Guanyu Zhou

Working Papers:

• The Worst and Best of Times: Stimulus and Credit Card Debt (Job Market Paper)

 Entering Covid, the stocks of credit card banks dropped by 60%, as the market expected defaults to rise to GFC levels. Instead, banks made record profits. I show that this was due to the fiscal stimulus (the "checks''). Using geographic variation in check recipient share, I show that they led to the largest decreases in credit utilization, and hence default, amongst the riskiest accounts—those with the highest credit utilization—consistent with borrowers being predominantly non-strategic.  I construct a model of optimal card lending to quantify the impact of the checks on banks. The checks have opposing effects on profitability: a decline in charge-offs, but also reduced demand for credit. The positive effect dominates during the crisis. Paydown of riskiest borrowers increases their survival and generates more profit that offsets softer demand from safer borrowers. I estimate that the $814B of stimulus checks transferred $75B ($99B risk adjusted) of value to credit card banks: $44B due to reduced average defaults, and $31B due to the covariance between account’s paydown and pre-crisis risk. My results show that providing fiscal support to consumers in a downturn involves a large transfer to credit card lenders.

• Credit Card Banking (with Itamar Drechsler, Hyeyoon Jung, Weiyu Peng, and Dominik Supera)

Credit card interest rates, the marginal cost of consumption for nearly half of households, currently average 23 percent, far exceeding the rates on any other major type of loan or bond. Why are these rates so high? To understand this, and the economics of credit card banking more generally, we analyze regulatory account-level data on 330 million monthly accounts, representing 90 percent of the US credit card market. Default rates are relatively high at around 5 percent, but explain only a fraction of cards' rates. Non-interest expenses and rewards payments are more than offset by interchange and non-interest income. Operating expenses, such as marketing, are very large, and are used to generate pricing power. Deducting them, we find that credit card lending still earns a 6.8 percent return on assets (ROA), more than four times the banking sector's ROA. Using the cross section of accounts by FICO score, we estimate that credit card rates price in a 5.3 percent default risk premium, which we show is comparable to the one in high-yield bonds. Adjusting for this, we estimate that card lending still earns a 1.17 percent to 1.44 percent "alpha" relative to the overall banking sector.

Media: U.S.News, MarketWatch, CNBC, American Banker

• The Rise and Fall of Fintech Lending (with Weiyu Peng)

Fintech lenders promise broader access to unsecured personal credit and more precise risk pricing than traditional banks. In this paper, we investigate whether these platforms improve risk pricing and if lenders on the platform are fairly compensated for their risk.  Analyzing the pricing and returns of 79 million personal loans originated between 2013 and 2024 on a major U.S. fintech platform, we find that the platform earns low returns and significantly underprices risky loans. Platform charges lower rates than credit cards but incurs higher losses. Consequently, the average default-adjusted spread and ROA are less than half of credit card issuers. Notably, returns for subprime borrowers are negative. We find that the platform ex-ante underestimates default rates by an average of 1%, and by 5% for subprime loans. This aggressive underpricing suggests that fintech lenders struggle to compete with incumbent lenders, likely explaining why many early fintech entrants have failed in the past five years.