Walter Sosa-Escudero

Individual heterogeneity in the returns to schooling: instrumental variables quantile regression using twins data

Abstract. Considerable effort has been exercised in estimating mean returns to education while carefully considering biases arising from unmeasured ability and measurement error. Recent work has investigated whether there are variations from the “mean” return to education across the population with mixed results. We use an instrumental variables estimator for quantile regression on a sample of twins to estimate an entire family of returns to education at different quantiles of the conditional distribution of wages while addressing simultaneity and measurement error biases. We test whether there is individual heterogeneity in returns to education and find that: more able individuals obtain more schooling perhaps due to lower marginal costs and/or higher marginal benefits of schooling and that higher ability individuals (those further to the right in the conditional distribution of wages) have higher returns to schooling consistent with a non-trivial interaction between schooling and unobserved abilities in the generation of earnings. The estimated returns are never lower than 9 percent and can be as high as 13 percent at the top of the conditional distribution of wages but they vary significantly only along the lower to middle quantiles. Our findings may have meaningful implications for the design of educational policies.

Tests for the error component model in the presence of local misspecification

It is well known that most of the standard speci¯cation tests are not valid when the alternative hypothesis is misspeci¯ed. This is particularly true in the error component model, when one tests for either random e®ects or serial correlation without taking account of the presence of the other e®ect. In this paper we study the size and power of the standard Raos score tests analytically and by simulation when the data is contaminated by local misspeci¯cation. These tests are adversely a®ected under misspeci¯cation. We suggest simple procedures to test for random e®ects (or serial correlation) in the presence of local serial correlation (or random e®ects), and these tests require ordinary least squares residuals only. Our Monte Carlo results demonstrate that the suggested tests have good ¯nite sample properties for local misspeci¯cation, and in some cases even for far distant misspeci¯cation. Our tests are also capable of detecting the right direction of the departure from the null hypothesis. We also provide some empirical illustrations to highlight the usefulness of our tests.

GENERAL SPECIFICATION TESTING WITH LOCALLY MISSPECIFIED MODELS

A well known result is that many of the tests used in econometrics, such as the Rao score (RS) test, may not be robust to misspecified alternatives, that is, when the alternative model does not correspond to the underlying data generating process. Under this scenario, these tests spuriously reject the null hypothesis too often. We generalize this result to generalized method of moments–based (GMM-based) tests. We also extend the method proposed in Bera and Yoon (1993, Econometric Theory 9, 649–658) for constructing RS tests that are robust to local misspecification to GMM-based tests. Finally, a further generalization for general estimating and testing functions is developed. This framework encompasses both likelihood and GMM-based results.

Deprivation and the Dimensionality of Welfare: A Variable‐Selection Cluster‐Analysis Approach

In this paper we tackle the problems of dimensionality of welfare and that of identifying the multidimensionally poor by first finding the poor using the original space of attributes, and then reducing the welfare space. The starting point is the notion that the ‘poor’ constitutes a group of individuals that are essentially different from the ‘non-poor’ in a truly multidimensional framwework. Once this group has been identified, we propose reducing the dimension of the original welfare space by solving the problem of finding the smallest set of attributes that can reproduce as accurately as possible the ‘poor/non-poor’ classification in the first stage.

Tests for skewness and kurtosis in the one-way error component model

This paper derives tests for skewness and kurtosis for the panel data one-way error component model. The test statistics are based on the between and within transformations of the pooled OLS residuals, and are derived in a moment conditions framework. We establish the limiting distribution of the test statistics for panels with large cross-section and fixed time-series dimension. The tests are implemented in practice using the bootstrap. The proposed methods are able to detect departures away from normality in the form of skewness and kurtosis, and to identify whether these occur at the individual, remainder, or both error components. The finite sample properties of the tests are studied through extensive Monte Carlo simulations, and the results show evidence of good finite sample performance.

The Distributive Effects of Education: An Unconditional Quantile Regression Approach

We use recent unconditional quantile regression methods (UQR) to study the distributive eects of education in Argentina. Standard methods usually focus on mean effects, or explore distributive effects by either making stringent modeling assumptions, and/or through counterfactual decompositions that require several temporal observations. An empirical case shows the exibility and usefulness of UQR methods. Our application for the case of Argentina shows that education contributed positively to increased inequality in Argentina, mostly due to the effect of strongly heterogeneous effects of education on earnings.

Testing for serial correlation in hierarchical linear models

This paper proposes a simple hierarchical model and a testing strategy to identify intra-cluster correlations, in the form of nested random effects and serially correlated error components. We focus on intra-cluster serial correlation at different nested levels, a topic that has not been studied in the literature before. A Neyman’s C(α) framework is used to derive LM-type tests that allow researchers to identify the appropriate level of clustering as well as the type of intra-group correlation. An extensive Monte Carlo exercise shows that the proposed tests perform well in finite samples and under non-Gaussian distributions.

A New Robust and Most Powerful Test in the Presence of Local Misspecification

ABSTRACT This article proposes a new test that is consistent, achieves correct asymptotic size, and is locally most powerful under local misspecification, and when any -estimator of the nuisance parameters is used. The new test can be seen as an extension of the Bera and Yoon (1993) procedure that deals with non maximum likelihood (ML) estimation, while preserving its optimality properties. Similarly, the proposed test extends Neymans (1959) C(α) test to handle locally misspecified alternatives. A Monte Carlo study investigates the finite sample performance in terms of size, power, and robustness to misspecification.

Testing for Persistence in the Error Component Model: A One-Sided Approach

This article proposes new simple testing procedures for the joint null hypothesis of absence of persistent effects, in the form of random effects and first-order serial correlation in the error component model. The fact that the presence of random effects is clearly of a one-sided nature, together with the fact that in many empirical applications researchers worry about positive serial correlation leaves room for a power gain that arises from restricting the parameter space under the alternative hypothesis, compared to existing procedures that allow for two-sided alternatives. A Monte Carlo experiment shows that the proposed statistics have good size and power performance in very small samples like those typically used in applied work in panel data. An empirical example illustrates the usefulness of the proposed statistics.