A note on model selection in (time series) regression models – general-to-specific or specific-to-general?

This article provides Monte Carlo evidence on the performance of general-to-specific and specific-to-general selection of explanatory variables in linear (auto)regressions. In small samples the former is markedly inefficient in terms of ex-ante forecasting performance.     

Aggregate behavior and microdata

It is shown how one can effectively use microdata in modelling the change over time in an aggregate (e.g. mean consumption expenditure) of a large and heterogeneous population. The starting point of our aggregation analysis is a specification of explanatory variables on the micro-level. Typically, some of these explanatory variables are observable and others are unobservable. Based on certain hypotheses on the evolution over time of the joint distributions across the population of these explanatory variables we derive a decomposition of the change in the aggregate which allows a partial analysis: to isolate and to quantify the effect of a change in the observable explanatory variables. This analysis does not require an explicit treatment of the unobservable variables.     

Macroeconomic environment,country risk and stock market performance: Evidence for Brazil

The paper aims at providing empirical evidence about (i) the influence of macroeconomic variables and economic policies on country risk and (ii) the influence of macroeconomic variables and country risk on the main Brazilian index of the stock market (Ibovespa). The study analyzes the role that macroeconomic fundamentals plays, but also the role that the credibility of the regime of inflation targeting and the reputation of the central bank play in lessening country risk and in the improvement of the stock market performance. The empirical evidence was obtained through the application of ordinary least squares (OLS), generalized method of moments (GMM) and GMM systems. The results found suggest that monetary policy and public debt management, as well as credibility and reputation affect country risk and the performance of the Brazilian stock market.     

Functional sensitivity of testing the environmental Kuznets curve hypothesis

Conventional tests for the environmental Kuznets curve (EKC) hypothesis mostly apply a quadratic equation in modeling the non-linear relationship between environmental indices (such as air pollutants) and welfare measures (such as income per capita). If their inverted-U shaped pattern is empirically accepted with two significant regressors, the income per capita and its square transformation, the EKC hypothesis is supported. Using an OECD sample, this paper shows that the validity of testing the EKC hypothesis is sensitive to how we transform income non-linearly in sulfur and carbon EKC regressions. This paper carries out experiments on different powers of γ for transforming income non-linearly and concludes that only when 0 < γ < 1 and 1 < γ < 2 will the EKC regression demonstrate a testable non-linear cointegration relationship between the two air pollutants and income per capita. In the generalized EKC regressions estimated in this paper, although we find sulfur and carbon EKC patterns in the OECD sample, none of the EKC regressions using different γ is a cointegrating equation. This finding implies an inside critique to the EKC literature that failure of cointegration of the conventional EKC regression is not because of using the quadratic functional form, but because of the fundamentally spurious relationship between the trends of pollutants and income levels.     

Functional sensitivity of testing the environmental Kuznets curve hypothesis

Conventional tests for the environmental Kuznets curve (EKC) hypothesis mostly apply a quadratic equation in modeling the non-linear relationship between environmental indices (such as air pollutants) and welfare measures (such as income per capita). If their inverted-U shaped pattern is empirically accepted with two significant regressors, the income per capita and its square transformation, the EKC hypothesis is supported. Using an OECD sample, this paper shows that the validity of testing the EKC hypothesis is sensitive to how we transform income non-linearly in sulfur and carbon EKC regressions. This paper carries out experiments on different powers of γ for transforming income non-linearly and concludes that only when 0 < γ < 1 and 1 < γ < 2 will the EKC regression demonstrate a testable non-linear cointegration relationship between the two air pollutants and income per capita. In the generalized EKC regressions estimated in this paper, although we find sulfur and carbon EKC patterns in the OECD sample, none of the EKC regressions using different γ is a cointegrating equation. This finding implies an inside critique to the EKC literature that failure of cointegration of the conventional EKC regression is not because of using the quadratic functional form, but because of the fundamentally spurious relationship between the trends of pollutants and income levels.     

Uncertainty of outcome versus reputation: Empirical evidence for the First German Football Division

This paper deals with the determinants of match attendance in the German premier football league. We analyse uncertainty measures of match outcome as well as uncertainty of championship outcome. Furthermore, we incorporate supporter clubs, reputation, performance measures and weather effects as explanatory variables. Due to the limited capacity of the stadiums, observations on attendance are right censored in our sample. While other authors use the ordinary least squares estimator, which is inconsistent in this framework, we take this restriction implicitly in consideration by using a Tobit model. In conclusion, we show that reputation and goodwill are more important for attendance levels than the thrill of outcome uncertainty.¹ First version received: September 1999/Final version received: January 2001     

A more general class of biased estimators of the co-efficients in linear regression

In a recent paper, Ullah and Ullah (1978) proposed a class of biased estimators, namely double k-class (k₁, k₂) for the coefficients in a linear regression model. Even though, this set of estimators contains James and Stein (1961) as a special case, in its present form, it does not contain the ridge type estimators. The aim of this note is to extend Ullah and Ullah set of estimators and then establish a relationship with the various operational ridge estimators. The conditions under which the extended set of estimators dominates the ordinary least squares estimator are analyzed.     

Censored quantile regressions and the length of unemployment periods in West Germany

In this paper, we estimate the effect of different macro and micro variables on the distribution of unemployment duration in West Germany using censored quantile regressions. We analyze unemployment periods of more than 91,000 observations from the years 1981 to 1997 drawn from the IAB employment subsample. The latter is an administrative data set that is representative with respect to the socially insured workforce. Surprisingly, we find that the educational degree and variables indicating the macroeconomic environment such as the unemployment rate have a weak effect only. On the other hand, variables reflecting the (un-)employment history of an individual such as the length of tenure, recall to the same employer in the past, recent unemployment, and the position in the population income distribution before unemployment have the strongest effects on unemployment duration. We conclude that work history variables are the ones most suitable in characterizing the unemployment duration of an individual. From a methodological point of view, it is interesting that some regression coefficients have a different sign depending on the quantiles of the unemployment duration distribution. This clearly is a violation of the classical proportional hazard assumption which is very common in unemployment duration analysis.     

Spatial econometric Monte Carlo studies: raising the bar

We discuss Monte Carlo methodology that can be used to explore alternative approaches to estimating spatial regression models. Our focus is on models that include spatial lags of the dependent variable, e.g., the SAR specification. A major point is that practitioners rely on scalar summary measures of direct and indirect effects estimates to interpret the impact of changes in explanatory variables on the dependent variable of interest. We argue that these should be the focus of Monte Carlo experiments. Since effects estimates reflect a nonlinear function of both \(\beta \) and \(\rho \), past studies’ focus exclusively on \(\beta \) and \(\rho \) parameter estimates may not provide useful information regarding statistical properties of effects estimates produced by alternative estimators. Since effects estimates have recently become the focus of inference regarding the significance of (scalar summary) direct and indirect impacts arising from changes in the explanatory variables, empirical measures of dispersion produced by simulating draws from the (estimated) variance–covariance matrix of the parameters \(\beta \) and \(\rho \) should be part of the Monte Carlo study. An implication is that differences in the quality of estimated variance–covariance matrices arising from alternative estimators also plays a role in determining the accuracy of inference. An applied illustration is used to demonstrate how these issues can impact conclusions regarding the performance of alternative estimators.     

Analysis of the determinants of sports participation in Spain and England

This article investigates the role of sociodemographic characteristics, educational and economic variables on sports participation in a comparative way in two European countries: Spain and England. Adopting a broad concept of sport, as in the common European approach, we analyse the determinants of sports participation in 40 different professional and nonprofessional sports and recreational activities in both countries. The research involves a comparative analysis between the data of England and Spain based on two logistic regressions. The regression equation of every country tests the effect of 17 binary explanatory variables on a dependent binary variable for participation. Higher education level, professional occupation, younger age and being male are all factors associated with more sports participation. Although there is no difference in the direction of the factor effects on participation between England and Spain, there is considerable variation in their relative strength, which has sport policy implications in the two nations.     

The effect of gender and race on student performance in principles of economics: the importance of personality type

This article analyses motion picture box-office gross revenue using a cross-section of films from 1997 to 2001. The dependent variable is total domestic box-office revenue. The independent variables investigated include: production budget; peak number of screens that the film was shown on in theaters; Consumer price index for movie tickets; personal income; season and year of the release in theaters; a measure of pre-existing audience; aggregate critic rating; MPAA rating; genre; word-of-mouth recommendation; the presence of popular stars and the award nominations. A distinction is made in the analysis between information available to the public prior to the release of the film in theaters (ex ante) and information available to the public after the film opens in theaters (ex post). Results for the ex ante ordinary least squares (OLS) regression reveal positive impacts of budget, summer and holiday release dates, critical reviews, sequels and several genres on gross revenue. Significant, positive determinants in the ex post OLS regressions include budget, the peak number of screens, sequels, critical reviews, summer and holiday releases, word-of-mouth, award nominations and star power.     

On the median regression for SURE models with applications to 3-generation immigrants data in Sweden

In this paper we generalize the median regression method to be applicable to system of regression equations, in particular SURE models. Giving the existence of proper system wise medians of the residuals from different equations, we apply the weighted median regression with the weights obtained from the covariance matrix of the equations obtained from ordinary SURE method. The benefit of this model in our case is that the SURE estimators utilise the information present in the cross regression (or equations) error correlation and hence more efficient than other estimation methods like the OLS method. The Seemingly Unrelated Median Regression Equations (SUMRE) models produce results that are more robust than the usual SURE or single equations OLS estimation when the distributions of the dependent variables are not normally distributed or the data are associated with outliers. Moreover, the results are also more efficient than is the cases of single equations median regressions when the residuals from the different equations are correlated. A theorem is derived and indicates that even if there is no statistically significant correlation between the equations, using SUMRE model instead of SURE models will not damage the estimation of parameters.     

Economic Development Indicators as Determinants of Medal Winning at the Sydney Olympics: An Extreme Bounds Analysis

This paper examines the variables that determine the performance of countries at the Olympic Games as measured by a weighted sum of the medals won at the Sydney 2000 Games. While previous studies have identified the importance of a country's economic size and the resources available to sport, this paper examines nine more variables including the number of athletes representing each nation and some development indicators. Based on 2310 regressions, both traditional and restricted extreme bounds analysis show that only two variables are robust: the number of athletes and national expenditure on health. Thus, the final model recognises four explanatory variables that include these two as well as GDP and population.     

Economic,psychological, and sociological determinants of voluntary turnover

Economists, psychologists, and sociologists have all contributed to the understanding of voluntary labor turnover. We argue for an integrated explanatory model which incorporates variables from each perspective. Such a model is presented and then estimated. Data from a cohort of 135 recently hired registered nurses employed by a university hospital are analyzed to assess the effects of the various explanatory variables on turnover during one year of employment. Turnover is measured by organization records for 12 months following the administration of the questionnaire designed to measure the independent variables. The integrated model portrays the work conditions, environmental conditions, and employee characteristics as primarily affecting turnover by impacting on the intervening variables of job satisfaction, organizational commitment and intent to stay. Ordinary least square (OLS) regression and logistic regression are used in the analysis. The data indicate support for sociological, economic, and psychological determinants of voluntary turnover. These findings are discussed from the perspective of Etzioni's claims about the importance of the moral dimension for explaining economic behavior such as turnover.     

Military Spending, Government Disarray, and Economic Growth: A Cross-Country Empirical Analysis

Recent years have seen a spate of empirical studies that have used cross-country regressions to examine a variety of possible determinants of long-term economic growth. The present study considers an additional, largely overlooked, explanatory variable: military spending. Consistent with Thompson’s (1974) hypothesis that enhanced national defense fosters economic growth by increasing the security of property rights, the military expenditures share of GDP is found to have a statistically significant positive impact on the growth rate of per capita GDP. This result is obtained for two alternative model specifications, a Barro-regression and a LISREL variant of that regression. The LISREL variant is motivated by Sala-i-Martin’s (1994) claim that the impact of government economic policies jointly, rather than extant government policies individually and separately, is "the phenomenon that really matters" for long-term economic growth.     

The societal burden of pain in Germany: Health-related quality-of-life,health status and direct medical costs

Abstract

Objectives:

The purpose of this paper is to estimate the impact of the severity and frequency of pain on health-related quality-of-life (HRQoL), self-reported health status, and direct medical costs in Germany.

Methods:

Data are from the internet-based 2010 National Health and Wellness Survey (NHWS). Estimates of the impact of pain experience are generated by a series of regression models. In the case of HRQoL the physical and mental summary scores from the SF-12, together with SF-6D utilities, are evaluated within an ordinary least squares framework. Health status is assessed through an ordered logit model. Direct medical costs are estimated through a semi-logarithmic healthcare cost function. Socioeconomic characteristics, health risk behaviors, and the Charlson Comorbidity Index (CCI) are introduced as control variables in all regressions.

Results:

An estimated 23.96% of the adult German population (16.39 million) reported experiencing pain in the last 30 days. Of these 13.16% reported severe pain. The experience of frequent severe and moderate pain has a significant deficit impact on HRQoL. For those experiencing severe daily pain, the deficit in the SF-12 physical component score (PCS) is ?17.930 (95% CI: ?18.720 to ?17.140), the SF-12 mental component score (MCS) is ?8.787 (05% CI: ?9.857 to ?7.716), and SF-6D absolute utilities ?0.201 (95% CI: ?0.214 to ?0.188); with self-reported health status the deficit impact of severe daily pain is also substantial (OR?=?29.000; 95% CI: 23.000–36.580). In the case of direct medical costs severe daily pain increases healthcare provider costs by 101.6% and total direct costs by 123.9%.

Limitations:

The NHWS is an internet survey. The principal limitation is that as a self-report there is no separate validation of pain severity or chronicity.

Conclusions:

The experience of pain has a substantial negative impact on HRQoL, health status, and resource utilization in Germany. If pain is considered as a disease in its own right, the experience of chronic pain presents policy-makers with a major challenge.     

THE SIZE DISTRIBUTION OF FAMILY INCOME IN U.S. SMSAs, 1959*

An attempt is made in this paper to identify and quantify the relative influence of several economic, social, and demographic factors on variations in the size distribution of family incomes in 208 Standard Metropolitan Statistical Areas (SMSAs) in the United States in 1959. Using a simple ordinary least squares model with Gini's concentration ratio (R) as the proxy for family income inequality, the estimating equations explain up to 89 percent of the SMSA-to-SMSA variation. The “best” explanatory variables are those having to do with size of nonwhite population, occupational structure, and median years of education. City size and region—which are represented by dummy variables—are also revealed as playing an important role, both on their own and in conjunction with other of the independent variables.     

Dimension reduction and model averaging for estimation of artists' age-valuation profiles

In hedonic regression models of the valuation of works of art, the age or period at which an artist produces a particular work is often found to have highly significant predictive value. Most existing results are based on regressions that pool many painters. Although the uniqueness of artists' career paths makes it interesting to estimate such regressions for individual artists, sample sizes are often inadequate for a model that would also include the large number of other relevant variables. We address this problem of inadequate degrees of freedom in individual artist regressions by using two statistical methods (model averaging and dimension reduction) to incorporate information from a potentially large number of predictor variables, despite relatively small samples. We find that individual age-valuation profiles can differ substantially from general pooled profiles, suggesting that methods that are more responsive to the unique features of individual artists may provide better predictions of art valuations at auction, and may be useful more generally in hedonic valuation problems.     

Financial frictions and investment: requiem in Q

Understanding the nature of financial frictions faced by firms is relevant for both monetary and fiscal policy experiments. Empirical investment studies commonly find that proxies for firms' internal funds are significant as explanatory variables, particularly in the Q-theory based regression framework. These findings are often interpreted as evidence of financial frictions. This paper investigates that inference by specifying and estimating a class of dynamic optimization models where imperfectly competitive firms face financial constraints. Market power induces the principal link between investment and internal funds. We find no evidence to support the argument that capital market imperfections contribute to the relationship between investment and profitability.     

Defining governance matters: A factor analytic assessment of governance institutions

This paper produces a new set of governance indices that enable estimation of the marginal economic benefits of particular categories of governance. The indices are constructed using a factor model, estimated on forty perceptions-based governance variables from eleven data sources. In contrast to previous research, a unified statistical framework is used to determine the number of governance indices to create, the conceptual content of each index, and the relative importance each has in accounting for the observable data. The four indices are labeled market infrastructure, downside governance risk, order, and civil liberties. Confidence intervals are calculated for each country, for each index to aid in the comparison of scores. As an application, the indices are used as explanatory variables in cross-country income regressions. Instrumental variables estimates show that market infrastructure and civil liberties both exert statistically and economically significant effects on per capita income, controlling for geography (malaria risk) and international trade.