Summary

  SUMMARY OF 2024 KANSAS ECONOMETRICS WORKSHOP

                                                             SUMMARY:

The 2024 Kansas Econometrics Workshop, organized by the Department of Economics of University of Kansas (KU), was successfully held on April 27 at The Oread Hotel on the Lawrence Campus with more than 30 attendees. This is one of the series of the Kansas econometrics workshops focusing on recent developments of econometric theories and methodologies as well as their applications in both economics and finance as well as data science. The main purpose of this series of workshops is to promote the advanced econometric research for faculty and graduate students at KU. The workshop included 12 invited speakers, including the famous econometricians and statisticians, and some young scholars. The following is the summary for each presentation.

Bill Barnett is currently the Oswald Distinguished Professor of Macroeconomics at the University of Kansas and Director at Center for Financial Stability in New York City. His research is in macroeconomics and econometrics, such as monetary asset demand modeling and extensions of index number theory to risk. At the 2024 Kansas Econometrics Workshop, he presented the paper entitled “Non-separability of Credit Card Services Within Divisia Monetary Aggregates”. In his presentation, he proposed the newly developed credit card augmented Divisia monetary aggregates to expand upon existing research in this area. The credit-card augmented Divisia monetary aggregate closely tracks the theoretical monetary aggregate, whereas the simple sum monetary aggregate does not. And it emerges as a preferable monetary policy indicator compared to the traditional federal funds rate within the recursive VAR framework. Furthermore, theoretical analysis reveals that the separability condition necessary for the existence of a monetary aggregator function would be violated if credit card deferred payment services were excluded from the monetary services component, unless all markets exhibit perfect characteristics.

Xuming He is the Kotzubei Beckmann Distinguished Professor of Statistics and Data Science and the Chair of the Department of Statistics and Data Science at Washington University in St. Louis. His research interests include robust statistics, quantile regression, Bayesian inference, and post-selection inference. At the 2024 Kansas Econometrics Workshop, he presented the paper entitled “Some Recent Developments in Expected Shortfall Regression”. In his presentation, he introduced a novel optimization-based semiparametric approach to estimating conditional expected shortfall. This approach is crafted to handle data heterogeneity while minimizing the need for model assumptions. It inherently adapts to diverse data structures across a range of models, often demonstrating superior performance compared to alternative methods. Notably, it does not necessitate the simultaneous modeling of quantile regression and can be implemented with only initial expected shortfall estimation.

Eric Zivot is the Robert Richards Chaired Professor of Economics at University of Washington and Chair of Department of Economics. His research interests include data science, econometrics, financial economics and risk management. At the 2024 Kansas Econometrics Workshop, he presented the paper entitled “Improving Price Leadership Share for Measuring Price Discovery”. In his presentation, he proposed an improvement to the information leadership measure of price discovery and the information leadership share measure. Their enhanced information leadership measure, which combines the price discovery share with the component share, effectively captures the proportion of initial reactions from competing markets to a permanent shock, even in scenarios where the residuals of reduced-form vector error correction models exhibit correlation. They showcased the efficacy of their enhanced measures by investigating price discovery dynamics for a range of Chinese stocks listed on both the Shanghai and Hong Kong exchanges (SH-HK), both pre and post the launch of the Shanghai-Hong Kong Stock Connect program.

Jeffrey Racine is a Professor at the Department of Economics, the Senator William McMaster Chair in Econometrics, and a Professor in the Graduate Program in Statistics in the Department of Mathematics and Statistics at McMaster University. His research areas include econometrics and statistics as well as monetary economics. At the 2024 Kansas Econometrics Workshop, he presented the paper entitled “Locally Adaptive Online Functional Data Analysis”. He believed that modern data is often functional in nature instead of point wise and he showed how people can learn the degree of (non)smoothness in functional data settings and they separate it from measurement noise. He introduced the functional data analysis and the corresponding functional data setting. This method is data-driven and locally adaptive to the regularity of the stochastic process. Then, he discussed about local regularity, the largest order fractional derivative admitted by the sample paths of regressor X as measured by the value of the local Holder exponent. Since this method requires nonparametric estimation, kernel functions are used here. The idea is to estimate by averaging over curves vertically. Estimation method for mean and covariate functions are provided as well. The estimation can be recursively updated online.

Runze Li is the Eberly Family Chair Professor in Statistics at Department of Statistics at Pennsylvania State University. His research interest includes variable selection and feature screening for high dimensional data, nonparametric modeling and semiparametric modeling and their application to social behavior science research. He is also interested in longitudinal data analysis and survival data analysis and their application to biomedical data analysis.  At the 2024 Kansas Econometrics Workshop, he presented the paper entitled “Feature-Splitting Algorithms for Ultrahigh Dimensional Quantile Regression”. In his presentation, he proposed feature-splitting algorithms for PQR in ultrahighdimension based on three-block ADMM and established the rate of convergence of theproposed algorithm and the theoretical convergence guarantee.There are some challenges in penalized quantile regression. For example, it encountersnon-smoothness due to the loss function and penalty function, ultrahigh dimensionality,and nonconvexity due to SCAD penalty. Estimation algorithms such as two-block ADMMand three-block ADMM have been suggested in the literature. While these methods couldsolve some of the aforementioned challenges, they can still be out of memory with a largerdimension. He proposed the so-called feature-splitting algorithm, which is a newthree-block semi-proximal ADMM framework that capacitates a parallel update of β. It splits the high dimensional variable to smaller blocks and speed up updates through parallelization. Based on the three-block semi-proximal ADMM, they come up with two algorithms: FSQRADMM-CD and FS-QRADMM-prox. Then, theories for the linear rate of convergence are established. Simulations are performed with the two algorithms with Lasso and SCAD, respectively, along with other methods used in history. Several criteria including average absolute error and size are used for comparison. In addition, the supermarket data are used as the realdata example. The results show that the proposed algorithms with SCAD penalty function perform better in general. He extended the analysis and algorithm for penalized quantile regression to the case where the dimension of independent variables is super high. The former methods will run out of memory with software available quickly under this case, while in this paper, splitting the ultrahigh-dimensional variables by feature will divide them into groups and solve the problem easier.

Le Wang is the David M. Kohl Chairprofessor of economics at Department of Economics at Virginia Tech University. His research areas include causal inferences, distributional and nonparametric econometrics, and data analytics and machine learning. At the 2024 Kansas Econometrics Workshop, he presented the paper entitled “Generalized Intergenerational Mobility Regressions”, which challenges conventional statistical approaches to intergenerational mobility research. He revealed subjective value judgments inherent in existing methods and proposed a new framework that allows for transparent economic preferences. This framework offers insights into the nuanced nature of mobility and its implications for various demographic groups and geographic regions. Through his work, he contributes significantly to advancing our understanding of economic dynamics and informing policy decisions.

Daniel Henderson is a Professor of economics at University of Alabama. His research focuses on applied nonparametric micro-econometrics, particularly in the economics of education and economic growth and development. At the 2024 Kansas Econometrics Workshop, he presented the paper entitled “Estimation and Inference for Varying Coefficient Multidimensional Fixed-effects Panel Data Models”. In his presentation, he introduced an estimation method and a set of hypothesis tests for varying coefficient multidimensional panel data regression models. The paper derives the asymptotic distribution of the proposed nonparametric estimator and develops central limit theory to conduct valid inference, considering the presence of multiple effects over differing dimensions. In this paper, they show the versatility of the inference methods for applied work by explicitly presenting test statistics for various hypothesis tests. They reject the random effects specification and find that the result obtained from the cross-sectional setting extends to the multidimensional panel data setting. The paper also includes an appendix containing detailed simulations supporting the asymptotic theory of the estimator and demonstrating that the testing infrastructure maintains correct asymptotic size and high power.

Wei Long is an Associate Professor of economics at the Department of Economics at Tulane University and specializes in econometrics and applied economics, particularly focusing on the economics of crime and financial econometrics. At the 2024 Kansas Econometrics Workshop, he presented the paper entitled "Estimation of Intergenerational Mobility: A Time-Varying Mixed Copula Method". In this study, he employed a time-varying mixed copula model to examine the trajectory of intergenerational mobility in the United States over the past three decades. Their analysis reveals a gradual decline in intergenerational mobility, especially concerning permanent incomes, between parents and children over time. This heightened income inequality persists across various racial and regional demographics. However, subgroup analyses uncover some heterogeneous patterns in the paths of intergenerational mobility. In general, this research sheds light on the evolving dynamics of intergenerational mobility and contributes to our understanding of income inequality in society.

Huixin Bi is a research and policy officer at the Economic Research Department of the Federal Reserve Bank of Kansas City. Her research interests are fiscal policy, sovereign debt and computational economics. At the 2024 Kansas Econometrics Workshop, he presented the paper entitled “Dynamics of Job Search Effort and Vacancies: Evidence from Classified Advertisements”. In this paper, they used machine learning techniques to construct measures of job posts, both from job seekers and firms, from digitized newspapers from the early twentieth century. They derived estimates for the elasticity of search effort costs and discipline a model of equilibrium search effort by the unemployed to examine its implications for labor market dynamics.

Shu Shen is an Associate Professor of economics at University of California at Davis. Her research focuses on econometric methods in the following topics: Regression discontinuity, GMM identification, Multiple testing, Treatment effect heterogeneity analysis, and Nonparametric, semiparametric distributional, and quantile Analysis. At the 2024 Kansas Econometrics Workshop she presented a paper entitled “Dynamics Panel Instrumental Variable Regression with Varying-Intensity Repeated Treatments: Theory and the China Syndrome Application”. She discussed dynamic treatment effects in panel settings with endogenous treatments and time-varying external instruments. Her work emphasized the importance of considering lagged effects and path-dependency in treatment effects, which are often overlooked in empirical research. She proposed a new benchmark model to incorporate treatment effect dynamics, addressing the limitations of existing models that ignore these dynamics. The model is applied to revisit the industry-level data from Acemoglu et al. (2016) to estimate the average contemporaneous effect of the China shock on U.S. employment. Overall, her work contributes to the literature on dynamic treatment effects and provides a more precise understanding of the impact of treatments over time in applied economics. Furthermore, it offers a theoretical framework that challenges the validity of existing empirical approaches when external instruments are serially correlated, and treatment effect dynamics are present.

Molin Zhong is a principal economist at Board of Governors of the Federal Reserve System. His main research areas include empirical macroeconomics and forecasting. He gave a talk on “Risk and Monetary Policy in a Data-Rich Model”. In this paper, they estimated a factor augmented VAR model with endogenous stochastic volatility to quantify the role of financial shocks and monetary policy in shaping sectoral and economywide risk dynamics. They uncovered evidence of correlations between the mean and volatility estimates of extracted factors, which indicates pervasive asymmetric risk dynamics. Further, they found wide heterogeneity in risk behavior across different variables at the aggregate and sectoral levels. Structural analysis suggests that a monetary policy shock has strong effects on the excess bond premium, which leads an easing of monetary policy to decrease the volatility of innovations.

John Keating is a Professor of macroeconomics at the University of Kansas. His research interests include macroeconomics, time series econometrics, monetary theory and policy, as well as DSGE modeling. At the 2024 Kansas Econometrics Workshop, he presented the paper entitled “Structurally Interpreting Misspecified Long-run Recursive Vector Autoregressions”. In this paper, he proved that misspecified long-run recursive VAR models can still be informative about the underlying structure, even under weaker assumptions than would be necessary to consistently estimate a structure. He also showed that some existing approaches in the literature can be generalized to reduced-form time series model of arbitrary size. In particular, the parameters of a general linear structure with an arbitrary number of endogenous variables are mapped into coefficients of a long-run recursive VAR model. Given assumptions about signs of some of the structural parameters, the sign of some other structural effect can be inferred using the long-run recursive model's estimates.