RESEARCH

The aging of U.S. population is believed to affect economic dynamism, housing markets as well as political balance. In this project, we uncover dramatic demographic divergence between low-density rural counties and high-density urban places. We identify two main phases: suburbanization in the 1950s and 1960s and urban revival since the 1990s. Their cumulated effects led to an increase in mean age of 8 years in rural counties between 1950 and 2010, while urban counties were only aging by 2 years.

We show that changing fertility and mortality can’t explain those patterns, which are entirely driven by migration across counties. Finally, we illustrate how these secular trends contributed to the decline in migration observed since the 1980s.

Published Papers

Generalized Pareto Curves: Theory and Applications (2022)

with Thomas Blanchet and Thomas Piketty, Journal of Income and Wealth

Full text Slides Online app

2022 Kendrick Prize
Method used to construct the World Inequality Database

We define generalized Pareto curves as the curve of inverted Pareto coefficients $b(p)$ , where $b(p)$ is the ratio between average income or wealth above rank $p$ and the $p$ -th quantile $Q(p)$ (i.e. $b(p) = E[X|X > Q(p)]/Q(p)$ ). We use them to characterize and visualize power laws. We develop a method to nonparametrically recover the entire distribution based on tabulated income or wealth data as is generally available from tax authorities, which produces smooth and realistic shapes of generalized Pareto curves. Using detailed tabulations from quasi-exhaustive tax data, we demonstrate the precision of our method both empirically and analytically. It gives better results than the most commonly used interpolation techniques. Finally, we use Pareto curves to identify recurring distributional patterns, and connect those findings to the existing literature that explains observed distributions by random growth models.

Generalized Pareto Curves: Theory and Application using Income and Inheritance Tabulations for France 1901-2012 (2015)

Master’s Thesis, Paris School of Economics

Full text

I develop a new nonparametric method to estimate shares of income and wealth accruing to the different deciles and percentiles of the distribution. Whereas methods usually employed in the literature are parametric, inasmuch as they are typically based on the assumption that the top of the income distribution follows a Pareto distribution, I am able to relax any assumption on the shape of the distribution. Namely, I evaluate non-parametrically the distribution after estimating the empirical “generalized Pareto curve”. It is defined as the curve of inverted Pareto coefficients $b(p)$ , where $p$ is the percentile rank and $b(p)$ is the ratio between the average income above percentile $p$ and the income threshold at percentile $p$ (i.e. $b(p)=\E{y | F(y) \geqslant p}/F^{-1}(p)$ ).

I exploit income tax tabulations from 1915 to 2012 in France to generate new series that I can compare to existing WTID series for top shares. I find that old and new series are almost equal throughout the period. This confirms that the Paretian form fits indeed relatively well the top of the distribution. However, the Pareto hypothesis is only valid locally, whereas the method elaborated here allows derivation of estimates for the whole income distribution. In particular, I provide computer codes that can be used to simulate reliable synthetic micro-files from income tabulations. Another potential application is the homogenization of series obtained for individual-based tax systems and for household-based tax systems, and for the income concept. Finally, I provide a preliminary application to French 1901-2000 inheritance tabulations and the distribution of wealth.