ECON 104 Lecture Notes - Lecture 52: Time Series, Panel Data, Marginal Cost
Document Summary
Causal effects that depend on the value of an observable variable, say wi, can be estimated by interacting the treatment variable with wi. In nonlinear regressions: iv regression model can be estimated only if the number of instruments is at least one larger than the number of endogenous regressors. False, as long as the validity assumptions are satisfied for each instrument, the equal number of instruments as endogenous regressor is sufficient. False, standard error estimates (and consequently p values and cis) of the 2nd stage results would be wrong. Some parameters (r^2, adjusted r^2, se, f stats) will be the same because they depend on estimate itself. You have a time fixed effect for each time period that you have, so you separately estimate each average time level. You are not using the information that a certain time period came after another. There are time fixed effects and country fixed effects.
