Structural econometric modeling and time series analysis

2016

Econometric modelling with time series: SpeciÞ cation, Estimation, and Testing is a graduate textbook covering a broad range of topics in time series econometrics. The book is unique and valuable in three aspects. First, the book tries to bridge the gap between the purely theoretical view of time series analysis such as the one offered by Hamilton (1994) and applied approach offered by Enders (2015). Indeed, every chapter in approximately 30 pages provides enough theoretical background and rigorous asymptotic theory which is followed by two empirical examples, usually one taken from economics and one from Þ nance. Second, the focus of the book is on maximum likelihood estimation which makes it stand out among other books on time series analysis. Third, authors offer solutions for all examples and exercises from the book in three different software: GAUSS, MATLAB, and R. Most of empirical examples are also available in RATS. In order to follow the content of the book with ease, mathe...

A growing field is related to automatized Time Series analysis, through complicated due to the dependence of observed and hidden dimensions often presented in these data types. In this report the problem is motivated by a Brazilian financial company interested in unraveling relation structure explanation of the Japanese' CPI ex-fresh Food \& Energy across 157 economical exogenous variables, with very limiting data. The problem becomes more complex when considering that each variable can enter the model with lags of 0 to 8 periods, as well as an additional restriction of admitting only a positive relationship. This report discusses three possible treatments involving models for structured time series, the most relevant approach found in this study is a Dynamic Regression Model combined with a Stepwise algorithm, which allows the most relevant variables, as well as their respective lags, to be found and inserted in the model with low computational cost.

Shown is a new method for estimating linear models with general time-varying structures such as the State Space Model based on the idea that the models can be represented as a classical regression model. The parameters are all estimated by OLS or GLS. An application of the smoothing to a time-varying AR model is presented.

Omega, 1998

Industrial production data series are volatile and often also cyclical. Hence, univariate time series models which allow for these features are expected to generate relatively accurate forecasts of industrial production. A particular class of unobservable components models Ð structural time series models Ð is used to generate forecasts of Austrian and German industrial production. A widely applied ARIMA model is used as a baseline for comparison. The empirical results show that the basic structural model generates more accurate forecasts than the ARIMA model when accuracy is measured in terms of size of error or directional change; and that the basic structural model forecasts better than the structural model with a cyclical component included on the basis of numerical measures, and tracking error for month-to-month changes.

indiana.edu

This mini course deals with recent developments in the econometrics of dynamic discrete choice structural models. The emphasis of the course is in the techniques, though these techniques will be illustrated in the context of actual applications. Some of the topics that we will cover are simulation and approximation methods, methods which avoid repeated full solution of the structural model in estimation, and techniques that allow researchers to estimate dynamic equilibrium models, both strategic and competitive.

Journal of Macroeconomics, 1992

We review the advancement of nonstationary time series analysis from the perspective of Cowles Commission structural equation approach. We argue that despite the rich repertoire nonstationary time series analysis provides to analyze how do variables respond dynamically to shocks through the decomposition of a dynamic system into long-run and short-run relations, nonstationarity does not invalid the classical concerns of structural equation modeling-identification and simultaneity bias. The same rank condition for identification holds for stationary and nonstationary data and some sort of instrumental variable estimators will have to be employed to yield consistency. However, nonstationarity does raise issues of inference if the rank of cointegration or direction of nonstationarity is not known a priori. The usual test statistics may not be chi-square distributed because of the presence of unit roots distributions. Classical instrumental variable estimators have to be modified to ensure valid inference. 1 The introduciton of intercept terms complicates algebraic manipulation without changing the basic message. For detail, see [28].

International Journal of Forecasting, 1992

The present volume contains an introduction and three sections: Tools to characterize parameter changes in linear regression models in various patterns of nonconstancy are studied by A.H. Westlund and B. Tornkvist in Chapter 8: "On the identification of time of structural changes by MOSUM-SQ and CUSUM-SQ procedures". Conditions for the CUSUM-SQ test to have nontrivial local power are given by W. Ploberger in "The local power of the CUSUM-SQ test against heteroscedasticity". J. Praagman's "Bahadur efficiency of tests for a shift in location of normal populations" Preface vii treats two generalized forms of the most relevant test statistics. Finally, in Chapter 11, Z. Wasilewski demonstrates "The use of graphical displays in the analysis of structural change" for the investigation of regression residuals. IlL Model Building in the Presence of Structural Change This section addresses models that are in some sense generalizations of constant-parameter models, so that they can assimilate structural changes. In Chapter 12, "Adaptive estimation and structural change in regression and time series models", J. Ledolter reviews heuristic and model-based approaches to adaptive estimation of regression parameters and discusses in detail the case where the parameters follow ARMA processes. The use of exponential weights for adaptive estimation is treated in "An adaptive method of regression analysis" by Y.P. Lukashin. J. Dziechciarz, in "Changing and random coefficient models. A survey", reviews comprehensively the related literature (about 200 references). P.M. Robinson discusses, in "Nonparametric estimation of time-varying parameters", the construction and properties of a kernel-based estimator of the regression coefficient. V.V. Fedorov, in "Latent variables in regression analysis", treats two types of regression models with unobservable variables, together with refornmlations that can be handled by traditional regression analysis techniques. L.D. Broemeling returns in Chapter 17, "Structural change and time series analysis", to demonstrate a Bayesian approach in analyzing a time series model for data so that the trend or the autocovariance function changes at an unknown time point. H. Tong, in "Thresholds, stability, nonlinear forecasting, and irregularly sampled data", examines threshold models, Le., global models composed of submodels for areas delineated by thresholds. "Forecasting in situations of structural change: A general approach" , by F .X. Diebold and P. Pauly, presents a method of combining forecasts to compensate for poor primary forecasts on the basis of time-varying weighting. J. Kleffe, in "Updating. parameters of linear change point models", discusses an algorithm for efficiently updating the residual slUll of squares applied in two-phase regression with a shifting change point. IV. Data Analysis and Modeling This section deals with real-life structural change situations. P.K. Sen, in "Change point problem relating to the poverty structure", constructs and analyzes poverty indices, based on income distribution. T. Ozaki and V.H. Ozaki, in "Statistical identification of nonlinear dynamics in macroeconomics using nonlinear time series models", describe a model representing both Keynesian and monetarist viewpoints vis avis the dynamics of the Hicksian IS-LM concept by a difference in model parameters. A. Keller's Chapter 23, "Econometrics of technical change: Techniques and problems", surveys studies concerning "technical progress"-an essential notion of economic growth with implications and pitfalls for data observations, model specifications, and estimation procedures. On the basis of interest rates for Austria, W. Polasek demonstrates, in "Local autoregression models for detection of changes in causality" , an approach to analyze nonstation-Vlll Statistical A nalysis and Forecasting of Economic Structural Change arity by applying local stationary autoregressive processes. Finally, in Chapter 25, J.-M. Dufour presents an empirical study, "Investment, taxation, and econometric policy evaluation: Some evidence of the Lucas critique", which discusses Lucas's arguments that parameters in econometric relationships reflect economic agents' decision rules. Foreword As Professor Hackl has already pointed out in his preface, this work on statistical identification of economic structural change was originally conceived as p~t of a common research project in which the other part was concerned with economic analysis and forecasting of economic growth and structural change. For reasons beyond our control, these projects had to be separated. This was unfortunate for several reasons. The most important one is that the results of the statistical project were to have been applied and tested in the economic project, and the practical problems encountered in the economic project were to have been analyzed and solved in the statistical project. Although this ideal arrangement was not possible, we strove to compensate for this loss. Within the Sonderforschungsbereich 303 (Special Research Unit 303) at Bonn University, three research projects have been carried out. The corresponding research reports are given by C. Weihs (1987) ("Auswirkungen von Fehlern in den Daten auf Parameterschatzungen und Prognosen", in K.-A. Scheffer

2007

Presents the main statistical tools of econometrics, focusing specifically on modern econometric methodology. The authors unify the approach by using a small number of estimation techniques, mainly generalized method of moments (GMM) estimation and kernel smoothing. The choice of GMM is explained by its relevance in structural econometrics and its preeminent position in econometrics overall. Split into four parts, Part I explains general methods. Part II studies statistical models that are best suited for microeconomic data. Part III deals with dynamic models that are designed for macroeconomic and financial applications. In Part IV the authors synthesize a set of problems that are specific to statistical methods in structural econometrics, namely identification and over-identification, simultaneity, and unobservability. Many theoretical examples illustrate the discussion and can be treated as application exercises. Nobel Laureate James A. Heckman offers a foreword to the work.