international econometric journal
in Russian language
We provide a non-technical summary of most of the recent results that have appeared in the econometric literature on instrumental variables estimation for the linear regression model. Standard inferential methods, such as OLS, are biased and inconsistent when the regressors are correlated with the error term. Instrumental variables methods were developed to overcome this problem, but finding instruments of good quality is cumbersome in any given situation and empirical researchers are often confronted with weak instruments. We review most of the recent studies on weak instruments and point to several methods that have been proposed to deal with such instruments, including “frugal” IV alternatives that do not rely on observed instruments to identify the regression parameters in presence of regressor–error dependencies.
The essay discusses reasons of correlatedness of explanatory variables and errors in regression applications, consequences of this correlatedness, and the method of instrumental variables aimed to resolve this problem.
Derivations are offered for the LIML and the 2SLS estimators of single equations of the classical econometric simultaneous-equation system that differ from the usual ones. By assimilating both estimators to the method of moments, their essential similarities are highlighted. The LIML estimator is derived from a least-squares criterion that exploits the interpretation of the structural equation as an error-in-variables model, and the 2SLS estimator is obtained by an approximation that is asymptoically valid. The LIML estimator may be calculated via an iterative procedure that begins with the 2SLS estimator. The conventional derivations of the 2SLS estimator are also reviewed.
This essay briefly surveys optimal instrumentation in linear and nonlinear models, both cross-sectional and stationary time series. Examples of judicious construction of instruments are given.
This essay is a selective guide to the literature on weak instruments. We use simple models to illustrate the problems raised by weak instruments and the suggestions that have been made to solve them. Because the literature is one that is still developing we only briefly touch on some recent developments.
We take a decision-theoretic view on the question of how to use instrumental variables and method of moments. Since prior beliefs play an inevitably strong role when instruments are possibly “weak”, or when the number of instruments is large relative to the number of observations, it is important in these cases to report characteristics of the likelihood beyond the usual IV or ML estimates and their asymptotic (i.e. second-order local) approximate standard errors. IV and GMM appeal because of their legitimate claim to be convenient to compute in many cases, and a (spurious) claim that they can be justified with few “assumptions”. We discuss some approaches to making such a claim more legitimately.
report contains impressions of a participant of the Canadian Econometric Study
Group meeting held in October,
report contains impressions of a participant of the Midwest Econometric Study
Group meeting held in
We evaluate the influence of interests of regional energy companies, consumers and governors on electricity regulation policy in Russian regions. We use panel data on electricity tariffs and electricity consumption in 77 regions during 1998–2003. We find evidence that governors tended to “freeze” the tariffs during governor elections, and that energy-intensive enterprises prefer buying electricity at the federal wholesale market to bargaining over lower tariffs with regional regulators. We conclude that regulation in 1998–2003 was socially oriented rather than was protecting interests of industrial consumers. However, an observed tendency of elimination of cross-subsidization is an evidence of reduction in social functions of electricity. Taking into consideration a high poverty level of population it is necessary to reconsider the norms of setting marginal tariffs so as to separate social and economic parameters of regulation.
aim of this paper is to explore the purchasing power parity between the
According to the Mehlum–Moene–Torvik model, the influence of institutions and natural resources on the level of GDP is ambiguous: depending on the value of a threshold function of institutional quality and natural resource endowment, the economy may be in one of the two equilibrium types – producer equilibrium or grabber equilibrium. In a grabber equilibrium, growth is negatively impacted by resource endowment and positively by institutions; in a producer equilibrium, more resources fosters economic growth, while institutions have no effect at all. Even though the empirical analysis supports the main result, the estimated specification does not fully correspond to the theoretical model. In this paper, we propose a different empirical testing strategy, more adequate to the Mehlum–Moene–Torvik model: the threshold function depends on both resources and institutions, and the regression specification more precisely reflects the influence of institutions and resources on the GDP growth rate. The econometric specification is a two-regime threshold regression, where a threshold value is also estimated. We show that the implications of the theoretical model are fully confirmed in the producer equilibrium, and only partly in the grabber equilibrium. We also discuss and compare various threshold and linear regression specifications.