Revenue Uncertainty and Enterprise Diversification in the Semi-Arid Tropics of India
Satheesh Aradhyula*, Uttam Kumar Deb@ and M.C.S. Bantilan@
*University of Arizona, @2 International Crops Research Institute for the Semi-Arid Tropics (ICRISAT)

  Semi-Arid Tropics(SAT)  
  • Characterized by scanty and highly uncertain rainfall.
    - Crop yields fluctuate a lot
    (more than in non-SAT areas).

  • SATs are home to about one sixth of the world’s population.

 

  Diversification  

  • Few degree of both yield and price uncertainty contribute to considerable revenue uncertainty.

  • Diversification is an important device for the SAT farmers to cope with revenue uncertainty.

 

  How is diversification  
measured?
Diversification can be measure in several different ways. In this study,we use these two measures:

a) number of crops grown by the farmer.

b) number of improved pearl millet     cultivars adopted by the farmer.

 

  Objectives of the  

study:
The study attempts to answer the following questions:
  • What is the extent of diversification in the SATs?

  • What factors influence on-farm enterprise diversification?

  • What factors influence number of improved cultivars grown by a farmer?

  Methodology  

  • Poisson regression models are used to account for the integer nature of the dependant variable.


      Data      

  • Farm Level Survey Data
  • Collected in 1996
  • Data from 7 districts (Ajmer, Alwar, Bharatpur, Churu, Nagaur, Sikar) in Rajasthan state
  • 331 farmers surveyed
   - Major crops in the sample
    area: Pearl Millet
  - Land quality is fairly     homogenous in the sample



  Sample Summary  


Variable Range Median
Farm Size(hectares) 0.8-200 6.0
Education (years of schooling) 0-17 5
Distnace from seed center(kms) 0-26 10
% land irrigated 0-100 0
Dependent variables:
# of crops cultivated 1-10 4
# pearlmillet HYVs adopted 0-3 1
Variable Value
% of farms with 1 crop (no diversification)   1
% of farms with no irrigation   96
% of farms with no HYVs   34
% of farms bigger than 50 hectares   2

    Results    

Maximum likelihoods estimates for the Poisson regression model
Explanatory Variable Estimated Coefficient t-ratio
Dependent Variable: # of crops cultivated
Intercept 1.5869 25.93*
Farm Size(hectares) 0.0037 2.63*
Distance from seed center(kms) -0.0093 -2.33*
%land irrigated 0.1410 1.13
Non-farm income dummy -0.0445 -0.84
Education(years of schooling) 0.0013 0.24

* Significant at 5% level

Ho: All slope parameters are zero
    - Estimated LR statistic: 13.44
      - Conclusion: Reject the Null at 2% level         of significance

Maximum likelihood estimates for the Poisson regression model
Explanatory Variable Estimated Coefficient t-ratio
Dependent Variable: # Pearlmillet HYVs adopted
Intercept -0.7018 -4.21*
Farm Size(hectares) 0.0072 3.03*
Distance from seed center(kms) -0.0069 -1.18*
%land irrigated -0.3578 1.29
Non-farm income dummy -0.0970 1.66@
Education(years of schooling) 0.3958 3.66*
Dummy for zone A 0.3958 3.66

* Significant at 5% level

Ho: All slope parameters are zero
    - Estimated LR statistic: 15.15
      - Conclusion: Reject the Null at 2% level         of significance

  Conclusion  

  • Large farms diversify more than small farms
  • Farms with easier access to seed centers diversify more. To encourage diversification, policy makers need to invest in seed delivery infrastructure.
  • Education is an important determinant of the adoption of improved cultivars
Presented at the American Agricultural Economics Association (AAEA) Annual Conference held at Tampa, USA on July 31- August 2, 2000