
on Econometrics 
By:  Martin Emil Jakobsen; Jonas Peters 
Abstract:  In causal settings, such as instrumental variable settings, it is well known that estimators based on ordinary least squares (OLS) can yield biased and nonconsistent estimates of the causal parameters. This is partially overcome by twostage least squares (TSLS) estimators. These are, under weak assumptions, consistent but do not have desirable finite sample properties: in many models, for example, they do not have finite moments. The set of Kclass estimators can be seen as a nonlinear interpolation between OLS and TSLS and are known to have improved finite sample properties. Recently, in causal discovery, invariance properties such as the moment criterion which TSLS estimators leverage have been exploited for causal structure learning: e.g., in cases, where the causal parameter is not identifiable, some structure of the nonzero components may be identified, and coverage guarantees are available. Subsequently, anchor regression has been proposed to tradeoff invariance and predictability. The resulting estimator is shown to have optimal predictive performance under bounded shift interventions. In this paper, we show that the concepts of anchor regression and Kclass estimators are closely related. Establishing this connection comes with two benefits: (1) It enables us to prove robustness properties for existing Kclass estimators when considering distributional shifts. And, (2), we propose a novel estimator in instrumental variable settings by minimizing the mean squared prediction error subject to the constraint that the estimator lies in an asymptotically valid confidence region of the causal parameter. We call this estimator PULSE (puncorrelated least squares estimator) and show that it can be computed efficiently, even though the underlying optimization problem is nonconvex. We further prove that it is consistent. 
Date:  2020–05 
URL:  http://d.repec.org/n?u=RePEc:arx:papers:2005.03353&r=all 
By:  Tobias Hartl; Rolf Tschernig; Enzo Weber 
Abstract:  We develop a generalization of unobserved components models that allows for a wide range of longrun dynamics by modelling the permanent component as a fractionally integrated process. The model does not require stationarity and can be cast in state space form. In a multivariate setup, fractional trends may yield a cointegrated system. We derive the Kalman filter estimator for the common fractionally integrated component and establish consistency and asymptotic (mixed) normality of the maximum likelihood estimator. We apply the model to extract a common longrun component of three US inflation measures, where we show that the $I(1)$ assumption is likely to be violated for the common trend. 
Date:  2020–05 
URL:  http://d.repec.org/n?u=RePEc:arx:papers:2005.03988&r=all 
By:  Jiaming Mao; Zhesheng Zheng 
Abstract:  We propose a novel method for modeling data by using structural models based on economic theory as regularizer for statistical models. We show that even if a structural model is misspecified, as long as it is informative about the datagenerating mechanism, our method can outperform both the (misspecified) structural model and unstructuralregularized statistical models. Our method permits a Bayesian interpretation of theory as prior knowledge and can be used both for statistical prediction and causal inference. It contributes to transfer learning by showing how incorporating theory into statistical modeling can significantly improve outofdomain predictions and offers a way to synthesize reducedform and structural approaches to causal effect estimation. Simulation experiments demonstrate the potential of our method in various settings, including firstprice auctions, dynamic models of entry and exit, and demand estimation with instrumental variables. Our method has potential applications not only in economics, but in other (social) scientific disciplines whose theoretical models offer important insight but are subject to significant misspecification concerns. 
Date:  2020–04 
URL:  http://d.repec.org/n?u=RePEc:arx:papers:2004.12601&r=all 
By:  Samuele Centorrino; Mar\'ia P\'erezUrdiales 
Abstract:  We provide a closedform maximum likelihood estimation of stochastic frontier models with endogeneity. We consider crosssection data when both components of the composite error term may be correlated with inputs and environmental variables. Under appropriate restrictions, we show that the conditional distribution of the stochastic inefficiency term is a folded normal distribution. The latter reduces to the halfnormal distribution when both inputs and environmental variables are independent of the stochastic inefficiency term. Our framework is thus a natural generalization of the normal halfnormal stochastic frontier model with endogeneity. Among other things, this allows us to provide a generalization of the BatteseCoelli estimator of technical efficiency. Our maximum likelihood estimator is computationally fast and easy to implement. We showcase its finite sample properties in montecarlo simulations and an empirical application to farmers in Nepal. 
Date:  2020–04 
URL:  http://d.repec.org/n?u=RePEc:arx:papers:2004.12369&r=all 
By:  Licht, Adrian; Escribano Saez, Alvaro; Blazsek, Szabolcs Istvan 
Abstract:  In this paper, the benefits of statistical inference of scoredriven statespacemodels are incorporated into the inference of dynamic stochastic general equilibrium (DSGE)models. We focus on DSGE models, for which a Gaussian ABCD representation exists. Precisionof statistical estimation is improved, by using a scoredriven multivariate tdistribution for theerrors. First, the updating term of the transition equation of the ABCD representation isreplaced by the conditional score of the loglikelihood (LL) with respect to location. Second,the timeconstant scale parameters of the error terms in the measurement equation of the ABCDrepresentation are replaced by a dynamic parameter that is updated by the conditional score ofthe LL with respect to scale. Impulse response functions (IRFs) and conditions of the maximumlikelihood (ML) estimator are presented. In the empirical application, a benchmark DSGE modelis estimated for real data on US economic output, inflation and interest rate for the period of19542019. The scoredriven ABCD representation improves the estimation precision of theGaussian ABCD representation. The scoredriven ABCD representation with dynamic scaleprovides the best description of the time series data, by identifying a structural change in thesample period and providing the most precise IRF estimates. 
Keywords:  BetaTEgarch; Generalized Autoregressive Score (Gas); Dynamic Conditional Score (Dcs); Dynamic Stochastic General Equilibrium (Dsge) 
Date:  2020–05–07 
URL:  http://d.repec.org/n?u=RePEc:cte:werepe:30347&r=all 
By:  Anna Mikusheva; Liyang Sun 
Abstract:  We develop a concept of weak identification in linear IV models in which the number of instruments can grow at the same rate or slower than the sample size. We propose a jackknifed version of the classical weak identificationrobust AndersonRubin (AR) test statistic. Largesample inference based on the jackknifed AR is valid under heteroscedasticity and weak identification. The feasible version of this statistic uses a novel variance estimator. The test has uniformly correct size and good power properties. We also develop a pretest for weak identification that is related to the size property of a Wald test based on the Jackknife Instrumental Variable Estimator (JIVE). This new pretest is valid under heteroscedasticity and with many instruments. 
Date:  2020–04 
URL:  http://d.repec.org/n?u=RePEc:arx:papers:2004.12445&r=all 
By:  Danilo LeivaLeon; Luis Uzeda 
Abstract:  We introduce a new class of timevarying parameter vector autoregressions (TVPVARs) where the identified structural innovations are allowed to influence — contemporaneously and with a lag — the dynamics of the intercept and autoregressive coefficients in these models. An estimation algorithm and a parametrization conducive to model comparison are also provided. We apply our framework to the US economy. Scenario analysis suggests that the effects of monetary policy on economic activity are larger and more persistent in the proposed models than in an otherwise standard TVPVAR. Our results also indicate that costpush shocks play an important role in understanding historical changes in inflation persistence. 
Keywords:  Econometric and statistical methods; Inflation and prices; Transmission of monetary policy 
JEL:  C32 E52 
Date:  2020–05 
URL:  http://d.repec.org/n?u=RePEc:bca:bocawp:2016&r=all 
By:  Samuel Brien; Michael Jansson (UC Berkeley and CREATES); Morten Ørregaard Nielsen (Queen's University and CREATES) 
Abstract:  We study largesample properties of likelihood ratio tests of the unit root hypothesis in an autoregressive model of arbitrary, finite order. Earlier research on this testing problem has developed likelihood ratio tests in the autoregressive model of order one, but resorted to a plugin approach when dealing with higherorder models. In contrast, we consider the full model and derive the relevant largesample properties of likelihood ratio tests under a localtounity asymptotic framework. As in the simpler model, we show that the full likelihood ratio tests are nearly efficient, in the sense that their asymptotic local power functions are virtually indistinguishable from the Gaussian power envelopes. 
Keywords:  Efficiency, Likelihood ratio test, Nuisance parameters, Unit root hypothesis 
JEL:  C12 C22 
Date:  2020–05 
URL:  http://d.repec.org/n?u=RePEc:qed:wpaper:1429&r=all 
By:  Sium Bodha Hannadige; Jiti Gao; Mervyn J. Silvapulle; Param Silvapulle 
Abstract:  This paper develops a method for forecasting a nonstationary time series, such as GDP, using a set of highdimensional panel data as predictors. To this end, we use what is known as a factor augmented regression [FAR] model that contains a small number of estimated factors as predictors; the factors are estimated using time series data on a large number of potential predictors. The validity of this method for forecasting has been established when all the variables are stationary and also when they are all nonstationary, but not when they consist of a mixture of stationary and nonstationary ones. This paper fills this gap. More specifically, we develop a method for constructing an asymptotically valid prediction interval using the FAR model when the predictors include a mixture of stationary and nonstationary factors; we refer to this as mixtureFAR model. This topic is important because typically time series data on a large number of economic variables is likely to contain a mixture of stationary and nonstationary variables. In a simulation study, we observed that the mixtureFAR performed better than its competitor that requires all the variables to be nonstationary. As an empirical illustration, we evaluated the aforementioned methods for forecasting the nonstationary variables, GDP and Industrial Production [IP], using the quarterly panel data on US macroeconomic variables, known as FREDD. We observed that the mixtureFAR model proposed in this paper performed better than its aforementioned competitors. 
Keywords:  bootstrap,generated factors, panel data, prediction interval. 
JEL:  C22 C33 C38 C53 
Date:  2020 
URL:  http://d.repec.org/n?u=RePEc:msh:ebswps:202019&r=all 
By:  Licht, Adrian; Escribano Saez, Alvaro; Blazsek, Szabolcs Istvan 
Abstract:  Relevant works from the literature on crude oil market use structural vector autoregressive(SVAR) models with several lags to approximate the true model for the variables change in globalcrude oil production, global real economic activity and log real crude oil prices. Those variables involveseasonality, cointegration, structural changes, and outliers. We introduce nonlinear Markovswitchingscoredriven models with common trends of the multivariate tdistribution (MSSeasonaltQVAR), forwhich filters are optimal according to the KullbackLeibler divergence. We find that MSSeasonaltQVAR provides a better approximation of the true data generating process and more precise shortrunand longrun impulse responses than SVAR. 
Keywords:  Markov RegimeSwitching Models; Outliers And Structural Changes; Nonlinear CoIntegration; ScoreDriven Models; Global Crude Oil Market 
JEL:  C52 C51 C32 
Date:  2020–05–07 
URL:  http://d.repec.org/n?u=RePEc:cte:werepe:30346&r=all 
By:  Niko Hauzenberger; Florian Huber; Gary Koop 
Abstract:  Timevarying parameter (TVP) regression models can involve a huge number of coefficients. Careful prior elicitation is required to yield sensible posterior and predictive inferences. In addition, the computational demands of Markov Chain Monte Carlo (MCMC) methods mean their use is limited to the case where the number of predictors is not too large. In light of these two concerns, this paper proposes a new dynamic shrinkage prior which reflects the empirical regularity that TVPs are typically sparse (i.e. time variation may occur only episodically and only for some of the coefficients). A scalable MCMC algorithm is developed which is capable of handling very high dimensional TVP regressions or TVP Vector Autoregressions. In an exercise using artificial data we demonstrate the accuracy and computational efficiency of our methods. In an application involving the term structure of interest rates in the eurozone, we find our dynamic shrinkage prior to effectively pick out small amounts of parameter change and our methods to forecast well. 
Date:  2020–05 
URL:  http://d.repec.org/n?u=RePEc:arx:papers:2005.03906&r=all 
By:  Thomas H. Joergensen (CEBI, Department of Economics, University of Copenhagen) 
Abstract:  Across many fields in economics, a common approach to estimation of economic models is to calibrate a subset of model parameters and keep them fixed when estimating the remaining parameters. Calibrated parameters likely affect conclusions based on the model but estimation time often makes a systematic investigation of the sensitivity to calibrated parameters infeasible. I propose a simple and computationally lowcost measure of the sensitivity of parameters and other objects of interest to the calibrated parameters. In the main empirical application, I revisit the analysis of lifecycle savings motives in Gourinchas and Parker (2002) and show that some estimates are sensitive to calibrations. 
Keywords:  Sensitivity, Transparency, Structural Estimation, Calibration, Savings Motives 
JEL:  C10 C52 C60 
Date:  2020–04–27 
URL:  http://d.repec.org/n?u=RePEc:kud:kucebi:2014&r=all 
By:  Adam McCloskey; Pascal Michaillat 
Abstract:  Statistical hypothesis tests are a cornerstone of scientific research. The tests are informative when their size is properly controlled, so the frequency of rejecting true null hypotheses (type I error) stays below a prespecified nominal level. Publication bias exaggerates test sizes, however. Since scientists can typically only publish results that reject the null hypothesis, they have the incentive to continue conducting studies until attaining rejection. Such $p$hacking takes many forms: from collecting additional data to examining multiple regression specifications, all in the search of statistical significance. The process inflates test sizes above their nominal levels because the critical values used to determine rejection assume that test statistics are constructed from a single studyabstracting from $p$hacking. This paper addresses the problem by constructing critical values that are compatible with scientists' behavior given their incentives. We assume that researchers conduct studies until finding a test statistic that exceeds the critical value, or until the benefit from conducting an extra study falls below the cost. We then solve for the incentivecompatible critical value (ICCV). When the ICCV is used to determine rejection, readers can be confident that size is controlled at the desired significance level, and that the researcher's response to the incentives delineated by the critical value is accounted for. Since they allow researchers to search for significance among multiple studies, ICCVs are larger than classical critical values. Yet, for a broad range of researcher behaviors and beliefs, ICCVs lie in a fairly narrow range. 
Date:  2020–05 
URL:  http://d.repec.org/n?u=RePEc:arx:papers:2005.04141&r=all 
By:  Guanyu Hu; Yishu Xue; Zhihua Ma 
Abstract:  In regional economics research, a problem of interest is to detect similarities between regions, and estimate their shared coefficients in economics models. In this article, we propose a mixture of finite mixtures (MFM) clustered regression model with auxiliary covariates that account for similarities in demographic or economic characteristics over a spatial domain. Our Bayesian construction provides both inference for number of clusters and clustering configurations, and estimation for parameters for each cluster. Empirical performance of the proposed model is illustrated through simulation experiments, and further applied to a study of influential factors for monthly housing cost in Georgia. 
Date:  2020–04 
URL:  http://d.repec.org/n?u=RePEc:arx:papers:2004.12022&r=all 
By:  Slichter, David 
Abstract:  This paper develops a simple diagnostic for the selection on observables assumption in the case of a binary treatment variable. I show that, under common assumptions, when selection on observables does not hold, designs based on selection on observables will estimate treatment effects approaching infinity or negative infinity among observations with propensity scores close to 0 or 1. Researchers can check for violations of selection on observables either informally by looking for a "smile" shape in a binned scatterplot, or with a simple formal test. When selection on observables fails, the researcher can detect the sign of the resulting bias. 
Keywords:  unconfoundedness, diagnostic test 
JEL:  C21 C29 
Date:  2020–04–25 
URL:  http://d.repec.org/n?u=RePEc:pra:mprapa:99921&r=all 
By:  Shalva Mkhatrishvili (Macroeconomic Research Division, National Bank of Georgia); Douglas Laxton (NOVA School of Business and Economics, Saddle Point Research, The Better Policy Project); Davit Tutberidze (Macroeconomic Research Division, National Bank of Georgia); Tamta Sopromadze (Macroeconomic Research Division, National Bank of Georgia); Saba Metreveli (Macroeconomic Research Division, National Bank of Georgia); Lasha Arevadze (Macroeconomic Research Division, National Bank of Georgia); Tamar Mdivnishvili (Macroeconomic Research Division, National Bank of Georgia); Giorgi Tsutskiridze (Macroeconomic Research Division, National Bank of Georgia) 
Abstract:  There has been an increased acceptance of nonlinear linkages being the major driver of the most pronounced phases of business and financial cycles. However, modelling these nonlinear phenomena has been a challenge, since existing solutions methods are either efficient but not able to accurately capture nonlinear dynamics (e.g. linear methods), or accurate but quite resourceintensive (e.g. stacked system or stochastic Extended Path). This paper proposes two new solution approaches that try to be accurate enough and less costly. Moreover, one of those methods lets us do Kalman filtering on nonlinear models in a nonlinear way, which is also important for this kind of models, in general, to be more policyrelevant. Impulse responses, simulations and Kalman filtering exercises show the advantages of those new approaches when applied to a simple, but strongly nonlinear, monetary policy model. 
Keywords:  Nonlinear dynamic models, Solution methods, Monetary policy 
JEL:  C60 C61 C63 E17 
Date:  2019–10 
URL:  http://d.repec.org/n?u=RePEc:aez:wpaper:01/2019&r=all 
By:  Benedikt M. P\"otscher; David Preinerstorfer 
Abstract:  We develop theoretical finitesample results concerning the size of wild bootstrapbased heteroskedasticity robust tests in linear regression models. In particular, these results provide an efficient diagnostic check, which can be used to weed out tests that are unreliable for a given testing problem in the sense that they overreject substantially. This allows us to assess the reliability of a large variety of wild bootstrapbased tests in an extensive numerical study. 
Date:  2020–05 
URL:  http://d.repec.org/n?u=RePEc:arx:papers:2005.04089&r=all 
By:  Franses, Ph.H.B.F. 
Abstract:  This paper introduces a new autoregressive model, with the specific feature that the lag structure can vary over time. More precise, and to keep matters simple, the autoregressive model sometimes has lag 1, and sometimes lag 2. Representation, autocorrelation, specification, inference, and the creation of forecasts are presented. A detailed illustration for annual inflation rates for eight countries in Africa shows the empirical relevance of the new model. Various potential extensions are discussed. 
Keywords:  Autoregression, Timevarying lags, Forecasting 
JEL:  C22 C53 
Date:  2020–04–01 
URL:  http://d.repec.org/n?u=RePEc:ems:eureir:126706&r=all 
By:  Davide Del Prete; Laura Forastiere; Valerio Leone Sciabolazza 
Abstract:  This paper presents a methodology to draw causal inference in a nonexperimental setting subject to network interference. Specifically, we develop a generalized propensity scorebased estimator that allows us to estimate both direct and spillover effects of a continuous treatment, which spreads through weighted and directed edges of a network. To showcase this methodology, we investigate whether and how spillover effects shape the optimal level of policy interventions in agricultural markets. Our results show that, in this context, neglecting interference may lead to a downward bias when assessing policy effectiveness. 
Date:  2020–04 
URL:  http://d.repec.org/n?u=RePEc:arx:papers:2004.13459&r=all 
By:  Candelaria, Luis E. (University of Warwick); Ura, Takuya (University of California, Davis) 
Abstract:  This paper considers a network formation model when links are potentially measured with error. We focus on a gametheoretical model of strategic network formation with incomplete information, in which the linking decisions depend on agents’ exogenous attributes and endogenous network characteristics. In the presence of link misclassification, we derive moment conditions that characterize the identified set for the preference parameters associated with homophily and network externalities. Based on the moment equality conditions, we provide an inference method that is asymptotically valid when a single network of many agents is observed. Finally, we apply our proposed method to study trust networks in rural villages in southern India. 
Keywords:  Misclassification ; Network formation models ; Strategic interactions ; Incomplete information JEL codes: C13 ; C31 
Date:  2020 
URL:  http://d.repec.org/n?u=RePEc:wrk:warwec:1258&r=all 
By:  Ruiz Ortega, Esther; Nieto Delfin, Maria Rosa 
Abstract:  Although the Basel Accords require financial institutions to report daily predictions ofValue at Risk (VaR) computed using tenday returns, a vast part of the literature deals withVaR predictions based on oneday returns. From the practitioner point of view, some ofthe conclusions about the best methods to estimate oneperiod VaR could not be directlygeneralized to multiperiod VaR. Consequently, in the context of twostep VaR predictors,we use simulated and real data to compare direct and iterated predictions of multiperiodVaR based on tenday returns assuming that the conditional variances of oneperiod returnsfollow a GARCHtype model. We show that multiperiod VaR predictions based on iteratingan asymmetric GJR model with normal or bootstrapped errors are often preferred whencompared with direct methods that are often biased and inefficient. 
Keywords:  Risk; MultiStep Forecasts; Gjr Model; Feasible Historical Simulation 
JEL:  C58 C53 C22 G17 
Date:  2020–05–07 
URL:  http://d.repec.org/n?u=RePEc:cte:wsrepe:30349&r=all 