Impact of RBI Repo rate changes on Call rates: How soon and How much

FIS project term IV

Submitted to Prof. Uday Damodaran

Group 22 Amoy Kumar Dutta | 262 Pooja Gandhi | 286 Sakshi Gupta | 296 Shipra Agarwal | 301

CONTENTS

Introduction ................................................................................................................................ 3 The Interest rate channel ........................................................................................................ 4 Objective of study ...................................................................................................................... 5 Description of data ..................................................................................................................... 5 methodology used ..................................................................................................................... 5 Descriptive statistics .............................................................................................................. 5 Correlation ............................................................................................................................. 5 Moving average ..................................................................................................................... 6 Event Study Analysis: ............................................................................................................ 6 Analysis...................................................................................................................................... 7 Conclusion ............................................................................................................................... 12 Assumptions & Limitations ..................................................................................................... 12 Future Scope ............................................................................................................................ 12 References ................................................................................................................................ 13

Group 22 | FIS Project

INTRODUCTION

What are the two most popular buzz words in the current Indian economy? GDP rate and Inflation! Nobody likes high inflation and thats where our Reserve Bank of India (RBI) comes in. One of the key objectives of the RBIs monetary policies is controlling the price level. The RBI and central banks in various other economies around the world use a variety of market and non-market oriented instruments like cash reserve ratio (CRR), statutory liquidity ratio (SLR), repo, bank rate, open market operations (OMO), foreign exchange operations, quantitative restrictions on credits etc. to name a few to control the monetary policies. The transmission mechanism of monetary policy is defined as how changes in the various monetary policy instruments affect changes in the price levels and also the aggregate output of an economy. Economists normally classify monetary policy transmission into five channels: 1. 2. 3. 4. 5. Interest rate channel Exchange rate channel Other asset effect channel Credit channel and Expectations

The success of any monetary policy is generally measured by how quickly and to what degree it is able to affect the economic variables in a country and thereby bring about changes in the price level and output. However, there is normally a lag associated with the changes in monetary policy and their effect on economic variables. According to Vaish (2000), [1] this lag can be broken down into broadly three types as shown in figure 1. The inside lag is due to the time elapsed as a result of the central banks decision making process, i.e. the time between when action is needed and when the action is actually implemented. The intermediate lag is the lag due to the time elapsed between when the monetary policy instruments are changed and the time it takes to significantly affect the credit rates and other variables that influence spending decisions of economic agents. Even after credit rates and deposit rates or other economic conditions change, people do not modify their spending habits immediately. This lag is called outside lag. It is thus the lag between change in the interest rates and corresponding change in the price levels and output.

Group 22 | FIS Project

Figure 1: Lags in Monetary Policy Transmission

THE INTEREST RATE CHANNEL The interest rate channel of monetary policy transmission is one of the popular channels employed by central banks of various countries to control their economies. According to this mechanism, changes in various key rates by the central bank, such as the repo rate, reverse repo rate and the bank rate to bring about changes in the prime lending rates and other lending rates and deposit rates. Repo rate is the rate at which commercial banks and primary dealers can borrow from the central bank by providing collateral in the form of government securities while reverse repo rate is the rate at which they can invest their surplus funds with the central bank for government securities. The bank rate is the rate at which the central bank provides credit to commercial banks to meet their short term temporary liquidity requirements. Changes in key policy rates is generally considered to initially affect the interbank call money rates which in turn affects the lending and deposit rates. This in turn affects the consumers spending and investment decisions, thereby affecting the price level and output. As mentioned above, in this channel too there is generally a lag and pass-through stickiness associated with changes in the policy interest rates and their effect on commercial banking rates. According to C Amarasekara [2], interest rate pass-through may be defined as the degree to which changes in the central bank policy rates and money market rates are reflected in the commercial bank interest rates, both in the short run and long run. A full pass-through is said to happen when there is one-to-one change in the policy rates and commercial bank rates. When the pass-through is sluggish or incomplete, it is referred to as stickiness. Central

Group 22 | FIS Project

banks need to be aware of this phenomenon to understand when and how much change in policy rates would bring about the desired change in future economic conditions. An efficient money market generally leads to speedy and complete pass-through while presence of inefficiencies can make the pass-through sticky.
OBJECTIVE OF STUDY

To analyze the impact of changes in repo rate on the call money market rate and size and speed of the pass-through from repo rates to call money market rates
DESCRIPTION OF DATA

Since the objective of the paper is to find the extent and the speed with which the change of Repo rate affects Call money rates, data used for the purpose were Repo rates and Call money rates. However, the periodicity of change in both of them is not consistent. Daily data of call money rates as obtained from historical data of RBI were used. The repo rate changes only at certain non-equal intervals and between the two changes it remains constant. The dates and the changed value of repo rate were also obtained from the historical data of RBI. Sample size The data taken is from 26-10-2005 till 26-7-2010 which includes 1737 observations. Our objective was to get enough number of data points for both increase and decrease of repo rate, so that both side changes can be accommodated in the analysis. In this sample, the repo rate changes by 19 times, which is a fair enough number.
METHODOLOGY USED

Mainly descriptive statistics, correlation, moving average and Event study analysis were used to find the type of relationship between the repo rate and the call money rates. They are explained as below: DESCRIPTIVE STATISTICS In order to find the nature of both these rates, mean, standard deviation, minimum and maximum are obtained. CORRELATION First the data was segregated according to nth day after the change of repo rates i.e. all the first days for all the 19 repo rate changes are taken as one series. Similarly the data points for all the second days in each window are taken as a series and so on till 15 days. Through observation of data and trial and error it could be inferred that the change of repo rate could impact call rates in the at the most 12-15 days after the change of repo rate and thus data till 15 days after the change in repo rate was taken. This was just used to find the extent to which the call rates are affected by the repo rates. Though this is a good first indicator to find the lag

Group 22 | FIS Project

of effect of change in repo rate on the call money rate, to find the extent of the the impact further analysis was required. MOVING AVERAGE As can be very evidently noticed, there are certain data points when the call money rate shows a very abnormal behavior and reaches to an inexplicable high or low value. So, to remove such abnormalities and smoothen the curve in order to observe the trend of call money rate, moving average was taken. Trial and error was done to find the appropriate number of data points over which moving average was to be found. EVENT STUDY ANALYSIS: An event study analysis is generally used to analyze the impact of various announcements such as quarterly earnings, merger & acquisition etc on the share price. Thus, it is model to evaluate time lag and the degree of impact that such shocks have on stock price. The objective of this paper is also similar i.e. to study the time lag and impact of shocks in repo rate on the call money market rates and hence we have used an adaptation of the same method. The steps involved in the process are as follows: Creation of windows to convert time series data into cross -sectional data: The data available was the time series of repo rate and call rates. The data was first bucketed into multiple windows such that data in tth window represents the call rates after a lag of t days/periods from the day of change in the repo rate. Windows corresponding to lag L of 115 days were created each containing 19 data points corresponding to the changes in repo rate. Thus, the time series data was sliced at uniform lag intervals from date of change in the rates to convert the time series data into cross-sectional data.

Group 22 | FIS Project

Linear regression in each window: Multiple regression runs were done -1 for each window with repo rate as the independent variable and the call money market rate as the dependent variable. The regression equation is th for the L window is given as follows:

Call Rate T+L = + *Repo Rate T +

Detailed calculations are contained in the following excel file.

22_EmpPaper.xlsx

Repo rate and call money rate when plotted for all the sample points, the following graph is obtained. It can be seen that the graph of repo rate is a step graph i.e. everytime the repo rate changes, it remains constant for quite some time before it changes again. While there is a lot of noise in the call rate as can be seen. The abnormal ebbs and troughs can also be seen.
30 25 20 15 10 5 0

38651 38698 38745 38792 38839 38886 38933 38980 39027 39074 39121 39169 39221 39268 39315 39362 39409 39456 39503 39550 39597 39644 39691 39738 39785 39832 39879 39926 39973 40020 40067 40114 40161 40208 40255 40302 40349
Repo rate call money rate Figure: Repo rate and call money rate

Group 22 | FIS Project

Given below are the graphs showing the trend of call money rates with moving average taken over different number of data points. Finally, 60 point moving average was found to be removing the noise to quite an extent and it could be quite easily inferred from it that the call rates follow the repo rate and changes in repo rate do affect the call rates.
30 25 20 15 10 5 0 38651 38704 38757 38810 38863 38916 38969 39022 39075 39128 39187 39240 39293 39346 39399 39452 39505 39558 39611 39664 39717 39770 39823 39876 39929 39982 40035 40088 40141 40194 40247 40300 40353 Repo rate call money rate 15 per. Mov. Avg. (call money rate) Figure: Repo rate and call money rate trendline with a moving average of 15 points 30 25 20 15 10 5 0 38651 38704 38757 38810 38863 38916 38969 39022 39075 39128 39187 39240 39293 39346 39399 39452 39505 39558 39611 39664 39717 39770 39823 39876 39929 39982 40035 40088 40141 40194 40247 40300 40353 Repo rate call money rate 30 per. Mov. Avg. (call money rate) Figure: Repo rate and call money rate trendline with a moving average of 30 points

Group 22 | FIS Project

30 25 20 15 10 5 0 38651 38704 38757 38810 38863 38916 38969 39022 39075 39128 39187 39240 39293 39346 39399 39452 39505 39558 39611 39664 39717 39770 39823 39876 39929 39982 40035 40088 40141 40194 40247 40300 40353 Repo rate call money rate 60 per. Mov. Avg. (call money rate) Figure: Repo rate and call money rate trendline with a moving average of 60 points

Using descriptive statistics, it can be seen that the average of call rate is less than that of the repo rate and thus is in line with the empirical result that the call rate lies within the reverse repo rate and the repo rate, though with higher volatility. Higher volatility can be seen in the form of noise in the above graphs. It can also be seen that the range within which the repo rate varied during this period is quite less compared to that of the call money rate. The abnormally high and low values of call rate can be accounted to some specific events on that particular day like liquidity, reporting Friday, tax effect, etc. The correlation value between the two rates across the sample is 0.6607, which is a considerably a high value of correlation. The remaining change in the call rates can be attributed to the specific events mentioned above.
Rate Repo rate Call rate Mean Std dev Minimum Maximum Correlation 6.61 1.304878 4.75 9.00 0.660754 5.771822 2.476138 0.13 25.13

Regression output The regression results for window 1 to 15 corresponding to a time lag of 1 to 15 days after the announcement of changes in the repo rate is as given in the following table.

Group 22 | FIS Project

Window Number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Time Lag(Days) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Adjusted R2 0.055975392 0.060090568 0.063011385 0.047600858 0.061957184 0.629670131 0.888273775 0.904589775 0.879354499 0.79918597 0.683386087 0.766326483 0.389126725 0.376948346 0.511130968

P-value 0.175649 0.160752 0.155388 0.185989 0.157302 3.05E-05 1E-09 2.6E-10 1.93E-09 1.53E-07 7.79E-06 5.63E-07 0.002567 0.003988 0.000757

To analyze the significance of linear relationship between repo rate and call rate we need to analyze the p-value for the coefficient of repo rate. At a level of significance of 2% the pvalue should be less than .02 for the linear relationship to be significant i.e. to reject the NULL hypothesis that =0 In addition we need to look at R2 and Adjusted R2 to understand the degree of variance in call money rate explained by the regression equation. Higher the R2 higher is the degree of fit between the data and the regression equation

p-value vs Time Lag

10

Group 22 | FIS Project

Adjusted R2 vs Time Lag

As clear from the graph the p-value for the coefficient of repo rate in the regression equation is significant for time delays greater than 7 days hence, the new repo rate significantly affects the call rate after a lag of more than 7 days. In addition, the adjusted R2 value increases first as the lag increases and then starts decreasing. The value is highest for a lag of 8 days hence, from the regression output it can be concluded that the call money rates are affected with a delay of 8 days The absolute change in the call money rate for a 1bps change in repo rate can be calculated using the coefficient for repo rate in the regression equation for the 8th window i.e. after a time lag of 8 days. The regression output for this window is as shown below: Regression output for Window 8 (Day 8)

Regression Statistics Multiple R 0.95388172 R Square 0.90989034 Adjusted R Square 0.90458978 Standard Error 0.59075868 Observations 19 ANOVA df Regression Residual Total SS MS 1 59.90828 59.90828 17 5.932929 0.348996 18 65.84121 F 171.659 Significance F 2.6E-10

11

Adjusted R2

Group 22 | FIS Project

Coefficients Intercept X Variable 1

Standard Error

t Stat

P-value

Lower 95%

Upper 95%

Lower 95.0%

Upper 95.0%

-3.3507173 0.741964 -4.51601 0.000305 1.42426376 0.108707 13.10187 2.6E-10

-4.91612 -1.78531 -4.91612 -1.78531 1.194912 1.653615 1.194912 1.653615

CONCLUSION

Changes in the repo rate affect the call money market rates with a lag of around 7-8 days and close to 90% of the change in repo rate is reflected in the call rates. Thus, repo rate significantly affect call money market rates but there is a lag in transmission.
ASSUMPTIONS & LIMITATIONS

It is assumed that the changes in repo rate precede the changes in call money market rates i.e. the repo rates cause change in call rates and not vice versa. Thus, we have not conducted statistical test for checking the direction of causality between repo rate and call rates. The relationship between repo rate and call rates is assumed to be linear and we have ignored other factors which might impact call rates such as systemic liquidity and seasonality There are a number of data points where the call rates seem to be abnormally high. This might be due to certain error in the data or due to certain unexpected events. Such data points can be removed as these can bias the results

FUTURE SCOPE

Instead of directly regressing call rates with repo rates we can use a moving average of call rates of the previous period as an estimator for the expected value of call rates and then regress the changes in call rates around its expected value with the changes in repo rate We can use advanced techniques such as finite period polynomial distributed lag analysis to more accurately model the impact of changes in the repo rate on the call money market rates Apart from repo rate there are a number of other factors that impact call rates. Most important factor being is the liquidity in the market which varies across the year due impact of reporting Friday, advanced tax payments and many such exogenous events. Thus, in order to more effectively model the call rates we can run multiple regression with liquidity as one independent variable or we can use dummy variables to remove the aberrations caused due to reporting Friday or advanced tax payments

12

Group 22 | FIS Project

REFERENCES

1. www.rbi.org.in 2. Monetary Theory by Vaish, M.C., (2000),. New Delhi: Vikas Publishing House Pvt. Ltd., 15th Edition 3. C Amarasekara , Interest rate pass through in Sri Lanka 4. Econometrics for Financial markets by Campbell, J.Y., Lo, A.W., MacKinlay, A.C, Princeton University Press

13

Group 22 | FIS Project