Dynamics of Rainfall Effect on Foodgrains Yield in West Bengal

Like most of the Indian states, specially those in the eastern region, foodgrains production in West Bengal is still predominantly rainfed. The eastern region in general and West Bengal in particular, ushered in a spectacular improvement in foodgrains production during the 1980s, which was again quickly followed by a sharp fall in production during the next decade. The question is how far rainfall influenced these ups and downs of foodgrains production. The present study is an attempt in this direction in the context of West Bengal. The period chosen is 1970-71 to 1999-2000. Rainfall effect on yield has been studied both indirectly and directly. The indirect method relates to the quadratic regression of deviation of actual yield from its parabolic trend using actual rainfall as the sole explanatory variable basically to find out the threshold rainfall, if any, up to which rainfall has favourable effect on the yield deviation. The direct method relates to the addition of a standardized rainfall variable in the trend equation of yield as a linear mediating variable to examine the impact of rainfall on the yield trend both in the whole and sub-periods. In the direct method regression has been done not only by Ordinary Least Squares (OLS) method but also by Durbin⬠"!s Ranking (DR) method to avoid the suspected problem of errors in the stochastic rainfall variable. Results show for the state of West Bengal, the threshold rainfall is quite close to its normal level. The inclusion of rainfall variable in the whole-period best fitted parabolic trend has improved significantly the value of `R2 with significant partial regression coefficient of yield on rainfall indicating a clear mediating effect for the period as a whole, rainfall-adjusted absolute and relative growth rates, on an average being moderately higher than the unadjusted growth rates. However, the rainfall coefficient is marginally significant by OLS and non-significant by DR method in the double kink linear model, which gave the best fit for sub-period analysis. Still it is observed in the sample that rainfall-adjusted absolute and relative growth rates are slightly/marginally lower during first/ second sub-periods and substantially higher during last sub-period as compared to unadjusted growth rates. Even after this adjustment the first trend-break continued to be significant, but it made the second trend-break non-significant.