Themes > Economic Briefs
03.04.2002
CSO's Shaky GDP Estimates
C Rammanohar Reddy
The past 10 days have been embarrassing for India's official statistics
machinery. First, the Central Statistical Organisation announced that the
economy grew by only 4 per cent in 2000-01, after saying a year ago that
growth would be 6 per cent and following that up in June 2001 with an
estimate of 5.2 per cent for 2000-01. If the extent of this downward
revision was not bad enough, the CSO quickly followed it up with its
Advance Estimate of growth in the current fiscal year 2001-02. This would
seem just what the doctor — the Government — ordered: GDP growth of 5.4
per cent, while everyone had been talking of growth of between just 4 and
5 per cent.
The cynics would say that the CSO has boosted growth in 2001-02 by lowering its new (Quick Estimate) of GDP in 2000-01. (If the CSO had kept its figure for GDP in 2000-01 at the previous estimate, growth in 2001-02 would have been just 3.9 per cent.)
Allegations of falsification are perhaps not warranted, though you would not expect the Government not to take advantage of what the CSO says about growth in 2001-02, while it chooses to keep mum about the latest figures showing a dismal performance in 2000-01.
The major revisions in the GDP data take place every year. Indeed, they give sufficient fodder for criticism by the media every February. (see Economic Outlook of February 24, 2001 and February 10, 2000 for similar analysis of past revisions.)
The CSO is not to blame for the embarrassing revisions it has to make to GDP estimates. It has defended its revisions by pointing out that its Advance Estimates of GDP are based on quantity/output indices of trends up to December which are supplied by the administrative machinery. This is true, the CSO relies entirely on output figures supplied by the others. Crop production estimates come from the Agriculture Ministry and industrial output from the Industry Ministry. It cannot question or revise these primary sources of information and has to build GDP data based on these estimates. Later, when the Agriculture Ministry revises its output figures (as it always does), the CSO has to make its adjustments. And only when the Annual Survey of Industries information is finally available, it can independently estimate value-added in industry. All this may be true. But the extent of revision — especially between the Advance Estimate and the Revised/Quick Estimate — is so large every year (see table), that one wonders if there is any point in the CSO putting out its Advance Estimate.
When the CSO began putting out its Advance Estimates in the mid-Nineties, it appeared that India's statistical machinery was coming of age in giving up to date GDP figures. But now it is time to ask if it is not better to wait for accurate, even if delayed information rather than use up to date but highly inaccurate estimates of GDP. As the table shows, only once (in 1997-98) was the final GDP figure not very different from the initial estimate. (And the achievement in 1997-98 was a coincidence because the sectoral GDP estimates were revised considerably!). In all other years, the GDP growth figures had to be considerably revised (up and down) between the Advance and Revised/Quick Estimates.
The solution is to improve the quality of information that the primary agencies supply to the CSO. But until that happens it may be worth taking a step back and temporarily abandoning the practice of putting out the Advance Estimates of GDP. Since so much of the projections in the Union budget for the next year are based on the Advance Estimates of GDP for the year that is to end, what we have now are budgetary projections that are invalidated even before the year has begun.
The cynics would say that the CSO has boosted growth in 2001-02 by lowering its new (Quick Estimate) of GDP in 2000-01. (If the CSO had kept its figure for GDP in 2000-01 at the previous estimate, growth in 2001-02 would have been just 3.9 per cent.)
Allegations of falsification are perhaps not warranted, though you would not expect the Government not to take advantage of what the CSO says about growth in 2001-02, while it chooses to keep mum about the latest figures showing a dismal performance in 2000-01.
The major revisions in the GDP data take place every year. Indeed, they give sufficient fodder for criticism by the media every February. (see Economic Outlook of February 24, 2001 and February 10, 2000 for similar analysis of past revisions.)
The CSO is not to blame for the embarrassing revisions it has to make to GDP estimates. It has defended its revisions by pointing out that its Advance Estimates of GDP are based on quantity/output indices of trends up to December which are supplied by the administrative machinery. This is true, the CSO relies entirely on output figures supplied by the others. Crop production estimates come from the Agriculture Ministry and industrial output from the Industry Ministry. It cannot question or revise these primary sources of information and has to build GDP data based on these estimates. Later, when the Agriculture Ministry revises its output figures (as it always does), the CSO has to make its adjustments. And only when the Annual Survey of Industries information is finally available, it can independently estimate value-added in industry. All this may be true. But the extent of revision — especially between the Advance Estimate and the Revised/Quick Estimate — is so large every year (see table), that one wonders if there is any point in the CSO putting out its Advance Estimate.
When the CSO began putting out its Advance Estimates in the mid-Nineties, it appeared that India's statistical machinery was coming of age in giving up to date GDP figures. But now it is time to ask if it is not better to wait for accurate, even if delayed information rather than use up to date but highly inaccurate estimates of GDP. As the table shows, only once (in 1997-98) was the final GDP figure not very different from the initial estimate. (And the achievement in 1997-98 was a coincidence because the sectoral GDP estimates were revised considerably!). In all other years, the GDP growth figures had to be considerably revised (up and down) between the Advance and Revised/Quick Estimates.
The solution is to improve the quality of information that the primary agencies supply to the CSO. But until that happens it may be worth taking a step back and temporarily abandoning the practice of putting out the Advance Estimates of GDP. Since so much of the projections in the Union budget for the next year are based on the Advance Estimates of GDP for the year that is to end, what we have now are budgetary projections that are invalidated even before the year has begun.
© MACROSCAN 2002