A Markov Chain Model of Inflation in India

A price rise per se is not inflationary; 'a price rise becomes inflationary only when every rise in the price level becomes the base for a further rise in the price level and the process becomes self-sustaining.' Thus a cumulative process of price rises characterises inflation; and this cumulation implies a very high long run probability of price rises. Modelling a Markov chain to the fluctuations of the price levels facilitates estimation of such long run probability. In this paper, we advance a Markov chain model for price movements in India, based on the general as well as a selected set of sectional wholesale price indices and on the three consumer price indices for 240 months from April 1981 to March 2001. The estimation period has been divided into two, viz., pre-liberalisation and post-liberalisation, of 10 years each. We consider 3 general states of monthly price changes - positive (tentatively suggesting inflation), negative (deflation), and zero (stable prices), and estimate the state transition probabilities and the steady state probabilities along with the return periods of these three states of price level changes. It is found that in general the state of inflation persists with a high probability and a much shorter return period in India. This indicates that price rises have been cumulative and hence inflationary in effect