Working Paper: NBER ID: w21917
Authors: Mariacristina De Nardi; Giulio Fella; Gonzalo Paz Pardo
Abstract: Earnings dynamics are much richer than those typically used in macro models with heterogenous agents. This paper provides multiple contributions. First, it proposes a simple non-parametric method to model rich earnings dynamics that is easy to estimate and introduce in structural models. Second, it applies our method to estimate a nonparametric earnings process using two data sets: the Panel Study of Income Dynamics and a large, synthetic, data set that matches the dynamics of the U.S. tax earnings. Third, it uses a life cycle model of consumption to compare the consumption and saving implications of our two estimated processes to those of a standard AR(1). We find that, unlike the standard AR(1) process, our estimated, richer earnings process generates an increase in consumption inequality over the life cycle that is consistent with the data and better fits the savings of the households at the bottom 60% of the wealth distribution.
Keywords: Earnings Dynamics; Consumption; Wealth; Inequality; Life Cycle Models
JEL Codes: D14; D31; E21; J31
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| richer earnings dynamics (J31) | increase in consumption inequality (F62) |
| richer earnings dynamics (J31) | better fit for wealth holdings of the bottom 60% of individuals (D14) |
| richer earnings dynamics (J31) | better forecast of future earnings (G17) |