The reserve bank of India frequently cites a survey of household inflation expectations in its monetary policy statements to support its inflation projections. The underlying assumption is that the survey is a useful indicator of future inflation. But that is not the case. Our analysis shows that the household survey on inflation is backward-looking. The central bank's reliance on these backward-looking inflation surveys is one of the key reasons why the RBI has been consistently failing to forecast the inflation downtrend over the past few years.
The target share of respondent groups in the RBI's inflation survey is heavily biased towards housewives. It is fairly reasonable to assume that housewives and even other respondent groups form their inflation expectations based on their experience. In our view, these inflation expectations are likely to be driven by the price levels of volatile items such as vegetables and gasoline, which have a sizeable and an immediate impact on household expenses. So, at best, these household expectations are likely to be only a weak measure of inflation for the generalised consumer basket as they are likely to be heavily biased by a few items.
Additionally, the household survey is restricted to cities, leaving out crucial input from rural areas. Even then, it is used as an indicative measure for combined urban and rural inflation. Over the past four months, rural inflation has slowed more sharply than urban inflation. November data shows rural inflation at 1.7 per cent year on year compared to urban inflation at 3.1 per cent. The household survey does not capture this full picture. The result: The survey has been overestimating inflation expectations in recent months.
We run a simple correlation analysis to find that the respondents in the household survey rely on the past experience of actual inflation when forming expectations about future inflation. We found that cross-correlations between inflation expectations and actual inflation increase when we lag the actual data and lead the expectations data. This result can also be observed in the data itself. For instance, a sharp drop in actual inflation in fiscal 2011 preceded a gradual and marginal easing of inflation expectations in the following periods. Again, a sharp drop in actual inflation in 2014 came ahead of a corresponding decline in inflation expectations a few quarters later.
Even the limited academic literature available on the subject supports our observation that inflation expectations in the RBI's survey of households are backward-looking and not reliable for use in predicting inflation. In a recent paper, a former employee of the RBI's monetary policy department has noted with co-authors that the quantitative expectations under the RBI's survey show a significant lag of more than two quarters in matching the turning points in actual inflation.
The RBI's reliance on the inflation expectations survey to measure the actual inflation has been a factor behind its undue hawkish bias, in our view. It has contributed to the RBI overestimating inflation consistently over the past few years, which has culminated in a policy error of keeping real interest rates too high.
As a case in point, minutes of the RBI's December policy review show that members expressed concerns regarding elevated year-ahead inflation expectations over the past one year. The monetary policy committee of the RBI essentially chose to overlook the hard data of a sharp correction in global crude oil prices, which was expected to ease input cost pressures and lower inflation ahead. Its reliance on elevated inflation expectations resulted in a policy error of sticking to its calibrated tightening stance. In our view, the RBI ought to pay less attention to the household survey to avoid future policy errors.
(This is an edited version of the research report first published on the Bloomberg Terminal on November 14, 2018)
The writer covers India for Bloomberg Economics in Mumbai.