The Myth of Bountiful Monsoons

Of the many hoary myths floating around, few are so widely believed as the notion that India has been fabulously lucky to get a string of good monsoons in the 1990s. Critics claim this is a lucky fluke that has made liberalisation look artificially good. In fact, a detailed look at met office data suggest that rainfall has not been exceptional in the 1990s. As the last column of the accompanying chart shows, rainfall in the seven years of liberalisation has been no more than 99.86 per cent normal. This is not bountiful rain, it is just about average. So our economic success in this period owes little to the rain gods.

Most claims of a lucky fluke are based on sub-divisional data released regularly by the met office. This divides the country into 35 sub-divisions, and categorises rain in each subdivision as excess, normal, deficient or scanty. ‘Normal’ rainfall is defined as precipitation ranging from 20 per cent above to 20 per cent below the average. So ‘deficient’ does not mean simply below average but more than 20 per cent below average — a big difference.

The accompanying table based mainly on the Economic Survey covers 11 years divided into pre-reform (1987-90) and post-reform years (1991-97). It shows India has not experienced a major drought since 1987. The absence of a major drought in the 1990s is often taken to be a sign of exceptional rain. In fact, this is not so, as we shall soon see.

The four pre-reform years in the table, on average, 76.4 per cent sub-divisions received normal/excess rainfall. In the post-reform period, this went up to 86.51 per cent. That is a substantial difference, and is a major reason for the widespread notion that India has received a string of good monsoons. In fact, there have been rough years as well as smooth. In 1991 only 27 of the 35 sub-divisions had normal rainfall. That means almost a quarter of the country had deficient/scanty rain. And much of the deficit was concentrated in the poorest and semi-arid areas, which is why coarse grain production (which is unirrigated) declined by a whopping fifth. In 1994, only 25 of the 35 sub-divisions, had normal/excess rain. By this yardstick, two of the 7 reform years had sub-standard monsoons. So talk of seven good monsoons in a row is at best dubious.

It appears even more dubious if we look at other relevant data. Sub-divisional data give only a partial picture of rainfall. Even if rain is normal in a division as a whole, some districts in it could receive deficient rain (and the converse can also be true). Nor do words like ‘excess’ and ‘deficient’ tell you how much the excess or deficiency was. So to get a truer picture we need to examine at least two more sets of data given in the Economic Survey. One is the district-wise break up. The other is average rainfall for the whole country.

Number of met office sub-divisions Normal/excess rain % of districts Rainfall as percentage of normal
Excess/Normal Deficient/Scanty Total
1987 14 21 35 43 81
1988 32 3 35 88 119
1989 29 6 35 72 101
1990 32 3 35 84 106
Average pre-reform 26.75 71.75 101.75
1991 27 8 35 68 91
1992 32 3 35 65 93
1993 31 4 35 78 100
1994 25 10 35 76 110
1995 33 2 35 79 100
1996 32 3 35 81 103
1997 32 3 35 n.a. 102
Average post-reform

30.28 74.5 99.86

Critics of the reforms attribute the high growth of the nineties to a run of good monsoons. Data prove them wrong, says Swaminathan S Anklesaria Aiyar

On average, each meteorological subdivision has more than a dozen districts. And district-wise data lead to significantly different conclusions. Normal/excess rainfall was received in 71.75 per cent of districts in the pre-reform years, and in 74.5 percent in the post-reform years. In both cases, the percentages were far lower than suggested by the sub-divisional data. Besides, the district-wise difference between the pre-reform and post-reform phase looks quite narrow.

Finally, let us go to the rainfall data for the country as a whole. This provides the biggest surprises. In the four pre-reform years, the average rainfall was 101.7 per cent of normal. It was actually a bit above average despite the 1987 drought This is because 1988 brought 119 per cent of normal rainfall, an excess as large as the deficit in the drought the preceding year. This drives home an important lesson. The existence or absence of a drought during a decade tells you nothing about whether the decade was good, bad or indifferent An exceptional agricultural decline in a bad year can be fully offset by exceptional growth the following year. So it is wrong to think that the pre-reform years were exceptionally bad because of the 1987 drought.

What of the post-reform era? Rainfall in these years averaged 99.86 per cent of normal. Far from the rains being exceptionally bountiful, they were a shade below normal. They were also lower than the average of the pre-reform years. So much for the thesis of exceptionally lucky monsoons. However, we must not jump to the opposite conclusion, that we have been unlucky. We need to treat rainfall data with caution because all three measures in the table look respectable but yield very different conclusions. Consider the following anomalies.

  • In both 1988 and 1992, 32 out of 35 sub-divisions received normal/excess rainfall. So you may think the two years were roughly on par. Yet on a district-wise basis we find that 1988 was the best year in the whole table (88 percent) and the 1992 the second-worst (65 percent).
  • In 1994, only 25 of the 35 sub-divisions received normal/excess rainfall. Judged by the is yardstick, it was the second worst year in our sample. But the district wise break-up in this year shows that 76 percent of districts received normal/excess rain, which was actually a bit above average. And, overall, the country received a whopping 110 per cent of normal rainfall, the second-best year in the whole table.

Some readers will be thoroughly confused by this. Which is the right measure, they will ask? Strictly speaking, none of the three. Ideally we need data tehsil by tehsil, we need to know how evenly the rain was spread over a season, where excessive rain spread over a season, where excessive rain caused flood or pest damage, where rains were too late or were punctuated by long dry spells that killed crops.

In the absence of such details, we have to make do with what we have. The data most regularly published in newspapers are the sub-divisional figures as the sole or even most important measure. The district wise data provide a much better estimate since they present a more detailed break-up. They show somewhat more rain in the post-reform period, but the difference is too modest to produce a significant trend difference in agricultural growth.

Let us now go one step further. Note that the definition of ‘normal ‘in met office jargon is somewhat misleading. Rainfall up to 20 percent below average is classified as normal. So neither the sub-divisional nor district-wise data will capture a deficiency of, say, 10 per cent in rainfall.

That makes it advisable to also look at the total rainfall data. And these show that, overall, rainfall has been close to 100 per cent of average in both periods. The post-reform period has averaged marginally less rain fall (99.86 per cent against 101.75 per cent), but this again is too small to amount to a change in trend.

The conclusion that emerges is that rainfall has been unexceptional before and after the reforms. The lay reader may think this amounts to sitting on the fence. But it leads to one unambiguous corollary: faster GDP growth in the 1990s is the result of better policies, not bountiful rainfall.

What do you think?