محدود کردن دسترسی به اعتبار مصرف کننده : شواهد بررسی خانوار بر اثرات محدوده نرخ Oregon

Many policymakers and some theories hold that restricting access to expensive credit helps consumers by preventing overborrowing. I examine some effects of restricting access, using household panel survey data on payday loan users collected around the introduction of binding restrictions on payday loan terms in Oregon. Borrowing fell in Oregon relative to Washington, with former payday borrowers shifting partially into plausibly inferior substitutes: bank overdrafts and late bill payment. Additional evidence suggests that restricting access caused deterioration in the overall financial condition of Oregon households. Overall the results are consistent with restricted access harming, not helping, consumers on average.

Expanding access to credit is a key ingredient of financial development strategies worldwide. The Small Business Administration and comparable small and medium-enterprise (SME) initiatives target billions of dollars of commercial credit in developed economies. The microcredit industry targets billions of dollars of commercial credit in developing economies. A widely shared presumption of these efforts is that expanding access to “productive” credit makes entrepreneurs and small business owners (weakly) better off. There is less consensus on whether access to consumer credit does borrowers more good than harm. Market forces have spurred dramatic growth in subprime nonmortgage consumer credit in the US; as others have noted, there are now more outlets offering small, two-week “payday loans” at 400% APR than McDonald’s and Starbucks outlets combined. 1 Revealed preference logic says that this growth should be welfare-improving: a consumer borrows only if she will benefit (weakly, in expectation). In contrast a growing body of work on psychological biases in household finance suggests that many consumers overborrow relative to an unbiased benchmark. 2 This work can motivate restricting access. Indeed, policymakers often raise concerns about “unproductive” lending at “usurious” rates in subprime markets. Usury laws have existed for millennia.3 At least 13 states currently have binding restrictions on payday loan terms. New Hampshire and Ohio enacted their restrictions in 2008, and several more states are considering legislation that would restrict access in this $40 billion market. A 36% APR federal interest rate cap on loans to military households took effect in 2007, and President Barack Obama seeks to “Cap Outlandish Interest Rates on Payday Loans” by extending that cap to all Americans.4 A growing empirical literature on the effects of access to expensive credit on borrowers has added fuel to this debate. Several studies find that access to expensive credit exacerbates financial distress (Campbell et al., 2008, Carrell and Zinman, 2008, Skiba and Tobacman, 2008a and Melzer, 2009). These findings suggest that psychological biases lead consumers to do themselves more harm than good when handling expensive liquidity, and hence that restricting access will help consumers by preventing overborrowing. But several other studies suggest otherwise. They find that, on average, access to expensive consumer loans helps borrowers make productive investments, broadly defined: smoothing negative expenditure shocks (Wilson et al., 2008 and Morse, 2009), preventing negative income shocks (Karlan and Zinman, forthcoming), or otherwise managing liquidity to alleviate financial distress (Morgan and Strain, 2008).5 These findings suggest that restricting access will harm borrowers by preventing them from financing valuable consumption smoothing and investment opportunities (e.g., in job retention).6 I add to this literature by examining the effects of restricting access to expensive consumer credit, using household survey data collected around new binding restrictions imposed by the state of Oregon in 2007 (the “Cap”, below).7 The neighboring state of Washington considered enacting similar restrictions but did not. Before- and after-Cap panel data, on a sample of Oregon and Washington respondents who were payday borrowers before-Cap, allow for difference-in-differences (DD) estimates of the effects of the Cap (and of access to expensive credit more generally) on borrower choices and outcomes. The data provide two key advantages over comparable studies on the effects of access to subprime credit in the US. First, it measures usage of several different types of expensive loan products, permitting analysis of substitution (or complementarity) between payday loans and other liabilities. Second, it permits construction of a summary measure of financial condition based on a combination of an objective measure (employment status), and two subjective respondent assessments of their financial condition over the last 6 months, and of their expected trend for the future.8 Employment status is a useful proxy for (financial) well-being here because unemployment is likely to be involuntary in this sample; subjects are relatively poor and credit constrained, and they have some recent attachment to the workforce (they are all recent payday loan users, and getting a payday loan requires a documented steady job). The subjective assessments help address the issue that financial condition may be difficult to infer from objective choices and outcomes without a complete accounting of the intertemporal optimization problem or strong related assumptions.9 The data and methodology have some disadvantages as well. Although the sample does seem to be representative of payday borrowers in most respects, my sample is considerably older than average, raising a question of external validity. With regards to internal validity, several issues complicate the DD estimation. Dissimilarities across treatment (Oregon) and control (Washington) groups in baseline characteristics and attrition motivate matching and weighting estimators (along with the standard simple means comparisons). The short-run follow-up period (5 months), and trend in lender exit from Oregon, motivate attempts to identify Oregon respondents who were most affected (i.e., most likely rationed) by the Cap. Overall, however, the results are robust to various DD estimation strategies. I find that the Cap dramatically reduced access to payday loans in Oregon, and that former payday borrowers responded by shifting into incomplete and plausibly inferior substitutes. Most substitution seems to occur through checking account overdrafts of various types and/or late bills (as in Morgan and Strain, 2008). These alternative sources of liquidity can be quite costly in both direct terms (overdraft and late fees) and indirect terms (eventual loss of checking account, criminal charges, utility shutoff). Under the broadest measure of liquidity in the data, the likelihood of any expensive short-term borrowing fell by 7–9 percentage points in Oregon relative to Washington following the Cap. This jibes with respondent perceptions, elicited in the baseline survey, that close substitutes for payday loans are lacking. Next I examine the effects of the Cap on the summary measures of financial condition that are available in the data: employment status, and respondents’ qualitative assessments of recent and future financial situations. Estimates on individual outcomes are noisy but consistent with large declines in financial condition. Estimates on a summary measure of any adverse outcome—being unemployed, experiencing a recent decline in financial condition, or expecting a future decline in financial condition—suggest large and significant deterioration in the financial condition of Oregon respondents relative to their Washington counterparts.10 As such the results suggest that restricting access harmed Oregon respondents, at least over the short-term, by hindering productive consumption smoothing and/or investment (e.g., in job retention). The paper proceeds with a brief overview of the payday loan market. Section 3 then details the Oregon policy change and subsequent lender exit. Section 4 describes the sample frame and survey data. Section 5 details my approaches to estimating treatment effects and related threats to identification. Section 6 presents the main results: estimates of the five-month impacts of the Cap on credit access, credit use, and financial condition. Section 7 discusses how and why longer-run impacts might differ, and presents results using predicted-rationed Oregon respondents as the treatment group, and Washington payday borrowers in the follow-up period as the control group. Section 8 concludes with brief discussion of directions for future research.

I examine some effects of restricting access to expensive consumer credit on payday loan users, using household survey data collected around the imposition of binding restrictions on loan terms in Oregon but not in Washington. The results suggest that the policy change decreased expensive short-term borrowing in Oregon relative to Washington, with many Oregon payday borrowers shifting into plausibly inferior substitutes. Oregon respondents were also significantly more likely to experience an adverse change in financial condition (where an adverse outcome is defined as being unemployed, or having a negative subjective assessment about one’s overall recent or future financial situation). The results suggest that restricting access to consumer credit hinders productive investment and/or consumption smoothing, at least over the short-term. Much work remains to address the questions of whether access to expensive credit improves (consumer) welfare, and why. The likelihood of additional policy changes at the state (and possibly federal) level seems high, suggesting that difference-in indifferences (DD) approaches like the one used in this paper will continue to be useful. Future studies would benefit from larger sample sizes and a richer set of proxies for consumer welfare and financial condition. Viable examples of proxies to collect from household surveys include postponed medical care and forced moves as used by Melzer (2009), shutoffs of heat or other utilities, dunning as used by Morgan and Strain (2008), and hunger and subjective well-being as used by Karlan and Zinman (forthcoming). Future studies would also do well to track outcomes of interest over longer horizons, since the costs and benefits of investment and consumption smoothing activities may have gestation periods, or compound over time. Finally, it is critical to begin reconciling findings across different studies. Are the differences due to methodology, market context, and/or borrower heterogeneity? Field experiments randomized at the individual level would help, by providing clean variation in credit access and more statistical power than statelevel natural experiments. Additional data collection on a richer set of outside options (for borrowing and economic activity) and decision inputs (for intertemporal choice models) would help address whether heterogeneity across consumers and markets drives the results.