Diffusion of MicroIrrigation Technologies in Gujarat, Western India: Do Institutions and Policies Matter? Chandra Sekhar Bahinipati & P. K. Viswanathan Gujarat Institute of Development Research (GIDR) Ahmedabad, India 12th Meeting of the International Water Resource Economics Consortium (IWREC) Washington D.C. September 13, 2016
AgroClimatic Zones of Gujarat Mostly Arid and Semiarid Agroclimatic zones Source: Authors Map
Introduction Gujarat, Western India Mostly Arid and Semiarid region Water scarcity & depletion of groundwater (Kumar 2005; Narula et al 2011) e.g., SGWD in state: 41% (2004), 75% (2009) & 67% (2011) 9 districts: overexploited (SGWD>100%) and critical (SGWD>85%) (GoI 2014) MI: Drip and Sprinkler groundwater extraction and energy use sustainable intensification (Fishman et al 2014) High likelihood of adoption water scarce and groundwater based irrigation (Palanisami et al 2011) In Gujarat: SPV: Gujarat Green Revolution Company Limited (GGRC) 200405 Subsidy Policy: caste, landholdings & geographical location Do Institutional Interventions and Subsidies enhance diffusion of MI technologies in Gujarat?
MI Subsidy Policy in Gujarat MIS Schemes Universal Tribal Taluka GWRDC Dark Zone Taluka SC/ST Marginal & Small Farmer INR 60,000/ ha or 50 % of the capital cost of MI cost (2005) Tribal farmers (43 tribal talukas): 75% or INR 90,000/ ha (2008) Public tubewells: marginal & small 100% cost of MI (2009) 54 dark zone talukas: 60% of the MI cost or INR 60,000/ ha (2012) All SC and ST: 75% cost of MI or INR 90,000/ ha (2015) 60% cost of MI or INR 70,000/ ha. Darkzone: 70% or INR 70,000/ ha (2015) Source: Authors based on various GOs of GoG
200607 200708 200809 200910 201011 201112 201213 201314 201415 Trends in MI adoption in Gujarat (200607 to 201415) 209.9 225.1 200.6 25.7 13.0 49.0 31.9 56.8 62.0 38.9 38.1 103.4 66.6 149.1 90.7 131.0 140.1 123.8 No. of farmers adopted MI (in '000) Total area under MI (in '000 ha) Source: Authors figure based on data collected from GGRC
Diffusion of MI in Darkzone Region Percentage of Farmers and NSA under MI % of Farmers % of total NSA 4.01 0.25 0.24 0.58 0.54 0.49 0.51 1.08 0.98 0.91 0.85 1.52 1.37 2.86 2.44 3.17 2.60 2.11 2006 2007 2008 2009 2010 2011 2012 2013 2014 Source: Authors figure based on data compilation from GGRC Note: the year 2006 represents the financial year, i.e., April 2006 to March 2007, and it is same up to 2013, and the data for 2014 covers between April and December; NSA Net Sown Area
Diffusion of MI in Tribal Talukas No. of farmers (in '000) Linear (No. of farmers (in '000)) Area (in '000 ha) Linear (Area (in '000 ha)) 4.22 4.84 y = 5.4761x 4.0761 R² = 0.7434 6 9.67 1.9 3.08 3.82 6.88 36.91 36 35.87 28.34 24.9 26.9 y = 4.1679x 3.9279 R² = 0.7424 200607 200708 200809 200910 201011 201112 201213 201314 31.02 22.8 Source: Authors figure based on data compilation from GGRC
Incentives and Diffusion of MI: Empirical Approach Additional subsidy sharp discontinuity in between dark and adjacent talukas Following RDD approach, the effects of treatment is estimated as: Y T X u vt 0 1 vt 2 vt vt Y vt Where adoption rate of MI and area under MI in village' v ' at time t T vt X vt treatment indicator (=1 if village access to extra subsidy) u vt captures other covariates and error term i. All the villages ii. Border villages share border with the adjacent talukas iii. Pairwise difference within border villages Unbalanced Panel model
Source: Authors Map Study Talukas, Gujarat
Data and Methods Darkzone Talukas (52) and Adjacent Talukas (58) 110 Talukas Villages & Towns: 8073 4019 in Darkzone and 4054 in Adjacent Border Villages & Towns: 1456 855 in Darkzone and 601 in Adjacent Data: GGRC : no. of farmers adopted MI and area (in ha) 200607 to 2014 Census (2011): no. of households & village area (in ha) Government reports: taluka wise SGWD (overexploited, semicritical, critical & safe categories)
Diffusion of MI in Study Talukas.01.02.03 0 Adoption Rate of MI (2006 to 2014) (Darkzone and Adjacent Talukas) Mean Adoption Rate of Area under MI 2006 2007 2008 2009 2010 2011 2012 2013 2014 Year.025.005.01.015.02 0 Adoption Rate of Area under MI (2006 to 2014) (Darkzone and Adjacent Talukas) 2006 2007 2008 2009 2010 2011 2012 2013 2014 Year 95% CI DarkZone Taluka Adjacent Taluka 95% CI DarkZone Taluka Adjacent Taluka Source: Authors Figure
Descriptive Statistics Full Sample Mean Parameter (SD) Adoption 0.011 Rate of MI (0.031) Area under 0.008 MI (0.021) Extra Subsidy 0.163 (0.370) Overexploited 0.304 (0.460) Critical 0.121 (0.326) SemiCritical 0.179 (0.383) Safe 0.396 (0.489) Darkzone Mean (SD) 0.014 (0.035) 0.011 (0.025) 0.328 (0.470) 0.61 (0.488) 0.124 (0.33) 0.096 (0.294) 0.17 (0.375) Adjacent Mean (SD) 0.008 (0.025) 0.005 (0.015) 0 (0) 0 (0) 0.118 (0.323) 0.261 (0.439) 0.621 (0.485) Border Village Mean (SD) 0.011 (0.030) 0.008 (0.019) 0.192 (0.394) 0.30 (0.459) 0.157 (0.364) 0.184 (0.387) 0.358 (0.48) Difference* Mean (SD) 0.005 (0.060) 0.005 (0.023) Source: Authors computation Note: Overexploited (SGWD> 100%); Critical (SGWD: 85100%), SemiCritical (SGWD: 6585%); Safe (SGWD< 65%); *village pairwise difference
Effect on Adoption Rate of MI Extra Subsidy 0.018*** (0.001) ARMI t 1 ARMI t 2 Region Dummy a Overexploited Source: Computed from primary data; Adoption Rate of MI 0.012*** 0.016*** (0.001) (0.001) 0.055 (0.137) 0.015 (0.070) 0.012*** (0.002) 0.334*** (0.074) 0.108*** (0.019) adoption rate 0.018*** (0.001) 0.096 0.001 (0.091) (0.022) Critical 0.251*** 0.051 (0.088) (0.048) SemiCritical 0.012 0.005 (0.068) (0.040) 2 2 R / Wald 0.124 1835.12*** 0.130 1158.07*** 0.018 AR(2) z statistics 0.766 0.782 (Pr>z) (0.444) (0.434) No. of Obs. 72597 56460 13080 10172 7443 Year FE Yes Yes Yes Yes No Model OLS(FE) AB OLS(FE) AB OLS(RE) Sample Full Full Border Border Border Note: a the omitted category is safe; the figures in the parentheses indicate village level cluster robust standard error in case of OLS and WC robust estimator for AB Model;*** p<0.01, ** p<0.05 and * p<0.1 respectively
Effect on Adoption Rate of Area Adoption Rate of Area under MI Extra Subsidy 0.013*** 0.008*** 0.010*** 0.007*** (0.000) (0.001) (0.001) (0.001) ARMI t 1 0.460*** 0.203*** (0.033) (0.077) ARMI t 2 Region Dummy a Overexploited Source: Computed from primary data; 0.137*** (0.026) 0.027 (0.046) adoption rate of Area 0.011*** (0.001) 0.003 0.030 (0.008) (0.026) Critical 0.044** 0.129*** (0.022) (0.048) SemiCritical 0.039*** 0.016 (0.015) (0.056) 2 2 R / Wald 0.126 5235.30*** 0.120 716.97*** 0.052 AR(2) z 0.493 2.424 statistics (Pr>z) (0.622) (0.015) No. of Obs. 71927 55944 13034 10136 7317 Year FE Yes Yes Yes Yes No Model OLS(FE) AB OLS(FE) AB OLS(RE) Sample Full Full Border Border Border Note: a the omitted category is safe; the figures in the parentheses indicate village level cluster robust standard error in case of OLS and WC robust estimator for AB Model;*** p<0.01, ** p<0.05 and * p<0.1 respectively.
Major findings: Concluding Observations I. Significantly increased adoption of MI technologies in the recent years, especially in darkzone and tribal talukas II. III. Institutional Reforms and Subsidy Policies have been played an important role for observed rapid diffusion Major Determinants: Additional Subsidy and Social Learning latter has higher impact IV. Less adoption in overexploited region: use it or lose it rule Policy Suggestion I. Promote the existing institutions and also provide subsidy for wide scale adoption in the state II. Scaling up this model across the country through recently launched Pradhan Mantri Krishi Sinchayee Yojana
Thank You Email: csbahinipati@gmail.com T: +91(O) 2717242366/ (M) 918128160893 GIDR Working Paper Series No. 231, Gujarat Institute of Development Research (GIDR), Ahmedabad, India