Dynamics of inflation in India: does the new inflation bias hypothesis provide an explanation?

Abstract

In this paper we estimate the Reserve Bank of India's (RBI) policy response to supply shocks. In particular, we exploit an important strand of the recent literature (the new inflation bias hypothesis) to understand why the two frequently cited measures of inflation in India have persistently diverged in recent years. Specifically, it is argued that the difference in coverage and weighting pattern between the indices interacting with policies pursued by the RBI to control its preferred inflation measure WPI turned out to be inappropriate with respect to stabilizing expected CPI-IW inflation. This in turn provides an explanation for the persistent divergence between the two measures of inflation.

Keywords:

• supply shocks
• monetary policy
• inflation divergence

Acknowledgements

We are grateful to an anonymous referee for valuable comments and suggestions which greatly improved this work.

Notes

 1. As such, WPI can be said to essentially measure price changes from the production side, and not from the consumption side. For policy formulation the WPI is the main measure of inflation since it has a broader coverage and is available at weekly frequency. The index however does not cover non-commodity producing sectors. From the viewpoint of a consumer, inflation estimates given by CPI-IW are considered more representative since it is used for indexation purposes for many wage and salary earners (including government employees). Moreover, the CPI-IW index captures the price increases in the service sector which have tended to outgrow the commodity producing sectors in recent years.

 2. We have divided the sample into two main sub-periods, for instance, 1993:4–1998:4 and 1998:12–2004:6. These sub-periods roughly correspond to the unstable and stable eras of recent inflation history. As Acharya (2001) points out in the light of the experience of the recent past, the year 1998–99 was rather unusual. There was an unexpected increase in food prices (especially onions and potatoes) in the period May–November 1998. By November 1998 the CPI-IW inflation had risen by an unprecedented 20% year-on-year. Nevertheless, the pattern is broadly similar even if one removes the influence of this period from the sample.

 3. After a long period of popularity this literature has recently been widely criticized. This is because the model's equilibrium inflation bias emerges from the monetary authority's desire to push the economy beyond its potential level. This is argued as being an inaccurate description of real life monetary policy-making (see Blinder 1998). Moreover, in formal econometric work, Ireland (1999) shows that while the long-run behaviour of inflation and unemployment in the USA is compatible with this hypothesis, the short-run dynamic behaviour is not.

 4. See Cukierman and Gerlach (2003), Gerlach (2003), and Ruge-Murcia (2003, 2004).

 5. In the special case where the conditional variance of supply shock is constant for all ‘t’ the effect of the variance of supply shock on mean inflation (see Equation (5)) cannot be identified separately from the constant. Hence, it is not possible to test the relationship between and mean inflation in the time-series dimension.

 6. We note that while the inflation (or deflation) bias result in (5) stems from the asymmetric response to supply shocks, we cannot exclude in principle that a nonlinear Phillips curve is responsible for this bias (see Ruge-Murcia 2004). In Appendix 1 we show that the specific form of nonlinearity in the Phillips curve that is required to generate the bias has no empirical basis. This makes our preference specification the only plausible source of the bias.

 7. Monetary authorities are assumed to choose interest rates in a discretionary fashion. This description of policy behaviour seems to be consistent with the actual practice of many central banks which rarely tie their hands over the course of future policy actions.

 8. The model has been kept deliberately simple since the intuition and main findings change little when richer dynamics such as persistence in the Phillips curve are added.

 9. Moreover, as in most of the empirical literature, we extend the model to include a partial adjustment mechanism for setting interest rate changes.

10. Furthermore, this formulation allows us to recover the asymmetric preference parameter ‘γ’. The preference parameter is given by γ = 2a₁₃/a₁₂.

11. Discussion is centred around the formulation of monetary policy since 1993. Prior to that, the unrelenting series of high fiscal deficits implied that monetary policy was essentially playing a subordinate role to fiscal policy.

12. Our empirical analysis maintains the assumption that all the variables in the reaction function are stationary. In addition to its empirical plausibility, stationarity of these variables is also a property of many of the theoretical models that rationalize the use of the kind of policy reaction function considered here (see Clarida, Gali, and Gertler 2000 for example).

13. Standard errors are reported in parentheses. Note that it is the structural parameters (a _1i s) of the model that are reported in the table. The reported standard errors are obtained by means of the delta method.

14. Note that our estimation strategy identifies only the constant term, but not R∗and π∗ separately. The standard approach in the literature is to fix the equilibrium interest rate a priori and then deduce the implied inflation target (see Clarida, Gali, and Gertler 2000). We do not follow this approach because the data used for estimation are all revised data and not ‘real-time’ data that were available to policy-makers while taking decisions. As the analysis in Orphanides (2001) suggests the difference between the real-time and revised data can have important consequences for the analysis of policy rules. Thus, by using revised data we could be wrongly inferring the RBI's inflation target.

15. This is corroborated by a recent study (see Srinivasan, Mahambare, and Ramachandran 2006). Evaluating the RBI's policy reaction function in an error correction framework suggests that policy-makers correctly diagnosed that supply shocks mainly have a transitory (and largely self-correcting) effect on the price level and therefore refrained from responding to them. This suggests that monetary policy had not provided the basis for a sustained change in the inflation process during the sample period.

16. Note that when such an exercise is carried out for advanced countries the results seem to be pretty robust irrespective of the price index used (see Clarida, Gali, and Gertler 2000). The message one gets is monetary policy is either appropriate or inappropriate regardless of which price index is used in the analysis. This is not surprising since the price indices in the USA for example, are highly correlated unlike in India.

17. Instead of using broad money growth as an additional instrument in the baseline regression we experimented with the level of real exchange rate. This proxy was meant to reflect the growing openness of the Indian economy in the 1990s.

18. Note that our Phillips curve specification nests the usual linear form with respect to shock ‘u’ as a special case. Formally, employing L’Hopital’s rule we find that as ‘b’ approaches zero, (e ^b(u) − 1)/b approximates ‘u' the linear form. That is, lim _b→0((e ^b(u) − 1)/b) = lim _b→0(ue ^b(u)/1) = u.