The folly of GDP predictions: MInt 16th November 2009

GDP forecasts can mislead, while their multitude can confuse. Governments should rather focus on targeting
Think of any variable and you will realize that its number is amenable to forecasting. Earlier, macroeconomic forecasts used to be accompanied by shoulder shrugs but today, there is a sense of finality as strong econometric models back them. Hence, we have a wide array of such forecasts provided by different agencies and, not surprisingly, the final number is quite different. All of which prompts confusion, and perhaps even imprudent decision-making.
Let us look at the forecast for gross domestic product (GDP) growth for India today. The Reserve Bank of India (RBI) put out a forecast of 6% for 2009-10 on 27 October, which was lower by 0.5 percentage point from what the Prime Minister’s economic advisory council had predicted a week before. Subsequently, the Planning Commission upped its own forecast from 6% to 6.5%. The International Monetary Fund (IMF) says 5.4%, the Asian Development Bank says 6.0%, the World Bank says 5.1%, the Organisation for Economic Co-operation and Development says 5.9%, the Indian Council for Research on International Economic Relations says 5.8-6.1%, while the National Council of Applied Economic Research says 7.2%.
So the range of forecasts is between 5.1% and 7.2%; quite clearly, the actual number will lie somewhere in between. All will claim, when the actual GDP figures are out, that they missed it by a certain margin. However, in monetary terms, this margin is substantial. India’s real GDP in 2008-09 was around Rs36.1 trillion. A variation of 1% in the forecast would actually mean around Rs36,000 crore (in constant terms). That amount is too large to be simply waved as a statistical error. Can we then choose the best forecast?
The answer is that it is inherently very difficult to gauge how the economy will behave by the end of the year.
First, we need to distinguish between statistical forecasts and targets. If it is the former, then we have a problem because when we have to forecast GDP, we have to take a call on agriculture, industry and services, which is a tough one. How do we know the behaviour of the monsoon in April or the performance of the rabi crop? Also, industrial growth is whimsical and dependent on agriculture. Further, around 40% of the service sector output comes from the unorganized sector—transport, hotels, retail trade, real estate—subject to varied imputations, thanks to lack of recorded data, and hence difficult to conjecture.
Econometric models use past data; but this is very unreliable because the performance of any sector is based on current conditions such as the monsoon, crop damage, government spending and so on. The variables that determine the forecast are themselves subject to forecasts, which make the entire process prone to error. Rough GDP scenarios are better hedges, but are not precise.
One can instead sit back and say: Agriculture accounts for around 20% of GDP and will show a growth of 2%; industry, with 20% of GDP, will grow by 8%; services, with 60%, by 8%—and then simply arrive at a GDP growth rate of 6.8%, which will not be far off from the final number!
Hence, pragmatically, it makes more sense to talk of GDP targeting instead, which governments should do. Ideally, they should target growth of, say, 7% and align policies for the same. They can then scale this target based on changing circumstances.
Second, we should remember that there are just too many imponderables during a year. Besides the varied domestic factors mentioned above, globalization has increased the shock effect of unknown variables—or the epsilons in econometrics jargon. Oil prices rising or falling, a war, the US Federal Reserve’s rate changes, housing boom and carry trade are some events which affect our economy indirectly through the trade and capital flow routes.
Third, this plethora of estimates is confusing, even though they are theoretically sound. Rarely does an estimator always get the number right, which means that the reader will never know which is the best estimate. So what is the reader to believe?
Curiously, IMF—which uses some of the most sophisticated models—also goes off the mark most of the time (see table). Based on the World Bank’s estimates of GDP in 2008, the world economy was sized at $60 trillion. A deviation of 0.1% of GDP here means going off the mark by $60 billion. The euro economy going off the mark by 0.7 percentage point means a $95 billion change! As can be seen in the table, the 2008 estimates were quite out of line, while 2007 was only marginally better.
So what are we to make of such statistical exercises in India? In general, RBI has been closer to the mark. Global agencies tend to be pessimistic, while agencies that try to “sell” India are usually optimistic. Politicians are overly sanguine when they take a call for future years, especially if the present looks cloudy.
But here’s the rub: Taking business decisions based on forecasts could upset the apple cart. Over-sanguine or over-gloomy forecasts could prompt over- or under-investment. So, even after businesses overcome the confusion of deciding whose estimate to rely on, they could be hurt when they realize that the estimate was far from reliable.
So while governments should consider targets, the average businessperson is best off glancing at one of these myriad forecasts, and then probably doing nothing about it.