Column: A shaky base

Written by Renu Kohli | Updated: Nov 14 2013, 22:23pm hrs
In its recent monetary policy statement on October 29, Reserve Bank of India published its first ever forecasts of consumer price inflation. This is a significant development. For one, it confirms the central banks intent to adopt the CPI as its monetary policy anchor. The appearance of retail inflation projections, concurrent with the WPI inflation as the official anchor, has muddled the monetary policy signals emerging from Mint Street. But this is a secondary matter. Primarily, the CPI inflation forecasts themselves raise considerable unease. Not because of the underlying model from which the forecasts might be obtained, or that they might be subjective predictions. The key point is the data available to forecasters: Is there sufficient information for guiding an optimal monetary policy

Here, the CPI forecasts raise a number of questions. The first of these is to do with data adequacy. The composite CPI is less than three years old, starting from January 2011. To date, i.e., until October, the CPI inflation data points numbered just twenty two. RBIs CPI-inflation forecast could then have been based upon twenty observations perhaps. Is that a long enough time-series for forecasting inflation two quarters ahead Perhaps the current series was extrapolated backwards with statistical techniques like bootstrapping to overcome data deficiency. How robust are these forecasts in that case

Other than insufficient data, the predictability of a significant chunk of headline CPI inflation is questionable. The CPI forecasts represent a guess of some prices, viz. food and fuel, which tend to be very volatile. By construction of weights, this variable component is as high as 60% in the price index. What view would the CPI forecasts incorporate of global oil prices and their pass-through to domestic consumers, not really an economic decision Moreover, food and vegetable prices fluctuate sharply from reasons as diverse as excess or deficit rainfall, other weather vagaries and whether some farm products should be imported or exported, again a political decision. A forecast of this component could be exceedingly complicated and embed a large capricious factor, potentially enlarging the gap vis--vis actual outcome.

Apart from the three-fifths variable component where supply shocks dominate, 9.77% weight of the balance CPI index is for house rents, which are largely imputed. This is an issue where there could be concerns about the efficacy of their measurement, noted the former RBI Governor, D Subbarao. He added that if these were to be excluded, as several other central banks do, this would leave the CPI significantly bereft in terms of coverage and information content. Then again, the CPI is geographically sensitive, according to Pronab Sen, chairman, National Statistical Commission and former chief statistician, a susceptibility which imparts inherently high risk for aggregate monetary

policy decisions.

By far, the greatest concern is of forecasting a price series that has yet to stabilise. This is a point noted by both current and former RBI Governors, Raghuram Rajan and D Subbarao in recent months, suggesting the CPI series is possibly still under observation by the central bank. It has also to prove robust and dependable. For example, even when prices of volatile components fluctuate sharply, they tend to return to the previous level within a relatively short span. We do not see such mean-reversion in the CPI series so far. Alternately, could there be a secular upward trend, i.e., a shifting mean Other than that complex processes such as these need pinning down if they exist, the danger is of forecasting inflation when the price level may be deviating from its long-run behaviour.

Apart from stability issues, the CPIs relationship vis--vis key macroeconomic variables central to the monetary policy decision is not yet known or understood. In the current macroeconomic juncture for instance, the fact that contractionary impulses from declining domestic demand are not reflected in retail price inflation, in marked contrast to the much-older WPI inflation series, raises questions about its coordination with other economic trends. The persistence of core-CPI inflation in the 8.2-8.6% range coexists rather uneasily with indications observed from the output-input price indices of HSBCs Purchasing Managers Index, lowered capacity utilisation and inventory levels, falling sales, etc. It defies the logic of pass-through dynamics where pricing power is smaller when demand conditions are weak; that output could be under-measured, e.g., industrial production, is countered by the higher-frequency, disaggregate measures of economic activity. The fact that RBI has not really been forthcoming in explaining the wholesale-to-retail prices transmission process, the magnitudes as well as the lags, suggests the second-round inflation dynamics is still under study at the central bank. All this lead one to conclude the new CPI series is still evolving.

A forecast of how the general price level will evolve over a specific horizon is an essential part of the central banks decision-making process. It forms the basis for framing the best possible monetary policy. This however depends upon optimal forecasts, consumer prices in this instance. The numerous data, information and linkage issues surrounding the new CPI series suggest the foundations of RBIs first-ever CPI-inflation projections could be rather shaky. An unequivocal answer whether these can guide optimal policy responses is hard in the light. Is the central bank not risking its credibility by this adventure then

The author is a New Delhi-based macroeconomist