Working Papers

1. Robust Hicksian Welfare Analysis under Individual Heterogeneity (with Sebastiaan Maes)

   Welfare effects of price changes are often estimated with cross-sections; these do not identify demand with heterogeneous consumers. We develop a theoretical method addressing this, utilizing uncompensated demand moments to construct local approximations for compensated demand moments, robust to unobserved preference heterogeneity. Our methodological contribution offers robust approximations for average and distributional welfare estimates, extending to price indices, taxable income elasticities, and general equilibrium welfare. Our methods apply to any cross-section; we demonstrate them via UK household budget survey data. We uncover an insight: simple non-parametric representative agent models might be less biased than complex parametric models accounting for heterogeneity.

2. Exact Inference from Finite Market Data (With Felix Kubler and Herakles Polemarchakis)

   (Revise and Resubmit (2nd Round) JET)

   We give conditions under which an individual’s preferences can be identified with finite data. First, we derive conditions that guarantee that a finite number of observations of an individual’s binary choices identify preferences over an arbitrarily large subset of the choice space and allow one to predict how the individual shall decide when faced with choices not previously encountered. Second, we extend the argument to observations of individual demand. Finally, we show that finitely many observations of Walrasian equilibrium prices and profiles of individual endowments suffice to identify individual preferences and, as a consequence, equilibrium comparative statics.

3. Rethinking Distribution: Introducing Market Segmentation as a Policy Instrument (with Yatish Arya)

   Inequality and skewed distribution of ‘essential’ goods are pertinent problems in the 21st-century world. We consider a general equilibrium framework with a division between essential and non-essentials, and only essential goods are relevant for distributional concerns. We then compare the effects of four policies on social welfare: subsidies, direct transfers, quantity rationing, and a fourth policy that we introduce and call Market Segmentation (MS). In MS, the market for essentials is segmented from non-essentials, i.e., they are not freely tradeable with each other. We find that if the relative number of low-income individuals in the economy is large and “essentials” are consumed in-elastically, MS outperforms direct transfers and subsidies. MS also weakly dominates quantity rationing. We discuss how market segmentation can help policymakers deal with issues such as automation and the superstar phenomenon (Scheuer and Werning, 2017).

4. Estimating the Nature of Technological Change: Exploiting Shifts in Skill Use Within and Between Occupations [Recording @ NBER SI Personnel] (With Costas Cavounidis, Vittoria Dicandia, and Kevin Lang )

   We exploit employment trends to uncover changes in skills’ productivities. Whereas Autor, Levy, and Murnane (2003) study the degree to which routine-intensity can rationalize employment trends, our reverse approach characterizes the kind of technological change that best explains shifts. We combine a tractable GE model with three DOT (dictionary of occupational titles) editions, the 1960, 1970, and 1980 Censuses, and the March CPS, to estimate changes in the relative productivities of skills. We find ‘skill bias’ - finger-dexterity productivity grew rapidly, while abstract-skill productivity lagged. With substitutability between abstract and routine inputs, these results also explain changing skill use within occupations.

5. Finite Tests for Functional Characterizations of Preference Families

   Historically, testing if decision-makers obey certain choice axioms using choice data takes two distinct approaches. The ’functional’ approach observes and tests the entire ’demand’ or choice’ function, whereas the ’revealed preference(RP)’ approach constructs inequalities to test finite choices. I demonstrate that a statistical recasting of the revealed enables uniting both approaches. Specifically, I construct a computationally efficient algorithm to output one-sided statistical tests of choice data from functional characterizations of axiomatic behavior, thus linking statistical and RP testing. An application to weakly separable preferences, where RP characterizations are provably NP-Hard, demonstrates the approach’s merit. I also show that without assuming monotonicity, all restrictions disappear. Hence, any ability to resolve axiomatic behavior relies on the monotonicity assumption.

6. Obsolescence Rents: Teamsters, Truckers, and Impending Innovations. (with Costas Cavounidis, Qingyuan Chai, and Kevin Lang)

   We consider large, permanent shocks to individual occupations whose arrival date is uncertain. We are motivated by the advent of self-driving trucks, which will dramatically reduce demand for truck drivers. Using a bare-bones overlapping generations model, we examine an occupation facing a risk of obsolescence. We show that workers must be compensated to enter the occupation - receiving what we dub obsolescence rents - while employment in the occupation falls and becomes older. We investigate the market for teamsters at the dawn of the automotive truck as an `a propos parallel to truckers themselves, as self-driving trucks crest the horizon. We find that as the widespread adoption of trucks drew nearer, the number of teamsters fell, the occupation became ‘grayer’, and teamster wages rose, as predicted by the model.

7. Beyond the Mean: Testing Consumer Rationality through Higher Moments of Demand

   We study a setting where an analyst has access to purely aggregate information about the consumption choices of a heterogenous population of individuals. We show that observing the statistical moments of market demand allows the analyst to test aggregate data for rationality. Interestingly, just the mean and variance of demand carry observable restrictions. This is in stark contrast to impossibility result of the Sonnenschein-Mantel-Debreu theorem, which shows that aggregate demand carries no observable restrictions at all. We leverage our approach to deliver a characterization of rationality in terms of moments for the common two-good case. We illustrate the usefulness of moment-based restrictions through two applications: (i) improving the precision of demand and welfare estimates; and (ii) testing for the existence of a welfare-relevant representative consumer.