A K Bhattacharya: Dealing with dodgy data

RAISINA HILL

It takes as many as five iterations spread over a period of two years to get a final fix on GDP numbers.

The Union government's numbers on its expenditure and receipt go through three stages before they become final. The first set comes almost a month before the year starts, along with the Budget that is usually presented on the last working day of February. These are called Budget Estimates (BE).

The BE numbers are premised on the government's assessment of how the economy will do in the year to come and the projections made by the departments of revenue and expenditure on the growth in tax collections and spending by different ministries. These are actually projections and, therefore, invariably get revised before the year ends. The revised estimates (RE) for the year are available about a month before that year comes to an end. And almost 11 months after the year ends, the government puts out the actual figures. These are final figures and are rarely revised.

There have been instances of the actual figures also getting revised, raising doubts about the sanctity of the Budget numbers that, mind you, are presented before Parliament. Some noise is made when such revisions get noticed a year and a half later. But on the whole, there is no great discomfort or disquiet over the government's finance numbers as these are available without much of a time lag and the revisions, if any, are rare.

It is, therefore, possible to get hold of only one-year old actual figures for the government's finances and also compare the REs with the BEs for the current year to assess how well the government has been able to spend its money or how efficiently it has been able to collect the taxes. This is an advantage that most analysts don't enjoy while dealing with other numbers, particularly, when they pertain to the state of the economy.

Data on the economy's growth continue to suffer from a maze of estimates that are put out periodically and with each release they get revised and in most recent cases they are only getting revised upwards. These obviously have other implications for the numbers on the government's finances. For instance, with the GDP growth figure for 2005-06 now being revised from an earlier 9 per cent to 9.4 per cent, the government's fiscal deficit as a percentage of GDP may look different. The only comfort the government might draw from this is that since these are upward revisions, the deficit numbers, a key indicator, can only look better.

But the fact is that the government's data on the economy's growth go through five iterations. And the time lag is quite unnerving. The first to come are the advance estimates. These are made public about seven weeks before the year comes to an end. So, the advance estimates provide the first indication of how the full year will look like from growth point of view. The problem is that there are four more revisions of this figure before it becomes final.

The advance estimates get revised a few months later and are rightly called revised estimates. Almost about 10 months after the year comes to an end, the revised estimates go through another revision. These are then called, strange as it may sound, quick estimates. The quick estimates are also revised and become provisional estimates. This process takes another twelve months. Thus, in January 2008, the country got to know the provisional estimates of growth for 2005-06, 22 months after the year had ended.

The problem is that even the provisional estimates are not final. One has to wait for a few more months to get the final estimates of growth. So, almost two years lapse before the final growth figures are actually available. In this period, various other estimates based on the GDP figures also keep changing.

Should data for GDP growth go through so many revisions and take almost two years before they become final? Like the government's Budget numbers, why can't growth numbers also be captured only in three stages "" advance estimates, revised estimates and final estimates? And why shouldn't the whole exercise be completed in a year? There may be practical difficulties. Perhaps the newly created National Statistics Commission should look at this issue as well, along with the several other changes it proposes to bring about in a bid to make India's economic data more robust.