Fabian Eckert

Visiting Scholar at the Federal Reserve Bank of Minneapolis

Research Interest: Trade, Economic Geography, and Macroeconomics

Update: I will be visiting the Minneapolis FED in July and August 2019, then I will be a postdoctoral associate in the IES section of the Economics Department at Princeton, before joining the UCSD Economics Department in Summer 2020 as an Assistant Professor!

contact information:
fabian dot eckert at yale dot edu

working papers:

Growing Apart: Tradable Services and the Fragmentation of the U.S. Economy [Job Market Paper]

Between 1980 and 2010, the college wage premium in U.S. labor markets with larger initial shares of high-skill service employment grew substantially faster than the nationwide average. I show how this trend can be explained within the con- text of a Ricardian model of interregional trade, where a reduction in communication costs magnifies regional specialization in high-skill services, raising the skill premium in service-exporting regions and reducing it in service-importing regions. Quantita- tively, I show that the decline in communication costs I infer from sectoral trade im- balances can explain a substantial part of the differential skill premium growth across U.S. labor markets in the data. These regional changes aggregate to account for 30 percent of the rise in the overall U.S. college wage premium between 1980 and 2010.

The Return to Big City Experience w/ Mads Hejlesen & Conor Walsh

In this paper, we provide evidence for higher returns to experience in big cities. We exploit a government policy of quasi-random settlement of political refugees in Denmark between 1986 and 1998, which generated plausibly exogenous variation in workers' initial local labor market. Detailed matched employer-employee datasets allow us to track workers' location and labor market experience, including their employers. We show that the slope of a refugee's lifetime wage path depends strongly and positively on initial placement in the country's capital, Copenhagen. Conditional on observables, settled refugees initially earn similar hourly wages across regions, but those placed in Copenhagen see their wages grow 0.81% faster than others with each year of experience they accumulate. We further show that this premium is driven by the greater acquisition of experience at high-wage establishments and by differential sorting across occupations. Estimating a statistical spatial model of earnings dynamics reveals that sorting on unobserved ability within cities plays an important role in explaining observed patterns.

Can We Save the American Dream? w/ Tatjana Kleineberg

Neighborhoods in the US differ substantially in the educational and economic oppor- tunities that they offer to children who grow up in them. We develop and estimate a structural spatial equilibrium model of residential and education choice to study the effects of school financing policies on education outcomes, intergenerational mobility, and welfare - at the local and aggregate level. Our model generates persistent effects of children’s neighborhoods on adult outcomes through local labor market access and local human capital formation. Local school funding is an important component of the latter. We estimate the model using a range of US Census datasets by fitting model predictions to regional data of the actual US geography. We use the estimated model to study the effects of counterfactual policy interventions, in particular, the equalization of school funding across all students. We find that general equilibrium responses in local prices and local skill compositions significantly dampen the partial equilibrium ef- fects of the policy, so that effects on education outcomes and intergenerational mobility are positive but only moderate in general equilibrium.

Spatial Structural Change w/ Michael Peters

This paper studies the spatial implications of structural change. The secular decline in spending on agricultural goods hurts workers in rural locations and increases the return to moving towards non-agricultural labor markets. We combine detailed spatial data for the U.S. between 1880 and 2000 with a novel quantitative theory to analyze this process and to quantify its macroeconomic implications. We show that spatial reallocation across labor markets accounts for almost none of the aggregate decline in agricultural employment. The reason is that population flows, while large in the aggregate, were only weakly correlated with agricultural specialization. Labor mobility nevertheless had important aggregate effects. Without migration income per capita would have been 15% lower. Moreover, spatial welfare inequality would have been substantially higher, especially among low-skilled, agricultural workers, which were particularly exposed to the structural transformation.

Combinatorial Discrete Choice [Python Codes] w/ Costas Arkolakis

Discrete choice problems with complementarities are prevalent in economics but the large dimensionality of potential solutions substantially limits the scope of their application. We define and characterize a general class that we term combinatorial discrete choice problems and show that it covers many existing problems in economics and engineering. We propose single crossing differences (SCD) as the sufficient condition to guarantee that simple recursive procedures can find the global maximum. We introduce an algorithm motivated by this condition and show how it can be used to revisit problems whose computation was deemed infeasible before. We finally discuss results for a class of games characterized by these sufficient conditions.

Imputing Missing Values in the US Census Bureau's County Business Patterns [Draft on request] w/ Peter Schott & Natalie Yang

The County Business Pattern data published by the US Census Bureau track employment and establishments by county and industry from 1946 to the present. While these data are extremely useful in many areas of research, they have two limitations: data for many cells are suppressed due to the Census Bureaus' need to protect confidentiality; and both industry and county classifications change over time. This paper addresses both issues. Building on techniques proposed by Isserman and Westervelt (2006), we develop a method for imputing missing observations using hierarchical adding-up constraints, and construct concordances that allow industries and counties to be compared over long periods of time.

A Consistent County-Level Crosswalk for US Spatial Data since 1790 [Materials] w/ Andres Gvirtz & Michael Peters

Both the number and the geographic boundaries of the counties covering the territory of the United States changed frequently and significantly since 1790. In order to study the spatial aspects of the United States' economic history it is frequently necessary to create a consistent panel of spatial units, which are consistent over time. We provide such a crosswalk, that enables researchers to aggregate historical US county data for every decade since 1790 to current US counties and commuting zones. This note describes the details of our data construction and how to use this crosswalk in practice. A data package containing the decadal crosswalks since 1790 and the accompanying GIS files is available on our websites.