Econometrics

Smoothing is an important statistical tool in applied analysis of political data. Smoothers can be use to incorporate nonlinearity, deal with duration dependence, and diagnose residual plots. But smoothers invariably require the analyst to make choices about the amount of smoothing to apply and the criteria for this choice amounts to visual examination. Such visual methods can lead to seemingly arbitrary selection of smoothing parameters. I argue that instead of using a visual method for the selection of smoothing parameters, analysts should use automated methods where the level of smoothing is estimated from the data. I outline why such methods are preferable and then demonstrate with both simulated data and in empirical settings how automatic smoothing is superior to visual selection of smoothing parameters.

While the derivation and estimation of heterogeneous probit models appears straightforward, the properties of such models are not well understood. It is assumed that the properties of these models are identical to those of discrete choice models, but this may not be true. While these models are formally identified, we argue that they suffer from fragile identification. We first look for evidence of fragile identification using a series of Monte Carlo experiments. We then reanalyze data on abortion attitudes and use profile likelihood plots and a convergence analysis to look for signs of fragile identification. We find strong evidence that heteroskedastic probit models tend to be tenuously identified.

We investigate under which conditions price competition in a market with matching frictions leads to sorting of buyers and sellers. Positive assortative matching obtains only if there is a high enough degree of complementarity between buyer and seller types. The relevant condition is rootsupermodularity; i.e., the square root of the match value function is supermodular. It is a necessary and sufficient condition for positive assortative matching under any distribution of buyer and seller types, and does not depend on the details of the underlying matching function that describes the search process. The condition is weaker than log-supermodularity, a condition required for positive assortative matching in markets with random search. This highlights the role competition plays in matching heterogeneous agents. Negative assortative matching obtains whenever the match value function is weakly submodular. The paper initially circulated under the title "The Sorting Effect of Price Competition". We would like to thank numerous colleagues and seminar participants for insightful discussions and comments. We greatly benefited from comments on an earlier draft

We examine inference in panel data when the number of groups is small as is typically the case for differences-in-differences estimation and when some variables are fixed within groups. In this case, standard asymptotics based on the number of groups going to infinity provide a poor approximation to the finite sample distribution. We show that in some cases the t-statistic is distributed as t and propose simple two-step estimators for these cases. We apply our analysis to two well-known papers.

This article examines nonsequential search when jobs vary with respect to nonpecuniary characteristics. In the presence of frictions in the labor market, the equilibrium job distribution need not show evidence of compensating wage differentials. The model also generates several pervasive features of labor markets: unemployment and vacancies, apparent discrimination, and market segmentation. When workers are homogeneous, restrictions on the range of job offers decrease welfare and cannot reduce unemployment. However, when workers have heterogeneous preferences, such restrictions may lower unemployment and can even lead to a Pareto-improvement in welfare. We consider the impact of policies banning discrimination and regulating working-conditions.

Although information visualization (infovis) technologies have proven indispensable tools for making sense of complex data, wide-spread deployment has yet to take hold, as successful infovis applications are often difficult to author and require domain-specific customization. To address these issues, we have created prefuse, a software framework for creating dynamic visualizations of both structured and unstructured data. prefuse provides theoretically-motivated abstractions for the design of a wide range of visualization applications, enabling programmers to string together desired components quickly to create and customize working visualizations. To evaluate prefuse we have built both existing and novel visualizations testing the toolkit's flexibility and performance, and have run usability studies and usage surveys finding that programmers find the toolkit usable and effective.

This paper considers a first price common value auction with one-sided private information among bidders. An object is rented each period to the highest bidder without revealing the bids made in previous periods. The paper extends single crossing arguments used in static games to show the existence of equilibrium in this repeated auction. The model predicts that bidders bid more aggressively in a repeated auction than in a static auction, because a higher bid increases the amount of information learned through bidding. This learning process occurs since bidders observe who won in previous periods and can update their beliefs about the signals of other bidders. This bidding behavior leads to a theoretical explanation for the winner's curse in repeated auctions. The aggressive bidding behavior also implies that the per period revenue of the repeated auction is higher than that of the oneperiod counterpart. In addition, the model yields predictions like an increasing wage schedule and the institution of a signing bonus that are consistent with labor market observations. We also extend the model to the case where the seller is strategic and maximizes her expected discounted revenue. * I am grateful to Wolfgang Pesendorfer for guidance and invaluable advice throughout this project. I am also indebted to Marco Battaglini and Eric Maskin for their useful suggestions. I would like to thank Attila Ambrus, Katalin Bognár, Norbert Maier, József Molnár, Shantanu Mukherjee, Daisuke Nakajima, Balázs Szentes, Steve Tadelis, Lawrence Uren as well as seminar participants at the University of Rochester, University of Rotterdam, University of Toronto and University of Virginia. All remaining errors are mine.

1 This is the seasonally adjusted civilian unemployment rate from the Bureau of Labor Statistics; http://stats.bls.gov. NOTE: All models also include cosine seasonal. Schwarz criterion calculated as ᑦ -(k/2)log(T) for ᑦ the log likelihood, k the number of parameters, and T = 682 the sample size.

Web site: http://www.pitt.edu/˜jfeigen. The results here can be simply generalized to the entire hyperbolic absolute risk aversion (HARA) family of preferences. However, analytic results already exist for the most interesting representatives of the HARA class not covered here, i.e. constant absolute risk aversion (CARA) (Caballero (1990)) and quadratic utility, although the preference shocks have to be turned off (m t = 0) in the case of CARA utility.

We propose a new method of testing stochastic dominance which improves on existing tests based on bootstrap or subsampling. Our test requires estimation of the contact sets between the marginal distributions. Our tests have asymptotic sizes that are exactly equal to the nominal level uniformly over the boundary points of the null hypothesis and are therefore valid over the whole null hypothesis. We also allow the prospects to be indexed by infinite as well as finite dimensional unknown parameters, so that the variables may be residuals from nonparametric and semiparametric models. Our simulation results show that our tests are indeed more powerful than the existing subsampling and recentered bootstrap.

Contrary to the usual presumption that welfare is maximized if consumers behave rationally, we show in a two-period overlapping generations model that there always exists a consumption rule that can weakly improve upon the lifecycle/permanent-income rule in general equilibrium.

The Svensson generalization of the popular Nelson-Siegel term structure model is widely used by practitioners and central banks. Unfortunately, like the original Nelson-Siegel specification, this generalization, in its dynamic form, does not enforce arbitrage-free consistency over time. Indeed, we show that the factor loadings of the Svensson generalization cannot be obtained in a standard finance arbitrage-free affine term structure representation. Therefore, we introduce a closely related generalized Nelson-Siegel model on which the no-arbitrage condition can be imposed. We estimate this new arbitrage-free generalized Nelson-Siegel model and demonstrate its tractability and good in-sample fit. . The Svensson extension adds a second curvature term, which allows for a better fit at long maturities. Following Diebold and Li , we first introduce a dynamic version of this model, the DNSS model, which corresponds to a modern four-factor term structure model. Unfortunately, we show that it is not possible to obtain an arbitrage-free "approximation" to this model in the sense of obtaining analytically identical factor loadings for the four factors. Intuitively, such an approximation requires that each curvature factor must be paired with a slope factor that has the same mean-reversion rate. This pairing is simply not possible for the Svensson extension, which has one slope factor and two curvature factors. Therefore, to obtain an arbitrage-free generalization of the Nelson-Siegel curve, we add a second slope factor to pair with the second curvature factor.

We argue for incorporating the financial economics of market microstructure into the financial econometrics of asset return volatility estimation. In particular, we use market microstructure theory to derive the cross-correlation function between latent returns and market microstructure noise, which feature prominently in the recent volatility literature. The cross-correlation at zero displacement is typically negative, and cross-correlations at nonzero displacements are positive and decay geometrically. If market makers are sufficiently risk averse, however, the cross-correlation pattern is inverted. Our results are useful for assessing the validity of the frequently-assumed independence of latent price and microstructure noise, for explaining observed crosscorrelation patterns, for predicting as-yet undiscovered patterns, and for making informed conjectures as to improved volatility estimation methods.

This paper analyzes partial identification of parameters that measure a distribution's spread, e.g. variance, Gini ratio, entropy, or interquartile range. The core result are tight bounds on such parameters for cases where a distribution is known to be a mixture between a known one and one whose c.d.f. can be bounded. More specifically, I derive joint identification regions for the mean of an incompletely observable random variable and parameters that increase in meanpreserving spreads -e.g. variance, entropy, or Gini ratio -as well as joint bounds on quantiles and quantile contrasts. A side result is a generalization of the Rothschild-Stiglitz notion of comparative riskiness.

In a recurring game, a stage game is played consecutively by di erent groups of players, with each group receiving information about the play of earlier groups. Starting with uncertainty about the distribution of types in the population, late groups may learn to play a correct Bayesian equilibrium, as if they know the type distribution. This paper concentrates on Selten's Chain Store game and the Kreps, Milgrom, Roberts, Wilson phenomenon, where a small perceived inaccuracy about the type distribution can drastically alter the equilibrium behavior. It presents su cient conditions that prevent this phenomenon from persisting in a recurring setting.

We show that an asymptotic envy-freeness condition is necessary for a form of robust approximate implementation in large economies. In settings where allocations are excludable, asymptotic envy-freeness is also su¢ cient for implementation, while in non-excludable settings it is not su¢ cient.

Recent analyses of social networks, both empirical and theoretical, are discussed, with a focus on how social networks influence economic behavior, as well as how social networks form. Some challenges of such research are discussed as are some of the important considerations for the literature going forward.

Theoretical computer science treats any computational subject for which a good model can be created. Research on formal models of computation was initiated in the 1930s and 1940s by Turing, Post, Kleene, Church, and others. In the 1950s and 1960s programming languages, language translators, and operating systems were under development and therefore became both the subject and basis for a great deal of theoretical work. The power of computers of this period was limited by slow processors and small amounts of memory, and thus theories (models, algorithms, and analysis) were developed to explore the efficient use of computers as well as the inherent complexity of problems. The former subject is known today as algorithms and data structures, the latter computational complexity.

Models of political participation have begun to incorporate actors who possess "social preferences". However, these models have failed to take into account the potentially incongruent political goals of different social preference types. These goals are likely to play an important role in shaping political behavior. To examine the effect of distinct social preferences on political activity we conducted an experiment in which participants played five rounds of a modified dictator game (Andreoni and Miller 2002). We used the decisions in these games to determine their preference type and mapped these types to reported political activity. Our results show that sub jects who were most interested in increasing total welfare in the dictator game were more likely to participate in politics than subjects with selfish preferences, whereas sub jects most interested in reducing the difference between their own well-being and the well-being of others were no more likely to participate.

Nonparametric derivative estimation has never attracted much attention as one gets the derivative estimates as "by-products" from a local polynomial or spline fit. However, these estimates often suffer from boundary effects and are very sensitive to outliers. Apart from this, the local polynomial estimators suffer from a systematic downward bias, as we will demonstrate. This article is intended to re-establish research interest in derivative estimation, and to guide the user who needs to work with one of the available packages.

We investigate design-weighted local smoothing and show that the optimal (bias-minimizing) weights have similar form and interpretation as the optimal weights given by the Horvitz-Thompson theorem known from sampling theory. We set forth that the hazards in using bias-minimizing weights apply to kernel smoothing, too, suggesting to be cautious with the application of biasminimizing weights in general.

When in Econometrica proposed a consistent test of stochastic dominance, they were silent about the asymptotic unbiasedness of their tests against p n-converging Pitman local alternatives. This paper shows that when we focus on …rst-order stochastic dominance, there exists a wide class of p n-converging Pitman local alternatives against which their test is asymptotically biased, i.e., having the local asymptotic power strictly below the asymptotic size. This phenomenon more generally applies to one-sided nonparametric tests which have a sup norm of a shifted standard Brownian bridge as their limit under p n-converging Pitman local alternatives. Among other examples are tests of independence or conditional independence. We provide an intuitive explanation behind this phenomenon, and illustrate the implications using the simulation studies.

When in Econometrica proposed a consistent test of stochastic dominance, they were silent about the asymptotic unbiasedness of their tests against p n-converging Pitman local alternatives. This paper shows that when we focus on …rst-order stochastic dominance, there exists a wide class of p n-converging Pitman local alternatives against which their test is asymptotically biased, i.e., having the local asymptotic power strictly below the asymptotic size. This phenomenon more generally applies to one-sided nonparametric tests which have a sup norm of a shifted standard Brownian bridge as their limit under p n-converging Pitman local alternatives. Among other examples are tests of independence or conditional independence. We provide an intuitive explanation behind this phenomenon, and illustrate the implications using the simulation studies.

We construct a framework for measuring economic activity at high frequency, potentially in real time. We use a variety of stock and flow data observed at mixed frequencies (including very high frequencies), and we use a dynamic factor model that permits exact filtering. We illustrate the framework in a prototype empirical example and a simulation study calibrated to the example. frequencies, including quarterly (e.g., GDP), monthly (e.g., industrial production), weekly (e.g., employment), and continuously (e.g., asset prices), and we want to be able to incorporate all of them, to provide continuously-updated measurements.

A strategy profile in a repeated game has bounded recall L if play under the profile after two distinct histories that agree in the last L periods is equal. proved that any strict equilibrium in bounded-recall strategies of a game with full support public monitoring is robust to all perturbations of the monitoring structure towards private monitoring (the case of almost-public monitoring), while strict equilibria in unbounded-recall strategies are typically not robust.

This paper studies inference in a continuous time game where an agent's decision to quit an activity depends on the participation of other players. In equilibrium, similar actions can be explained not only by direct influences but also by correlated factors. Our model can be seen as a simultaneous duration model with multiple decision makers and interdependent durations. We study the problem of determining the existence and uniqueness of equilibrium stopping strategies in this setting. This paper provides results and conditions for the detection of these endogenous effects. First, we show that the presence of such effects is a necessary and sufficient condition for simultaneous exits. This allows us to set up a nonparametric test for the presence of such influences which is robust to multiple equilibria. Second, we provide conditions under which parameters in the game are identified. Finally, we apply the model to data on desertion in the Union Army during the American Civil War and find evidence of endogenous influences.

Experimental evidence suggests that individuals are more risk averse when they perceive risk gradually. We address these …ndings by studying a decision maker (DM) who has recursive preferences over compound lotteries and who cares about the way uncertainty is resolved over time. DM has preferences for one-shot resolution of uncertainty (PORU) if he always prefers any compound lottery to be resolved in a single stage. We establish an equivalence between dynamic PORU and static preferences that are identi…ed with the behavior observed in Allais-type experiments. The implications of this equivalence on preferences over information systems are examined. We de…ne the gradual resolution premium and demonstrate its magnifying e¤ect when combined with the usual risk premium. In an intertemporal context, PORU captures "loss aversion with narrow framing".

We examine the timing of firms' operations in a formal model of labor demand. Merging a variety of data sets from Portugal from 2003, we describe temporal patterns of firms' demand for labor and estimate production-functions and relative labor-demand equations. The results demonstrate the existence of substitution of employment across times of the day/week and show that legislated penalties for work at irregular hours induce firms to alter their operating schedules. The results suggest a role for such penalties in an unregulated labor market, such as the United States, in which unusually large fractions of work are performed at night and on weekends.

We investigate a multiple hypothesis test designed for detecting signals embedded in noisy observations of a sensor array. The global level of the multiple test is controlled by the false discovery rate (FDR) criterion recently suggested by Benjamini and Hochberg instead of the classical familywise error rate (FWE) criterion. In the previous study [3], the suggested procedure has shown promising results on simulated data. Here we carefully examine the independence condition required by the Benjamini Hochberg procedure to ensure the control of FDR. Based on the properties of beta distribution, we proved the applicability of the Benjamini Hochberg procedure to the underlying test. Further simulation results show that the false alarm rate is less than 0.02 for a choice of FDR as high as 0.1, implying the reliability of the test has not been affected by the increase in power.

Given the location of a relative maximum of the log-likelihood function, how to assess whether it is the global maximum? This paper investigates a statistical tool, which answers this question by posing it as a hypothesis testing problem. A general framework for constructing tests for global maximum is given. The characteristics of the tests are investigated for two cases: correctly specified model and model mismatch. A finite sample approximation to the power is given, which gives a tool for performance prediction and a measure for comparison between tests. The sensitivity of the tests to model mismatch is analyzed in terms of the Renyi divergence and the Kullback-Leibler distance between the true underlying distribution and the assumed parametric class and tests that are insensitive to small deviations from the model are derived. The tests are illustrated for three applications: passive localization or direction finding using an array of sensors, estimating the parameters of a Gaussian mixture model, and estimation of superimposed exponentials in noise -problems that are known to suffer from local maxima.

This paper analyzes firm panel data to examine how export demand shocks associated with the 1997 Asian financial crisis affected Chinese exporters. We construct firm-specific exchange rate shocks based on the pre-crisis destinations of firms' exports. Because the shocks were unanticipated and large in magnitude, they are a plausible instrument for identifying the impact of exporting on firm productivity and other aspects of firm performance. We find that firms whose export destinations experience greater currency depreciation have slower growth in exports and that export growth increases firm productivity as well as other measures of firm performance. Consistent with the "learning-by-exporting" hypothesis, the productivity impact of export growth is greater when firms export to more developed countries. Southern California, the World Bank, and the NBER conference "Firms and the Evolving Structure of Global Economic Activity".

Innovation is the creation of intellectual assets; intellectual assets are distinguished by nonconvexities. But different nonconvexities affect outcomes differently. Provided knowledge is finitely expansible, nonrivalry causes no difficulties for the existence and expost efficiency of competitive equilibrium. However, the minimum-size indivisibility in creating intellectual assets renders competitive equilibrium socially inefficient in general, even though existence is maintained. The inefficiencies have the size of Dupuit triangles. Innovation patterns are invariant within a range of legal structures regulating intellectual asset dissemination.

This two-phase study examines a physician's use of one of five different types of technology to note a patient's symptoms during the medical interview. In this between-subjects design, 342 undergraduates viewed one of several videos that demonstrated one condition of the doctor/patient interaction. After viewing the interaction, each participant completed a series of questionnaires that evaluated their general satisfaction with the quality of care demonstrated in the medical interview. A main effect of technology condition was present in both phases.

This paper uses the Panel Study of Income Dynamics to provide some of the first direct evidence that wealth is systematically higher for consumers with predictably greater income uncertainty. However, the apparent pattern of precautionary wealth is not consistent with a standard parameterization of the life cycle model in which consumers are patient enough to begin saving for retirement early in life: wealth is estimated to be far less sensitive to uncertainty than implied by that model. Instead, our results suggest that over most of their working lifetime, consumers behave in accordance with the "buffer-stock" models of saving described in or , in which consumers hold wealth principally to insulate consumption against near-term fluctuations in income.

This paper analyzes the effects of dividend taxation on corporate behavior using the large tax cut on individual dividend income enacted in 2003. Using data spanning 1980 to 2004-Q2, we document a sharp and widespread surge in dividend payments following the tax cut, along several dimensions.

We consider the open economy consequences of U.S. monetary policy, extending the identification approach of Romer and Romer [2004] and adapting it for use with asset prices. Intended policy changes are orthogonalized against the economy's expected future path, which captures any effects from open economy variables. Estimated from a set of bilateral VARs, the dynamic responses of the exchange rate, foreign interest rate, and foreign output are consistent with recent work that identifies U.S. policy via futures market changes and a priori impulse response bounds. We compare the two approaches, finding important commonalities. We also outline some advantages of our approach. JEL Classification: E52, F31, F41 Keywords: open economy monetary policy identification, exchange rate adjustment, interest rate pass-through † This paper supersedes Nuffield College Working Papers 2005-w18 and 2006-w04. The authors acknowledge financial support from British Academy post-doctoral fellowships. We would like to thank

Using a panel of 439 German regions we evaluate and compare the performance of various Neural Network (NN) models as forecasting tools for regional employment growth. Because of relevant differences in data availability between the former East and West Germany, the NN models are computed separately for the two parts of the country. The comparisons of the models and their ex post forecasts are carried out by means of a non-parametric test: viz. the Friedman statistic. The Friedman statistic tests the consistency of model results obtained in terms of their rank order. Since there is no normal distribution assumption, this methodology is an interesting substitute for a standard analysis of variance. Furthermore, the Friedman statistic is indifferent to the scale on which the data are measured. The evaluation of the ex post forecasts suggests that NN models are generally able to correctly identify the fastest-growing and the slowest-growing regions, and hence predict rather well the correct ranking of regions in terms of their employment growth. The comparison among NN models – on the basis of several criteria – suggests that the choice of the variables used in the model may influence the model’s performance and the reliability of its forecasts.

Some probabilistic aspects of the number variance statistic are investigated. Infinite systems of independent Brownian motions and symmetric α-stable processes are used to construct new examples of processes which exhibit both divergent and saturating number variance behaviour. We derive a general expression for the number variance for the spatial particle configurations arising from these systems and this enables us to deduce various limiting distribution results for the fluctuations of the associated counting functions. In particular, knowledge of the number variance allows us to introduce and characterize a novel family of centered, long memory Gaussian processes. We obtain fractional Brownian motion as a weak limit of these constructed processes.

The credible identification of endogenous peer group effects-i.e. social multiplier or feedback effects-has long eluded social scientists. We argue that such effects are most credibly identified by a randomly assigned social program which operates at differing intensities within and between peer groups. The data we use are from Project STAR, a class size reduction experiment conducted in Tennessee elementary schools. In these data, classes were comprised of varying fractions of students who had previously been exposed to the Small class treatment, creating class groupings of varying experimentally induced quality. We use this variation in class group quality to estimate the spillover effect. We find that when allowance is made for this 'feedback' effect of prior exposure to the Small class treatment, the peer effects account for much of the total experimental effects in the later grades, and the direct class size effects are rendered substantially smaller.

This paper summarizes our recent research on evaluating the distributional consequences of social programs. This research advances the economic policy evaluation literature beyond estimating assorted mean impacts to estimate distributions of outcomes generated by different policies and to determine how those policies shift persons across the distributions of potential outcomes produced by them. Our approach enables analysts to evaluate the distributional effects of social programs without invoking the "Veil of Ignorance" assumption often used in the literature in applied welfare economics. Our methods determine which persons are affected by a given policy, where they come from in the ex-ante outcome distribution and what their gains are. We apply our methods to analyze two proposed policy reforms in American education. These reforms benefit the middle class and not the poor.

This paper tests whether manufacturing exporters pay more to educated workers in an e¤ort to ascertain whether the productivity of human capital is raised by exports. Using a panel of matched employer-employee data from Morocco, we …nd no evidence that the education wage premium is higher in exporting sectors and …rms. Although exporters pay more on average, much of the wage di¤erential can be explained by the fact that exporters have a larger workforce and more capital. Educated workers who start working for an exporter do not experience a larger wage increase relative to their previous job. We …nd a mild positive association between exports, technology, and product quality, part of which is due di¤erences in …rm size. We discuss why our results di¤er from those obtained using di¤erent countries and methodologies.

Does capital-embodied technological change play an important role in shaping labor market inequalities? This paper addresses the question in a model with vintage capital and search-matching frictions where costly capital investment leads to heterogeneity in productivity among vacancies in equilibrium. The paper Þrst demonstrates analytically 1) how technology growth affects equilibrium wage inequality, income shares and unemployment, and 2) how its impact depends on the presence of labor market institutions (welfare beneÞts, Þring taxes). Next, it applies the model to a quantitative evaluation of vintage capital as an origin of wage inequality: through labor market frictions, a 10-year age differential in capital translates into a 4% wage gap among exante equal workers. Finally, the model allows a U.S.-continental Europe comparison: a common embodied technological acceleration interacted with different labor market institutions can explain a substantial part of the differential rise in unemployment and capital share and some of the differential dynamics in wage inequality.

It is commonly asserted by economists, comparative political scientists and international policymakers that government effectiveness is important. The liberal tradition holds that good government performance leads to competitive markets and to the efficient delivery of goods and services. These ideas derive from numerous studies emphasising the benefits of good governance. These studies suggest that the incentives for good governance are created by a country’s formal political rules. Political and economic “success are often attributed to a set of formal institutions, social infrastructure, or even mental frameworks. The underlining rationale of this approach is based on the assumption that "institutions matter". Although the literature stressing the benefits of institutions for good governance is prolific, no one has actually been able to prove which institutions matter for social and economic success or to what extent. At the same time, and more fundamentally, most scholars have difficulty agreeing what effective government really is. Added to this conundrum is the reality that fully effective governments and markets have not been attained in any country. In summary, this vast literature is casual and even contradictory in defining which institutions actually matter for good governance and is generally lacking in empirical foundation.