nep-ecm New Economics Papers
on Econometrics
Issue of 2021‒05‒31
twenty-two papers chosen by
Sune Karlsson
Örebro universitet

  1. Sample Selection in Linear Panel Data Models with Heterogeneous Coefficients By Alyssa Carlson; Riju Joshi
  2. Specification tests for GARCH processes By Giuseppe Cavaliere; Indeewara Perera; Anders Rahbek
  3. Intercept Estimation in Nonlinear Selection Models By Arulampalam, Wiji; Corradi, Valentina; Gutknecht, Daniel
  4. The Influence Function of Semiparametric Two-step Estimators with Estimated Control Variables By Jinyong Hahn; Zhipeng Liao; Geert Ridder; Ruoyao Shi
  5. Identification and Estimation of Average Partial Effects in Semiparametric Binary Response Panel Models By Laura Liu; Alexandre Poirier; Ji-Liang Shiu
  6. A new test for common breaks in heterogeneous panel data models By Jiang, Peiyun; Kurozumi, Eiji
  7. Small Sample Adjustment for Hypotheses Testing on Cointegrating Vectors By Canepa, Alessandra
  8. Comparison of Local Projection Estimators for Proxy Vector Autoregressions By Martin Bruns; Helmut Lütkepohl
  9. Inference for multi-valued heterogeneous treatment effects when the number of treated units is small By Marina Dias; Demian Pouzo
  10. Vector autoregression models with skewness and heavy tails By Karlsson, Sune; Mazur, Stepan; Nguyen, Hoang
  11. Non Parametric Classes for Identification in Random Coefficients Models when Regressors have Limited Variation By Gaillac, Christophe; Gautier, Eric
  12. Identification and Estimation of a Partially Linear Regression Model using Network Data: Inference and an Application to Network Peer Effects By Eric Auerbach
  13. Uniform Theory for CCE under Heterogeneous Slopes and General Unknown Factors By Stauskas, Ovidijus
  14. Comparing Forecast Performance with State Dependence By Odendahl, Florens; Rossi, Barbara; Sekhposyan, Tatevik
  15. Quasi-Experimental Shift-Share Research Designs By Borusyak, Kirill; Hull, Peter; Jaravel, Xavier
  16. Multivariate Hermite polynomials and information matrix tests By Dante Amengual; Gabriele Fiorentini; Enrique Sentana
  17. Deep Kernel Gaussian Process Based Financial Market Predictions By Yong Shi; Wei Dai; Wen Long; Bo Li
  18. Flexible empirical Bayes estimation of local fertility schedules. reducing small area problems and preserving regional variation By Stefan Leknes; Sturla A. Løkken
  19. A two-stage pooled panel data estimator of demand elasticities By Thomas von Brasch; Arvid Raknerud
  20. Signaling, Random Assignment, and Causal Effect Estimation By Chemla, Gilles; Hennessy, Christopher
  21. Historical Econometrics: Instrumental Variables and Regression Discontinuity Designs By Valencia Caicedo, Felipe
  22. Forecasting with Shadow-Rate VARs By Andrea Carriero; Todd E. Clark; Massimiliano Marcellino; Elmar Mertens

  1. By: Alyssa Carlson (Department of Economics, University of Missouri); Riju Joshi (Department of Economics, Portland State University)
    Abstract: In this paper, we propose a parametric estimation procedure for linear panel data models with sample selection and heterogeneous coefficients. We allow heterogeneous coefficients in both outcome model as well as in the selection model. Our two-step estimation procedure accounts for endogeneity from the selection process as well as endogeneity due to correlation between the individual unobserved heterogeneity and the observed covariates. Conditional linear projections are used to establish a tractable control function approach that builds upon the original Heckman correction to sample selection. We derive the consistency and √n-asymptotic normality of the proposed two-step procedure and provide an asymptotic variance formula that allows for arbitrary correlation over time. Monte Carlo simulations illustrate the finite sample properties of our estimator and demonstrate that our proposed estimator outperforms standard estimators that ignore the presence of parameter heterogeneity and heteroskedastic sample selection. We apply the proposed approach to estimate gender differences in high-stakes time constrained decisions using Elo ratings data from the World Chess Federation. When addressing both sources of endogeneity, we find a much larger gender skill gap and substantial differences across the gender in strategically selecting into time constrained matches.
    Keywords: panel data, sample selection, heterogeneous coefficients, control function, conditional linear projections
    JEL: C23 C33
    Date: 2021
  2. By: Giuseppe Cavaliere (Department of Economics, Exeter Business School, Department of Economics, University of Bologna,); Indeewara Perera (Department of Economics, The University of Sheffield); Anders Rahbek (Department of Economics, University of Copenhagen)
    Abstract: This paper develops tests for the correct specification of the conditional variance function in GARCH models when the true parameter may lie on the boundary of the parameter space. The test statistics considered are of Kolmogorov-Smirnov and Cramér-von Mises type, and are based on a certain empirical process marked by centered squared residuals. The limiting distributions of the test statistics are not free from (unknown) nuisance parameters, and hence critical values cannot be tabulated. A novel bootstrap procedure is proposed to implement the tests; it is shown to be asymptotically valid under general conditions, irrespective of the presence of nuisance parameters on the boundary. The proposed bootstrap approach is based on shrinking of the parameter estimates used to generate the bootstrap sample toward the boundary of the parameter space at a proper rate. It is simple to implement and fast in applications, as the associated test statistics have simple closed form expressions. A simulation study demonstrates that the new tests: (i) have excellent finite sample behaviour in terms of empirical rejection probabilities under the null as well as under the alternative; (ii) provide a useful complement to existing procedures based on Ljung-Box type approaches. Two data examples are considered to illustrate the tests.
    Keywords: GARCH model, bootstrap, specification test, Kolmogorov-Smirnov test, Cramér-von Mises test, marked empirical process, nuisance parameters on the boundary,
    Date: 2021–05–25
  3. By: Arulampalam, Wiji (University of Warwick); Corradi, Valentina (University of Surrey); Gutknecht, Daniel (Goethe University Frankfurt)
    Abstract: We propose various semiparametric estimators for nonlinear selection models, where slope and intercept can be separately identifed. When the selection equation satisfies a monotonic index restriction, we suggest a local polynomial estimator, using only observations for which the marginal distribution of instrument index is close to one. Such an estimator achieves a univariate nonparametric rate, which can range from a cubic to an 'almost' parametric rate. We then consider the case in which either the monotonic index restriction does not hold and/ or the set of observations with propensity score close to one is thin so that convergence occurs at most at a cubic rate. We explore the finite sample behaviour in a Monte Carlo study, and illustrate the use of our estimator using a model for count data with multiplicative unobserved heterogeneity.
    Keywords: irregular identification, selection bias, local polynomial, trimming, count data
    JEL: C14 C21 C24
    Date: 2021–05
  4. By: Jinyong Hahn (UCLA); Zhipeng Liao (UCLA); Geert Ridder (USC); Ruoyao Shi (Department of Economics, University of California Riverside)
    Abstract: This paper studies semiparametric two-step estimators with a control variable estimated in a first-step parametric or nonparametric model. We provide the explicit influence function of the two-step estimator under an index restriction which is imposed directly on the unknown control variable. The index restriction is weaker than the commonly used identification conditions in the literature, which are imposed on all exogenous variables. An extra term shows up in the influence function of the semiparametric two-step estimator under the weaker identification condition. We illustrate our influence function formula in a mean regression example, a quantile regression example, and a sample selection example where the control variable approach is applied for identification and consistent estimation of structural parameters.
    Keywords: control variable approach, generated regressors, influence function, semiparametric two-step estimation
    JEL: C14 C31 C32
    Date: 2021–05
  5. By: Laura Liu; Alexandre Poirier; Ji-Liang Shiu
    Abstract: Average partial effects (APEs) are generally not point-identified in binary response panel models with unrestricted unobserved heterogeneity. We show their point-identification under an index sufficiency assumption on the unobserved heterogeneity, even when the error distribution is unspecified. This assumption does not impose parametric restrictions on the unobserved heterogeneity. We then construct a three-step semiparametric estimator for the APE. In the first step, we estimate the common parameters using either conditional logit or smoothed maximum score. In the second step, we estimate the conditional expectation of the outcomes using local polynomial regression given generated regressors that depend on first-step estimates. In the third step, we average this conditional distribution over a subset of conditioning variables to obtain a partial mean which estimates the APE. We show that this proposed three-step APE estimator is consistent and asymptotically normal. We then evaluate its finite-sample properties in Monte Carlo simulations, and illustrate our estimator in a study of determinants of married women's labor supply.
    Date: 2021–05
  6. By: Jiang, Peiyun; Kurozumi, Eiji
    Abstract: In this paper, we develop a new test to detect whether break points are common in heterogeneous panel data models where the time series dimension T could be large relative to cross-section dimension N. The error process is assumed to be cross-sectionally independent. The test is based on the cumulative sum (CUSUM) of ordinary least squares (OLS) residuals. We derive the asymptotic distribution of the detecting statistic under the null hypothesis, while proving the consistency of the test under the alternative. Monte Carlo simulations and an empirical example show good performance of the test.
    Keywords: CUSUM test, panel data, structural change, common breaks
    JEL: C12 C23
    Date: 2021–05
  7. By: Canepa, Alessandra (University of Turin)
    Abstract: Johansen (2000) Bartlett correction factor for the LR test of linear restrictions on cointegrated vectors is derived under the i.i.d. Gaussian assumption for the innovation terms. However, the distribution of most data relating to financial variables are fat-tailed and often skewed, there is therefore a need to examine small sample inference procedures that require weaker assumptions for the innovation term. This paper suggests that using a non-parametric bootstrap to approximate a Bartlett-type correction provides a statistic that does not require specification of the innovation distribution and can be used by applied econometricians to perform a small sample inference procedure that is less computationally demanding than estimating the p-value of the observed statistic.
    Date: 2021–03
  8. By: Martin Bruns (University of East Anglia); Helmut Lütkepohl (DIW Berlin and The Free University of Berlin)
    Abstract: Different local projection (LP) estimators for structural impulse responses of proxy vector autoregressions are reviewed and compared algebraically and with respect to their small sample suitability for inference. Conditions for numerical equivalence and similarities of some estimators are provided. A new LP type estimator is also proposed which is very easy to compute. Two generalized least squares (GLS) projection estimators are found to be more accurate than the other LP estimators in small samples. In particular, a lag-augmented GLS estimator tends to be superior to its competitors and to perform as well as a standard VAR estimator for sufficiently large samples.
    Keywords: Structural vector autoregression, local projection, impulse responses, instrumental variable
    JEL: C32
    Date: 2021–05–28
  9. By: Marina Dias; Demian Pouzo
    Abstract: We propose a method for conducting asymptotically valid inference for treatment effects in a multi-valued treatment framework where the number of units in the treatment arms can be small and do not grow with the sample size. We accomplish this by casting the model as a semi-/non-parametric conditional quantile model and using known finite sample results about the law of the indicator function that defines the conditional quantile. Our framework allows for structural functions that are non-additively separable, with flexible functional forms and heteroskedasticy in the residuals, and it also encompasses commonly used designs like difference in difference. We study the finite sample behavior of our test in a Monte Carlo study and we also apply our results to assessing the effect of weather events on GDP growth.
    Date: 2021–05
  10. By: Karlsson, Sune (Örebro University School of Business); Mazur, Stepan (Örebro University School of Business); Nguyen, Hoang (Örebro University School of Business)
    Abstract: With uncertain changes of the economic environment, macroeconomic downturns during recessions and crises can hardly be explained by a Gaussian structural shock. There is evidence that the distribution of macroeconomic variables is skewed and heavy tailed. In this paper, we contribute to the literature by extending a vector autore- gression (VAR) model to account for a more realistic assumption of the multivariate distribution of the macroeconomic variables. We propose a general class of generalized hyperbolic skew Student's t distribution with stochastic volatility for the error term in the VAR model that allows us to take into account skewness and heavy tails. Tools for Bayesian inference and model selection using a Gibbs sampler are provided. In an empirical study, we present evidence of skewness and heavy tails for monthly macroe- conomic variables. The analysis also gives a clear message that skewness should be taken into account for better predictions during recessions and crises.
    Keywords: Vector autoregression; Skewness and heavy tails; Generalized hyper- bolic skew Students t distribution; Stochastic volatility; Markov Chain Monte Carlo
    JEL: C11 C15 C16 C32 C52
    Date: 2021–05–20
  11. By: Gaillac, Christophe; Gautier, Eric
    Abstract: This paper studies point identification of the distribution of the coefficients in some random coefficients models with exogenous regressors when their support is a proper subset, possibly discrete but countable. We exhibit trade-offs between restrictions on the distribution of the random coefficients and the support of the regressors. We consider linear models including those with nonlinear transforms of a baseline regressor, with an infinite number of regressors and deconvolution, the binary choice model, and panel data models such as single-index panel data models and an extension of the Kotlarski lemma.
    Keywords: Identification; Random Coefficients; Quasi-analyticity; Deconvolution
    Date: 2021–05
  12. By: Eric Auerbach
    Abstract: This paper provides additional results relevant to the setting, model, and estimators of Auerbach (2019a). Section 1 contains results about the large sample properties of the estimators from Section 2 of Auerbach (2019a). Section 2 considers some extensions to the model. Section 3 provides an application to estimating network peer effects. Section 4 shows the results from some simulations.
    Date: 2021–05
  13. By: Stauskas, Ovidijus (Department of Economics, Lund University)
    Abstract: A recent study proposed by Westerlund (CCE in Panels with General Unknown Factors, Econometrics Journal, 21, 264-276, 2018) showed that a very popular Common Correlated Effects (CCE) estimator is significantly more applicable than it was thought before. Contrary to the usual stationarity assumption, common factors can in fact be much more general and not unit root only. This also helps to alleviate the uncertainty over deterministic model components since they can be treated as unknown, similarly to unobserved stochastic factors. While very promising, these theoretical results concern only the pooled (CCEP) version of the estimator for the homogeneous parameters, which does no take heterogeneous effects into account. Therefore, it is natural to generalize these findings to the case of unit-specific slopes. It is especially interesting, because many previous studies on heterogeneous slopes did not rigorously account for the usual situation when the factors are proxied by more explanatory variables than needed. As a result, the current setup introduces more uniformity to the CCE theory. We demonstrate that save for some regularity conditions, CCEP and the mean group (CCEMG) estimators are asymptotically normal and unbiased under heterogeneous slopes and general unknown factors.
    Keywords: Panel data; CCE; Non-Stationarity; Factors; Heterogeneity
    JEL: C12 C23 C33
    Date: 2021–05–18
  14. By: Odendahl, Florens; Rossi, Barbara; Sekhposyan, Tatevik
    Abstract: We propose a novel forecast comparison methodology to evaluate models' relative forecasting performance when the latter is a state-dependent function of economic variables. In our bench¬mark case, the relative forecasting performance, measured by the forecast loss differential, is modeled via a threshold model. Importantly, we allow the threshold that triggers the switch from one state to the next to be unknown, leading to a non-standard test statistic due to the presence of a nuisance parameter. Existing tests either assume a constant out-of-sample forecast performance or use non-parametric techniques robust to time-variation; consequently, they may lack power against state-dependent predictability. Importantly, our approach is applicable to point forecasts as well as predictive densities. Monte Carlo results suggest that our proposed test statistics perform well in ï¬ nite samples and have better power than existing tests in selecting the best forecasting model in the presence of state dependence. Our test statistics uncover "pockets of predictability" in U.S. equity premia forecasts; the pockets are a state-dependent function of stock market volatility. Models using economic predictors perform signiï¬ cantly worse than a simple mean forecast in periods of high volatility, but, in periods of low volatility, the use of economic predictors may lead to small forecast improvements.
    Keywords: Forecast evaluation; Pockets of Predictability; State Dependence
    JEL: C52 C53 G17
    Date: 2020–08
  15. By: Borusyak, Kirill; Hull, Peter; Jaravel, Xavier
    Abstract: Many studies use shift-share (or "Bartik") instruments, which average a set of shocks with exposure share weights. We provide a new econometric framework for shift-share instrumental variable (SSIV) regressions in which identification follows from the quasi-random assignment of shocks, while exposure shares are allowed to be endogenous. The framework is motivated by an equivalence result: the orthogonality between a shift-share instrument and an unobserved residual can be represented as the orthogonality between the underlying shocks and a shock-level unobservable. SSIV regression coefficients can similarly be obtained from an equivalent shock-level regression, motivating shock-level conditions for their consistency. We discuss and illustrate several practical insights of this framework in the setting of Autor, Dorn, and Hanson (2013), estimating the effect of Chinese import competition on manufacturing employment across U.S. commuting zones.
    JEL: C18 C21 C26 F16 J21
    Date: 2020–08
  16. By: Dante Amengual (CEMFI, Spain); Gabriele Fiorentini (Università di Firenze, Italy; Rimini Centre for Economic Analysis); Enrique Sentana (CEMFI, Spain)
    Abstract: We show that the information matrix test for a multivariate normal random vector coincides with the sum of the two moment tests that look at the means of all the different third- and fourth-order multivariate Hermite polynomials, respectively. We also explain how to simulate its exact, parameter-free, finite sample distribution to any desired degree of accuracy for any dimension of the random vector and sample size. Specifically, we exploit the numerical invariance of the test statistic to affine transformations of the observed variables to simulate draws extremely quickly.
    Keywords: Exact test, Hessian matrix, Multivariate normality, Outer product of the score
    JEL: C30 C46 C52
    Date: 2021–05
  17. By: Yong Shi; Wei Dai; Wen Long; Bo Li
    Abstract: The Gaussian Process with a deep kernel is an extension of the classic GP regression model and this extended model usually constructs a new kernel function by deploying deep learning techniques like long short-term memory networks. A Gaussian Process with the kernel learned by LSTM, abbreviated as GP-LSTM, has the advantage of capturing the complex dependency of financial sequential data, while retaining the ability of probabilistic inference. However, the deep kernel Gaussian Process has not been applied to forecast the conditional returns and volatility in financial market to the best of our knowledge. In this paper, grid search algorithm, used for performing hyper-parameter optimization, is integrated with GP-LSTM to predict both the conditional mean and volatility of stock returns, which are then combined together to calculate the conditional Sharpe Ratio for constructing a long-short portfolio. The experiments are performed on a dataset covering all constituents of Shenzhen Stock Exchange Component Index. Based on empirical results, we find that the GP-LSTM model can provide more accurate forecasts in stock returns and volatility, which are jointly evaluated by the performance of constructed portfolios. Further sub-period analysis of the experiment results indicates that the superiority of GP-LSTM model over the benchmark models stems from better performance in highly volatile periods.
    Date: 2021–05
  18. By: Stefan Leknes; Sturla A. Løkken (Statistics Norway)
    Abstract: Reliable local demographic schedules are in high demand, but small area problems pose a challenge to estimation. The literature has directed little attention to the opportunities created by increased availability of high-quality geo-coded data. We propose the use of empirical Bayes methods based on a model with three hierarchical geographic levels to predict small area fertility schedules. The proposed model has a flexible specification with respect to age, which allows for detailed age heterogeneity in local fertility patterns. The model limits sampling variability in small areas, captures regional variations effectively, is robust to certain types of model misspecification, and outperforms alternative models in terms of prediction accuracy. The beneficial properties of the model are demonstrated through simulations and estimations on full-count Norwegian population data.
    Keywords: small area estimation; hierarchical linear models; empirical Bayes method; shrinkage; age-specific fertility
    JEL: J13 R58 C13 C18
    Date: 2021–04
  19. By: Thomas von Brasch; Arvid Raknerud (Statistics Norway)
    Abstract: In a seminal paper, Feenstra (1994) developed an instrumental variable estimator which is becoming increasingly popular for estimating demand elasticities. Soderbery (2015) extended this estimator and created a routine which was shown to be more robust to data outliers when the number of time periods is small or moderate. In this paper, we extend the Feenstra/Soderbery (F/S) estimator along two important dimensions to obtain a more efficient estimator: we handle the cases where there are no simultaneity problems, i.e. when supply is either elastic or inelastic, and we generalize the current practice of choosing a particular reference variety by creating a pooled estimator across all possible reference varieties. Using a Monte Carlo study, we show that our proposed estimator reduces the RMSE compared to the F/S estimator by between 60 and 90 percent across the whole parameter space.
    Keywords: Demand elasticity; Panel data; Two-stage estimator
    JEL: C13 C33 C36
    Date: 2021–04
  20. By: Chemla, Gilles; Hennessy, Christopher
    Abstract: Causal evidence from random assignment has been labeled "the most credible." We argue it is generally incomplete in finance/economics, omitting central parts of the true empirical causal chain. Random assignment, in eliminating self-selection, simultaneously precludes signaling via treatment choice. However, outside experiments, agents enjoy discretion to signal, thereby causing changes in beliefs and outcomes. Therefore, if the goal is informing discretionary decisions, rather than predicting outcomes after forced/mistaken actions, randomization is problematic. As shown, signaling can amplify, attenuate, or reverse signs of causal effects. Thus, traditional methods of empirical finance, e.g. event studies, are often more credible/useful.
    Keywords: Causal effect; CEO; Corporate Finance; Government Policy; household finance; investment; random assignment; selection; signal
    JEL: D82 E6 G14 G18 G28 G3 J24
    Date: 2020–08
  21. By: Valencia Caicedo, Felipe
    Abstract: This chapter surveys the usage of Instrumental Variables (IVs) and Regression Discontinuity Designs (RDDs) in economic history. I document the positive trends of economic history articles employing these methods using three different samples: top 20 journals in economics, top 5 journals in economic history and top five general interest journals in economics from 2000-2020. I detail two broad phases: seminal articles published from 2001 to 2011, and a second wave of studies refining these techniques appearing from 2012 to today (2020). I discuss some methodological refinements that have appeared recently in the econometrics field-in the IV and RDD fronts. I then present a practical guide on regression diagnostics, acknowledging that there are other useful sources of identification available to tackle potential endogeneity issues.
    Keywords: econometrics; economic history; instrumental variables; Regression Discontinuity Designs; survey
    JEL: A33 C1 C26 C36 N01
    Date: 2020–08
  22. By: Andrea Carriero; Todd E. Clark; Massimiliano Marcellino; Elmar Mertens
    Abstract: Interest rate data are an important element of macroeconomic forecasting. Projections of future interest rates are not only an important product themselves, but also typically matter for forecasting other macroeconomic and financial variables. A popular class of forecasting models is linear vector autoregressions (VARs) that include shorter- and longer-term interest rates. However, in a number of economies, at least shorter-term interest rates have now been stuck for years at or near their effective lower bound (ELB), with longer-term rates drifting toward the constraint as well. In such an environment, linear forecasting models that ignore the ELB constraint on nominal interest rates appear inept. To handle the ELB on interest rates, we model observed rates as censored observations of a latent shadow-rate process in an otherwise standard VAR setup. The shadow rates are assumed to be equal to observed rates when above the ELB. Point and density forecasts for interest rates (short term and long term) constructed from a shadow-rate VAR for the US since 2009 are superior to predictions from a standard VAR that ignores the ELB. For other indicators of financial conditions and measures of economic activity and inflation, the accuracy of forecasts from our shadow-rate specification is on par with a standard VAR that ignores the ELB.
    Keywords: Macroeconomic forecasting; effective lower bound; term structure; censored observations
    JEL: C34 C53 E17 E37 E43 E47
    Date: 2021–03–29

This nep-ecm issue is ©2021 by Sune Karlsson. It is provided as is without any express or implied warranty. It may be freely redistributed in whole or in part for any purpose. If distributed in part, please include this notice.
General information on the NEP project can be found at For comments please write to the director of NEP, Marco Novarese at <>. Put “NEP” in the subject, otherwise your mail may be rejected.
NEP’s infrastructure is sponsored by the School of Economics and Finance of Massey University in New Zealand.