Now Reading:

Identifying Inflation in India: An Autoregressive Analysis

Identifying Inflation in India: An Autoregressive Analysis

By Kalyani Chaudhary, Sagar Wadhwa, Chandan Chintu

Edited by Nidhi Singh, Junior Editor, The Indian Economist


This paper attempts to study the determinants of annual inflation in India holistically, over the time period 1970-2010. We look at several factors, including both conventional (money supply, interest rates, unemployment, food production) and non-conventional ones (political factors like election year, pay commission year), and our study indeed throws up some interesting results. In the recent debate between the government and RBI on who has a greater role to play in terms of controlling Inflation; while RBI has tossed the ball in the government’s court by saying that it is only a stable government policy which can tame inflation, our study shows that RBI has a substantial role to play in tackling inflation through means of its weapons of interest rates and money supply.”


Rationale of the study: Inflation has been a subject of major interest in macroeconomic theory and policy. Lately, it has begun to pinch consumers and producers alike, especially with steep rises in food inflation. In our paper, we consider the movement of the Consumer Price Index (Industrial Workers) as a measure of headline inflation over a period of roughly four decades, and try to study the factors affecting consumer’s inflation over this time period. With inflation becoming a household name and an electoral agenda, it is imperative that dynamics and mechanism of inflation in India be understood for policy purposes.

This paper has found strong evidence in support of our former chief Economic Advisor, Mr. Kaushik Basu. According to Mr Basu, “there was no magic solution to inflation and prices could not be controlled by a Government decision, but rather had to be addressed through a combination of monetary and fiscal measures.”


A fair amount of literature exists on the determinants of inflation in India, and more broadly, in the developing countries. A lot of studies use an autoregressive approach to estimate current inflation, because present inflation is considered to significantly respond to its previous years’ values. Among the major demand-side factors affecting inflation, many studies find Money Supply to be the most significant. Among the supply-side factors, it is GDP (or Index of Industrial Production, which is found to be strongly correlated with GDP) that is a major determinant of inflation.

The papers that we have referred to are mainly empirical studies trying to examine the primary, or ‘proximate’ causes of inflation in India in the post-independence era, maximum over a period of two to three decades. In this sense our study contributes significantly to the literature by considering a larger timeframe of forty years, moreover we also consider some political factors (in the form of binary variables) specific to the Indian context, which are also unique to our attempt of studying inflation. The following paragraphs give a brief outline of the papers we referred to for background theory, variables, modelling ideas and estimation techniques.

Rakesh Kumar (2013) tries to comprehend the multifaceted relationships that determine Indian inflation using a Restricted Vector Autoregressive technique, over the past twenty years. He finds that Money supply (broad money M3), followed by Import Index and Index of Industrial Production (IIP) are the most significant variables in explaining inflation. Anuradha Pattanaik (2010) also studies inflation in the post-liberalisation era of 1990s, and finds that IIP, Money supply and Import Index are the most important factors, along with lags.

Patra and Ray (2010) find past years’ inflation to be a major factor in determining expectations of future inflation, and even amongst them, food inflation to be the most significant. Loungani and Swagel (2001) also use a Vector Autoregressive Technique to study the relationship between the exchange rate regime and the sources of inflation for developing countries in particular. They find that money growth and exchange rate changes are more important in countries with floating exchange rate regimes than in fixed exchange regimes.


We use several established theories of macroeconomics, along with some personal intuitions for the choice of our explanatory variables. Following is a brief description of the theories and insights that we used for our study:

  •   Phillips curve: a historical inverse relationship between the rate of unemployment and the rate of inflation in an economy, originally observed and verified for the US economy in the short run. The theory was revised in many forms later, but the starting point was of a short- run tradeoff between unemployment and inflation.
  •   Fisher equation: estimates the relation between nominal (i) and real interest rates (r) under inflation (π), by the simple approximation given by Irving Fisher in his theory of interest rates, i≈r+π
  •   Quantity theory of money, which essentially says that the price level is directly proportional to the money supply, at least in the long run, wherein we assume that money velocity is stable and prices are fully flexible. Here, Money supply is taken as a fiscal phenomenon.
  •   Total Production: as we know, prices are determined by demand and supply. Therefore we also consider a measure of aggregate production in the economy as an explanatory variable, to account for supply side factors in inflation.
  •   Prices of oil or Oil imports: also taken as a measure of supply factors in the economy.
  •   Additionally, we also test for the impact of the political environment in the country on overall inflation, by including a dummy to indicate an election year, and another for a Pay Commission year.page3image27224 page3image27384\



Consumer Price Index (Industrial Workers): For measuring Inflation – our dependent variable in the model – we have used CPI (IW) data compiled by Labor Bureau, Ministry of Labor and Employment (GOI) and Ministry of Statistics and programme implementation (GOI). The Indices given was for varying base years (1960, 1982, 1987, 2001 & 2010). The CPI (IW) was then converted into a common base year of 1960 to make the data comparable.

Choosing CPI over WPI:

To quote our present RBI governor Dr. Raghuram Rajan, “probably we should focus more on CPI than wholesale price index (WPI) because CPI is what actually the common man sees and based on many decisions including wage decisions”.

The Wholesale Price Index (WPI) is used extensively as a measure of inflation and important monetary and fiscal policy changes are often linked to it. The weekly index numbers of wholesale prices have acquired considerable significance over time, since this is the only index which gives an idea of the week-to-week fluctuations in the prices of all the traded commodities. However, the WPI in its role as a guide to policy formulation has several critical limitations.

Limitations of WPI:

 non-inclusion of services
 following a fixed weighting scheme while the economy is undergoing major structural

 use of gross transactions data rather than data on final purchases.

In principle, inflation requires to be managed with respect to changes in prices of final goods or consumer prices. A number of consumer price indices like Consumer Price index for Industrial Workers (CPI-IW), for Agricultural Labourers (CPI-AL), and for Urban Non-Manual Employees (CPI- UNME) are compiled on a monthly basis.

The Labor Bureau of Government of India prepares the CPI-IW. It attempts to measure changes in the retail prices of fixed baskets of goods and services being consumed by the target group (namely the average working class family). Based on the Income and Expenditure survey by NSSO in 78 selected centers, this index is constructed on a monthly basis.

This index is also used for determining the dearness allowances to be paid to Central and State Government employees and to industrial workers besides fixation and revision of minimum wages to scheduled employments. The coverage of CPI-IW is broader than that for CPI-AL and for CPI- UNME. The CPI-AL and CPI-UNME are designed for specific groups of population with the main objective of measuring the impact of increase in prices on rural and urban poverty. The CPI-IW captures to some extent the price increase in the service sector.

Based on these ideas, we find that CPI (IW) is much more comprehensive and covers most of the goods and services consumed by average Indians.


For our study we have used, checked and applied almost 10 independent variables and their possible meaningful variations like one year lag, two year lag, changes from previous period to current period, etc. to make the estimation as significant as possible.

1) Rate of Interest (Key Lending Rates as Prescribed by RBI- All Commercial Banks Including SBI)

The source of data is the Reserve Bank of India website. The Average lending rates for all the commercial banks as prescribed by RBI has been used for the data on Interest rates. For most of the years, Interest rate was given in the min-max form, e.g. for the year 1973-74 Interest rate was given as 12-13. So mean of the value is used for all these years to test the Implications of ROI on India’s inflation.

2) Unemployment Rate (estimated):

As we have tradeoff between price rise and unemployment: Inflation and jobs go hand-in-hand. India is quite notorious when it comes to availability of timely and quality data and when it comes to unemployment more so. Till recently, the NSSO conducted the employment survey only every five years.

To see the impact of unemployment rate on India’s inflation we have tried to incorporate Unemployment rate in our model. But due to severe data crisis in term of Unemployment rate in India we have to drop it. As NSSO is the sole source of data collection for unemployment rates in India but with a lag of 5 year. For our time-series data, because of data insufficiency we have to drop that variable.

3) Oil imports (Rs. Billion):

Oil imports (in US dollar terms) accounted for 33 per cent of India’s total imports in 2007-08, as compared with the average 23.2 per cent in the 1990s, 27.2 per cent in the 1980s and 21.2 per cent in the 1970s. (Source-RBI) All oil-importing countries faced the specter of the ‘oil shock’ and India, being a net oil importer, is no exception. Although the burden of the price increase at the international market is not fully passed on to domestic consumers in India, the combined oil pool deficit exerted pressure on government finances, affecting the macroeconomic outlook and inflation in the ensuing year.

As we know our domestic production of Oil is only 15% and we have to import the rest of the oil to fuel our economy. This direct causes a heavy burden on our foreign reserves and holds a major share of trade deficit in our country. The larger the import of oil, the larger we have to buy foreign assets and it can change the price level of our domestic goods and will further fuel inflation. As Oils are consumed in transportation, electric production, Agriculture, etc. it could have been one of the determinants of Inflation In our economy. The source of data is RBI which gives the data on foreign trade and the amount spent on buying Oil.

4) Minimum Support Price (MSP):

For many years, MSP has been a very central parameter in the policy debate and its implication on Indian economy. As we know The Minimum Support Price (MSP) is a scheme of the Government of India to safeguard the interests of the farmers. Under this Scheme the GOI declares the minimum support Prices of various agricultural produces and assures the farmers that their agricultural produce (of FAQ) will be purchased at the MSP, thereby preventing its distress sale.

But in recent years this populist policy has received very sharp criticism among many economists. MSPs higher than input costs, this should benefit farmers. “However, at a time when inflation is already high and growth weak, rising MSPs will add to the structural uptrend in food price inflation, raise the floor on inflation and complicate matters for monetary policy. To check its impact in our model we have collected the data sheet from Source: 1. Ministry of Agriculture, Government of India. 2. Commission on Agriculture Costs & Prices (CACP).

5) Annual Production (Million Ton):

Inflation in India is said to be very sensitive to the agricultural production and particularly the food grains production. The originating cause in Indian case is said to be the fall in agricultural or food grains production, creating a situation of food shortage relative to demand and thus leading to what is called as food-shortage inflation, which through its spread to other sectors, causes the general price level to move up.

Natural calamities like flood or drought, and infrastructural bottlenecks in agriculture cause the production and thus the supply of agricultural/ food articles to fall short of demand, causing their prices to inch up. The price rise so began gets further momentum when organized labour succeeds in securing higher wages to compensate for the increase in their cost of living. This means increased industrial costs which breed what is called as cost-push inflation. So price rise begins in the agricultural or food sector whereby food prices rise on account of shortfall in production and get transmitted to non-food articles, resulting in inflation in the general level of prices.

The source of data for Annual production of food grains in India is again, 1. Ministry of Agriculture, Government of India, and 2. Commission on Agriculture Costs & Prices (CACP).

6) Money Supply (M1, Rs. Billion)

Milton Friedman, the father of monetarism and Nobel laureate in Economics, said that “inflation is always and everywhere a monetary phenomenon and argued that the changes in overall price level are only brought about by the changes in monetary stock or money supply.”

M1 is a measure of money’s function as a medium of exchange. As we are using CPI (IW) as our dependent variable, M1 will be the most suitable for our model. It includes all coins, publicly held currency, traveler’s checks, and money account balances for checking accounts, credit union accounts, NOW accounts, and automatic transfer service accounts. The source of data is RBI.

7) Import Index:

The Import indices measure average changes in prices of goods and services that are imported. The Import of many essential items on large scale like Petroleum, Drugs, Machines, technology widens the CAD. It affects the health of the economy. The raising of the CAD can fuel inflation in the long period. Government need to borrow foreign assets (particularly USD) to finance Import. In the next stage to pay debt, RBI print money and we can easily apply the quantity theory of money to analyze the impact of fresh money supply.

To measure the external influence on internal prices has been represented by levels of Import Indices (IMP).The data is taken from RBI and it was given for different base year. So the common base (Base: 1978-1979 = 100) method was used to make the data comparable. Source- RBI website (

8) Index Of Industrial Production (IIP):

This is an index which details out the growth of various sectors in an economy Index of Industrial Production (IIP) is an abstract number, the magnitude of which represents the status of production in the industrial sector for a given period of time as compared to a reference period of time.

The all-India IIP is a composite indicator that measures the short-term changes in the volume of production of a basket of industrial products during a given period with respect to that in a chosen base period. It is compiled and published monthly by the Central Statistics Office (CSO) with the time lag of six weeks from the reference month. IIP has been used as a representative for the volume of supply in the economy. The source of data is RBI with base year 1980-81.

9) Exchange rate with dollars average:

According to Mundell-Fleming model, an increase in interest rate is necessary to stabilize the exchange rate depreciation and to curb the inflationary pressure and thereby helps to avoid many adverse economic consequences. Appreciation of Rupee makes import cheaper and hence lowers inflation. A depreciation of exchange rate increases prices of imported consumption and intermediate goods. However the effect is not confined only to import prices. It affects prices of domestically produced goods through supply and demand channels. The rise in intermediate goods prices increases costs of production and hence prices of domestically produced goods. On the other hand because of rise in import prices, demand shifts to domestically produced goods, leading to further increase in domestic prices. The source of data is

10) GDP at Market Prices, at 2004-05 prices:

An increase of GDP, or growth in the amount of goods and services, should equate to a reduction in the level of prices for those items, or that deflation should occur. GDP and inflation are often associated with one another because governments and central banks often make decisions based on these figures and they attempt to manipulate them. If an economy is not growing or is not growing fast enough, a central bank may lower interest rates to make borrowing more attractive. The logic behind this is that it will encourage spending, which will lead to a rise in GDP. The drawback of this move is that, according to many popular beliefs, it will also prompt inflation. The Source of Data is RBI (



In a move to woo the voters ahead of the upcoming Lok Sabha elections, the government’s interim budget is the last before the party goes to face the electorate. As in prior elections, the economy is set to play a key role, and the government is under pressure to make sure growth is on track. Many tax duties are cut to appease the voter and for giving a populist budget. The good thing about the interim last Budget was that it did not throw up any negative surprises and made a point that India’s economy is on a firm ground.

The opposition always termed it as the election budget. At least the Budget did not make things worse and the industry expect a much larger package from the new government. To test, how the maximum targeted populist interim budget just before election affects inflation, we have incorporated a dummy variable that takes value (D=1) for the year in which last interim budget was presented just before the general election and D=0 for rest. For example, the 15th Lok Sabha election was held in April-May, 2009. And before this election the last Budget was presented in Feb. 2009. So, year 2009-2010 is assigned D=1.


Since India’s Independence, seven pay commissions have been set up on a regular basis to review and make recommendations on the work and pay structure of all civil and military divisions of theGovernment of India. Pay Commission set up for central government employees (and subsequently for state government) always leads to additional burden on the exchequer and increases inflation by pushing demand.

“Hefty pay hikes for the central government employees (as has been the experience of the previous pay commissions) act as a big drain on government finances and add to an inflationary pressure,” For ex- The cost of hikes in salaries after the recommendations of 6th pay commission was anticipated to be about Rs. 20,000 crore for a total of 5.5 million government employees. The implementation of Fifth Pay Commission cost about Rs. 17,000 crores to central government.

To quote the World Bank’s view on Pay commission for India, “the single largest adverse shock to the public finance of the nation.” We assign D=1 for the year in which pay commission recommendations were implemented instead of the year in which it was setup (i.e. D=1 for 1973- 74, 86-87, 96-97 and 2008-09) and D=0 for rest of the observations.





Standard Deviation

Minimum value

Maximum value


d-CPI (IW)








d- Money SS








Real r.o.i








d- IIP








5. MODELLING AND ESTIMATION: We began our study with the following model in mind:

∆CPIt = B1 + B2 ∆Money Supplyt + B3 ∆GDPt + B4 Real rate of interestt + B5 Unemployment Ratet + B6 Oil importst + B7 Exchange rate (with U.S. Dollar)t + B8 Minimum Support Pricet + B9 Dummy for Election Year + B10 Dummy for year of implementation of Pay Commission + B11 ∆Import Indext + ut

5.1 Evolution of the model:

But our initial model did not give us significant results as it suffered from various problems. Since we had only five-yearly data for unemployment rate for India (instead of annual data), we estimated annual rate of unemployment by assuming an instantaneous rate of growth of unemployment rate. But using this did not give us the correct sign of the coefficient, and we had to drop the variable. We also found a high degree of correlation between change in Money Supply and change in GDP. As a result, we used change in Index of Industrial Production (IIP) which is highly correlated with change in GDP but not with change in money supply.

Also, as we read in the literature, inflation is affected by not just the present value of money supply, but also past years’ money supply. We thus decided to use the lag of the first difference of Money Supply as the changes in Money Supply take some time in translating into inflation. And we also went ahead with dropping the variables which were either not turning out to be significant, or made other variables insignificant, like oil imports, exchange rate, Minimum Support Price, both the dummies and the Import Index.

Apart from this, on further research, we found that the current year’s inflation significantly depends on last year’s inflation and therefore, we decided to use the lag of ∆CPI as an independent variable as well. So, our model took the following form:

∆CPIt= B1 + B2 ∆Money Supplyt-1 + B3 ∆IIPt + B4 Real rate of interestt + B5 ∆CPIt-1 + ut

page9image62120 page9image62280


5.2 Estimation method

To accommodate the effect of last year’s inflation on this year’s, we resorted to autoregressive estimation. Since using OLS for estimating an autoregressive model leads to inconsistent and biased estimates, we have used the Cochrane-Orcutt method of estimation. Following is a brief description of the method:

In the estimation by the Cochrane-Orcutt method , at time “t”, say, we have the following model



Subtracting ρ times the one-period lagged version of equation (1) from equation (1) itself, we get:




The procedure proposed by Cochrane and Orcutt was essentially as follows:


= xt’ β + εt ; t = 1, 2, 3, ……, T (1)


= ρεt-1

+ut ; |ρ|<1

E[ut] = 0 ; var.[ut] = σu2

; E[ut

,us]=0 (t≠s).


-ρyt-1)=(xt’-ρxt-1′)β+ut ;t=2,3,….,T (2)

(i) Estimate (1) by OLS and obtain the residuals, et

( t = 1, 2, …., T).

(ii) Obtain a consistent estimate of ρ by regressing et on et-1(with no intercept), using OLS. Call the estimate “r”.

(iii) Replace ρ with r in equation (2), and estimate that model by OLS. This yields a new estimate of β.

(iv) Using this estimate of β, re-compute the residuals from (1), and re-estimate ρ, as in step (ii). (v) Use this new value of r in place of ρ in equation (2), and re-estimate that model by OLS.

(vi) Go to (iv), and iterate to convergence. The iterations stop when successive estimates of rho differ by less than 0.001.





















5.3 Testing basic assumptions

Since we have used the Cochrane-Orcutt method, the problem of autocorrelation has been removed. Also, to be sure that there is no problem of multicollinearity or heteroskedasticity, we have performed the following tests:

Testing for multicollinearity:


Variance Inflation Factors
Minimum possible value = 1.0
Values > 10.0 may indicate a collinearity problem

d_IIP 3.288 d_MONEY_SS__M1__Rs__BILLION_1 5.575 realroi 1.153 d_CPI_IW_1 3.002

VIF(j) = 1/(1 – R(j)^2), where R(j) is the multiple correlation coefficient between variable j and the other independent variables

Properties of matrix X’X:

1-norm = 21362388
Determinant = 2.971813e+020
Reciprocal condition number = 3.549176e-007

Testing for heteroskedasticity:

By plotting the residuals against the fitted values of dependent variable (which is a weighted sum of all independent variables), we can clearly see that there is no relation between the two:







Having checked for the basic assumptions, we proceed with our results.


So, our final equation becomes:




Out of these, the constant, ∆Money Supplyt-1 and ∆CPIt-1 are significant at 1% significance level while the Real rate of interest and ∆IIP are significant at 10 % significance level. Our estimated line is displayed below:









The coefficients can be interpreted as follows:

  •   If the value of all the explanatory values is 0, then the change in CPI from last year to current year is 42.78.
  •   Keeping other variables constant, an increase in Money Supply by 1 billion rupees in the previous year (i.e. from year t-2 to t-1) leads to 0.12 units increase in CPI.
  •   Keeping other variables constant, an increase in Index of Industrial Production by 1 unit leads to a fall in CPI by 1.19 units.
  •   Keeping other variables constant, an increase in real rate of interest by 1 % point leads to a fall in CPI by 2.10 units.
  •   Keeping other variables constant, an increase in CPI by 1 unit in the previous year leads to an increase in current year’s CPI by 0.55 units.

The R2 for this model turns out to be 0.85 (and adjusted R2 is 0.83) which is very significant as the p value of F test is tending to 0.


Our study finally concludes by offering a long term perspective on inflation in India that finds it to be significantly responsive to both demand-side and supply-side factors, like money supply, real interest rate and industrial production, respectively.

  •   The lack of significance of political and other factors like oil imports and exchange rate might be pointing to the fact that inflation does not respond to such factors in the long term (e.g. our analysis of forty years), even if it might be possible to explain a relation between these in a shorter time frame.
  •   We also find inflation to be affected by its own values in the previous year, highlighting the importance of expectation formation in actual realization of inflation. So for targeting, every year’s inflation is important.
  •   The insufficiency of data for testing unemployment still remains a loophole in our study. Availability of data on this front might throw up different results with regard to inflation.
  •   The significance of Index of Industrial Production shows that supply side in India plays a big role. Industrial production not only keeps growth engine moving ahead but also inflation low. It is not only policy variables in the hands of the RBI to impact inflation, but supply side considerations which are primarily in the hands of the government.
  •   Since, money supply and rate of interest can be controlled by RBI, this just re-emphasizes the fact that policies of RBI always play a very crucial role. Against this finding, the statement by the Finance Minister ‘to walk alone’ may be no more than just political rhetoric, not an economic possibility.
  •   Monetary policy has increasingly resorted to use of key policy rates in order to achieve objectives of monetary policy, including taming inflation. Our project finds unambiguous support for the potency of such an approach.


  • Kumar, R. (2013) “A Study of Inflation Dynamics in India: A Cointegrated Autoregressive Approach”, IOSR Journal Of Humanities And Social Science, Volume 8, Issue 1
  •   Patra, Michael D. and Ray, P. (2010) “Inflation Expectations and Monetary Policy in India: An Empirical Exploration” IMF working paper WP/01/84
  •   Pattanaik, A. (2010) “Study of Inflation in India: A Cointegrated Vector Autoregression approach”, Journal of Quantitative Economics, Vol. 8 No.1
  •   Loungani, P. and Swagel, P. (2001) “Sources of inflation in developing countries”, IMF working paper WP/01/198
  •   Chand, Sheetal K. (1996) “Fiscal and other determinants of the inflation rate”
  •   “Measuring inflation in India: Limitations of the WPI”, Madras School of Economics
  •   Raj J, Dhal S. and Jain R. (2008) “Imported Inflation: The Evidence from India” , ReserveBank of India Occasional Papers
  •   “Money Supply & Inflation”: Shodhganga , a reservoir of Indian theses
  •   Basu K. (2011) “Understanding Inflation and Controlling It”, Economic and Political Weekly,September 2011
  •   “Agricultural Output and Inflation”: Shodhganga, a reservoir of Indian theses

page15image14624 page15image14784


The authors of the paper are graduates in Economics from St. Stephen’s College, Delhi University. All three are set to pursue higher education in the field. You can reach them at [email protected], [email protected] and [email protected] respectively.

Leave a Reply

Your email address will not be published. Required fields are marked *

Input your search keywords and press Enter.