﻿*******************************************
EXPLANATIONS OF DATA AND FILES
*******************************************
This readme is written to help the reader in understanding the data and the analysis behind the 
paper "Liquidity Coinsurance and Bank Capital" by 
Fabio Castiglionesi Fabio Feriozzi, Gyöngyi Lóránth
 and Loriana Pelizzon.



In order to run the do files, one have to change the directory over the \, i.e. \folder\filename.dta




DATA SOURCE
**********************


The source for the dataset is public and the website is:

http://www2.fdic.gov/sdi/download_large_list_outside.asp

. These are Pre-Zipped large Download Files. Each of the downloadable ZIP files listed below contains all report items 
in all SDI reports for all FDIC insured institutions. Each ZIP file contains data for one reporting quarter. 

Since we collect almost all the data and then select some variables, all the pre-zipped files are not included.
 

The size of all these single-quarter folders (in csv format) is around 22.14Gb, too big to be reported. 

All the single reports are then merged and the final database has been cleaned following the indication in the paper. 


In the folder there are 2 database: 


The first "1_PCA_DATA.dta" is used to perform the Principal Component Analysis using the routine "1_PCA_PROGRAM.do".


The second "2_SDI_DATA.dta" is used to perform panel regression and descriptive statistics. File "2_SDI_PROGRAMS.do" 
contains the code to replicate the results of the paper.





XLSX FILE
**********************


The excel file "Figure 3 - Table 3" contains the data for the cited figure and table. 
Data comes from the PCA routine (see "1_PCA_PROGRAM.do") and the aggregate annual data comes from
 the SDI Repository
 http://www2.fdic.gov/sdi/main.asp




REQUIREMENTS FOR RUNNING THE PROGRAMS



FURTHER NOTES:
**********************
1. All the analysis is done in MAC OSX 10.7 operating system.
2. Merge and append operations may require large blocks of memory. We actually dedicate 4GB RAM for the tasks. Maybe 2GB should be sufficient for running the programs.

3. PCA require from 15 to 20 minutes to be completed.
4. Analysis is done with STATA version 10 for Mac. Older versions may not be able to run certain do files.

 
