We describe some recent developments in PcGets, and consider their impact on its performance across different (unknown) states of nature. We discuss the consistency of its selection procedures, and examine the extent to which model selection is non-distortionary at relevant sample sizes. The problems posed in judging performance on collinear data are noted. We also describe how PcGets has been extended to assist non-experts in model formulation, handle more variables than observations and tackle non-linear models.

The recent controversy over model selection in the context of ‘growth regressions ’ has led to some remarkably numerous ‘estimation ’ strategies, including 4 million regressions by Sala-i-Martin (1997b). Only one regression is really needed, namely the general unrestricted model, appropriately reduced to a parsimonious encompassing congruent representation. Such an outcome was achieved in one run on PcGets, within 15 minutes of receiving from Professor Ley the data set in Fernández et al (2001). We reproduce that equation, and corroborate the findings in Hoover and Perez (2004), who also adopt an automatic general-to-simple approach.

Most social science disciplines seek an interaction between theoretical ideas and empirical evidence,

When the DGP is nested in the model, PcGets delivers high performance selection across different (unknown) states of nature. One of its steps involves sub-sample post-selection assessment, and here we consider its properties and investigate its practical application. The simulation results show that conditional on retaining a variable, sub-sample information cannot discriminate between substantive and adventitious significance. The Monte Carlo experiments also reveal that the sub-sample selection method suggested by Hoover and Perez (1999) is dominated by procedures selecting only on full-sample evidence, when both approaches are evaluated at a given size. Nevertheless, although the sub-sample procedures do not result in a genuinely beneficial trade-off between size and power, they are particularly successful in controlling the size for selection problems that were previously deemed almost intractable.

Human capital plays an important role in the theory of economic growth, but it has been difficult to measure this abstract concept. We survey the psychological literature on cross-cultural IQ tests and conclude that modern intelligence tests provide one useful measure of human capital. Using a new database of national average IQ along with a methodology derived from Sala-i-Martin [1997a], we show that in growth regressions that include only robust control variables, IQ is statistically significant in 99.7 % of these 1330 regressions. A 1 point increase in a nation’s average IQ is associated with a persistent 0.16 % annual increase in GDP per capita.

We examine the properties of automatic model selection, as embodied in PcGets, and evaluate its performance across different (unknown) states of nature. After describing the basic algorithm and some recent changes, we discuss the consistency of its selection procedures, then examine the extent to which model selection is non-distortionary at relevant sample sizes. The problems posed in judging performance on collinear data are noted. The conclusion notes how PcGets can handle more variables than observations, and hence how it can tackle non-linear models.

Abstract: This paper reviews and compares twenty-one different model selection algorithms (MSAs) representing a diversity of approaches, including (i) information criteria such as AIC and SIC; (ii) selection of a “portfolio ” or best subset of models; (iii) general-to-specific algorithms, (iv) forward-stepwise regression approaches; (v) Bayesian Model Averaging; and (vi) inclusion of all variables. We use coefficient unconditional mean-squared error (UMSE) as the basis for our measure of MSA performance. Our main goal is to identify the factors that determine MSA performance. Towards this end, we conduct Monte Carlo experiments across a variety of data environments. Our experiments show that MSAs differ substantially with respect to their performance on relevant and irrelevant variables. We relate this to their associated penalty functions, and a bias-variance tradeoff in coefficient estimates. It follows that no MSA will dominate under all conditions. However, when we restrict our analysis to conditions where automatic variable selection is likely to be of greatest value, we find that two general-tospecific MSAs, Autometrics, do as well or better than all others in over 90 % of the

This survey reviews both theoretical and empirical papers that examine the economic effects of labour mobility with a particular reference to intra-European migration. We address three broad sets of issues: firstly, the effect that immigration has on the host country’s labour market. Although the possible adverse effects that immigration can have on the wage and employment levels of natives are typically examined, immigration may also have a role to play in raising skill levels. This leads to the second broad issue: the effect of migration of a particular skill composition on the long-term (endogenous) growth of the host country. Finally, immigration can have a major economic impact on the source country. These effects can either be positive or negative depending on the interplay between the effects of growth, remittances and the brain drain.