Big Data in Macro and Labor Economics: New Insights into Open Questions

Abstract: The PI will use 'big data' to measure important economic variables in a unique way. The project uses a data set drawn from government administrative records; the data include over 30 years of annual earnings for a sample of millions of American men. With these data, the PI will be able to measure the year-to-year risk of changes in annual earned income. In particular, he will be able to determine whether or not common assumptions used in other economic models (for example, that the risk of an income shock is normally distributed) are correct.

This is potentially transformative research; the PI is making innovative use of existing data and his results could overturn the conventional understanding of how business cycles, industry-specific factors, occupation, and skill set determine the risk of large year-to-year changes in income.

The results will be broadly useful as well for economic policy. To understand how recessions affect American families, we need to understand how recessions affect earnings. Our current evidence here is based on small samples of the US population, and the data from these samples are not rich enough to answer the full range of questions.