32 A^riculiural Research and Prodiniivily 



And we will not be "testing" explicit hypotheses regarding this 

 investment behavior. 



The approach taken is to specify a simple general investment 

 relationship and utilize data at the commodity level to identify the 

 major features of investment behavior.*' Publications on crop /' in 

 country J are related to crop-level and country-level variables. 

 The equation estimated is 



Pii=f<^r^''r'^r"r''j-P>^i^''' 0.2) 



where f( ) is again estimated in the double-log form. The newly 

 defined variables are: 



q.. Value of product / (/ = 1,. . . ,1 1) in 

 country,/ 



Ex.. Share of export in value of product / 



H. Share of farm labor in total labor force 



w. Number of agricultural workers 

 (males) 



j3. A "product effect," a dummy variable 

 measuring ease of publication or other 

 biases in publishing. 



Note that only the eleven commodities are included in the 

 analysis; soil science and phytopathology are omitted. 



The regressions presented in table 2.8 indicate that larger prod- 

 uct values entail more publications in their respective areas, but 

 that the elasticity is only of the order of magnitude of .4. The posi- 

 tive coefficient of the export variable indicates that countries 

 direct comparatively more research to export crops. In some in- 

 stances, this is the result of the structure of the research institu- 

 tions inherited from colonial times. This may also be econom- 

 ically justified, as the demand for exports is elastic. 



A comparison of the regressions reveals that the decision to 

 support related biological research (as measured by plant 

 physiology publications) tend to dominate the socioeconomic 

 variables in explaining agricultural research investment. Of 



6. See p. 20 and n. 2. 



