Price and Income Elasticity of Demand for Services in India: A Macro Analysis
SatyanarayanKishanKothe
Assistant Professor (Sr)
Department of Economics
University of Mumbai, Mumbai (INDIA)
+91 9699200509
Abstract
India has experienced phenomenal growth in recent period which is attributed by services revolution. India emerged as prominent service provider to the global consumers. Besides exports of services, Indian services are also consumed by domestic consumers. Domestic demand for services equally contributes in India’s services led growth paradigm.A systematic analysis of domestic demand for services is needed to understand services revolution in India.The paper estimates domestic demand for services. The attempt is made to estimate price and income elasticity of demand for services for 1951 to 2010. The Study endeavours to find structural changes in demand for services during pre and post liberalized period (1951 to 1990 and 1991 to 2010) through the most important determinants of demand for any product/service i.e. price and income.
Keywords: Demand for Services, Price Elasticity of Demand, Income Elasticity of Demand, Consumption Expenditure and Elasticity of Demand for Services.
Introduction:
India’s services revolution and services led growth have raised various issues for discussion. The studies that have comprehend the sources of recent services revolution include Kothe and Kouthe (2009) entails role of exports of services in economic growth, the impact of FDI on services growth in India is discussed by Kothe and Sawant (2010), Kothe (2012a) endeavored to bring out the linkage of globalization and services revolution in India andKothe (2012b) connoted that rise in demand for services with rise in per capita income induced more employment in services sector. Kothe (2013) also pronounced that increase in demand for services promoted capital formation in services sector in India. The standard export demand function for India’s services is estimated and theorized in Kothe (2014) that India’s services exports are highly income elastic and less immune to the changes in domestic prices. The discussion encourages the desire to find more insights of services revolution in the context of demand for services.
Demand for services function can be as conventional as the demand for goods. Perhaps we find enough discussion about demand for goods functions, but not on demand for services. Engle’s Law (1857) is however more conventionalized by Clark (1951) that the employment in services sector increase with increase in income which implies that demand for services respond to the changes in income. Baumol (1967) explicitly modeled that the demand for services increases with the increase in income. Summers (1985) estimated price and income elasticities of demand for services classified under SNA for thirty four countries. And he also discussed service specific elasticities and their nature of response to price and income. Falvey and Gemmell (1991) stated that service income elasticities in aggregate tend to be statistically greater than, but numerically close to, unity. Further Falvey and Gemmell (1996) have been inclined to reject the hypothesis that income-elastic demand for overall services but found income elasticity estimates above and below unity for different types of services. Mahadevanand Kalirajan (2002) did find the income inelastic demand for services in Singapore. Hansda (2001)also remarked about the empirically found income elasticity of demand for services and also talked about the demand for services in India. Since there have not been attempts to find structural changes in the demand for services in India, the present study tries to discover the same if exist and compare the measures for pre and post liberalization period.
Methodology and Data:
As noted in Falvey and Gemmell (1996), Summers (1985) used following three equations for the estimation of income and price elasticity of demand for services.
(1)
(2)
(3)
Where is expenditure per capita on services and is GDP per capita, both converted to $ at nominal exchange rates; and are respectively “real” expenditure per capita on services and “real” GDP per capita (i.e. converted to $ using category specific PPP exchange rates). and are the (domestic) price of services and GDP respectively, and is a random error term. The nominal and real share of services rise with GDP per capita if and respectively, and services may be deemed to be income-elastic in demand if . Real expenditure () here are equivalent to quantities - in national accounting terms - and thus equation (3) may be viewed as a simple demand function. It may attract objections that it omits ‘taste’ variable. Therefore a more complete specification is:
(4)
Where, Z is a vector of ‘test’ variables. It is to be noted that in (4) service and commodity prices (and ) appear separately allowing testing of homogeneity condition . Equations (3) and (4) can be used to estimate the income elasticities of aggregate or individual services with a ‘composite’ of commodity and other service prices in the later case. Therefore it is very much agreeable that equation (4) after omitting Z could be of help in estimation of price and income elasticity in our case. Z is a vector of ‘test’ variables that can be omitted to compare the results with that of Summers (1985).
Though the above methodology does not accomplish the necessary data requirements are accepted by economists as general form of simple demand for services function. Hence for the computation of conventional demand function of goods/services one need to have the data on quantities of services, prices of services and GDP. There are few issues in defining quantities of services and prices of services. These issues have been sorted in the following discussion. As the quantities/units of services cannot be defined as structured as it is required and also are not readily available. Summers (1985) suggested that consumption expenditure are equivalent to quantities in national accounting terms, perhaps, as discussed above have estimated elasticity of demand for services with the help of consumption expenditure function for services.
Prior to the estimation of demand for services, it is required to define the demand for services. But discussion by Hill (1999) makes it easier task to define the demand for services. The whole debate on conceptualization of services is carried out by Hill (1999) in which he discussed the issues in conceptualization of services and also about the tangibility and intangibility features of goods and services. In the deliberation he stated that services cannot exist independent of their producer, they consist of change brought about in the condition of one economic unit by the activities of another economic unit. Many services are capable of bringing material changes in persons or products. Since services are not entities, they cannot be stored at large. Therefore every unit of service is produced and consumed within a stipulated time. Many times the production and consumption of services is simultaneous. A very few services can be stored in physical nature. Having known this we can conclude the discussion about the storability characteristic of services and assume that every unit of service produced in the economy is consumed by either an individual as consumer service or an institution as producer service. In that sense, therefore, services GDP is same as the consumption expenditure on services.
However the consumer expenditure on services does not interpret the demand for services truly. Though the consumer expenditure on services do not represent demand for services truly, the studies by Summers (1985), Falvey and Gemmell (1991) (1996) and Mahadevan and Kalirajan (2002) have considered the former as demand for services for the estimation of elasticities of demand for services. And also in the present study, elasticities are measured for demand for services function. Hence defining consumption expenditure on services in India is not that difficult though no official data is published on consumption of services. Having some arithmetic, we can get the data on consumption expenditure on services. The consumption expenditure on services would be equal to the services GDP (+/-) net services trade.
In India, services price index is not measured, therefore it would not be appropriate to measure price elasticity using WPI or CPI both of which do not represent the changes in prices of services of which weights are too small and also do not cover all the services. The GDP deflator could be better measure in which prices of all the goods and services are approximated with their respective shares in GDP. Perhaps in the present study we have computed services output deflator i.e. the product of the ratio of nominal services output to real services output and hundred and considered that to be Services Price Index (Ps). Therefore the income and price elasticity of demand for services is measured by employing log linear regression of demand for services (CONSERi) on GDP (Yi) as income and Services Price Index (Ps) as price proxy for services. We have not only measured the price and income elasticity but also attemptedto find structural changes, if any, in the aggregate demand for services function in pre and post liberalization period i.e. 1951 to 1990 and 1991 to 2010. Therefore unlike equation (3) equation (5) estimates price and income elasticity of demand for services for the entire sample period.
In the present study relatively simple demand for services function, unlike Summers (1985), Falvey and Gemmell(1996) is estimated with the help of following equation.
(5)
Where is the demand for services, represents the GDP at constant prices, is the Services Price Index and is the Price Index of Rest of the Services GDP, further and t represent time. And t =1951 to 2010.
Perhaps, to discover the structural changes in demand for services, existing methods suggest the use of Chow test (1960) on priory basis. Two different regressions could have been run and applying Chow test (1960), useful to examine the structural stability of a regression model, which would help in deciding whether the two periods give two different demand functions or there is no need to do so. And also is possible to find structural change if any in two different periods. But Chow test has its own limitations, as it is powerless to provide reasons for the structural change if any. Such a structural change is due to difference in slopes or intercept or both from each of the regressions of a model. But instead of that we have used dummy variable approach by Gujarati (1970) that is more capable of providing specification of any such difference. However after reviewing the available methodologies the final call is taken on to use the following function form of the model. Therefore equation (6) attempts to find structural changes in demand for services during pre and post liberalized period through the most important determinants of demand for any product/service i.e. price and income.
(6)
And t =1951 to 2010.
Where is the demand for services, represents the GDP at constant, is the Services Price Index and is the Price Index of Rest of the Services GDP, further t represents time. is dummy variable associated with and which define for pre and for post liberalization period respectively.
Similar way for the sub-sectors in services, demand functions are estimated and price and income elasticities are measured. India’s national accounting system facilitate the disaggregation of output of services into four categories, that are 1) Construction, (CNSTN) 2) Trade, Hotel, Transportation and Communication, (THTC) 3) Finance, Insurance, Real Estate and Business Services (FIRB) and 4) Community, Social and Personal Services (CSPS). For this purpose equation (5) and (6) are formed as base equations and the following generalized form of equations are used to estimate the price and income elasticities of demand for group of services stated above.
(7)
where is consumption expenditure in i sub service, is the GDP, and is the Services Price Index of ith sub service and is the Price index of rest of the sub services GDP and lastlyt represents time. And t = 1951 to 2010.
And to evaluate the structural changes in demand for sub services as earlier the following equation is used.
(8)
And t =1951 to 2010.
where is consumption expenditure in i sub service i.e. CNSTN, THTC, FIRB and CSPS, is the GDP, is the Services Price Index of i sub service i.e. , , and and is the Price index of output of rest of the i sub services, further t represents time. is dummy variable associated with and which define for pre and for post liberalization period respectively.
All the variables at level i.e. the demand for services () demand for CNSTN, demand for THTC, demand for FIRB and demand for CSPS, GDP (), ratio of the Price Indices and , dummy variable (D1) and its product with GDP (D1) and ratio of Services Price Indices and alls are tested for stationarity, as the variables form time series, with the help of augmented Dickey-Fuller (ADF) (1979) and Phillips-Perron (1988) test for unit root. And both the tests confirmed that all the variables found to be non-stationary at level.
According to Granger (1986), if variablesare individually non-stationary and are I(1), that is, they have stochastic trends, their linear combination is I(0). The linear combination cancels out the stochastic trends in the two series. As a result, a regression of such non-stationary variables at level would be meaningful (i.e., not spurious). It is said that the two variables are cointegrated. Economically speaking, two variables will be cointegrated if they have a long-term, or equilibrium, relationship between them. Here the meaning of equilibrium is not as same as it is used in economic theories. Hence, equation (5) and (6) are tested with the help of Engle-Granger test (1987) for a cointegrating regression and also observed that these regressions are not spurious, even though individually the two variables are non-stationary.
Therefore we run log linear regression to find the demand for services function which ultimately estimates the income and price elasticity of demand for services. Such a regression may be called the static or long run demand for services function and interprets its parameters as long run parameters. Therefore the parameters represent long run income and price elasticity of demand for services. The present study also attempts to find the change in elasticity are due to structural change, if any, between pre and post liberalized period by including dummy variable to distinguish the period.
Results and Observations
We estimated the above equations (5) and (6) to measure the response in demand for services due to changes in price and income. We obtain three estimates from above equation (5) and (6) that are summarized in the table (1).
Table 1: Income and Price elasticity of Demand for Services
Period / Income Elasticity of Demand= / Price Elasticity of Demand
=
1951 to 2010 / 1.169 / -0.29
1951 to 1990 / 1.218 / -0.219
1991 to 2010 / 1.125 / 0.683
Summers (1985)** / 0.977 / -0.06
Falvey and Gemmell (1996)** / 0.979 / -0.32
* All the measures are significant at 5 per cent level of significance
Source: Author’s estimation and ** are from Summers (1985) and Falvey and Gemmell (1996)
Income Elasticity of Demand for Services (Total)
The income elasticity of demand for services (Total) is obtained from regression equations (5) and (6). Results reported in Table (1) were obtained using ordinary least squares (OLS) method.
Our long run estimate of income elasticity of demand for the period of 1951 to 2010 found to be 1.169 which suggests that demand for services is elastic in response to change in income. Therefore the empirical hypothesis that services are always supposed to be income-elastic in demand is accepted. The results also suggest that services were more income elastic in pre liberalized period in comparison to post liberalization. Decline in income elasticity of demand for services in the post liberalized period suggests that service that were once luxurious now tend to become comparatively less luxurious. Therefore that is the change occurred in the behavior of the consumers in post liberalized period that services once were luxuries have tend to become relatively normal good but it remained luxurious by and large. Perhaps, it would be more appealing if we disaggregate the services and see how the behavior of income elasticity of demand is found to be for the group of services for the entire sample period and in also pre and post liberalization period. The reason for that is services are heterogeneous and the need for services also differs from group to group. Though not comparable with that of Summers (1985) and Falvey and Gemmell (1996) our estimated results in long run are greater than that of from these two studies imply that services in India are luxurious in total and found to income elastic in nature.
The national database does not classify each service separately rather it groups the services. Therefore eventually our analysis limits ourselves to the group of services rather than each service. But still with the given limitations our estimates provide enough substance to elaborate broader perspective about income elasticity of demand for group of services.
We estimated the equations (7) and (8) to measure the response in demand for services, for each group separately, due to changes in price and income. We obtain one estimate from above equation (7) and two from equation (8) that are summarized in the table (2).
The Income Elasticity of Demand for Construction Services (CNSTN)
The general definition of demand for construction services include demand for housing, demand for commercial and industrial buildings, demand for infrastructure such as roads, hospitals, schools, bridges, ports, airports, dams, etc., demand for repair and maintenance. With increase in income the need for all such demand for construction services also rise. That indicates that income elasticity of demand for construction always is greater than 1.
Table 2: Group wise Price and Income Elasticity of Demand for Services
Period / Price and Income Elasticity of Demand for group of Sub ServicesConstruction (CNSTN) / Trade, Hotel, Transportation and Communication (THTC) / Finance, Insurance, Real Estate and Business Services (FIRB) / Community, Social and Personal Services (CSPS)
Income / Price / Income / Price / Income / Price / Income / Price
1951 to 2010 / 1.378 / -1.108 / 1.325 / -0.279 / 1.042 / -0.604 / 1.093 / -0.735
1951 to 1990 / 1.478 / -1.207 / 1.398 / -0.388 / 0.962 / -0.499 / 1.174 / -0.497
1991 to 2010 / 0.956 / 0.660 / 1.402 / 0.141 / 0.958 / -0.136 / 0.922 / 1.193
* All the measures are significant at 5 per cent level of significance