1. Inflation forecasting: To get serious divergent views, high time economists think in terms of real-time data analytics

Inflation forecasting: To get serious divergent views, high time economists think in terms of real-time data analytics

It is high time economists start thinking in terms of real-time data analytics.

By: | Published: May 22, 2017 5:06 AM
In the Indian context, standard univariate ARIMA (autoregressive integrated moving average) time series forecasting or even macro modelling may not work.

Economics as a subject has always been blessed with diversity of opinions. In 2013, for the first time (and perhaps the only time in history) Robert Shiller and Eugene Fama were both awarded Nobel Prize in Economics for sharing diametrically opposite views on market efficiency.

Unfortunately, in India there is now an increasing consensus that economists (particularly those related to financial sector) practice herd mentality. Interestingly, only a few days back, chief economic advisor lambasted such rare unanimity in voices among economists in terms of giving a call on interest rate decisions by RBI and likewise.

The most recent example of retail inflation undershooting the market and RBI forecasts have opened a pandora’s box. Interestingly, just as the data has been released there are opinions that inflation will ultimately go up. Some have even suggested possibility of rates hardening, but cleverly shifted the dates into next fiscal. Herein lies our concern. Even a cursory look at the inflation data suggests that inflation is going to decline in the next couple of months and will stay within reasonable limits in second half of current fiscal. Our current forecast based on seasonal trends now show inflation averaging in the lower band of 4-4.5% for FY18, with a significant probability of the average being closer to 4% (possibly undershooting it). It is intriguing that we are unable to spot such a decline in prices (watch out for inflation readings close to 2% soon!), which brings us to the larger question of inflation or macro-modelling in India.

You May Also Want To Watch:

In the US, members of the Board of Governors and Federal Reserve Bank presidents come up with their projections of the most likely outcomes for real output growth, the unemployment rate, and inflation for each year from 2016 through 2019, and over the longer run, based on their individual assessments of the appropriate path for the federal funds rate. Whereas, in India the MPC reviews the surveys conducted by RBI to gauge consumer confidence, households’ inflation expectations, corporate sector performance, credit conditions, the outlook for the industrial, services and infrastructure sectors and the projections of professional forecasters. The Committee also reviews, in detail, staff’s macroeconomic projections, and alternative scenarios around various risks to the outlook. Drawing on the above and after extensive discussions on the stance of monetary policy, the MPC adopts the resolution.

But even after all this, what is most important is the uncertainty surrounding the CPI forecast made by MPC. It is interesting to see what forecasting technique the MPC members adhere to. In the Indian context, standard univariate ARIMA (autoregressive integrated moving average) time series forecasting or even macro modelling may not work. For instance, generating forecasts under (and often unstated) assumptions about exogenous variables such as oil prices, government spending, and global growth will throw up illusory or elusive results. Under such circumstances, it may be better for the MPC to work with short-term forecasts for next three to six months as macro-variables like oil prices are almost difficult to predict. Remember the oil price crash in FY15 that had resulted in inflation undershooting RBI projection by more than 300 basis points in December 2014.

Interestingly, RBI’s CPI projections were off the mark most of the times (particularly in period of volatile oil prices and off course vegetable) with actual inflation being much less than what the central bank had projected. One can however argue that as a central bank, inflation forecast tend to be on the higher side and this is done ostensibly with the purpose of tempering inflationary expectations. However, if future inflation projection is kept higher, it may prevent inflation expectations from declining significantly, even as actual inflation may continue to fall. This may keep the rates elevated for a longer period of time than otherwise.

Let us now objectively assess the inflation outlook of MPC in context of the recent developments as outlined in April policy.

First, is the issue of stickiness of core inflation. We believe RBI to be fair in inflation assessment, but we must distinguish between core inflation stickiness and core inflation stability (average at 4.7% for the 24 month period ended April 2017).

Second, in terms of food inflation, the fears of increase in cereals, sugar, and protein rich items other than pulses are unwarranted to say the least, even if one looks at non-seasonally adjusted m-o-m increase in these items for FY17 and compares it with trend averages.

Third, the fear of El Nino is also unwarranted. IMD predicted that this year monsoon would be “Normal” or around 96% of Long Period Average (LPA) with an error of ± 5% and with a fair distribution of rainfall across major parts of country. The establishment phase of the monsoon north of the equator has already started, and the Indian Ocean Dipole phenomenon—this is what counters the impact of an El Nino—will have an incremental positive effect on the Indian monsoon. Australia’s Bureau of Meteorology also said there were signs of concerns easing over El Nino.

Fourth, fears of imported inflation has now largely subsided with crude prices expected to stage a smart recovery and rupee appreciation. Our analysis of past few years (since 2011) indicate that oil prices in the second half are always lower than the first half of the year.

Fifth, the potential second round impact of the CPC adjustment continues to form a key argument for higher inflation numbers. This is weird when implementation of higher allowances have not even taken place at the Centre. We believe the government is meaningfully aware of this and will do its bit in this regard. Also, the first round statistical impact of the adjustment is going to be spread out.

To sum up, it is now high time that economists start looking at the data more often and perhaps refine their thinking in terms of real-time data analytics. Let us get some serious divergent views on macro variable forecasting.

(Tapas Parida and Sumit Jain co-authored this article)

  1. No Comments.

Go to Top