This paper focuses on the econometric methodology developed by Professor David Hendry, Hendry’s Methodology (HM), with his associates over more than 50 years. It discusses the statistical foundations of the methodology and formulates a framework that structures the content of HM into four main planks, incorporating 11 basic methodology features which are then explained in detail. They include: the Local Data Generation Process (LDGP); the General Unrestricted Model (GUM); the reduction process from the GUM to a final congruent model; the question of having more variables than observations; the problem of the contamination of the data by outliers, location shifts, truncated trends, changes in estimated parameters and possible changes with specific shape; and the statistical problems in selecting economic variables. A selection of the many applications of HM is discussed in this paper. This methodology has been incorporated into Climate Econometrics and the paper illustrates the connection between problems in Climate Science and Economics. Moreover, a big component of HM is forecasting and the pivotal findings incorporated into robustifying forecasting methods developed by Hendry and associates are outlined and discussed in this paper.
This study aims to examine econometric models that are used to describe and analyze individual choice behavior. Choice models examine relationships between attributes of choices, characteristics of decision makers and the decisions that choosers make. The hybrid choice model (HCM) adds latent attitudes and perceptions to the set of influencers of choice outcomes and integrates latent variable models with discrete choice models. The models considered here extend two leading platforms for contemporary choice modeling in Transportation Research: (1) the mixed (random parameters) or latent segmentation (latent class) model, and (2) stated preference analysis. This paper will survey some of the received and contemporary theoretical and Econometric aspects of this corner of discrete choice modeling. The authors will present several applications from the received literature for illustration.
We argue that frequentist hypothesis testing - the dominant statistical evaluationparadigm in empirical research - is fundamentally unsuited for analysis of the non-experimental data prevalent in economics and other social sciences. Frequentisttests comprise incompatible repeated sampling frameworks that do not obey theLikelihood Principle (LP). For probabilistic inference, methods that are guided bythe LP, that do not rely on repeated sampling, and that focus on model compari-son instead of testing (e.g., subjectivist Bayesian methods) are better suited forpassively observed social science data and are better able to accommodate thehuge model uncertainty and highly approximative nature of structural models inthe social sciences. In addition to formal probabilistic inference, informal modelevaluation along relevant substantive and practical dimensions should play a leading role. We sketch the ideas of an alternative paradigm containing these elements.
