Surprisingly, neither inflation nor GDP growth plays a role in how the repo rate is set - this is overwhelmingly explained by just money supply.
Stimulus 2.0 was announced on January 2, 2009. The joint coordinated action of the Ministry of Finance and the RBI was widely applauded in the media. This bonhomie was not always present. As stated as recently as December 23 in the The Indian Express: “The RBI and the government, it is learnt, are not on the same page on the pace and extent of monetary actions required to stimulate the slowing economy”. This story documented a rift between the RBI and all others in the government. Specifically, that the RBI was objecting strenuously to the lowering of interest rates any further, especially after the previously large 250 basis point reduction in the repo rate.
Analysis of widely available macro-economic data confirms the suspicion that the RBI has been on a different page (planet?) for some time now. In its policy statement of October 2008, when economic activity in the world, and India, collapsed like never before, the RBI did now lower its policy rate, the repo, and stated this as a major reason: “It is necessary to moderate monetary expansion and plan for a rate of money supply growth in the range of around 17 per cent in 2008-09 in consonance with the outlook on growth and inflation so as to ensure macroeconomic and financial stability in the period ahead”.
This statement provided a major clue to humble practitioners of the art of deciphering the policy actions of the RBI (or the Fed or the Bank of England etc). Could it be that the RBI determined its policy actions overwhelmingly on just the rate of growth of money supply? But money supply growth, within broad ranges, has been repeatedly shown to be ineffective in either affecting GDP growth or inflation. Having zero policy importance in the rest of the world (the bi-annual IMF flagship publication, World Economic Outlook has no data on money supply growth for any country in the world) it was unlikely that money supply growth would be an important predictor, of as important a policy decision as the repo rate. Further, most practitioners of monetary policy agree that it has to attempt to be forward-looking and, to be effective, at least occasionally contain an element of surprise. A repo rate entirely determined by money supply growth would not serve any purpose.
All central bankers look at the trend in inflation, and real output before making their decisions. Hence the popular Taylor rule which looks at both before deciding on the policy interest rate. Though there is some debate, it is likely that inflation has a higher weight in the calculations. The repo interest rate was introduced by the RBI in the final quarter of 2000 (November).
Using quarterly data, I estimated various “models” of repo rate determination. Some are reported in the table. The models were estimated for two time-periods: from inception of the policy rate till end-2007 (the Jalan-Reddy period named after the Governors in charge), and from September 2003 to end-2007 (the Reddy period). Data for the year 2008 were not included in the sample for two reasons: first to obtain out-of-sample predictions and second, because 2008 was a special year for all!
Surprisingly, for neither period is the inflation rate significant. All the noise in 2008 was about containing inflation, yet this variable has never mattered. GDP growth does matter for the entire period (but has a wrong sign!), but has zero relevance during the Reddy period. This is also surprising since an explicitly stated goal during the Reddy period was to arrest “overheating”.
What does matter in both periods is the rate of growth of money supply, lagged one quarter. Equations 2 and 4 are estimated on the basis of just the lagged one-quarter growth in money supply. Note that knowledge of GDP growth or WPI inflation has zero relevance for the Reddy period.
The predictions of the M3 only equation (model 4) are plotted against the actual repo rate for the period since 2003, third quarter. Note the close fit. The error (difference between the predicted and actual rate) for the entire period 2003 to 2008 is minuscule — -0.064 per cent or minus 6.4 basis points. As close to a perfect prediction as possible. Indeed, investment banks need not have hired RBI-watchers during the Reddy period — the simplest of models would have generated a fortune of predictions.
|Repo Rate Determinants: A Simple Model (Quarterly data)|
|Joint Jalan-Reddy period (Nov 2000 to Dec 2007)||Reddy period (Sept 2003 to Dec 2007)|
|All determinants||Only money supply||All determinants||Only money supply|
|Lagged % change GDP||-0.22***||—||-0.027||—|
|Lagged % change WPI||0.013||—||-0.01||—|
|Lagged % change M3||0.25***||0.24***||0.24***||0.24***|
|Note: 1) Absence of * indicates not significant even at 5% level of confidence; * indicates significant at 5%, ** at 1% and *** at .1% 2) The Reddy period ended in September 2008 but the model was estimated till the end of 2007 to obtain out-of-sample predictions for 2008 as reported in the graph 3) The repo rate was introduced in November 2000|
The model errors during the first two quarters of 2008 are also minuscule — positive 12 and negative 18 basis points. The largest error is 140 basis points for Q3 (July-September 2008) — a repo rate forecast of 7.6 per cent, but an actual rate of 9 per cent. This rate was set at Reddy’s last policy meeting of July 27th 2008 against a backdrop of high oil prices but steeply declining oil and commodity prices. Most analysts at that time were predicting a world growth slowdown; most commodity prices were 20 to 40 per cent off their peaks, and oil itself was trading at $125/barrel, some 15 per cent of its own July 2 peak of $147. The one time RBI went completely against its own simple model was in July; the model predicted that it should have lowered rates by 50 basis points — from 8 per cent to 7.5 per cent. Instead, the RBI (goaded on by the even more simple Congress party worried about inflation?) raised rates by 50 basis points to 9 per cent. A difference of 150 basis points between what the model predicted and what the RBI ended up doing.
Which brings us to the fourth quarter of 2008. After much reluctance and obstinacy, the RBI (now under Subbarao, a different master but the same simple model?) reduced the repo rate to 7.5 per cent thus bringing down the quarterly average to 7.33 per cent. The predicted repo rate for the quarter — 7.34 per cent. It never gets closer than this!
So now we know why the RBI thinks one way, and the rest of India, and the world, thinks differently. The RBI is a money-making and rate-setting machine; it cares little for those people who change their mind about interest rates on the whimsical basis that 12-month industrial growth has become negative for the first time in more than a decade; it cares little about a global financial crisis with world GDP declining at the fastest rate, ever! It is made of sterner stuff, even though the weight is no more than that of paper money. Further food for thought — the RBI has been zealous in pricking positive asset bubbles and reluctant to even consider pricking negative asset bubbles (as in September 2008 onwards). Why the asymmetry? Because simple models cannot pick up the difference.
The author is Chairman, Oxus Investments, a New Delhi-based asset management company. The views expressed are personal