Charts of the day

FinTech Lending with LowTech Pricing (with M. Johnson, I. Ben-David and Jason Lee)

This figure shows the sharp discontinuity at FICO 660 for interest rate (APR), to a much less degree in default.
This figure shows in the overlapped market where prime and nonprime loans intersect, the rate gap is about 10 percentage points for the same risk.
This figure plots the policy-targeted industries of individual provinces' 13th Five-Year Plans. The left-hand side of the figure lists each of the provinces, and right-hand side of the figure lists each of the targeted industries.
This figure plots the number of new startups in treated and control industries by SOE and private incubators by year. The x-axis is the first year when these startups entered the incubators.

Deadly Subprime Credit (with G. Li and H. Ma).

This figure plots the binned scatter plot between cross-year average subprime/prime auto loan originations and traffic fatalities (per 100,000 people) from 2001 to 2018. Subprime (prime) borrowers are defined as those with an Equifax Risk Score below 620 (above 720). The correlation coefficient between subprime (prime) auto loan originations and the number of traffic fatalities is 0.49 (-0.30).
This figure displays the regression coefficient estimates and 95% confidence intervals from our event studies using fintech lender (Autofi) entries as events. The dependent variable is the standardized number of subprime auto loans originated in county i and quarter t, scaled by the total number of borrowers in the county. The coefficients capture the differential subprime auto loans between ex ante high-subprime and low-subprime concentration counties.
This figure plots the time series of number of automobile loan originations in the U.S. by different lender types.
This figure plots the market share of auto loans originated by autofi---fintech lenders in automobile market---in the treated and control counties. Treated are counties entered by autofi in 2014.
This figure plots the estimates of daily pattern of auto loans with respect to note rate based on loan-level data. X axis is day relative to the last day of the month (end of month) within year month. Y axis is the estimated coefficient from regression of note rate on the auto loan while controlling for dealer x year-month, lender, county, year/make/model fixed effects.
This figure plots the dynamic estimates of the effect of auto loan origination on borrowers' credit outcomes (total balances) based on borrower by quarter panel data. X axis is the credit report quarter relative to the origination date of the auto loan. Coefficients, plotted in the figures, are on the interaction term of the indicator whether the loan is originated in the next day after the month-end and different quarters relative to the origination date, while controlling for individual, relative quarter, calendar quarter, dealer x year-month, and make/model/year of the collateral.

Surviving the Fintech Disruption (with W. Jiang, Y. Tang and R. Xiao).

The figure plots the relation between occupation-level fintech exposure and cumulative change in job posting shares. The y-axis is the cumulative change in job posting shares from 2007 to 2018 based on BGT data, and the x-axis is the time-invariant occupation-level fintech exposure percentiles. The fintech exposure measure is constructed by the authors.
The figure plots state-level average of occupational fintech exposure percentiles weighted by job postings in 2007. The fintech exposure measure is constructed by the authors based on fintech patent filings from 2003 to 2017.
This figure plots the home price indices estimated using the repeat sales methods based on residential housing transactions for two types of land leases in Hong Kong from 2004 to 2020. The first type of land lease expires on 6/30/2047, while the second type expires between 7/1/2047 and 12/31/2064. The 95% confidence intervals of price indices are indicated by the gray shaded area.
This figure presents the model-implied discounts and discounts estimated from hedonic regressions across twelve lease groups (including colonial British and HKSAR ones). The Y axis is the coefficient on various lease group.

Corporate Climate Risk: Measurements and Responses (with Q. Li, H. Shan and Y. Tang).

The figure shows the time-average of firm-level acute risk and chronic risk (divided by its standard deviation in the time series), respectively. We label each spike with the corresponding topics discussed in the conference calls which contribute to the increase in each type of climate risk.
The figure shows the time-average of firm-level transition risk (divided by its standard deviation in the time series), respectively. We label each spike with the corresponding topics discussed in the conference calls which contribute to the increase in each type of climate risk.

Fintech Borrowers: Lax-Screening or Cream-Skimming? (with M. Di Maggio).

The Review of Financial Studies, 2021.

The figure plots the coefficient on the interaction of fintech loan indicator and relative month dummies from the regressions that examine the difference in the loan performance dynamics between the fintech and non-fintech loans.
The figure plots the coefficient on the interaction of fintech loan indicator and relative month dummies from the regressions that examine the difference in the borrower credit score dynamics between the fintech and non-fintech loans.

Second Chance: Life Without Student Debt (with M. Di Maggio and A. Kalda).

Journal of Finance.

The figure plots geographic distribution, at state level, of delinquent student loan borrowers based on complete credit bureau data.

Mortgage Refinancing, Consumer Spending, and Competition: Evidence from the Home Affordable Refinancing Program, (with S. Agarwal, G. Amromin, S. Chomsisengphet, T. Piskorski, and A. Seru).

The Review of Economic Studies.

This figure shows following HARP refinancing, individuals significantly increase their auto purchase using the savings from mortgage payment compared to similar HARP eligible population.

Relational Contracts in the Housing Market (with S. Agarwal and C. Song).

This figure plots the relative frequency of appraisals over appraisal--contract or AVM--contract value differences. The green bars denote the frequency of appraisals over the difference between the appraisal and contract values for arm's length transactions. The white bars denote the frequency of appraisals over the differences between appraised and ex post contract values for distressed properties owned by lenders, so-called real estate--owned properties. The red bars denote the frequency of appraisals based on the difference between appraisals and an AVM benchmark value.

Interest Rate Policy Pass-Through: Mortgage Rates, Household Consumption and Voluntary Deleveraging, (with M. Di Maggio, A. Kermani, B. Keys, T. Piskorski, R. Ramcharan, and A. Seru).

American Economic Review, 2017.

This figure shows the identification of monetary policy in mortgage data. The control group consists of 7/1 ARM loans whose interest rate remains fixed through first 84th months while the treatment group includes 5/1 ARMs who interest rates are reset to lower level due to monetary policy.

Systemic Mistakes of Borrowers in the Mortgage Market and Lack of Financial Sophistication, (with S. Agarwal and I. Ben-David).

Journal of Financial Economics, 2017.

The figure plots the simulated pre-tax NPV, ex ante, of the points-takers using a sample of mortgage applicants where we observe whether the borrower takes out a point or not. In order to estimate the ex ante NPV, we estimate a proportional hazard model to obtain the base survival curve along with sensitivities of exits with respect to borrower/loan characteristics. We then randomly draw home price and interest rate scenarios using Monte Carlo simulations and apply to mortgage origination date to predict the ex ante hazards. This is used to calculate the expected NPV based on different I/O price.

The Foreclosure Discount: Myth or Reality? (with J. Harding and E. Rosenblatt).

Journal of Urban Economics, 2012.

The figure compares the estimated contagion effect by phase of foreclosure for all four rings. Ring 1 contains all properties within 300 feet of the non-distressed sale. Ring 2 contains all properties greater than 300 feet and less than 500 feet from the non-distressed sale. Ring 3 includes properties between 500 feet and 1000 feet and Ring 4 contains properties from 1000 feet to 2000 feet. The plotted phase effects represent the average estimated effect over the seven different MSAs.