Advanced Econometrics II: 2021
Assignment on Panel Data
Advanced Econometrics II代写 Empirical Replication Study：In this section, we replicate and extend some of the empirical results in Adema and Pozzi (2015).
Empirical Replication Study
In this section, we replicate and extend some of the empirical results in Adema and Pozzi (2015). The dataset together with the data description is provided on Canvas. In particular, we are interested in replicating the results in Tables 2-3 of that paper. Before proceeding with the tasks described in this section, read carefully Adema and Pozzi (2015).
For your convenience I also mention which lecture each question corresponds to.
Remark 1. As it is mentioned by the authors, for some countries data in the beginning of the sample is not available. To simplify your analysis, please consider only data from 1971-2012 for 14 countries, i.e. drop Denmark and Ireland from your sample. Notice that because the empirical model has leads and lags of RHS variables the effffective length of your panel variables is only 41 + 1 − 2 = 40.
Discuss the following aspects Advanced Econometrics II代写
1.Given that sample size is limited to N = 14, can one expect to get reasonable statistical precision based on pooled estimators? Mean-group estimator?
2.Consider empirical specififications in Column 4 of Tables 2 and 3. Provide estimation (no testing for now) results for:
- (L1-L2) Pooled (homogenous estimators): FE (without time-effffects). FE (with time-effffects) – 2W-FE.
- (L3) Pooled Half Panel Jackknife estimator for FE and 2W-FE.
- (L3) Mean group estimators: MG and HPJ-MG. (both without time-effffects).
Do you obtain similar conclusions as the authors, when you use the bias-corrected versions of the estimators as opposed to standard versions of the estimators?
3.Consider testing.
- (L1) For all pooled estimators provide percentile bootstrap confifidence intervals with B = 400.
- (L1) You can also try normal approximation based CI using the CCM variance-covariance estimator.
- (L3) For MG estimators use : i) the standard “non-parametric” variance esti-mator; ii) bootstrap CI with B = 400. NOTE: for bootstrap procedure you do not have to re-estimate βi in each bootstrap replication.
4.Do you still fifind a signifificant negative effffect of ∆ ln Yi,t irrespective of the procedure considered?
References Advanced Econometrics II代写
Adema, Y. and L. Pozzi (2015): “Business Cycle Fluctuations and Household Saving in OECD countries: A panel data analysis,” European Economic Review, 79, 214 –233.
Kapetanios, G. (2008): “A Bootstrap Procedure for Panel Data Sets with Many Cross-sectional Units,” Econometrics Journal, 11, 377–395.
Appendix A. Procedures Advanced Econometrics II代写
Appendix A.1. HPJ
For each estimator, we need to compute the following three quantities:
Appendix A.2. Cross-sectional bootstrap
See slides From Lecture 1, and Kapetanios (2008).
其他代写：code代写 app代写 algorithm代写 作业加急 北美代写 assembly代写 CS代写 Data Analysis代写 data代写 澳大利亚代写 essay代写 assignment代写 analysis代写 homework代写 加拿大代写 英国代写 作业代写
