Inflation expectations: What to expect when you are (not) expecting

Published: April 4, 2019 2:05:10 AM

There are, therefore, both economic and political-economy reasons to expect food inflation to gradually re-accelerate, though this could take a while.

We, therefore, run Granger causality tests and find that observed inflation “Granger-causes” inflation expectations but not the other way around. This finding reinforces the legitimacy of the specification above.

By Sajjid Z Chinoy & Toshi Jain 

What factors influence household inflation expectations in India? Expectation formation is typically a hybrid of rational (forward-looking) and adaptive (backward-looking) impulses. That said, in India’s case, several studies find: (1) a meaningful adaptive component and, (2) more importantly, that food and fuel prices are key drivers of household inflation expectations.
The obvious implication is that the sustained disinflation of food over the last few years should have pulled down household inflation expectations, potentially significantly. Lower inflation expectations, in turn, should have a salutary impact on core and headline inflation, through the aforementioned Phillips Curve. So the conditions should be in place for core inflation to soften towards much lower headline inflation. Right?

Not so quick. RBI’s household survey of inflation expectations reveals that, contrary to popular perception, inflation expectations have been remarkably, and disconcertingly, sticky. To be sure, expectations saw a sharp downshift in 2014-15, ostensibly responding to headline CPI inflation halving between 2013 and 2015. Since then, however, inflation expectations have been remarkably sticky, despite inflation trending even lower and food inflation, in particular, collapsing in recent years. Yes, much can be made about the fact that 12-month-ahead expectations have softened by 150bps over the last two quarters. But, at 8.5% in December 2018, they are at exactly the same level they were 12 months ago and 24 months ago. Similarly, 3-month-ahead expectations softened last quarter but, as of December 2018, were still higher than both 12 and 24 months ago. This, despite food inflation falling by almost 400bps over the last three years.

What drives inflation expectations?

Why haven’t inflation expectations responded to lower food and fuel prices—as is commonly presumed? To investigate this puzzle, we explicitly model household inflation expectations as a function of the different components of inflation—food, fuel and core. This acknowledges both the adaptive nature of expectations as well as tries to ascertain the relative importance of different subcomponents of inflation in influencing expectations—a question of particular saliency, given the wide divergence across food and core.
We estimate this equation:

Inflation Expectations = Constant + a(Food) + ß(Core) +d(Fuel) + ?(Gasoline) + e

We alternately use both the 3-month and 12-month-ahead expectations. Food and Core refer to food inflation and core inflation, respectively. Fuel refers to the fuel index in the CPI basket. But because this index does not include gasoline and diesel—which are instead included within services—we separately include a “Gasoline Index”, comprising a weighted average of petrol and diesel prices.

Methodological housekeeping

There are, however, a few methodological issues to contend with. First, running the above regression in year-on-year growth rates creates nonstationarity concerns (since all the variables have a unit root). Therefore, we run the regression in differences, similar to the approach followed by New York Fed.

Second, there is a potential endogeneity issue. We hypothesise that observed inflation impacts inflation expectations (because expectations are adaptive). But, equally, inflation expectations influence inflation (through the postulated Phillips Curve channel). We, therefore, run Granger causality tests and find that observed inflation “Granger-causes” inflation expectations but not the other way around. This finding reinforces the legitimacy of the specification above.

Results: Core inflation matters thrice as much as food!

What do we find? First, contrary to conventional wisdom, core inflation influences household inflation expectations three times as much as food! Second, gasoline inflation has no statistically (or economically) significant impact on expectations. These results are particularly true of the 12-month-ahead expectations, thereby overturning the conventional wisdom that food and fuel are the key determinants of household inflation expectations in India. This result is robust to choice of different control variables in our estimating equation as well as to different time periods (see graphic).

We also get much of the same results using 3-month-ahead expectations, instead of the 12-month-ahead expectations. While the coefficients on food and core are slightly lower, core still matters three times as much as food. That said, core inflation is statistically significant at 12% level, rather than 10% levels when using 12-month-ahead expectations.

Finally, since the household survey is an urban one, we replicate the same results using Urban CPI. Qualitatively, we get the same results when using 12-month-ahead expectations. The coefficient on core is almost three times as much as that on food, though it is significant at a slightly lower level (12% level versus the 10% level in the baseline result, for the combined CPI). The Gasoline Index remains economically and statistically insignificant across all specifications.

One implication of these results is that while food inflation may shape expectations at the 3-month horizon (since the role of core is marginally less statistically-significant at this horizon), it is core inflation that decisively shapes expectations at the 12-month horizon—the horizon that likely matters for central banks given the “long and variable lags” associated with monetary policy. In other words, household inflation expectations are formed much more broadly than commonly presumed.

Our findings can, therefore, help explain why inflation expectations have been so sticky in recent years. Food inflation has fallen by more than 300bps over the last two years which—even by our own results—should have had a depressive impact on expectations. But core inflation has accelerated by about 100bps during that time period, and because the impact of core is three times that of food, our results suggest that accelerating core has largely undone the impact of food, thereby keeping expectations sticky.

A number of implications flow from our findings. First, they help explain why core inflation has been so sticky. The fact that core inflation plays a crucial role in shaping expectations which, in turn, impact the future out-turn of core inflation helps explain the persistence of core.

Second, they help us make sense of our earlier finding—that in recent years it is headline that converges to core and not the other way around. This is because, if headline inflation is ultimately driven by “inflation expectations” and “slack”, and (i) slack influences core; and (ii) core influences inflation expectations, it is no wonder that headline eventually converges to core.
Third, the forces that are commonly assumed—that food and fuel will pull down inflation expectations which, in turn, will pull down core—are unlikely to work, given our findings. For core to soften materially, growth will have to slow and output gaps will have to open up.

All this raises the bar for core to soften to headline.

The key is what dynamics could push food towards core inflation? The typical transmission should occur through wages. As output gaps close, wages should start firming up. This should put pressure on food prices through both a cost-push channel (as agricultural wages firm up) and through a demand channel (as higher wages boost demand for food). Are we seeing first signs of that already? Rural wages have stopped decelerating and particularly agricultural wages have begun to tick up. Also, food prices have sequentially increased (on a seasonally-adjusted basis) for three consecutive months, for the first time since the last quarter of 2017. WPI food inflation has accelerated much more sharply. Will CPI prices follow suit?

Economic channels apart, as the last few months have revealed, the political-economy of India is unlikely to tolerate the terms of trade confronting farmers to remain at these depressed levels. One can, therefore, reasonably presume that efforts will be made—whether through cash transfers, minimum income guarantees or opening up exports—by the next government to push up food prices, and thereby alleviate the stress confronting farmers. There are, therefore, both economic and political-economy reasons to expect food inflation to gradually re-accelerate, though this could take a while.

All this suggests we should not take the currently benign inflation trajectory for granted. The current may not be a particularly good predictor of the future.

(Excerpted from JP Morgan Asia Pacific Emerging Markets Research dated April 3, 2019.)

Get live Stock Prices from BSE and NSE and latest NAV, portfolio of Mutual Funds, calculate your tax by Income Tax Calculator, know market’s Top Gainers, Top Losers & Best Equity Funds. Like us on Facebook and follow us on Twitter.

Switch to Hindi Edition