The Political Economy of Immigration: Limits of the Median Voter Model We compare the predictions of the standard median-voter model of immigration policy with data from the last century. We begin by extending the model to allow for technological change and endogenous schooling. This makes it possible for the model to be consistent with the large increase over the second half of the 20th century in education among native-born Americans. However, the model is still unable to match other patterns observed in the data such as the simultaneous increases in the proportion of immigrants who have attended college. We conclude that the median voter is not appropriate for analyzing US immigration policy. This failure of the median voter model provides the context for the model of immigration policy in the second chapter of my thesis.

Technological Change and Immigration Policy We propose a dynamic general equilibrium model to address the effects of technological progress on immigrant skill composition. Our results from this positive model suggest that neutral and skill-biased technological change imply essentially different immigration policies. On the one hand, skill-neutral change implies an immigrant skill distribution that is dominated by the native skill distribution; on the other hand, skill-biased change implies an immigrant overrepresentation at the top and bottom of the skill distribution. This result is interesting because of its unexpected nature. It implies that if technology changes as it has in the last decades and education has an increasing cost, then it is optimal to allow some low-skill immigration along with high-skill immigration. We show consistency of our model's predictions with data from the United States and Canada.

 

Discrimination and Freedom of Speech This essay shows that political correctness (self-censorship) may benefit society by effectively avoiding discrimination in a fairly general class of social interactions. The rationale for this effect is that speech, despite being per se harmless, may permit others to gauge the viability of any discriminatory action in a social context. That is, speech may signal whether the individual will approve or disapprove of discriminatory actions taken by others and therefore influence the perceived cost of undertaking such actions. Thus political correctness, an implicit convention of restrained speech, need not have purely negative effects by suppressing welfare enhancing information. Instead, political correctness may be an effective instrument for reducing discrimination.

Work in progress:

 

Introducing a New Estimator for Errors-In-Variables Models

The typical solution to the measurement-error problem in linear models is the IV estimator. In theory this estimator has remarkable properties but in practice, despite the ubiquitous use of it throughout the profession, there are very few instances were it is actually a feasible alternative. Indeed, good instruments are rare and the soundness of the rest is often impossible to accurately judge. On the other hand, non-parametric results are available but their application is severely limited when it comes to linear models and normal disturbances. We would like to propose an instrument-free estimator with good large-sample properties.

 

Government Design: Insights from Behavioral Economics (joint with Jörgen Weibull)

We consider the phenomenon of time-inconsistency in individual preferences and explore it's implications for political economy in democracies (i.e. majority rule) as well as for utilitarian governments. We introduce heterogeneity (initially only) in the myopic behavior for more general results. Finally, we put forth some examples of "stationary" inefficiencies that can frequently happen in governments and propose a mechanism to address/solve/minimize them.

 

Language and Vagueness

We consider the question posed by Lipman(2006): "Why is language vague?". We try to answer this question in the simplest possible framework. We then show how, indeed as Lipman conjectures, bounded rationality is necessary to account for such a phenomenon. Finally we point towards future extensions that would benefit from a more general framework for language; and we postulate some characteristics that such a framework must include.