Establishing Causality in Welfare Research: Theory and Application: Interim Report of the California Welfare Reform Impact Study

Chapter 2 Experimental and Non-Experimental Approaches For Determining Causality2.1 Examples of Welfare Experiments Page 2-1 2.2 Assessing Randomized Experiments Page 2-2 2.3 Observational Studies Page 2-7 2.4 Quasi-Experiments Page 2-10 2.5 Natural Experiments Page 2-10 2.6 What are the Best Methods for Establishing Causality?Page 2-11 Chapter 3 Estimation, Forecasting, and Model Uncertainty 3.1 Estimation, Model Selection, and Hypothesis Testing Page 3-1 3.2 Consistency and Efficiency -Specification Error and Violating the Assumptions of OLS Page 3-7 3.3 Pooling/Panel Data: Correcting for Specification Error or Lack of Data Page 3-14 3.4 Forecasting and Model Checking Page 3-19 3.5 Model Uncertainty Page 3-23 Chapter 4 Data Description 4.1 Welfare Participation Variables Page 4-1 4.2 Employment Variables Page 4-8 4.3 County Demographics and Characteristic Variables Page 4-10 Chapter 5 Empirical Section: Estimation 5.1 Correlation and Bivariate Analysis Page 5-2 5.2 Multivariate Analysis Page 5-3 5.3 Panel/Pooled Data Analysis Page 5-8 5.4 Notes on Employment Variables and Coefficients Page 5-15 5.5 Summary Page 5-27 Chapter 6 Unconditional Out-Of-Sample Forecasting 6.1 Time Series Forecasting Page 6-1 6.2 Panel Data Forecasting Page 6-2 6.3 Varying the Estimation Sample: The Effect on Forecasts Page 6-5 6.4 Moving Forecasts and Estimation Sample Page 6-10 6.5 Optimal Estimation Lengths for Varying Forecast Years Page 6-13 6.6 Calculating the Minimum Required Impact Page 6-17 6.7 AMSFE: The Variance-Bias Tradeoff Page 6-19 6.8 Summary Page 6-20 Conclusions Appendix Bibliography Table 1.1 Four Theories of Causality Neo-Humean Regularity Theory Counterfactual Theory Manipulation Theory Mechanisms and Capacities