Omitted Variables Bias Research Articles

Trade Dependence or Size Dependence?

The gravity model of trade states that the volume of trade between two countries is proportional to the product of the sizes of the two countries and the inverse of the distance between them. The gravity model, however, was initially suggested for other types of social interactions, and it also predicts well the probability of militarized disputes. This raises a concern about omitted-variable bias in the literature on trade and conflict: if the gravity model is correlated with both trade and conflict, it is necessary to control for the size of both countries in the dyad. In this article, I analyze the nature of this bias both empirically and theoretically. I find that a correct model specification leads to a stronger relationship between trade and conflict than in the traditional models. However, the empirical finding must be reinterpreted: what is crucial is not the relationship between the value of the bilateral trade and the size of the larger economy in the dyad (`trade dependence'), but rather `trade efficiency'—the extent to which the two countries trade more than expected from the gravity model.

THE HEALTH EFFECTS OF MILITARY SERVICE: EVIDENCE FROM THE VIETNAM DRAFT

There is a strong cross‐sectional association between military service and adverse health. However, veterans differ very significantly in their observable characteristics from nonveterans, suggesting that some of the association between military service and adverse health may be due to omitted variables bias. To address this problem, we use draft eligibility as an instrument for military service. Despite a very strong first‐stage relationship between draft eligibility and military service, the two‐stage least squares estimates of the difference in health between veterans and nonveterans are statistically insignificant and too imprecise to preclude the differences in health found in the cross‐sectional regressions. (JEL H56, I12, I18)

Proxying Ability by Family Background in Returns to Schooling Estimations is Generally a Bad Idea*

AbstractWhen schooling is measured with error and data on ability are lacking, return to schooling estimates will be subject to positive omitted variable bias (OVB) and negative measurement error bias (MEB). We investigate how these biases are affected when ability is proxied by family background variables. We show that the effect on OVB is uncertain, while MEB invariably increases in magnitude. Our empirical analysis demonstrates that MEB generally dominates OVB. With more background variables or increased measurement error, the total bias rapidly becomes negative and increasing in magnitude, thereby driving the return estimate further and further away from the true value.

The Principal Components of Growth in the Less Developed Countries

SUMMMARYThis paper re‐examines the international evidence on the sources of growth in less developed countries (LDCs) using exploratory factor analysis (EFA). Although EFA was first used in the development context by Adelman and Morris (1967) it has rarely been used since, despite being ideally suited to a context in which a large number of latent factors have been hypothesized to determine growth, and in which an even greater number of imperfectly measured and multicollinear proxies have been used to measure these latent factors. This paper uses EFA to minimize these problems of omitted variables biases, multicollinearity and measurement error, by reducing a large array of hypothesized growth determinants into a parsimonious and non‐collinear set of composite indices. The paper then provides theoretical interpretations of the derived indices, tests their statistical significance and quantitative importance in otherwise conventional growth regressions, and uses these results to reappraise the usefulness of cross‐country empirics in deriving robust, policy‐relevant knowledge of the principal components of growth in LDCs, including the so called ‘economic miracles’.

Do child care centers benefit poor children after school entry?

Attendance in preschool centers can yield short-term benefits for children from poor or middle-class families. Yet debate persists in Europe and the United States over whether centers yield gains of sufficient magnitude to sustain children's cognitive or social advantages as they move through primary school. We report on child care and home environments of 229 children in the US who were 2½ years of age (on average) at entry to the study. Among children attending a center at 2½ or 4½ years of age, cognitive proficiencies were significantly higher at 7½ years of age, compared with children in home-based care, after taking into account prior proficiency levels, maternal attributes, and other covariates. No relationship between center attendance and social development, positive or negative, was detected at 7½. A priori selection factors modestly helped to explain the likelihood that mothers enrolled their child in a center. But associations between center exposure and higher cognitive proficiency at age 7½ remained after controlling for selection factors and testing for omitted variables bias.

How Many Hours Would You Want to Work a Week? Job Quality and the Omitted Variables Bias in Labour Supply Models: A Longitudinal Analysis of Working Time Preference Change

This paper sets out to provide an understanding of how individuals form their preferences over the extent of their paid work involvement - their working time preferences - in different work environments and societal contexts. The main objective of the empirical analysis is to investigate how preferences of this kind are constructed at the individual level and adapted over time following changes in work - and family-related circumstances. The consideration of the Old and New Lander of Germany as cases for comparative analysis allows for a test of common factors in different contexts of economic conditions and gender relations. The empirical findings from a longitudinal analysis of the German Socio-Economic Panel (1993-2003) run counter to the predictions of neoclassical labour supply theory. This owes to a fundamental difference in terms of theoretical approach. While (most) economists tend to view paid work in instrumental terms - as something that people perform only for its monetary rewards, this study takes account of intrinsic work rewards as central determinants of work motivation. We find the qualitative experience of work to exert an independent influence on individuals' preferences over work hours, and therefore argue for the inclusion of work quality as a central factor in labour supply decisions.

Foreign Direct Investment and Structural Reforms: Evidence from Eastern Europe and Latin America

This paper investigates the role of structural reforms - privatization, financial reform and trade liberalization- as determinants of FDI inflows based on newly constructed dataset on structural reforms for 19 Latin American and 25 Eastern European countries between 1989 and 2004. Our main finding is a strong empirical relationship from reforms to FDI, in particular, from financial liberalization and privatization. These results are robust to different measures of reforms, split samples, and potential endogeneity and omitted variables biases.

The Impact of Workplace Conditions on Firm Performance

This paper estimates the impact of work environment health and safety practice on firm performance, and examines which firm-characteristic factors are associated with good work conditions. We use Danish longitudinal register matched employer-employee data, merged with firm business accounts and detailed cross-sectional survey data on workplace conditions. This enables us to address typical econometric problems such as omitted variables bias or endogeneity in estimating i) standard production functions augmented with work environment indicators and aggregate employee characteristics and ii) firm mean wage regressions on the same explanatory variables. Our findings suggest that improvement in some of the physical dimensions of the work health and safety environment (specifically, internal climate and repetitive and strenuous activity) strongly impacts the firm productivity, whereas internal climate problems are the only workplace hazards compensated for by higher mean wages.

Biases of OLS and Spatial Lag Models in the Presence of an Omitted Variable and Spatially Dependent Variables

A number of authors have suggested that omitted variables affect spatial regression methods less than ordinary least-squares (OLS). To explore these conjectures, we derive an expression for OLS omitted variable bias in a univariate model with spatial dependence and show that positive dependence in the disturbances, regressand, and regressor magnifies the magnitude of conventional omitted variables bias. Moreover, we show that spatial dependence in the regressor exacerbates the usual bias that occurs when using OLS to estimate a spatial autoregressive data generating process (DGP).

Does advanced medical technology encourage hospitalist use and their direct employment by hospitals?

In the United States, inpatient medical care increasingly encompasses the use of expensive medical technology and, at the same time, is coordinated and supervised more and more by a rapidly growing number of inpatient-dedicated physicians (hospitalists). In the production of inpatient care services, Hospitalist services can be viewed as complementary to sophisticated and expensive medical equipment in the provision of inpatient medical care. We investigate the causal relationship between a hospital's access to three types of sophisticated diagnostic and therapeutic medical equipment - intensity-modulated radiation therapy, gamma knife, and multi-slice computed tomography - and its likelihood of using hospitalists. To rule out omitted variables bias and reverse causality, we use technology-specific Certificate of Need regulation to predict technology use. We find a strong positive association, yet no causal link between access to medical technology and hospitalist use. We also study the choice of employment modality among hospitals that use hospitalists, and find that access to expensive medical technology reduces the hospital's propensity to employ hospitalists directly.

Heightened mortality after the death of a spouse: Marriage protection or marriage selection?

We test whether the heightened mortality after the death of a spouse represents correlation or causation by examining the heterogeneity in the bereavement effect based on the spouse's cause of death. Some causes of death are correlated with socioeconomic characteristics while others are not. Equality in the bereavement effect across these two types of death would signal a causal relationship while no bereavement effect for uncorrelated causes of death would indicate an omitted variables bias. Results indicate that the observed effect for women is subject to an omitted variables bias but the estimates for men indicate a causal relationship.

Interdependence in Comparative Politics

Interdependence is ubiquitous and often central across comparative politics. Indeed, as the authors show first, theoretically, any situation involving externalities from one unit's actions on others' implies interdependence. Positive or negative externalities induce negative or positive interdependence, which spurs competitive races or free riding, with corresponding early or late-mover advantages, and thus strategic rush to act or delay and inaction. The authors show next how to model interdependent processes empirically, how not doing so risks omitted-variable biases favoring domestic and exogenous-external accounts over interdependence, but how doing so naïvely risks simultaneity biases with the opposite substantive implications. They then discuss how to estimate properly specified interdependence models and, finally, how to interpret and present estimated spatio-temporally dynamic effects, response paths, and long-run steady-states, with associated standard errors. They illustrate by replicating a noteworthy earlier, nonspatial study of capital tax competition. Web appendices contain further technical details, literature survey, data, statistical software code, and spreadsheet templates for all suggested procedures.

Do Differences in Legal Protections Explain Differences in Ownership Concentration?

Existing research on cross-country differences in ownership concentration analyzes country averages of ownership concentration instead of firm-level data. This creates omitted-variable and aggregation biases. Most papers also use small samples of large firms. This makes inferences to country populations problematic because ownership concentration is inversely related to firm size and firm size varies across countries. When I analyze firm-level observations; control for firm-level determinants of ownership concentration, including size; and use a broad sample of firms from 32 countries, there is no support for the widely held theory that large shareholders are a response to weak legal protections for investors.

Foreign Direct Investment and Structural Reforms: Evidence from Eastern Europe and Latin America

This paper investigates the role of structural reforms –privatization, financial reform and trade liberalization– as determinants of FDI inflows based on newly constructed dataset on structural reforms for 19 Latin American and 25 Eastern European countries between 1989 and 2004. Our main finding is a strong empirical relationship from reforms to FDI, in particular, from financial liberalization and privatization. These results are robust to different measures of reforms, split samples, and potential endogeneity and omitted variables biases.

Foreign Direct Investment and Structural Reforms: Evidence From Eastern Europe and Latin America

This paper investigates the role of structural reforms -financial reforms, trade liberalization, and privatization- as determinants of FDI inflows based on newly constructed dataset on structural reforms for 19 Latin American and 25 Eastern European countries between 1989 and 2004. Our main finding is a strong empirical relationship from reforms to FDI, in particular, from financial liberalization and privatization. These results are robust to different measures of reforms, split samples, and potential endogeneity and omitted variables biases.

Salários e risco de acidentes de trabalho: evidências de diferenciais compensatórios para a indústria manufatureira

O objetivo deste trabalho é fornecer evidências empíricas para a hipótese de diferenciais compensatórios de salários para a indústria brasileira de transformação. O argumento básico é o de que indústrias que expõem seus trabalhadores a maiores riscos de acidentes de trabalho deveriam compensá-los com maiores salários. Os resultados obtidos neste trabalho corroboram os fatos estilizados da literatura, ou seja, a ausência do controle da heterogeneidade não observada dos trabalhadores subestima fortemente os valores dos prêmios compensatórios e, após tal controle, a hipótese de diferenciais compensatórios é amplamente corroborada.

Mispricing in Linear Asset Pricing Models

In the framework of a reduced form asset pricing model featuring linear-in-z betas and risk premiums with lagged macro instruments, I propose a clean measure of mispricing that is free from the omitted-variable bias due to either missing priced factors or missing instruments. Applying the model to U.S. stock returns for 1927-2005, I nd that momentum and contrarian strategies are related to the mispricing measure that does not vary with the macro variables. Moreover, a zero-dollar strategy intersecting the two eects is highly protable

Correcting for biases when estimating production functions: an illusion of the laws of algebra?

This paper shows that the endogeneity bias that allegedly appears when estimating production functions using value data, and which the literature has tried to deal with since the 1940s, is simply the result of omitted-variable bias due to a poor approximation to an accounting identity. This problem has no econometric solution. As a result, recent attempts to solve the problem by developing new estimators are questioned. The only possible way to estimate the technological parameters of the production function is to use physical quantities.

The degree of precautionary saving: A reexamination

Extending Dynan's methodology [Dynan, K., 1993. How prudent are consumers? Journal of Political Economy 101, 1104–1113], we show that a significant fraction of the prudence parameter puzzle can be explained by a downward omitted-variable bias. Further, the estimated prudence is substantially higher for liquidity-constrained households.

Finding an Impact of Preservation Policies: Price Effects of Historic Landmarks on Attached Homes in Chicago, 1990-1999

The impact of landmark designation on prices of the property and its neighbors sits at the core of the policy debate and empirical research on historic preservation. Yet these studies suffer from serious methodological limitations and biases. First, as important unobserved characteristics likely correlate with landmark designation, an omitted-variable bias results. Second, if designations depend on property values or neighborhood housing market conditions, the endogenous selection process further undermines inferences about preservation policies’ effects. This article outlines more robust empirical strategies and presents new evidence on landmark designation effects on property values. For a sample of Chicago home sales during the 1990s, a hedonic price analysis suggests that landmark buildings and districts sell at a small premium. To address the omitted-variable bias, a repeat-sales approach demonstrates significant spillover effects of landmark designation on prices. These estimates are also robust to sample-selection bias and some forms of spatial autocorrelation.