15. Januar 2020
Abstract: Since the introduction of the hedonic model, economists have developed increasingly robust methodological tools for inferring the monetary value of regionally varying environmental attributes like climate or air quality. In recent years, national equilibrium sorting models have become a popular alternative to the conventional Rosen-Roback-Albouy framework. In the US context, this class of models typically uses a set of Metropolitan Statistical Area (MSAs) as the objects of discrete choice and explains households’ residency in an MSA as a function of wages, housing rents, local amenities, and household characteristics. By observing trade-offs between amenities, wages and rents across the landscape, it is possible to infer the marginal willingness to pay (MWTP) for an amenity based on Tiebout logic. Moreover, these models (i) relax assumptions of free mobility, yielding MWTP estimates that adapt to incomplete amenity capitalization into wages and rents; and (ii) allow for relatively straightforward linear IV strategies when dealing with endogenous amenities.
In this paper, we link the national sorting paradigm to a broader economics literature explaining urban migration patterns. Our model builds from the standard random utility framework, with households selecting their optimal MSA from a choice set of US cities. We innovate by explicitly accounting for households’ origin city in the decision-making process. This modification yields a gravity-like estimation structure that nicely accommodates the use of publicly available aggregate- and micro-data.
In spatial equilibrium, under the literature’s standard baseline assumption of linear-in-variables utility, we derive MWTP for an amenity as a ratio of utility function parameters and the estimated marginal utility of income. Building from the model’s novel reliance on both origin and destination for preference revelation (and thus parameter identification), we estimate MWTP for (exogenous) climate elements and (endogenous) air quality using a database constructed from 2012-2016 Census annual migration flow estimates, ACS microdata, and various MSA-level measures of natural and anthropogenic amenities.
In initial analysis, we turn to econometric tools of the trade and international migration literatures: our gravity-like PPML estimation of models using aggregate MSA to MSA counts of migration flows results in average MWTP estimates. We also allow for limited heterogeneity by a city’s observable demographic characteristics. Preliminary baseline estimates of average MWTP for climate and air quality are of similar magnitude to previous findings from national sorting models. However, when we allow for heterogeneous migration costs by origin city, our estimates nearly double in magnitude. In line with recent labor and urban economics research highlighting differences in migration frictions across space, our model’s results highlight the importance of accounting for cities’ “stickiness” or “churn” when inferring nonmarket amenity values.
In ongoing work, we extend this aggregate analysis to the household-level, using a 5-year sample of ACS microdata to model individual households’ location decisions. We link the gravity estimation structure of our aggregate analysis to the more conventional approach of previous RUM-based national sorting models. This link allows us to bifurcate household indirect utility functions into (i) an average origin-destination specific component; and (ii) a household-specific component which permits rich heterogeneity in HH income, rent expenditures, and demographic-driven migration costs. We estimate our model by maximum likelihood, nesting a BLP-style contraction mapping inside the procedure to extract a vector of average origin-destination utilities. In the second stage of the model, we run IV regressions of average origin-destination utility on destination amenities and an origin fixed effect, again allowing us to account for spatially heterogeneous migration frictions and endogenous amenities. Using parameters from the first and second stage of our model, we infer MWTP for our amenities of interest, and propose our model as an alternative demand-side base for richer counterfactual migration patterns in a GE setting.
Zeit: 12:15 - 13.45 Uhr
Raum: 0079 VMP 5