with David Scharfstein
We show that the use of algorithms to predict race has significant limitations in measuring and understanding the sources of racial disparities in finance, economics, and other contexts. First, we derive theoretically the direction and magnitude of measurement bias in estimates of unconditional disparities that use predicted instead of actual race. If their prediction errors were random, existing algorithms such as BIFSG (Voicu, 2018) would underestimate disparities in credit access for Black borrowers by 30–50%. In practice, the algorithms are systematically biased toward identifying minority borrowers who are likely to experience worse outcomes. Second, we show that in many applications the accuracy of predicted race is illusory, as many empirical methodologies call for the inclusion of location fixed effects and comparison of white and minority individuals within a given geography. As a result, estimates of conditional disparities can be dramatically underestimated, in some of our analyses, by up to 60%. While underestimating conditional disparities, predicted race overstates the importance of location in explaining disparities. Finally, because algorithm accuracy can vary across subsamples, predicted race can under- or overestimate interaction effects meant to measure cross-sectional variation in disparities.
with David Scharfstein
Abstract: We document significant racial disparities in the utilization of the Paycheck Protection Program (PPP). In a large sample of Florida restaurants, Black- and Hispanic-owned restaurants are 20.5% and 7.0% less likely to receive PPP loans than white-owned restaurants within the same ZIP code. Although a part of these disparities stems from differences in firm size and age, Black- and Hispanic-owned restaurants are still 14.5% and 4.9% less likely to receive PPP loans after controlling for these characteristics. Lower PPP utilization is driven by differences in bank lending: relative to white-owned restaurants, Black- and Hispanic-owned restaurants are 19.9% and 4.8% less likely to receive PPP loans from banks, while Black-owned restaurants are 5.5% more likely to receive PPP loans from nonbank lenders and Hispanic-owned restaurants are equally likely to receive nonbank PPP loans. Minority-owned restaurants are also more likely to use the Economic Injury Disaster Loan program, in which firms applied directly to the Small Business Administration for subsidized loans that had less attractive terms than PPP loans. While white-owned firms with prior lending relationships are more likely to receive PPP loans, this is not the case for Black-owned firms. Further, we show that Black-owned restaurants are significantly less likely to receive bank PPP loans in counties in which white people exhibit greater implicit and explicit racial bias towards Black people. Finally, extending our analysis to all industries, we show that conditional on receiving emergency loans, minority-owned firms are less likely to receive them from banks than are white-owned firms that operate in the same industry and ZIP code.
Abstract: We use the 2020 Small Business Credit Survey to study the sources of racial disparities in use of the Paycheck Protection Program (PPP). Black-owned firms are 8.9% less likely than observably similar white-owned firms to receive PPP loans. About 55% of this take-up disparity is explained by a disparity in application propensity, as Black-owned firms are 4.9% less likely to apply for PPP loans. While administrative data covering the universe of PPP recipients indicate that Black-owned firms were less likely than white-owned firms to borrow from banks and more likely to borrow from fintech lenders, we show that this fact is driven entirely by application behavior. Conditional on applying for PPP, Black-owned firms are 9.9\% less likely than white-owned firms to apply to banks and 7.8% more likely to apply to fintechs. However, Black-owned firms face similar average approval disparities at banks and fintechs: 7.4% and 8.4%. Sorting by Black-owned firms away from bank and toward fintech applications is significantly stronger in more racially biased counties, and the bank approval disparity is also larger in more racially biased counties. Thus, while automation at fintechs may reduce the direct impact of racial bias on both application behavior and approval outcomes, we find no evidence that fintech lenders are able to mitigate racial disparities in approval rates.
with Adi Sunderam
R&R at the Journal of Finance
Abstract: We propose a novel measure of market liquidity that does not depend on transaction or quote data: the strength of the cross-sectional relationship between mutual fund cash holdings and flow volatility. Our measure captures funds’ expectations of future liquidity and can be applied in settings where transactions are rare or data are not available. We show that municipal bonds and syndicated loans are perceived to be twice and five times as illiquid as corporate bonds respectively. The perceived liquidity of speculative-grade and Rule 144A bonds is significantly lower than investment-grade bonds and deteriorated significantly following the 2008-9 financial crisis.
with Viet-Dung Doan
Abstract: Using novel trade-level data, we study how municipal bond mutual funds trade in response to daily flows. When forced to sell bonds to satisfy redemptions, funds prearrange fewer trades, sell more liquid bonds, and trade with more central dealers, who offer faster execution. Funds are especially likely to turn to more central dealers when trading lower rated bonds, when funds have low cash buffers, and when trading after periods of aggregate outflows. More central dealers charge higher markups when funds demand fast execution.
with Viet-Dung Doan
Abstract: We develop a novel measure of the dollar value of liquidity created by open-end mutual funds. Our measure compares the costs investors would have incurred had they traded on their own in response to liquidity shocks with the actual costs incurred by open-end mutual funds when trading to satisfy investor redemptions. Applying this measure to municipal bond mutual funds, we show that during the 2008-2017 period the average fund provides liquidity services worth 1.80 cents per dollar of gross redemptions or 50 basis points of fund assets per year. The aggregate value of liquidity services provided during this period was $14-22 billion. We decompose liquidity creation into three components: 1) flow netting, 2) liquidity management, and 3) trade execution, and explore the cross-sectional and time-series variation in liquidity creation.
with Adi Sunderam
Joint winner of the ESRB research prize in memory of Ieke van den Burg, 2016.
Abstract: We study liquidity transformation in mutual funds using a novel data set on their cash holdings. To provide investors with claims that are more liquid than the underlying assets, funds engage in substantial liquidity management. Specifically, they hold substantial amounts of cash, which they use to accommodate inflows and outflows rather than transacting in the underlying portfolio assets. This is particularly true for funds with illiquid assets and at times of low market liquidity. We provide evidence suggesting that mutual funds’ cash holdings are not large enough to fully mitigate price impact externalities created by the liquidity transformation they engage in.
Review of Financial Studies 35(11): 4902–4947
Abstract: Analyzing hand-collected credit agreements data for a random sample of middle-market firms during 2010-2015, we find that a third of all loans is extended directly by nonbank financial intermediaries. Nonbanks lend to less profitable and more levered firms that undergo larger changes in size around loan origination. The probability of borrowing from a nonbank jumps by 34% as EBITDA falls below zero, an effect that is largely due to bank regulation. Controlling for firm and loan characteristics, nonbank loans carry 190 basis points higher interest rates, suggesting that access to funding, rather than prices, is why firms borrow from nonbanks.
Review of Financial Studies 34(5): 2362-2410
Abstract: “Founder-friendly” venture financings and nontraditional venture investors have both flourished over the past decade. Using detailed contract data, we study open-end mutual funds investing in private venture-backed firms. We posit that conflicts between early-stage venture investors and liquidity-constrained later-stage ones influence the classic agency problems affecting entrepreneurs and investors. We find that mutual funds with more stable funding are more likely to invest in private firms and that financing rounds with mutual fund participation have stronger redemption, stronger IPO-related rights, and less board representation. These findings are consistent with our conceptual framework.
Internet appendix with SAS code to extract mutual fund holdings of unicorns from CRSP Mutual Fund Database. Note that starting around September 2018, T. Rowe Price funds stopped reporting holdings of private securities to commercial data providers such as CRSP and Morningstar. These holdings continue to be reported in the funds' SEC filings.
with Adi Sunderam
Journal of Financial Economics 135(3): 602-628
Abstract: We develop three novel measures of the incentives of equity mutual funds to internalize the price impact of their trading. We show that mutual funds with stronger incentives to internalize their price impact accommodate inflows and outflows by adjusting their cash buffers instead of trading in portfolio securities. As a result, stocks held by these funds have lower volatility, and flows out of these funds have smaller spillover effects on other funds holding the same securities. Our results provide evidence of meaningful fire sale externalities in the equity mutual fund industry.
Journal of Finance 72(5): 1893-1936
Abstract: I study the incentives of the collateral managers who selected securities for ABS CDOs—securitizations that figured prominently in the financial crisis. Specialized managers that did not have other businesses that could suffer negative reputational consequences invested in low quality securities underwritten by the CDO's arranger. These securities perform significantly worse than observationally similar securities. Managers that invested in these securities were rewarded with additional collateral management assignments. Diversified managers that did assemble CDOs suffered negative reputational consequences during the crisis: institutional investors withdrew from their mutual funds. Overall, the results are consistent with a quid pro quo between the collateral managers and CDO underwriters.
Journal of Financial Economics 122(2): 248-269
Abstract: Many have argued that overoptimistic thinking on the part of lenders helps fuel credit booms. We use new micro-data on mutual funds’ holdings of securitizations to examine which investors are susceptible to such boom-time thinking. We show that firsthand experience plays a key role in shaping investors’ beliefs. During the 2003-2007 mortgage boom, inexperienced fund managers loaded up on securitizations linked to nonprime mortgages, accumulating twice the holdings of more seasoned managers. Moreover, inexperienced managers who personally experienced severe or recent adverse investment outcomes behaved more like seasoned managers. Training and institutional memory can serve as partial substitutes for personal experience.
with Adi Sunderam
Review of Financial Studies 27(6): 1717-1750
Abstract: We document frictions in money market mutual fund lending that lead to the transmission of distress across borrowers. Using novel security-level holdings data, we show that funds exposed to Eurozone banks suffered large outflows in mid-2011. These outflows had significant spillovers: non-European issuers relying on such funds raised less short-term debt financing. Issuer characteristics do not explain the results: holding fixed the issuer, funds with higher Eurozone exposure cut lending more. Due to credit market frictions, funds with low Eurozone exposure provided substitute financing only to issuers they had pre-existing relationships with, even though issuers are large, highly rated firms.
with Adi Sunderam
Review of Financial Studies 25(7): 2041-2070
Review of Financial Studies Young Researcher Prize, 2012.
Abstract: We study the real effects of market segmentation due to credit ratings by using a matched sample of firms just above and just below the investment-grade cutoff. These firms have similar observables, including average investment rates. However, flows into high-yield mutual funds have an economically significant effect on the issuance and investment of the speculative-grade firms relative to their matches, especially for firms likely to be financially constrained. The effect is associated with the discrete change in label from investment- to speculative-grade, not with changes in continuous measures of credit quality. We do not find similar effects at other rating boundaries.
with C. Fritz Foley and Robin Greenwood
Financial Management 41: 885-914
2nd place Pearson Prize for the best paper published in Financial Management between Autumn 2012 and Autumn 2014
Abstract: Standard theories of ownership assume insiders ultimately bear all agency costs and therefore act to minimize conflicts of interest. However, overvalued equity can offset these costs and induce listings associated with higher agency costs. We explore this possibility by examining a sample of public listings of Japanese subsidiaries. Subsidiaries in which the parent sells a larger stake and subsidiaries with greater scope for expropriation by the parent firm are more overpriced at listing, and minority shareholders fare poorly after listing as mispricing corrects. Parent firms often repurchase subsidiaries at large discounts to valuations at the time of listing and experience positive abnormal returns when repurchases are announced.
with Michael Faulkender
Journal of Financial and Quantitative Analysis 46: 1727-1754
Abstract: Existing cross-sectional findings on nonfinancial firms’ use of derivatives that are usually interpreted as the result of hedging may alternatively be due to speculation. Panel data examinations can distinguish between derivatives practices that endure over time and are therefore more likely to result from hedging, and those that are more transient, thus more consistent with speculation. Our decomposition results indicate that hedging of interest rate risk is concentrated among high-investment firms, consistent with costly external finance. Simultaneously, firms appear to use interest rate swaps to manage earnings and to speculate when their executive compensation contracts are more performance sensitive.
with Alain P. Chaboud and Jonathan H. Wright
Journal of the European Economic Association 6: 589-596
Abstract: This article introduces a new high-frequency data set that includes global trading volume and prices over five years in the spot euro-dollar and dollar-yen currency pairs. Studying the effects of US macroeconomic data releases, we show that spikes in trading volume tend to occur even when announcements are in line with market expectations, in sharp contrast to the price response. There is some evidence that the volume after announcements is negatively related to the ex ante dispersion of market expectations, contrary to the standard theoretical prediction. At very high frequency, we find evidence that much of the immediate jump in prices in reaction to an announcement occurs before the surge in volume.
with David W. Berger, Alain P. Chaboud, Edward Horowka, and Jonathan H. Wright
Journal of International Economics 75: 93-109
Abstract: We analyze the association between order flow and exchange rates using a new dataset representing a majority of global interdealer transactions in the two most-traded currency pairs at the one minute frequency over a six-year time period. This long span of high-frequency data allows us to gain new insights about the joint behavior of these series. We first confirm the presence of a substantial association between interdealer order flow and exchange rate returns at horizons ranging from 1 min to two weeks, but find that the association is substantially weaker at longer horizons. We study the time-variation of the association between exchange rate returns and order flow both intradaily and over the long term, and show that the relationship appears to be stronger when market liquidity is lower. Overall, our study supports the view that liquidity effects play an important role in the relationship between order flow and exchange rate changes. This by no means rules out a role for order flow as a channel by which fundamental information is transmitted to the market, as we show that our findings are quite consistent with a recent model by Bacchetta and Van Wincoop (2006: Can information heterogeneity explain the exchange rate determination puzzle? American Economic Review, 96, pp. 552–576.) that combines both liquidity and information effects.