It is a relief now to be able to confess that it has been very tough explaining to potential foreign investors how on Rs. 50,000 a year, a family of four can live, eat, educate children, and still buy consumer durables. Or how 60% of those with an income of Rs 10,000 to 12,000 per month have two wheelers and 25% have cars. It is also tough convincing MNC consulting firms that they should not be making forecasts on India applying their empirical data from other countries (e.g. the threshold level of income beyond which consumption "takes off ") to our income statistics.
How have we been managing so far? Has the whole marketing community really been batting blind? Not quite blind. There are two popular sources of income data, NCAER's MISH survey and ORG-MARG's IRS survey. Both these surveys produce income data which is reliable (if you do it again and again on different samples, you will get the same result), but not valid (is a survey determined income of Rs.50,000 really that or is it closer to Rs.1 lakh? And does this factor of 2 remain the same if the survey income is Rs. 30,000)? We have learnt to calibrate survey income and make some sense of it through understanding consumption pattern in each stated income group. Consumption, like maternity is a certainty. Income, like paternity, is merely a matter of opinion. (A great step in this direction was the development of the consumption based model of consuming classes by NCAER in 1991-92. However NCAER has always been rather vague on the specifics of how this model was arrived at, and it is hard therefore to understand how exactly to use this model in various business contexts.)
So far, income distribution data of NCAER and ORG-MARG have been reasonably similar - despite slightly different income classifications, the approximate number of households in each income group has been the same. However this comfort is also vanishing. The 2000 IRS data from ORG-MARG is sharply different in the story it tells about the income structure of the country from the 1999 projections of NCAER, which shows a richer Urban Consumer India as compared to IRS.
It certainly is time to say "enough of fuzzy data and personal correction factors mixed with common sense and experiential gut feel". It is time to get real data that can be used as is. And it would also help to get data that is both timely and comparable over time. IRS data does not adjust income data collected over time for inflation, making it not too good on comparability. NCAER does, but uses it based on a pan Indian wholesale price index, whereas inflation in small and large towns, and in different parts of the country could be quite different. On timeliness, IRS provides, through a rolling sample done each year, data which is maximum six months old. NCAER however are still working with five or six year old survey data and projections based on GDP growth (begs the question which GDP "official" number do they use!) In an environment as volatile as the one we live in, the maxim "better late than never' certainly does not apply. There is no market for data that is only useful for the luxury of conducting post mortems. NCAER uses the explanation of its massive sample size compared to that of commercial market research agency surveys. However the value of such a sample size, indeed the cost benefit of such a sample size in terms of improved confidence levels is not clear, at least to me. And eventually, it is better to be approximately timely and useable than perfectly late and unusable.
I went to the Guru of this subject, the former Director General of NCAER, SL Rao, and asked him what he thought. Could we for, example, use NSS data , which is a household expenditure survey and hence avoids all the speculative paternity type issues that go with income data? His view was that NSS was not very helpful at all. It is data that comes out five years too late. Also it looks only at expenditure and not at savings, though it does cover interesting data like spend on health care for the family. Further, it was created 40 years ago, for use in the policy issues of those days, like for example whether the public distribution system was working etc., it does not cover too many of the manufactured goods that consumers buy today. Net net, it does not provide (nor was it designed to provide) the expenditure patterns that would help businesses today target consumers better.
SL Rao feels that doing a proper income survey for a country as heterogenous and large as India is quite complex. Black money, rural wage patterns, applying the concept of purchasing power parity (PPP) to incomes in big and small towns etc. need to be factored in. The last time such a study was done, according to SL Rao, was in 1972 by NCAER and cost Rs.1 crore. To do it again today would cost at least four times as much. It involves far more intensive measurement than the current method of waving a card with income classes at the respondents and asking them which class they belong to. It involves intensive questioning, observation of all durables owned in the household, when they were bought, what was paid for them etc. and a lot of expenditure, saving and earnings questions, which then get summed up into a final income number. This costs money, and the question he asks is "do you need mammoth samples" to do this? Or can a better, more efficient design take care of this?
So who is to embark on this mission of providing the most basic data that consumer goods and financial services marketers need? And who will fund it? My vote would be for a consortium of commercial market research agencies to execute it, with NCAER as technical consultants, funded by the Government (who need this data in any case to assess the impact of reforms on the people of this country).