Keith Finlay and Leandro M. Magnusson, "Two Applications of Wild Bootstrap Methods to Improve Inference in Cluster-IV Models", Journal of Applied Econometrics, Vol. 34, No. 6, 2019, pp. 911-933. All files are zipped in the file fm-files.zip. The text files are in DOS format, but there are also binary (.mat) files. Introduction: ------------- We use MATLAB software to estimate confidence intervals and sets and to perform Monte Carlo simulation exercises. Details about the algorithms underlying the tests are in the Supplement that accompanies the manuscript. The main folder is called cluster_iv_JAE. It contains 4 folders: - boot_simulation, - Civil_war_cr, - Colonial_Origins, and - functions. The first folder has the scripts for generating the Monte Carlo simulation results, while the second and third folders contain the scripts for obtaining the confidence sets and intervals for the applications. The last folder contains several functions that are called by the script files in the other folders, and therefore must be in the MATLAB search path. This can be done by adding the following command at the beginning of the MATLAB script file: addpath('[path name]/functions','-end') After executing a script, it is recommended to remove that path: rmpath('[path name]/functions') One should also download the MATLAB data file table_CLR.mat (42MB approximately) from the link https://www.dropbox.com/s/6yqqguq2cryw0df/table_CLR.mat?dl=0 and save it in the functions folder. Application 1: The Colonial Origins of Comparative Development -------------------------------------------------------------- The data for this application were obtained from the American Economic Review site using the link https://www.aeaweb.org/aer/data/oct2012/20110390_data.zip The original dataset, a Stata data file, was converted into the tab-delimited text file data_colonial_tab.txt, which is also available in excel xls format. The description of the data is written at the data_colonial_origins_description.[m or tex]. It contains 64 observations and 29 columns. The MATLAB script files are labeled T[1 or 3]_C[2 or 4 or 6]_parallel.m, where T stands for Table and C for Columns. For example, T3_C2_parallel.m corresponds to the script file that generate the results of Table 2, Columm 1 of Acemoglu et. al (2012) paper. The script files with ending _pval return the p-values for testing the null assumption H0 : \theta_0 == 0. The saved results can be used to fill in the values of Table 2. Table 1: Confidence Intervals T1_C2_parpool.m Column(1) T3_C2_parpool.m Column(2) T3_C4_parpool.m Column(3) T3_C6_parpool.m Column(4) Table 2: p-values T1_C2_parpool_pval.m Column(1) T3_C2_parpool_pval.m Column(2) T3_C4_parpool_pval.m Column(3) T3_C6_parpool_pval.m Column(4) Each script file will generate 8 txt files, which will be saved in the newly created results folder. The txt files contain the confidence intervals and the p-values to be used to fill in Tables 2 and 3. References: >Acemoglu, Daron, Simon Johnson, and James A. Robinson 2012. The Colonial Origins of Comparative Development: An Empirical Investigation: Reply. American Economic Review 102(6): 3077-3110. >Albouy, David Y. 2012. The Colonial Origins of Comparative Development: An Empirical Investigation: Reply. American Economic Review 102(6): 3059-3076. Application 2: Civil War ------------------------ The data for this application were obtained from Prof Miguel's personal webpage: http://emiguel.econ.berkeley.edu/assets/miguel_research/47/Data.zip The original data is a Stata data file (mss_repdata.dta), which was converted into a MATLAB data file (extension .mat) and text file (extension .txt). The description of the data is detailed in the data_civil_war_description.[m or txt] file. It contains 743 observations and 108 columns. One should first run the files ending in parpool.m. Each of these script files will save a MATLAB data file in the subfolder named results. Afterwards, run the Confidence_Regions_Civil_War_test.m file to generate the confidence regions which will be saved as pdf files. Figure 1: Confidence Sets T6_C1_parpool.m top row TC3_C4_parpool.m middle row TC3_C5_parpool.m bottom row Reference: >Miguel, Edward, Shanker Satyanath, and Ernest Sergenti. 2004. Economic Shocks and Civil Conflict: An Instrumental Variables Approach. J ournal of Political Economy 112(4): 725???53. Monte Carlo Simulation ---------------------- The folder boot_simulation contains the subfolder Power. Files in the root of the boot_simulation perform the computation of the rejection frequencies, while the files inside the subfolder Power compute the power curves. Rejection Frequencies: To obtain the rejection frequencies reported in Tables 3, 4, and 5 in the manuscript, as well as all the tables in the supplementary material, one should first run the script files RUN_lognormal_nG.m and RUN lognormal nG20.m. After completing the execution of these files, several folders and subfolders will be created. The first folders will be nG_10, nG_20, nG_40, and nG_80, where nG indicates the number of clusters. In each of the nG folder sits the subfolder lognormal, and, inside the lognormal subfolder, the subfolders kz_2, kz_5, kz_10, and kz_15 will be generated. The initials kz refer to the number of excluded instruments. The MATLAB data files will be saved in the kz subfolders. By running the script files beginning with ???table??? several text files will be generated and saved in the results folder. Those text files contain tables of the results extracted from the data files. The main manuscript and its supplement use only selected results displayed in the those text files. Power Curves The RUN_boot_power_ G20 mu180_rho_fix_het.m file located in the subfolder Power should be executed to replicate the Power comparision figures. After executing this script, the results/lognormal/kz_5/nG_20_theta [xxx] subfolders will be created with several MATLAB data files in them. The [xxxx] indicates the values of the parameters under the true DGP, which range from -2.750 to +2.750. Next, one should run the file power_G20_het_fix_log to generating the pdf files with the plotted graphs of the power curves.