This readme file explains how to use the computer codes for the data manipulation/results and the calibration used in Sections 2, 6 and 7 of the paper "Search Capital" by Carlos Carrillo-Tudela and Eric Smith. Paper contains descriptive statistics and reduced form estimations using NLSY and two calibrations of the model. We now detailed each of them. We suggest that the reader executes first the files related to the NLSY data and then the ones related to the calibration. However, since the two set of files are independent of each other the order can be reversed without affecting the results. NLSY DATA Sections 2 and 7 of the paper deal with the NLSY data. We use STATA 13 to manipulate the data and compute the tables in these sections. To obtain the statistics described in Tables 1 to 3 of section 2 and Tables 6 and 7 of section 7, the reader will need to unzip the file ?STATA_code.zip?. By unzipping the latter a folder of the same name will be created. Inside the folder one will find the main.do file and three subfolders: data, data_for_tables and do. To reproduce these results the first step is to change the working directory on line 11 in the file main.do and execute using STATA. The main.do file calls the three do-files in the folder ?do?: do1samplepreparation.do, do2descriptive_statistics.do, do3mlogit_and_lpm.do. The first do-file prepares the data, the second one generates the descriptive statistics presented in tables 1-3 and the last do-file executes the estimation for tables 6 and 7. The full set of results from these estimates can be found in the tables in the appendix of the paper. The do-file main.do uses the data-file sample_estimation.dta, which is available inside the directory "data_for_tables". In order to generate this data file anew, place the previously generated datasets 18-4 monthly panel - complete.dta, 16-4 spell-monthly panel - all spells.dta, 16-8 spell-level complete dataset.dta and 16-9 return-level.dta dataset inside the directory "data", uncomment the "do" command on line 71 in the file main.do and execute. All the do files have comments that will guide the reader when using these codes. To construct the data files 18-4 monthly panel - complete.dta, 16-4 spell-monthly panel - all spells.dta, 16-8 spell-level complete dataset.dta and 16-9 return-level.dta use the procedure described on the appendix of the paper. The computational time needed to obtain these estimates is less than an hour based on a PC with the Intel processor INTEL(R) Core(TM) i5-4570 CPU @ 3.20GHz with an installed memory of 16.0 Gb and using a 64-bit Operating System (Windows 7). CALIBRATION Section 6.4 deals with a calibration of the model. To perform this calibration we us Matlab 2016a. To obtain the results described in tables 4 and 5, the reader will need to unzip the file ?MATLAB_code.zip?. By unzipping the latter a folder of the same name will be created. Inside the folder one will find four m-files: SMD_SC.m, SC_F.m, f_and_lambda_twotypes.m and comparative_statics_s.m. The m-file SMD_SC.m contains the code to execute the simulation minimum distance procedure to calibrate the model. One can do it using fminsearch command which will use the Nedler-Meade algorithm or, by uncommenting the last line of the code, on can do this using the pattern search algorithm. Both give the same answer. The m-file calls the function file SC_F.m, which in turn calls the function file f_and_lambda_twotypes.m. We have provided comments in these m-files to guide the reader when using this code. Since we provide a calibration for homogenous and heterogeneous firms, the reader should set manually x2 to be equal to x1 to obtain the parameters for the homogenous case. Executing the SMD_SC.m file would produce the parameter values described in table 5. One can then modify the m-file SC_F.m to obtain the corresponding moments as shown in table 4. This modification requires the reader to create a new m-file with code inside the ?function? command, substitute the parameter values estimated and then execute. The overall computational time for this part is less than 10 minutes based on a Mac Book Pro (13-inch) with a processor 2.7 GHz Intel Core i7 and installed memory of 8.0 Gb. The m-file comparative_statics_s.m is to be used to compute the data for Figure 1. Here one has to change manually the values of s and then produce graphs. This is straightforward but can take up to an hour to do as we transferred the results to an excel spreadsheet and then used Matlab to graph these results using the subplot function.