Lies, damned lies and statistics

Updated: Feb 7 2002, 05:30am hrs
This seems to be the season to remember Mark Twains dictum. For starters, Enrons implosion has triggered a global wave of scepticism regarding the financials of even blue chip corporates. Back home, the growth numbers of the Central Statistical Organisation for 2000-2001 and 2001-2002 ahead of the Union budget has led observers to smell a scandal of sorts and has drawn attention to the pathetic state of official data collection.

It is a sad commentary that just when a flurry of second generation reforms are taking off, official number crunching has hit a nadir. The rot in fact set in during the 1990s perversely highlighting the fact that numbers were a tad more credible during the licence raj times than in the era of reforms! The attention grabber this time was CSOs downward revision of GDP growth for 2000-2001 by two percentage points to 4 per cent.

Actually, this is nothing new. Even during the mid-1990s, the credibility of CSOs growth numbers was so low that a leading pink paper dubbed it as the Congress Statistical Organisation and argued that it was tailoring these numbers with the sort of urgency often associated with manifesto committee of political parties. Back to the present, and the agency followed this up with a bullish advance estimate of 5.4 per cent for 2001-2002.

Ahead of the Union budget, there are no prizes for guessing who is pleased as punch with such numbers. Of course, it is finance minister Yashwant Sinha. The nations television networks lost no time in breathlessly proclaiming that CSOs advance estimate showed that India was one of the fastest growing countries in the world! An honourable exception of course has been f40The Financial Express which noted in a recent editorial that having ensured lower growth in 2000-2001, Mr Sinha has ensured higher growth in 2001-2002 and a more respectable fiscal deficit to GDP ratio.

It is easy to see why. With the downward revision in overall growth in 2000-2001, the fiscal deficit straightaway goes up to 5.4 per cent of GDP. But thanks to the bullish advance estimate for the next fiscal, the deficit immediately goes down to 5.1 per cent of GDP. So, if for some miraculous reason the fiscal deficit is similar to the budget estimate for 2001-2002, the denominator effect gives it a more respectable cast.

Of course, there is bound to be slippage in the fiscal deficit, but CSOs advance estimate makes the problem more manageable than otherwise. The same applies to the revenue deficit which is the indicator most budget watchers track closely. Lower growth in 2000-2001 pushes up the revenue deficit to 3.7 per cent of GDP, while the higher GDP growth the next fiscal lowers the budgeted deficit to 3.4 per cent of GDP.

For such budgetary numbers to be credible, it is necessary that the government assumes a more realistic growth rate. There is no point framing the budget with a rosy 5.4 per cent GDP growth and reworking the numbers afresh when the latter is lowered dramatically soon thereafter. A 4 to 4.5 per cent growth rate ought to serve the purpose, especially given the low credibility of CSOs numbers throughout the 1990s till now.

Interestingly, the worry today is over downward revisions but during the 1990s the agency frequently revised upwards its GDP forecasts. Thus during 1994-95, its advance estimate of 5.3 per cent was marked up to 6.2 per cent and 6.3 per cent and then finally to 7.2 per cent. This charade was enacted for three straight years in a row, ending in 1996-97. There is, however, no point smelling a conspiracy in all of this. To be sure, CSO does have good professionals on its rolls.

What is at fault is the shaky foundation of primary data. Take the advance estimates of 2001-2002, for instance. The Index of Industrial Production during April-November 2000 is used to make the forecast for the year as a whole. But is this data reliable No, say researchers like R Nagaraj of the Indira Gandhi Institute of Development Research who fault the poor and inadequate quality of primary data. The IIP relies on voluntary reporting of monthly output by firms and reportedly 18 different agencies supply the primary data. Even revisions to the index (with 1993-94 as base) has made no difference whatsoever in its ability to correlate with time trends of the Annual Survey of Industries data on value added in registered manufacturing. Interestingly, it did so during the licence raj but not the era of reforms.

The foundation for service sector data is even shakier. For a sector that accounts for around 50 per cent of GDP, CSO uses measures like gross trading index, a combined index of railway net tonne kilometers and passenger kilometers, cargo handled in major ports, stock of commercial vehicles, outstanding phone connections, growth in bank credit among others. But what about the all important public administration and defence

The impact of the Fifth Pay Commission in terms of higher incomes of employees in public administration and defence clearly did pad up GDP growth during the second half of the 1990s. There is a need to strengthen the official data base in this regard to enable CSO to churn out more reliable advance estimates for service sector growth. It is no accident that lower growth in banking forced the two percentage point revision in 2000-2001.

With the poor quality of CSO numbers, the borderline between economic fact and fantasy has certainly been getting blurred over time. The government in power may hear the music that it likes, but there is no warrant for the finance minister to rely on such slippery estimates for his budget. The effort should be instead directed towards beefing up the statistical machinery so that official numbers correctly reflect the reforms era rather than force observers to hark back to the old licence raj.