Revised: 5/18/2019 This README file describes how to replicate the simulation results in "Relationships in the Interbank Market" by Chiu, Eisenschmidt and Monnet. -------------------------------------- Files included: -------------------------------------- Folder "stata_files" containing - Analysis.do - figure1a.dta - figure1b.dta - table1.dta - Data stored in Stata files.doc Folder "matlab_files" containing - Codes1.m - Codes2.m - Codes3.m Subfolder "para" containing - id.txt: deposit rate - il.txt: lending rate - beta.txt: discount rate - big_theta.txt: bargaining power in core market - theta.txt: bargaining power in peripheral market - gamma.txt: participation cost - kappa_L.txt: search cost for L banks - kappa_S.txt: search cost for S banks - sig.txt: separation rate - mu_F.txt: mean reserve balances for S banks - mu_G.txt: mean reserve balances for L banks - sig_F.txt: variance of reserve balances for S banks - sig_G.txt variance of reserve balances for L banks - n.txt: number of L banks - max_x.txt: max grid point for L bank reserve balances - max_e.txt: max grid point for S bank reserve balances - tol.txt: max tolerence Remarks: - The files in "stata_files" are used to replicate the results in Figure 1 and Table 1 & 3 in the paper (See section A below). - The files in "matlab_files" are Matlab files used to replicate the other tables and figures (See section B below). - The subfolder "para" contains parameter values for running the Matlab codes - Unzip the files to one folder, keeping the subfolder structure unchanged ---------------------------------------------------------------------------------------------------- A. To replicate Figure 1 and Table 1 & 3, Go to folder "stata_files" and use Stata to run the files ---------------------------------------------------------------------------------------------------- A document file "Data stored in Stata files" is included to describe the details of the data Instruction: The dofile "Analysis.do", combined with the dta files "figure1a", "figure1b" and "table1", allows to replicate the results in Figure 1 and Table 1 in the paper. The statistics shown in Table 3 (data column) are taken from Table 1, with the exception of the median values which are reported in the dataset "table1". The dta "figure1a" is made of three variables: 1. "cluster" identifies a group composed of at least 3 counterparties trading with reporting agents in MMSR; 2. "volume" identifies the overall volume of the transactions taking place between a cluster of couterparties and reporting agents in MMSR; 3. "count" identifies the number of reporting agents each counterparty in a cluster transacts with. The dta "figure1b" is made of three variables: 1. "cluster" identifies a group composed of at least 3 reporting agents reporting MMSR; 2. "volume" identifies the overall volume of the transactions taking place between a cluster of reporting agents and their counterparties in MMSR; 3. "count" identifies the number of counterparties each reporting agents in a cluster transacts with. The dta "table1" is made of seven variables: 1. "direction" accounts for the direction of the transaction from the perspective of the reporting agent in MMSR. Accordingly, BORR accounts for Non-RA to RA transactions, while LEND accounts for RA to non-RA transactions; 2. "below" is a dummy variable identifying whether the observation refers to transactions taking place below the DFR; 3. "volume" identifies the overall voume of the transactions; 4. "rate" is the transaction-weighted average rate on the trades; 5. "counter" accounts for the number of transactions by direction; 6. "counter2" accounts for the number of transactions taking place below the DFR by direction; 7. "semester" is the half-year period during which the transactions took place. To replicate the results, the dofile takes each of the dta files as input and: 1. Lines 6-10 generate the variable sh_vol (share of volume), which accounts for the share of volume of non-RA by number of RA counterparties in Figure 1a; 2. Lines 12-13 allow to display the median number of relationship of banks S, as reported in Table 3; 3. Lines 15-19 generate the variable sh_vol (share of volume), which accounts for the share of volume of RA by number of counterparties in Figure 1b; 4. Lines 21-51 replicate the summary statistics shown in Table 1. Lines 23-26 reproduce the average rates of Non-RA to RA transactions (BORR) and RA to non-RA transactions (LEND); lines 29-34 reproduce the average size of "Non-RA to RA" transactions (BORR) and RA to non-RA transactions (LEND); lines 37-44 reproduce the fraction of trades below the DFR for Non-RA to RA transactions (BORR); lines 47-51 reproduce the average rate ----------------------------------------------------------------------------------------------------------------------------- B. To replicate the other figures and tables, go to folder "matlab_files" and use Matlab to run Codes1.m, Codes2.m, Codes3.m ----------------------------------------------------------------------------------------------------------------------------- Instruction: [1] Run Codes1.m Produce 1) Figures 2-3 2) Tables 2 and 3 (Model column) 3) Data for plotting Figure 4 using software Gephi (Saved in Figure_4.mat) Run time: 43.428237 seconds. [2] Run Codes2.m Produce Figures 5-8, 10, 11 Run time: about 200 seconds for each figure. [3] Run Codes3.m Produce Figures 9 Run time: 614.582650 seconds.