Financial Inclusion and Financial Stability

Financial Inclusion and Financial Stability Rochelle M. Edge Office of Financial Stability Federal Reserve Board October 17, 2012 The views expressed here are my own and do not necessarily reflect the views of the Board of Governors or the staff of the Federal Reserve System. 1

Initial thoughts The financial sector provides key services to economic agents. The inability to access these services prevents production and spending from taking place and reduces economic welfare. In time-series data, periods in which significant segments of the economy are unable to access financial services due to the sector s not functioning properly are episodes of financial instability. Financial stability policies promote the prudent and consistent provision of financial services over time. In the cross section, individuals and households that are unable to access financial services typically, those with lower incomes are financially excluded (or must rely on less traditional sources). Financial inclusion policies promote the prudent and consistent provision of financial services over the cross section. 2

Initial thoughts, continued This leads to two questions. Can policies that promote financial inclusion compromise financial stability? If yes, do policies that promote financial inclusion necessarily compromise financial stability? To answer these questions, we must examine: What makes the financial system vulnerable to shocks that could compromise financial stability; and, Whether financial inclusion policies can lead to the emergence or escalation of such vulnerabilities. 3

Initial thoughts, continued Characteristics that make the financial system vulnerable to potentially destabilizing shocks include: The size, complexity, and interconnectedness of financial institutions; The leverage of financial institutions and the financial system; The maturity mismatch of financial institutions and the financial system and the stability of funding sources; and, The financial system s tendency toward procyclicality, as reflected in: Procyclical leverage and maturity mismatch; Procyclical credit growth and asset valuations; and, Deteriorating credit underwriting standards in buoyant times. 4

Approach Take a somewhat U.S.-centered perspective Describe U.S. financial inclusion policies and initiatives Consider whether these policies could potentially compromise financial stability Consider whether any of these policies have compromised financial stability in the recent past; that is, contributed to the financial crisis (as has been claimed) Consider possible financial stability policy responses to increased risks arising from financial inclusion policies. 5

U.S. financial inclusion policies and initiatives Key services provided by the financial sector include transactions, credit-intermediation, and risk-management services. U.S. public-sector initiatives on expanding the provision of transactions services to unbanked and underbanked households include: The U.S. Federal Deposit Insurance Corporation s Model Safe Accounts Pilot Program; and, The U.S. National League of Cities Bank On initiative. U.S. policies to expand access to credit to low- and moderateincome borrowers include: The Community Reinvestment Act (CRA); The affordable housing goals (AHGs) imposed on U.S. government-sponsored enterprises (GSEs); and, Programs operated by the Community Development Financial Institutions (CDFI) Fund at the U.S. Treasury. 6

Unbanked and underbanked households In the 2010 FRB SCF, 7.5% of U.S. households were unbanked. In the lowest income quintile, 23.8% of households were unbanked. In the 2011 FDIC Survey, 8.2% of U.S. households were unbanked and 20.1% of U.S. households were underbanked. 7

Services to the unbanked and financial stability The provision of safe, low-cost transactions and savings accounts to underserved consumers likely has limited implications for stability. Bank deposits of underserved customers are likely to be insured and represent a stable form of bank funding. This makes sudden withdrawals of this form of funding unlikely. Bank deposits of underserved customers are unlikely to represent a sizable fraction of overall bank deposits or liabilities. Thus, even if a sizable fraction of these deposits were suddenly withdrawn, they would probably not result in large funding or liquidity strains on banks. Also, effects on bank profits are not likely to be material enough to alter bank risk-taking incentives. 8

Expanded credit access and financial stability Expanded credit access to low- and moderate-income (LMI) borrowers has greater potential to compromise financial stability. Expanded credit access to LMI borrowers could result in: Laxer underwriting standards on loans to LMI borrowers; Excess credit growth that leads to an escalation in home prices and/or over-indebtedness for LMI borrowers; or, Reduced bank profits due to enhanced but costly underwriting. Whether U.S. financial inclusion policies such as the CRA and GSE AHGs caused the financial crisis is currently a topic of debate. In his dissent from the Majority Financial Crisis Inquiry Commission (FCIC) Report, Wallison (2011) argued for this explanation. The debate has initiated a large research literature. 9

Community Reinvestment Act The CRA encourages banks and thrifts to meet the credit needs of their entire community consistent with safety and soundness. CRA compliance is evaluated based on lending in assessment areas. Residential mortgages play a key role in evaluating CRA compliance due to the availability of loan-level data on originations and purchases (the HMDA data). Since the mid-1990s, CRA compliance has used numerical measures. These include the share of loans originated/purchased in low- and moderate-income (LMI) census tracts or made to LMI borrowers. A LMI census tract has a median family income less than 80% of the median family income of the surrounding area. A LMI borrower has an income less than 80% of the median family income of the surrounding area. HMDA data are typically used for assessment area calculations. 10

GSE Affordable Housing Goals GSE AHGs were imposed by Congress on Freddie Mac and Fannie Mae as part of the 1992 GSE Act. The goals establish annual percentage business requirements in terms of GSEs purchases of mortgages falling into three categories: Loans to LMI borrowers; Loans to underserved areas; and, Loans to special affordable populations. LMI borrowers (for GSE purposes) in urban areas have incomes below the median family income of their MSA. Underserved areas in urban areas are census tracts with Median family incomes less than 90% that of its MSA; or, Median family incomes up to 120% of the MSA median and more than 30% minority population. 11

GSE Affordable Housing Goals, continued Special affordable populations in urban areas are borrowers that have incomes: Below 60% of their MSA s median family income; or, Below 80% of their MSA s median and who live in tracts with median family incomes less than 80% of the MSA median. Numerical targets for GSE lending are set in advance each year by the GSE regulator. A mortgage purchase could count toward more than one goal. Non-conforming or jumbo loans, subprime loans, and governmentbacked loans (i.e., FHA and VA loans) are not generally considered. 12

GSE Affordable Housing Goals, continued The target ratios for all three goals have been rising over time. This increase and mid- 1990s changes in CRA evaluation compliance methods raises the question of whether financial inclusion policies caused the housing bubble and subsequent crisis. 13

Wallison s dissent from the FCIC report Wallison issued a dissent from the majority report arguing that: the U.S. government s housing policies were the major contributor to the financial crisis of 2008. These policies fostered a massive housing bubble and the creation of $27 million of subprime and Alt-A loans, which were ready to default as soon as the bubble began to deflate. Deregulation, lack of regulation, predatory lending or other factors cited in the report were not determinative factors. Many responses to Wallison s dissent have raised concerns about the statistics on which he based his conclusions (see Min, 2011). His subprime and Alt-A loan definitions more than quintuple the volume of these loans relative to standard (e.g., GAO) estimates. Researchers have also noted that time-series co-movements do not establish causality and have moved to using cross-sectional data. 14

Policy analysis on govt. policy & the housing crisis Bhutta and Canner (2009) document two facts that run counter to the claim that CRA contributed significantly to the housing crisis. Only a small portion of subprime mortgages originated in 2005-06 were related to CRA. Only 6 percent of subprime mortgage in 2005-06 were extended by CRA-associated lenders to LMI borrowers or neighborhoods. Although CRA-associated lenders could have purchased loans from other lenders, the link between these lenders is weak, especially for high-priced loans. CRA-related loans perform comparably to other types of subprime loans. 15

Research on government policy & the housing crisis Research on this topic has considered how government policies have affected targeted areas Borrower delinquency rates; Underwriting standards; House price appreciation and subsequent depreciation; and, Lending activity. Identification of the effects of policies relies on: CRA applying only to banks and thrifts (not credit unions or independent mortgage companies) and the persistence over time of institution-types lending shares in geographic areas; and, Discrete cutoffs for targeted versus non-targeted areas. Analysis is indirect but is a significant improvement on examining time-series co-movements. 16

Government policy & housing outcomes Avery & Brevoort (2011): If CRA policies caused the crisis, LMI census tracts with more CRA-associated lending and loan purchases should have higher delinquency rates, more marginal (i.e., high PTI ratio) loans, and larger house price appreciations and depreciations. This approach relies on the fact that over time the share of loans in a tract accounted for by different institution types changes little. Avery & Brevoort: If GSE AHGs caused the crisis, LMI census tracts with more loans sold to GSEs should have worse outcomes. Avery & Brevoort find that tracts with more CRA institution withinassessment-area lending and more GSE sales have better housing outcomes. (Although, the GSE effect is insignificant.) Evidence for loan purchases is mixed but wherever a greater share implies worse housing outcomes the effect is small. 17

Government policy & housing outcomes, continued Avery & Brevoort also compare outcomes for CRA- and GSE-eligible versus non-eligible census tracts. Tracts with lower relative incomes have (in 2008) higher delinquency rates and (over 2004-06) a greater share of loans with high PTI ratios. Loan growth between 2001-03 and 2004-06 was more rapid in tracts with lower relative incomes. There is little evidence (visually or through regression discontinuity analysis) that this relationship is influenced by policy thresholds. Relative tract income 18

Government policy & housing outcomes, continued The share of loans sold to GSEs and the share of CRA-covered lenders lending in their assessment area increases with relative tract income. If GSE and CRA policies were the cause of faster loan growth in tracts with lower relative incomes we would not find these results. Note also that there is no discontinuity at the policy thresholds. Loans purchased by CRE-covered lenders is the only variable with a discontinuity but this does not appear to have affected loan quality. Relative tract income 19

Government policy & housing outcomes, continued Bhutta (2010) and Bolotnyy (2012) examine the effects of one of the GSE AHGs the underserved areas goal (UAG) on GSE mortgage purchases in lower-income neighborhoods. Both authors use regression discontinuity analysis, with the relevant discontinuity being for census tracts with a median family income less than 90% that of its MSA. Both authors use the fact that some census tracts gained or lost UAG eligibility between 2005 and 2006 (due to the switch to using the 2000 Decennial Census to determine eligibility). Bhutta uses Home Mortgage Disclosure Act (HMDA) data, which allows him to consider loan volumes as well. Bolotnyy uses HUD s GSE Public Use database, which includes GSE purchases of seasoned loans (not captured in the HMDA data). 20

Government policy & housing outcomes, continued Using regression discontinuity analysis, Bhutta and Bolotnyy both find that the UAG boosts GSE purchases modestly. Bhutta estimates the increase to be 3 to 4 percent. Bolotnyy finds the increase to be 0 to 3 percent (but insignificant). Using the fact that some census tracts gained or lost UAG eligibility between 2005 and 2006, Bhutta and Bolotnyy both find somewhat larger effects of the UAG on GSE purchases. Bhutta estimates the increase to be 6 percent for tracts whose relative incomes fell by just enough to be UAG eligible, although the increase dissipates the larger the fall in relative income. Bolotnyy finds that the increase is 5 percent and dissipates much more rapidly with larger declines in relative incomes. These magnitudes do not seem large enough to imply that the UAG led to a large increase in credit supply to lower-income households. 21

Responding to financial stability risks This cross-sectional research suggests that policies to expand credit to low- and moderate-income borrowers did not drive the housing bubble and subsequent crisis. Nonetheless, hypothetically, these types of financial inclusion policies could still potentially jeopardize financial stability. Putting financial stability policies into practice entails: Monitoring the build-up of risks in the financial and non-financial sectors that could result from financial inclusion policies and that could jeopardize financial stability; Implementing financial stability policies to stem the build-up of these risks; and, Establishing a governance structure for implementing financial stability policies given financial inclusion objectives. 22

Responding to financial stability risks: Monitoring Since financial inclusion policies can adversely affect underwriting standards, credit growth, and indebtedness among LMI borrowers, these variables should be monitored. Variables should also be monitored relative to non-lmi borrowers. Analysis should be undertaken to ensure that financial inclusion policies are not driving these developments. FR System staff undertake a quarterly assessment of financial system vulnerabilities, which includes an analysis of the financial condition of the household sector. Specifically, the staff monitor: Credit growth to households by credit score; and, Lending standards on consumer loans and residential mortgages. FR Board staff undertake a detailed review of HMDA data each year. Higher-priced loans are among the variables reviewed. 23

Responding to financial stability risks: Policies Note: None of these tools are readily available to U.S. bank regulators. If policies to expand credit to targeted borrowers lead lenders to weaken underwriting standards rather than to undertake more costly underwriting to identify less-obviously creditworthy applicants Higher capital risk weights could be applied to loans with weaker underwriting standards, since this would make more-costly underwriting a more-attractive option. If policies to expand credit to targeted borrowers lead to excessive credit growth or indebtedness, it could be appropriate to review targets to ensure they can be met prudently. Fine-tuning targets to expand credit to targeted borrowers while not causing over-indebtedness can be challenging. Caps on borrower DTI or PTI ratios could be imposed on loans to prevent over-indebtedness. 24

Responding to financial stability risks: Policies, ctd. Policymakers with financial stability responsibilities are currently assessing how well different policy tools can address risks like declining underwriting standards and excess credit growth. Fixed and time-varying caps on LTV and DTI ratios have been used by several East Asian and Scandinavian countries to address housing bubbles. Fewer countries have used variable capital risk weights. LTV and DTI ratio caps are more forceful policies, as they limit what a household or firm can borrow. They are also more politically sensitive, especially in countries with less interventionist governments. The Bank of England s (interim) Financial Policy Committee elected not to request these tools and decided that more public debate was needed before doing so. 25

Community Development Financial Institutions Community Development Financial Institutions (CDFIs) are missiondriven lenders that provide financial services to markets and borrowers that are underserved by traditional financial institutions. CDFIs can take several institutional forms BHCs, banks, thrifts, credit unions, loan funds, and venture-capital funds. U.S. Treasury s CDFI Fund certifies CDFIs based on seven criteria. The capital and financing of CDFIs varies by institutional form. CDFIs obtain funds mainly from banks, foundations, socially responsible investors, and public sources. Funds can take the form of equity, loans, deposits, and grants. Banks can meet CRA goals by loans and investments in CDFIs. CDFIs obtain funds at market and below-market rates, which enables them to provide lower rates than traditional lenders. 26

Community Development Financial Institutions, ctd. The majority of CDFIs are unregulated and as a result publicly disclosed and externally verified data are lacking. There is strong desire by both private and philanthropic investors for greater transparency and better metrics for evaluating the social return on investments. The CDFI industry lacks an industry-wide regulator. CDFIs that are depository institutions are regulated by their respective regulators. The CDFI Fund ensures CDFIs are meeting their objectives but does not offer examinations, supervision, etc. A stronger regulatory infrastructure would enhance transparency through consistent reporting obligations. 27

CDFIs and financial stability Lack of data limits somewhat our ability to assess fully the financial stability issues that surround CDFIs. CDFIs funding sources can be variable (though not prone to flight). Philanthropic funding declines when endowments face losses. S&L government funding is affected by fiscal strains. Funding that receives investor tax credits loses value in recessions. To improve the reliability of CDFIs funding sources, HERA 2008 allows CDFIs to apply for membership in Federal Home Loan Banks. Although CDFIs are a small part of the overall financial system, they are important for many low- and moderate-income communities. Individually, CDFIs are also small and localized lenders. If a CDFI fails, other lenders will likely not be able to replace the credit the CDFI provides and so locally the impact could be severe. 28

Summing up Although research for the U.S. suggests that financial inclusion policies did not drive the housing bubble and subsequent crisis, these types of policies can potentially jeopardize financial stability. In formulating financial inclusion policies an assessment of how financial stability risks could arise needs to be undertaken. Given these channels, regular monitoring should be performed. For example, since financial inclusion policies can potentially weaken loan under-writing and lead to excessive credit growth, these variables should be regularly monitored. Policies to address any observed risks should be implemented. Figuring out the toolbox of policies for addressing emerging risks is the first step in this process. Time-varying capital risk weights or caps on borrower debt-toincome ratios or on loan-to-value ratios may be possible policies. 29

References 1. Avery, R. B., and K. P. Brevoort (2011). The subprime crisis: Is government housing policy to blame? FEDS Working Paper Series, No. 2011-36. 2. Bernanke, B. (2009). Speech at the Global Financial Literacy Summit Community Development Financial Institutions. June 2009. 3. Bhutta, N. (2010). GSE activity and mortgage supply to lower-income and minority neighborhoods: The effect of the affordable housing goals. Journal of Real Estate Finance and Economics, Vol. 45, pages 238-61. 4. Bhutta, N., and G.B. Canner (2009). Did the CRA cause the mortgage market meltdown? Community Divided, Federal Reserve Bank of Minneapolis, March 2009. 5. Bolotnyy, V. (2012). The GSEs and the mortgage crisis: The role of the affordable housing goals. FEDS Working Paper Series, No. 2012-25. 6. Bricker, J., A.B. Kennickell, K.B. Moore, and J. Sabelhaus (2012). Changes in U.S. Family Finances from 2007 to 2010: Evidence from the Survey of Consumer. Federal Reserve Bulletin, Federal Reserve Board, June 2012. 7. Corporation for Enterprise Development (2011). What do we know about the unbanked. Research brief, March 2011. 8. Federal Deposit Insurance Commission (2012). 2011 FDIC National Survey of Unbanked and Underbanked Households. September 2012. 9. King, M. (2012). Speech at the FRB Conference Central Banking: Before, During and After the Crisis. March 24, 2012. 10. Min, D. (2011). Faulty conclusions based on shoddy foundations. Center for American Progress, February 2011. 11. Wallison, P.J. (2011). Dissent from the majority report of the financial crisis inquiry commission. American Enterprise Institute, January 2011. 12. Weicher, J. C. (2010). The affordable housing goals, home ownership, and risk: Some lessons from past efforts to regulate GSEs. Mimeo, Hudson Institute. 30