Data and algorithms for the project of sterilized intervention effectiveness evaluation on Russian data 2014-2017
This project presents tests for sterilized intervention effectiveness evaluation for small open commodity exporting economy. It corresponds to submission of the paper: "Sterilized interventions in the form of FX repo auctions: VECM analysis of Russian case" prepared for Russian Journal of Money and Finance in April, 2018. Three Vector error correction models are estimated and analyzed (VECM1, VECM2, VECM3). Used in the estimation time series include 618 daily observations from 6 November 2014 to 20 April 2017. The oil price is the price of the OPEC basket in US dollars per barrel. The exchange rate is the average spot rate (with delivery TODAY) on the Moscow Exchange (MOEX) measured in Russian rubles per one US dollar. Both oil price and exchange rate are introduced in the models in natural logarithm to measure short term and long term elasticity of exchange rate with respect to oil price. The policy interest rate is the BoR key interest rate in percent. I introduce the cumulated sterilized interventions variable as total claims of BoR to commercial banks in USD, created by foreign currency repo auctions . Sterilized intervention variable is the volume of US dollars distributed through the repo auction at day t minus the volume of US dollars paid by commercial banks to the BoR at day t. Data are contained in: All graphs for RJMF reestimated2.xls for Excel VECM_reestimated2.jsc for JMulti JM_data_VECM1_reest2.mat JM_data_VECM2_reest2.mat JM_data_VECM13_reest2.mat for Matlab.
Steps to reproduce
To reproduce results you should unpack Data and algorithms archive and run JMulti (see http://www.jmulti.de/ ). Then open VECM_reestimated2 project and repeat standard steps for VECM estimation in JMulti. Speps needed (for VECM1 estimation): 1. Choose "Two stage procedure" and Specify "Estimate coint relations by S2S with possible restrictions on beta" 2. Specify restrictions on beta. (2 restrictions: betas, corresponding to Zneg and Zpos should be zeros) 3. Run estimation 4. Estimate SVEC model ("Estimate with boot Std. Err.") 5. Bootstrap confidence intervals with Hall and Efron algorithms for 50 periods 6. Display Impulse Responses 7. If you wand to retreive IRF series you may use Control/Symbol Control button in JMulti The archive contains algorithms of different shock contribution in exchange rate dynamics on the base of estimated VECMs. To do that you should enter Scenario folder and run files in Matlab: Scenario_VECM1 for calculations for VECM1 Scenario_VECM2 for calculations for VECM2 Scenario_VECM3 for calculations for VECM3