Georges Bresson

Fixed effects, random effects or Hausman-Taylor?: A pretest estimator

Abstract This paper suggests a pretest estimator based upon two Hausman tests as an alternative to the fixed effects or random effects estimators for panel data models. The bias and RMSE properties of these estimators are investigated using Monte Carlo experiments.

Comparison of forecast performance for homogeneous, heterogeneous and shrinkage estimators: Some empirical evidence from US electricity and natural-gas consumption

Abstract Maddala et al. [Journal of Business and Economic Statistics, 15 (1997) 90] obtained short-run and long-run elasticities of energy demand for each of 49 US states over the period 1970–1990. They showed that heterogeneous time series estimates for each state yield inaccurate signs for the coefficients, while panel data estimates are not valid because the hypothesis of homogeneity of the coefficients was rejected. Their preferred estimates are those obtained using the shrinkage estimator. This paper contrasts the out-of-sample forecast performance of heterogeneous, panel and shrinkage estimators using the Maddala et al. [Journal of Business and Economic Statistics 15 (1997) 90] electricity and natural gas data sets. Our results show that the homogeneous panel data estimates give the best out-of-sample forecasts.

THE MAIN DETERMINANTS OF THE DEMAND FOR PUBLIC TRANSPORT: A COMPARATIVE ANALYSIS OF ENGLAND AND FRANCE USING SHRINKAGE ESTIMATORS

This study analyses the impacts of changes in fares, service supply, income and other factors on the demand for public transport on the basis of panels of English counties and French urban areas. The analysis is based on dynamic econometric models, so that both short- and long-run elasticities are estimated. Conventional approaches (i.e. fixed- and random-effect models) rely on the hypothesis that elasticities are the same for all areas. Having shown that this hypothesis is not valid for these data sets, the heterogeneity amongst areas is accounted for using a random-coefficients approach, and Bayesian shrinkage estimators. Estimated elasticities for France and England are compared, by using a common set of variables, similar time period and a common methodology. The results show a considerable variation in elasticities among areas within each country. The major conclusion is that public transport demand is relatively sensitive to fare changes, so that policy measures aimed at fare reduction (subsidisation) can play a substantial role in encouraging the use of public transport, thus reducing the use of private cars.

Forecasting with spatial panel data

Various forecasts using panel data with spatial error correlation are compared using Monte Carlo experiments. The true data generating process is assumed to be a simple error component regression model with spatial remainder disturbances of the autoregressive or moving average type. The best linear unbiased predictor is compared with other forecasts ignoring spatial correlation, or ignoring heterogeneity due to the individual effects. In addition, the root mean squared error performance of these forecasts is examined under misspecification of the spatial error process, various spatial weight matrices, and heterogeneous rather than homogeneous panel data models.

Heterogeneous Labor and the Dynamics of Aggregate Labor Demand: Some Estimations Using Panel Data

In this article we pay attention to the conditions that make an aggregate labor demand equation consistent with the underlying model at a more disaggregated level when heterogeneity exists across firms or across workers. It is argued that this consistency rests on the condition that employment evolves in the same direction in all firms and for all skill levels.Moreover, it is shown empirically that even though the above condition is satisfied, satisfactory estimations of an aggregated model can hide misspecification problems that become apparent when one also estimates the underlying disaggregated models.

To Pool or Not to Pool

For panel data studies with large N and small T , it is usual to pool the observations, assuming homogeneity of the slope coefficients. The latter is a testable assumption which is quite often rejected. Moreover, with the increasing time dimension of panel data sets, some researchers including Robertson and Symons (1992), Pesaran and Smith (1995), and Pesaran, Smith and Im (1996) have questioned the poolability of the data across heterogeneous units. Instead, they argue in favor of heterogeneous estimates that can be combined to obtain homogeneous estimates if the need arises. Maddala, Trost, Li and Joutz (1997) on the other hand argue that the heterogeneous time series estimates yield inaccurate estimates and even wrong signs for the coefficients, while the panel data estimates are not valid when one rejects the hypothesis of homogeneity of the coefficients. If one is after reliable coefficient estimates, Maddala, Trost, Li and Joutz (1997) argue in favor of shrinkage estimators that shrink the heterogeneous estimators towards the pooled homogeneous estimator. Proponents of the homogeneous panel estimators have acknowledged the potential heterogeneity among the cross-sectional units, but have assumed that the efficiency

Dynamic Labour Demand Models

When firms face a change in their environment, they do not necessarily adjust immediately their level of employment to the new business conditions. Since the pioneering work of Oi [1962], it is well-known that labour is not a fully flexible production factor, due to the existence of adjustment costs, namely hiring and firing costs. In his article, Oi presented some evaluations of hiring and firing costs. From these evaluations, two important features can be noticed. First of all, hiring and firing costs were not equal. Oi evaluated average hiring costs to about three weeks of pay and average firing costs to be about two weeks of pay. Moreover, the discrepancy between costs associated with the hiring of skilled workers and those induced by hiring unskilled workers was large, whereas firing costs did not depend as much on the skill level of workers, having mainly institutional origins, such as explicit or implicit, legal or contractual, arrangements.

Crowding-Out Effects of Cruise Tourism on Stay-Over Tourism in the Caribbean: Non-Parametric Panel Data Evidence:

Based on a non-parametric panel data estimation, this study provides an econometric analysis of the relationship between stay-over and cruise tourism for 15 Caribbean islands over the period 1985–2004. It reveals heterogeneity of tourism flows across different destinations. It also emphasizes the crowding-out effects of cruise tourism on stay-over tourism. Using a static Cournot game framework, the authors show that, for most of the destinations, stay-over tourism and cruise tourism are strategic substitutes. Moreover, a ‘crowding-out trap’ seems to have emerged since the beginning of the 1990s and is tending to spread. Policymakers have a weak negotiating position vis-à-vis an oligopolistic cruise ship tourism industry, resulting in an uncontrolled and increasing share of cruise tourism flows. As the Caribbean authorities have managed their growth process in the belief that there is complementarity between cruise tourism and stay-over tourism, these results suggest that the strategic vision of a growth process based on tourist activity has to be revised.

ADAPTIVE ESTIMATION OF HETEROSKEDASTIC ERROR COMPONENT MODELS

ABSTRACT This paper checks the sensitivity of two adaptive heteroskedastic estimators suggested by Li and Stengos (1994) and Roy (2002) for an error component regression model to misspecification of the form of heteroskedasticity. In particular, we run Monte Carlo experiments using the heteroskedasticity setup by Li and Stengos (1994) to see how the misspecified Roy (2002) estimator performs. Next, we use the heteroskedasticity setup by Roy (2002) to see how the misspecified Li and Stengos (1994) estimator performs. We also check the sensitivity of these results to the choice of the smoothing parameters, the sample size, and the degree of heteroskedasticity. We find that the Li and Stengos (1994) estimator performs better under this type of misspecification than the corresponding estimator of Roy (2002). However, the former estimator is sensitive to the choice of the bandwidth.

A Robust Hausman–Taylor Estimator

This chapter suggests a robust Hausman and Taylor (1981), hereafter HT, estimator that deals with the possible presence of outliers. This entails two modifications of the classical HT estimator. The first modification uses the Bramati and Croux (2007) robust Within MS estimator instead of the Within estimator in the first stage of the HT estimator. The second modification uses the robust Wagenvoort and Waldmann (2002) two-stage generalized MS estimator instead of the 2SLS estimator in the second step of the HT estimator. Monte Carlo simulations show that, in the presence of vertical outliers or bad leverage points, the robust HT estimator yields large gains in MSE as compared to its classical Hausman–Taylor counterpart. We illustrate this robust version of the HT estimator using an empirical application.