***************************************************************************
Explanation of Codes and Data for "Reason for Reserve? Reserve Requirements and Credit"
Author: Nada Mora
Email:  namora@alum.mit.edu
***************************************************************************

The data is divided in 3 datasets in comma-delimited text files:

1. bilan.csv
2. bdl.csv
3. macro.csv



There are 5 associated STATA text do-file codes:
(Note that these do files include documentation)
Note also that bilan.do is the main do-file and it calls mergers, banktype and macro
The 5th do-file (bdl.do) simply runs the figures and produces some summary stats based on the central bank (BDL) aggregate monthly data

1. bilan.do
2. mergers.do
3. banktype.do
4. macro.do
5. bdl.do



The general set-up of the codes is as follows:

1. bilan.do is the main do-file which reads in the main csv text file, bilan.csv, which is the bank panel data from Bilanbanques. As mentioned above, it calls all of mergers, banktype, and macro to produce the summary statistics and regression tables.


Additional information:
 
2. Note that bilan.csv is a panel data set of the over 60 banks operating in Lebanon from 1996 to 2004.
The data are entered by hand by RAs based on hard copy editions of Bilanbanques from 1998 to 2005 (each hardcopy has data for the previous two years).
The bank codes range from B02 to B83 (note that B01 is the consolidated aggregate). These codes were assigned by the author. Pure renames were easily dealt with directly in the bank codes. Examples of pure renames were Allied Business Bank (B06) that became Allied Bank in 2000 (also coded B06). Similar case for Banque Audi that became Bank Audi SAL - Audi Saradar Group (B13) The British Bank of the Middle East became HSBC Bank Middle East (B53). More substantial work on mergers was done in Stata program after the data was read in (see #3 below).
   
The variables in bilan.csv are largely self-explanatory. For example all variables with a_X refer to asset-side variables and l_X to liability-side variables (e.g., a_total, a_loan, l_deposit). There are also some variables with extensions pl_X which refer to variables from the income statement (e.g., pl_interestinc). And finally, a lot of variables end with extensions _ll or _fc, which reflect lira denominated and foreign currency-denominated terms respectively (e.g., l_deposit_ll and l_deposit_fc). 
 
3. Note that mergers adjusts the panel so that banks involved in merger activity during the sample are pre-merged so that their structure throughout the sample reflects the group's composition at the end of the sample.

4. Note that banktype do-file assigns banks to foreign or domestic based on majority shareholder information (underlying data from Bilanbanques as well as from direct correspondence with the BDL)

5. Note that macro do-file reads in the macro data to be used in regressions as controls (e.g., interbank rate and coincident index) See information below on sources.


More detailed information on sources for macro data:

1) The first two series are bb_erate and bb_inf (both are from Bilanbanques)
Data on Bilanbanques exchange rate used for item reporting, and Bilanbanques inflation rate:
This is Annual Exchange Rate and Inflation Rate Used by Bilanbanques (reported on first page of "Analyses of Balance Sheets and Profit/Loss Accounts" (e.g. p.91 of 2001 ed.)

2) imf_inf: Also inputted in this file is an alternative inflation rate obtained from IMF PINs on Lebanon, various issues for Beirut and suburbs, period average.  I use this series to create imf_cpi index in order to deflate nominal b.s. items for robustness check 

3) imf_growth, % Change in real GDP also from IMF PINs various issues

4) imf_tbill24m .Primary market rate, eop also from IMF PINs.  It seems this is the 24 mo rate from comparing with BDL data

Then there are a number of series that are based on raw monthly data from the BDL, US St. Louis FRED and JP Morgan for EMBI spreads.  These series are based on using a collapse (mean) var1  , by(year) command in Stata on monthly data in bdl_irates and data in fred US. Except for interbank, where I also collapse by median and max, respectively (b/c highly variable).  

5) bdl_coincident (Source: Economic and Financial Data, Banque du Liban website)

6) bdl_idep_ll   Interest Rates : Commercial Banks -LBP: Average Rate on Deposits (Source: Economic and Financial Data, Banque du Liban website)

7) bdl_ilend_ll   Interest Rates : Commercial Banks -LBP: Discount and Loans(Source: Economic and Financial Data, Banque du Liban website)

8) bdl_interbank ; bdl_medinterbank ; bdl_maxinterbank  Interbank rate, eop (Commercial Banks LBP) (Source: Economic and Financial Data, Banque du Liban website)

9) bdl_idep_fc Interest Rates : Commercial Banks -US$: Average Rate on Deposits  (Source: Economic and Financial Data, Banque du Liban website)

10) bdl_ilend_fc   Interest Rates : Commercial Banks -US$: Discount and Loans (Source: Economic and Financial Data, Banque du Liban website)

11) bdl_tbill3m  Interest Rates : Treasury Bills - 3 months Note : PRIMARY MARKET RATES ON TREASURY BILLS (Source: Economic and Financial Data, Banque du Liban website)

12) fred_uscpi   Consumer Price Index for All Urban Consumers: All Items, series id: CPIAUCNS.  Originally from U.S. Department of Labor: Bureau of Labor Statistics

13) fred_ustbill3m  3-Month Treasury Bill: Secondary Market Rate, series id: TB3MS

14) fred_usfedfunds  Effective Federal Funds Rate, series id: FEDFUNDS

15) jpm_ssprd (note data series begins in 1998)  It is the average of the daily series JPM EMBI GLOBAL LEBANON -STRIPPED SPREAD, Datastream code: JPMGLEB(SSPRD)


