Skip to main content
eScholarship
Open Access Publications from the University of California

Four essays in econometrics

  • Author(s): Lu, Xun (Sean)
  • et al.
Abstract

This dissertation contains four self-contained essays in econometrics. In Chapter 1, we give natural definitions of direct and total structural causality applicable to both structural VARs and recursive structures representing time -series natural experiments. These concepts enable us to forge a previously missing link between Granger (G-) causality and structural causality by showing that, given a corresponding conditional form of exogeneity, G- causality holds if and only if a corresponding form of structural causality holds. We illustrate with studies of oil and gasoline prices, monetary policy and industrial production, and stock returns and macroeconomic announcements. Chapter 2 provides nonparametric tests for various hypotheses about the effects of a continuous treatment variable in a nonseparable structural equation. These hypotheses have direct policy implications. Specifically, we consider marginal effects and various average effects and test (i) whether these effects depend on the level of treatment; (ii) whether average effects are heterogeneous across different subpopulations defined by covariates; and (iii) whether marginal effects are subject to unobservable heterogeneity. Our tests are based on consistent procedures of Bierens (1982, 1990) and Stinchcombe and White (1998). We apply our tests to study interest rate elasticities of loan demand in microfinance and the impact of education on wages. In Chapter 3, we show that careful examination of the structure determining treatment choice and outcomes is central to proper choice of covariates. We demonstrate how causal diagrams can play a key role in this examination by using these methods to give a detailed analysis of the choice of efficient covariates.Chapter 4 is about a common exercise in empirical studies -"robustness check," where the researcher examines how certain "core" regression coefficient estimates behave when the regression specification is modified by adding or removing regressors. If the coefficients are plausible and robust, this is commonly interpreted as evidence of structural validity. Here, we study when and how one can infer structural validity from coefficient robustness and plausibility. We discuss how critical and non-critical core variables can be properly specified and how non-core variables for the comparison regression can be chosen to ensure that robustness checks are indeed structurally informative

Main Content
Current View