High-dimensional Fixed Effects Research Articles

Institutions and bilateral trade in Africa: an application of Poisson’s estimation with high-dimensional fixed effects to structural gravity model

ABSTRACT This study provides a comparative analysis of the effect of political and economic institutions on bilateral trade of emerging and low-income African countries. Using the structural gravity model estimated by the Poisson pseudo-maximum likelihood with high-dimensional fixed effects (PPMLHDFE) estimator, the study reveals that both political and economic institutions are fundamental determinants of bilateral trade in Africa. However, the effect differs by type and components of the institutions and levels of income of the countries. Generally, the findings show that institutions have greater impact on trade of the emerging African economies (EE) than the low-income countries (LICs). Relatively, economic institutions affect trade more than political institutions in the LICs, while the reverse is the case for EE. Hence, the design of an institutional framework that aimed at facilitating bilateral trade should consider the peculiarities of the countries in terms of their level of income and the various facets of institutions.

The effect of pollution on crime: Evidence from data on particulate matter and ozone

We estimate the effect of short-term air pollution exposure (PM2.5 and ozone) on several categories of crime, with a particular emphasis on aggressive behavior. To identify this relationship, we combine detailed daily data on crime, air pollution, and weather for an eight-year period across the United States. Our primary identification strategy employs extremely high dimensional fixed effects and we perform a series of robustness checks to address confounding variation between temperature and air pollution. We find a robust positive effect of increased air pollution on violent crimes, and specifically assaults, but no relationship between increases in air pollution and property crimes. The effects are present in and out of the home, at levels well below Ambient Air Pollution Standards, and PM2.5 effects are strongest at lower temperatures. The results suggest that a 10% reduction in daily PM2.5 and ozone could save $1.4 billion in crime costs per year, a previously overlooked cost associated with pollution.

Foregone Consumption and Return-Chasing Investments

We show individuals investments chase stock index returns and are financed by foregoing consumption, even after controlling for individual-account level portfolio income effects and other high dimensional fixed effects. This effect only exists for positive stock index returns with no effects for negative returns and are driven by the most salient stock index rather than other stock indices that are more representative of total stock market returns. To finance the investment, the foregone consumption occurs in the week after stock index returns are realized and are more pronounced for luxuries than necessary goods, and durables versus non-durables and services. Contrary to predictions from rational models of investor behavior with financial constraints, wealthy and more liquid accounts exhibit a larger return-chasing investment and foregone consumption, more consistent with theories of extrapolative expectations and loss aversion.

Performance characteristics of profiling methods and the impact of inadequate case-mix adjustment

Profiling or evaluation of health care providers involves the application of statistical models to compare each provider’s performance with respect to a patient outcome, such as unplanned 30-day hospital readmission, adjusted for patient case-mix characteristics. The nationally adopted method is based on random effects (RE) hierarchical logistic regression models. Although RE models are sensible for modeling hierarchical data, novel high dimensional fixed effects (FE) models have been proposed which may be well-suited for the objective of identifying sub-standard performance. However, there are limited comparative studies. Thus, we examine their relative performance, including the impact of inadequate case-mix adjustment.

Recentered Influence Functions in Stata: Methods for Analyzing the Determinants of Poverty and Inequality

Recentered influence functions (RIFs) are statistical tools popularized by Firpo, Fortin, and Lemieux (2009) for analyzing unconditional partial effects on quantiles in a regression analysis framework (unconditional quantile regressions). The flexibility and simplicity of these tools has opened the possibility of extending the analysis to other distributional statistics using linear regressions or decomposition approaches. In this paper, I introduce three Stata commands to facilitate the use of RIFs in the analysis of outcome distributions: rifvar() is an egen extension used to create RIFs for a large set of distributional statistics; rifhdreg facilitates the estimation of RIF regressions, enabling the use of high-dimensional fixed effects; and oaxaca_rif to implement Oaxaca-Blinder type decomposition analysis (RIF decompositions).

How Do Technical Barriers to Trade Affect Exports? <br/>Evidence from Egyptian Firm-Level Data

The paper examines the impact of Technical Barriers to Trade on firms’ exports in Egypt over the period 2005~2011. Firm-level data for Egypt are combined with the WTO’s TBT-specific trade concerns database. Employing a variant of a gravity model with highdimensional fixed effects, the impact of Technical Barriers to Trade on firms’ intensive and extensive margins, and exit and entry probabilities is estimated, as well as impacts on product and market diversification. Regressions examine the heterogeneous effect of Technical Barriers to Trade by firm size. Results indicate an insignificant effect of Technical Barriers to Trade on firms’ intensive margin. On the other hand, the extensive margin and entry probability are negatively affected by Technical Barriers to Trade, whereas exit probability is positively affected. Accordingly, Technical Barriers to Trade mainly induce an increase in the fixed costs of exporting. Importantly, smaller firms are more adversely affected by Technical Barriers to Trade in their export participation and entry and exit decisions. The effect of Technical Barriers to Trade on firms’ product diversification is found to be sector-dependent:positive for agricultural sectors and mixed for non-agricultural ones. Finally, firms generally tend to increase their market diversification in response to Technical Barriers to Trade. This is especially true for large firms within their set of African and Asian destination markets. By contrast, there are lower prospects of firm diversification into less stringent destinations within the European region.

International trade, exchange rate regimes, and financial crises

The main objective of this paper is to study the impact of different exchange rate regimes on international trade and to analyze their performance during crises. To this end, a gravity equation for bilateral trade is estimated for a sample of 191 countries over the period 1970–2016 by adding a set of regressors built from a de factoclassification of exchange rate arrangements and the dates of recognized financial crises. Moreover, we differentiate between anchor currencies and direct and indirect exchange rate arrangements. The gravity model is consistently estimated by including three different types of high-dimensional fixed effects and using PPML estimates. The main empirical findings are: (i) other intermediate exchange rate regimes, between completely fixed and completely flexible, promote flows of goods between countries; (ii) results depend on the anchor currency and indirect arrangements do not have any significant impact on international trade; (iii) systemic banking crises negatively affect trade flows between countries; and (iv) the impact of the exchange rate regimes on trade during crisis depends on the anchor currency and whether the crisis takes place in the exporting or the importing country.

Spillovers from regulating corporate campaign contributions

Populist clamor and recent Supreme Court decisions have renewed calls for increased regulation of corporate money in politics. Few empirical estimates exist, however, on the implications of existing rules on firms’ political spending. Exploiting within firm-cycle cross-candidate variation and across firm-cycle variation, we demonstrate that the regulation of PAC campaign contributions generates large spillovers into other corporate political expenditures such as lobbying. Using both high dimensional fixed effects and regression discontinuity designs, we demonstrate that firms constrained by campaign contribution limits spend between $549,000 and $1.6M more on lobbying per election cycle, an amount that is more than 100 times the campaign contribution limit. These results demonstrate that, similar to regulations in other domains of the economy, constraining specific corporate political activities often yields unintended effects.

The sources of wage variation and the direction of assortative matching: Evidence from a three-way high-dimensional fixed effects regression model

This paper estimates a wage equation with three high-dimensional fixed effects, using a longitudinal matched employer-employee dataset covering virtually all Portuguese private sector wage earners over a 26-year interval. First, the variation in log real hourly wages is decomposed into three components reflecting worker, firm, and job title characteristics and a residual element. It is found that worker permanent heterogeneity is the most important source of wage variation accounting for one third of the wage variance, while firm permanent effects contribute one fourth. Job title fixed effects still explain a considerable one fifth of wage variance. Second, having established that high-wage workers tend to match with high-paying firms, worker fixed effects from the wage equation are next correlated with firm fixed effects from sales and value-added production equations to provide unambiguous evidence on the sign and strength of assortative matching. The correlations are positive and large, indicating that higher productivity workers tend to match with higher productivity firms.

Do electric vehicle incentives matter? Evidence from the 50 U.S. states

We estimate the effectiveness of policy incentives for adoption of electric vehicles (EVs) in the 50 U.S. states. We employ a rich dataset of semi-annual state-level new EV vehicle registrations by make and model from 2010 to 2015 and state-level policy instruments that could affect new EV model registrations. We construct two measures of policy, one which aggregates policy instruments that can be assigned a value and a second that aggregates those without explicit values. Using a within model difference-in-difference estimator with high-dimensional fixed effects, we find that a $1000 increase in the value of a state’s model-specific EV policies increases registrations of that model within the state by 5–11%.

UNIFORM INFERENCE IN HIGH-DIMENSIONAL DYNAMIC PANEL DATA MODELS WITH APPROXIMATELY SPARSE FIXED EFFECTS

We establish oracle inequalities for a version of the Lasso in high-dimensional fixed effects dynamic panel data models. The inequalities are valid for the coefficients of the dynamic and exogenous regressors. Separate oracle inequalities are derived for the fixed effects. Next, we show how one can conduct uniformly valid inference on the parameters of the model and construct a uniformly valid estimator of the asymptotic covariance matrix which is robust to conditional heteroskedasticity in the error terms. Allowing for conditional heteroskedasticity is important in dynamic models as the conditional error variance may be nonconstant over time and depend on the covariates. Furthermore, our procedure allows for inference on high-dimensional subsets of the parameter vector of an increasing cardinality. We show that the confidence bands resulting from our procedure are asymptotically honest and contract at the optimal rate. This rate is different for the fixed effects than for the remaining parts of the parameter vector.

Hybrid marriages and phenotypic heterosis in offspring: Evidence from China

In genetics, heterosis refers to the phenomenon that cross-breeding within species leads to offspring that are genetically fitter than their parents and exhibit improved phenotypic characteristics. Based on the theory of heterosis and existing genetic evidence, offspring of “hybrid” marriages (spouses originating from different states/provinces/countries/areas), though relatively rare due to physical boundaries, may exhibit greater genetic fitness in terms of intelligence, height, or physical attractiveness (the “distance-performance” hypothesis). This study explores whether heterosis is a contributing factor to offspring’s educational attainment in China by applying a high-dimensional fixed effects (HDFE) modelling framework to the unique 0.1% micro-sample of the 2000 Chinese Population Census data. Concerning potential endogeneity of hybrid marriages, we conduct a series of robustness checks. Reassuringly, the estimated heterosis effect remains significantly positive across various measurements, after controlling for parental educational attainments/height, environmental influences, and over a thousand region and region-by-year fixed effects. The effects in male and higher-educated offspring are found to be stronger. Results are replicated when analyzing body height using data from the China Health and Nutrition Survey. Although endogeneity of “hybrid marriages” may not be completely ruled out, the current study sheds light on the potentially beneficial effects of interprovincial migration on population-level human capital accumulation, and we hope that this paper can intrigue future studies that further address endogeneity. The implied heterosis effect could, therefore, be profound for Homo sapiens as a species from an evolutionary point of view. An additional important implication is that the overall genetic influences of parents on offspring’s performance may be further decomposed into a conventional heredity effect and a heterosis effect that has been neglected previously.

Cultural Change and the Migration Choice

Cultural differences play an important role in shaping migration patterns. The conventional proxies for cross country cultural differences - such as common language, ethnicity, genetic traits or religion - implicitly assume that cultural proximity between two countries is constant over time and symmetric, which is far from realistic. This paper proposes a tractable model for international migration which explicitly allows for the time varying and asymmetric dimensions of cultural proximity. Similarly to Disdier et al (2010) we assume that the evolution of bilateral cultural affinity over time is reflected in the intensity of bilateral trade in cultural goods. Our empirical framework includes a comprehensive set of high dimensional fixed effects which enables for the identification of the impact of cultural proximity on migration over and beyond the effect of pre-existing cultural and historical ties. The results are robust across different econometric techniques and suggest that positive changes in cultural relationships over time foster bilateral migration.

Decomposing the structure of wages into firm and worker effects

PurposeThe purpose of this paper is to show that for some key topics on labour economics such as the effect of seniority and job mobility in wages, it is important to explicitly consider firm fixed effects. The author also wants to test whether the importance of firm in explaining wage dispersion is higher or lower in Spain than in other European countries.Design/methodology/approachThe author estimates an individual wage equation where firm and workers effects are considered and the estimation process control for censored wages. This exercise is performed for the Spanish economy over the course of a whole business cycle, i.e., 2000-2015.FindingsThe author demonstrates that Spanish firms contribute to explain around 27 per cent of the individual wage heterogeneity but more importantly around 74 per cent of inter-industry wage differentials. In both cases, this contribution is mainly related to large dispersion in firm’s wage policies. The process of positive sorting of workers across firms or industries does not play an important role. Interestingly, the importance of firm’s wage policies in explaining individual wage dispersion has increased over the current Big Recession.Practical implicationsThe results confirm that firms set wages and, henceforth, are partially responsible for individual wage heterogeneity but more importantly for inter-industrial wage dispersion.Originality/valueThe exercise is performed under optimal conditions because the author uses a longitudinal matched employer-employee data set, observed wages are at a monthly frequency, and implements an estimation method suitable for censored models with two high-dimensional fixed effects. This is the first study that looks deeply into the role of firms in explaining wage heterogeneity at the individual and industry level in Spain and along the current Big Recession.

AN ESTIMATION OF WORKER AND FIRM EFFECTS WITH CENSORED DATA

ABSTRACTIn this paper, we develop a new estimation method that is suitable for censored models with two high dimensional fixed effects and that is based on a sequence of least squares regressions, yielding significant savings in computing time and hence making it applicable to frameworks in which standard estimation techniques become unfeasible. We propose to apply this estimation method to investigate the role of firms in individual wage variation. Using a longitudinal match employer‐employee dataset from Spain, we show that the analysis of wage determination can be misleading when wages are censored. In particular, the role of firm wage policies in wage dispersion is overestimated by more than ten percentage points, while the role of time‐invariant individual characteristics is underestimated by fifteen percentage points. Hence, controlling for censored wages appears to reinforce the idea that when explaining individual wage dispersion, what workers ‘are’ is more important than what workers ‘do’.

A Friendly Turn: Advertising Bias in the News Media

This paper investigates whether newspapers report more favorably about advertising corporate clients than about other firms. Our identification strategy based on high-dimensional fixed effects and high frequency advertising data shows that advertising leads to more positive press coverage. This advertising bias in reporting is found among local and national newspapers. Further results show that advertising bias manifests particularly in less negative reporting after bad news events such as negative earnings surprises or extremely negative stock returns. Our findings cast doubt on the independence of the press from corporate pressure and hint at important information frictions.

Industry localization, distance decay, and knowledge spillovers: Following the patent paper trail

This paper investigates Alfred Marshall’s hypothesis that knowledge spillovers increase where industries are localized. At the same time, we take a fresh look at the role of distance in the diffusion of knowledge spillovers. Relying on a cited-citing gravity-like equation with high-dimensional fixed effects that control for multiple sources of observed and non-observed heterogeneity, we implement a Poisson pseudo-maximum-likelihood (PPML) estimator. We find that knowledge spillovers correlate positively with industry localization and that the agglomeration of an industry can offset the adverse effect of distance. The results also corroborate the distance decay effect uncovered in earlier research. Our new approach to estimate the PPML with two high-dimensional fixed effects should prove valuable in applications to a variety of other problems in economics, such as the estimation of gravity equations widely used in modeling migration, trade and other flows among countries and regions.

Decomposing the Wage Losses of Displaced Workers: The Role of the Reallocation of Workers into Firms and Job Titles

Using an unusually rich matched employer-employee-job title data set for Portugal, this paper evaluates the sources of wage losses of workers displaced due to firm closure based on the comparison of workers' wages differentials before and after displacement. Potential wage losses of displaced workers can be related to firm, job title, and match heterogeneity in the pre- and post-displacement jobs. In this vein, we estimate a three-way high-dimensional fixed effects regression model that enables us to decompose the sources of the wage losses into the contribution of firm, job title, and match fixed effects. The worker-firm match plays a very sizable role. We found that the allocation of workers into poorer matches accounts for 38 percent of the total average wage loss. Sorting among firms accounts for 36 percent. Job downgrading also plays a significant role in explaining the wage loss of displaced workers, accounting for the remaining 26 percent.

Firm–worker matching in industrial clusters

In this paper we use a novel approach and a large Portuguese employer-employee panel data set to study the hypothesis that industrial agglomeration improves the quality of the firm-worker matching process. Our method makes use of recent developments in the estimation and analysis of models with high-dimensional fixed effects. Using wage regressions with controls for multiple sources of observed and unobserved heterogeneity we find little evidence that the quality of matching increases with firm’s clustering within the same industry. This result supports Freedman’s (2008) analysis using U.S. data.(This abstract was borrowed from another version of this item.)

The Sources of Wage Variation: A Three-Way High-Dimensional Fixed Effects Regression Model

This paper estimates a wage equation with three high-dimensional fixed effects, using a longitudinal matched employer-employee dataset covering virtually all Portuguese wage earners over a little more than two decades. The variation in log real hourly wages is decomposed into different components related to worker, firm, and job title characteristics (both observed and unobserved) and a residual component.It is found that worker permanent heterogeneity is the most important source of wage variation (36.0 percent) and that the unobserved component plays a more important role (21.0 percent) than the observed component (15.0 percent) in explaining wage differentials. Firm permanent effects are less important overall (28.7 percent) and are due in almost equal parts to the unobserved component and the observed component. Job title effects emerge as the least important dimension but they still explain close to 10 percent of wage variation. We found definitive evidence of positive assortative matching.