Income and Charitable Giving

Are Americans with higher incomes more generous? An OLS study with robust inference

The question. Research agrees that richer households give more dollars — but disagrees on whether they give a larger share. Testing the popular “U-shaped generosity” hypothesis, we modeled charitable-giving proportion against income using county-aggregated 2022 IRS Statistics of Income data. UC Berkeley MIDS statistics final project.

How it works. OLS regression in R with linear, quadratic, and logarithmic AGI specifications, robust standard errors (sandwich/lmtest), full CLM assumption diagnostics, and stargazer regression tables, in a reproducible renv project with peer-reviewed drafts.

Results. The data rejected the U-shape: we found a J-shaped positive relationship — giving proportion grows non-linearly with mean AGI. The preferred model (AGI terms plus dividend, capital-gains, and rent/royalty income) explains 78.5% of the variance in charitable contributions per return (adjusted R² = 0.785).

The model progression (stargazer, robust standard errors): from a linear AGI baseline (adj. R² = 0.602) through the quadratic specification to the preferred model (4), which adds dividend, capital-gains, and rent/royalty income — adj. R² = 0.785, n = 2,089 counties.

Team. With Fatema Alsaleh and WooJung Kim.

Links. Repository · Final report (PDF)