72 Auriciillural Research and Produciivily 



With the area deflator, d = A (disregarding within-country varia- 

 tions in area), and 73' the regression coefficient, measures total 

 contribution in the country. This is also true if the correct deflator 

 is not equal but only proportional to the area, since the factor of 

 proportionality will be incorporated in the estimate of 73- Hence, 

 in the estimates where the area deflator was used, the marginal 

 (= average, since this is in the linear model) productivity of 

 research is the same for all countries. 



The regional deflator is not proportional to the area (in many 

 countriesc?2= 1), and if it is the correct deflator, the contribution 

 of research varies with the crop's area, in many cases propor- 

 tionally. 



In the double-log model the regression coefficients are 

 elasticities, measuring percentage increase in yield resulting from 

 percentage changes in the stock of knowledge (whatever the 

 deflator). With country dummies included in the regressions, the 

 constant-per-country deflator ''washes out.'' The marginal con- 

 tribution of a publication is exactly proportional to total produc- 

 tion level of the crop. 



In the rate regressions model of table 4.1 (all area deflated), ad- 

 ditional research will shift yield to new growth paths from the 

 average rate of growth of 2 percent per year in wheat to 3.6 per- 

 cent (see figure 4.3 where t* marks the time in which research 

 findings are put into practice), and in maize from 3 to 5.9 percent. 

 In absolute values, the marginal contribution of a publication de- 

 pends, therefore, on the rate of change of yield and on the time 

 elapsed from the year in which the knowledge was created. In ta- 

 ble 4.4 the values of the marginal contribution are $1,581 in 

 wheat and $2,330 in maize in the first year (the mean year of the 

 sample, since these values are calculated for average yields) and 

 $20,287 and $30,822 ten years later,"^ respectively. 



The economic contribution of a scientific publication is com- 

 posed, according to the model in the background of equation 4.1 3, 



I 



10. These figures, the second pair, are biased upward to some extent since, as 

 the area unit is 1,000 hectares, they measure the average contribution of a paper 

 when a group of 1 paper per 1 ,000 hectares is added to the stock of knowledge in a 

 country. 



