEFI serves households with lower levels of income than those served by popular banks is its diversity of financial products. It has, for example, developed a deposit account, similar to the standard account (Tandahorro) at a popular bank that allows small savers to set up a medium-term savings goal with a specified set of periodic deposits to meet that goal. But the account at BANSEFI is guaranteed a yield above inflation. It has also developed an account for housing linked to the government-housing program "INFONAVIT" as well as savings accounts for children and a savings account that offers a chance to win a lottery. Another element that has helped BANSEFI's success has been the location of its branch offices, which are commonly located to be convenient to the residents of low-income communities. It could be argued that Popular Banking under the coordination of BANSEFI has become the cornerstone of the institutional framework for the unbanked.

H. Microfinance companies
A number of microfinance companies were launched in Mexico in the 1980s, but they still have only a small role in the overall financial system. The companies received their funds from non-governmental organizations, and usually distributed loans through solidarity groups of about 5 people that the borrowers are required to form, à la Grameen Bank in Bangladesh. Some microfinance organizations require or encourage their members to make small deposits with the organizations. They do so in an effort to promote a savings culture.
In 2001, the federal government established a program (Programa Nacional de Financiamiento al Microempresario, PRONAFIM) to support microfinance companies. Since that time, 37 companies have been accepted into the program and have been granted resources to increase the size of their operations. The size of the loan ranges from U.S. $27 up to $454 per person in a solidarity group.

I. Interactions among outreach initiatives
Although there are a variety of initiatives to increase the access that lower-income households have to formal-sector and semi-formal sector financial services, one weakness of these efforts is a lack of coordination and interaction. One exception should be noted. A microfinance company (CAME) and a bank (BANAMEX) teamed up to issue debit cards to members of CAME's solidarity groups in order to reduce the members' transaction costs. This sort of interaction is promising, as it builds on the technological sophistication of the large banks and the grassroots presence of informal institutions. If such models were to become more common, they could greatly strengthen outreach efforts.

IV. Assessing of the Outreach Efforts Underway in the U.S. and Mexico
In this final section, we briefly review the rationale and challenges facing the strategies that both countries are employing to improve the financial services available to lower-income households. In doing so, we do not mean to imply that government agencies and private sector institutions in either country share an overall strategic vision. Often, in fact, efforts to improve financial services for lower-income households are piecemeal and uncoordinated. Nevertheless, the combination of these efforts provides a general strategic outline, even if no organization has enunciated the components and rational for that outline.
In the case of the U.S., the major efforts have two goals: first, to reduce the cost of financial services for households that continue to live from paycheck to paycheck and, second, to help more of these households build financial savings. As discussed earlier, efforts to reduce the cost of financial services for households living without financial savings include state laws that limit the fees that non-bank financial institutions, such as check-cashing outlets or small-loan firms, can charge for payment or credit services. They also include the use of technology, such as payroll cards and automated check-cashing machines, that can lower the cost of providing a basic financial service. The measures are undoubtedly positive, but in our view they are likely to have a modest impact. For one, the evidence indicates that state laws that simply set ceilings on the fees that non-bank financial institutions can charge for financial services can only lower those fees a small degree before they adversely impact the willingness of those institutions to provide the services. In addition, the cost savings from the use of technology in payments have been modest as a percentage of the incomes of even poor households and technology has done little to lower the cost of high-risk lending. Such lending remains labor intensive and, given the small value of the advances, this results in high interest rates. Technological advances, such as the computers that manage credit cards, have lowered credit costs for low-and moderate-risk borrowers, have done much less for labor-intensive high-risk credit.
The one exception to our claim that gains from lowering the cost of payment services have been modest is international remittances. As noted earlier, fees for remittances, especially for transfers between the U.S. and Mexico, have declined substantially. But much of the gain here came from making what had been a nearly monopolistic and opaque wire transfer service much more transparent and competitive.
In recent years, there has been an increasing focus on helping lower-income U.S. households to build financial assets. A variety of factors have led to this shift in emphasis. First, numerous studies have found that a lack of financial savings is the major barrier preventing lowerincome households from maintaining bank deposits. Second, households without savings have no financial margin of safety, so any adverse financial shock can cause them to miss payment obligations, placing them in a high credit risk profile. Third, some researchers argue that the accumulation of savings will affect the psychological orientation of these households. They will become more able to plan for the future and feel more in control of their lives. More generally, there is broad recognition that the possession of savings makes one less vulnerable to financial shocks and reduces the stress in one's life.
As noted earlier, asset-building initiatives take a variety of forms. One set of initiatives attempts to make deposit accounts financially and culturally accessible to a broader set of individuals. Accordingly, banks have begun to promote basic savings accounts accessible by ATM and on-line debit cards rather than basic checking accounts. Since the savings accounts cannot be overdrawn, banks can offer them to almost anyone with little risk. In addition, banks have formed partnerships with check-cashing outlets to enable the banks to deliver deposit services through the outlets in which many of the unbanked may feel more comfortable. The motivating idea is that financial and cultural access to a deposit account may help many individuals to build savings, which will ultimately give them access to a broader array of lower-cost financial services. As an additional inducement to people to build savings, some financial institutions promote small value time deposit accounts in which individuals pledge to make regularly scheduled deposits similar to the "Christmas Club" accounts that were common decades ago.
Financial education and Individual Development Accounts (IDAs) are a second set of initiatives intended to promote asset accumulation by lower-income households. In the case of IDAs the challenge is to demonstrate that they can be cost effective. As noted earlier, the IDAs that have been studied most carefully have had very high operating costs relative to the savings their participants generated. In the case of financial education, there is also a pressing need to validate with empirical studies the notion that low-cost educational efforts will actually help lower-income households to build savings or improve credit histories. Finally, there is also no rigorous evidence that supports the view that simply providing the unbanked with financial and cultural access to a deposit account will help them build savings. Research on all of these topics would be helpful.
Although much attention has gone in recent years to the unbanked in the U.S., the problem is far more important in Mexico where a minority of the population uses the formal financial system. As noted earlier, this may not only be costly to those outside of the banking system, but could even slow the growth rate of the Mexican economy. An underdeveloped formal financial system could impede the channeling of saving to productive investment opportunities.
In the case of Mexico, there are three likely reasons that the problem is so much larger in scale. First, as discussed earlier, incomes are much lower in Mexico so a much higher percentage of the population lives without financial savings. Second, a large segment of the population works in the informal sector and lacks the documentation and employer or union structures that would give them access to the banking system. Third, the banking system in Mexico is highly concentrated. Perhaps as a result of the lack of competition (table 27), the required minimum balance to avoid fees is much greater relative to incomes in Mexico than in the U.S (table 28). This observation undoubtedly contributes to the high percentage of the unbanked who cite minimum balance requirements as the barrier to opening a banks account.  moderate-income households. A simple way to begin is to promote the use of payroll cards as a way to make people comfortable with non-cash means of making payments and maintaining short-term savings. In addition, the government might consider implementing a variation on the Community Reinvestment Act (CRA), the U.S. legislation that puts mild pressure on banks to serve moderate-income communities. The entry into the Mexican banking system of Banco Azteca, with it focus on middle-and moderate-income households, is also a positive development. The government should encourage other investors to compete with Banco Azteca in this market, while ensuring that they maintain sound banking controls and balance sheets. Finally, the effort to strengthen the oversight and financial stability of the popular banks, while also aiding their outreach, could bring many benefits.
The growth and stabilization of the popular banking sector can bring two major benefits to the Mexican financial system. First, it provides more competition for the large commercial banks which could lower their fees. Second, the popular banks, which mainly serve moderateincome consumers and small businesses, can operate much more informally than traditional banks. In simple terms, they do not need lobbies made of marble. This means that they can be financially and culturally accessible to moderate-income households. If their offices have a similar appearance to other businesses or non-profit organizations that serve moderate-income households, then their target clientele will feel comfortable in this setting. In addition, by having a secure but informal setting, their operating costs will be lower which will allow them to serve moderate-income communities profitably. While we support the growth of popular banks, we do not do so at the cost of the stability of the financial system. But we do not believe that this tradeoff is necessary. The U.S. experience with community development banking has demonstrated the stability and profitability of "popular banking." It has also proved a source of investment for traditional commercial banks.
As in the U.S., future Mexican policy would benefit from more research. The ENDESFI survey was invaluable in providing a baseline survey of the use of financial services by households in Mexico City. Additional research should develop this further and also focus on the business models of financial institutions, both banks and non-banks that currently serve moderate-income households. This will provide policy makers and private sector institutions a better understanding of how moderate-income households obtain financial services, what they pay form then, and what might be done to lower the cost or improve the quality of these services.