Agricultural Economics Research Review 2019, 32 (2), 239-246 DOI: 10.5958/0974-0279.2019.00035.1

Efficiency gains from micro-irrigation: a case of sprinkler irrigation in wheat

Prabhat Kishore ICAR-National Institute of Agricultural Economics and Policy Research, New Delhi 110012, Email: [email protected]

Abstract This paper assesses the efficiency gains from the adoption of sprinkler irrigation using data from a farm survey in the water-scarce, drought-prone region of . Our analysis goes beyond the conventional comparison of physical and economic parameters between micro-irrigation with flood irrigation. We estimate gains in crop yield, water productivity, and technical efficiency, correcting for omitted variable bias. Our findings show that sprinkler irrigation significantly improves yield (21%), water productivity (34%), and technical efficiency (20%); in other words, it saves water (15%) and diesel (8%). These findings are confirmed by our econometric estimates.

Keywords Sprinkler irrigation, yield, water productivity, technical efficiency

JEL classification Q12, Q15, Q25

Water is a critical input in agricultural production. Of in about 15% of blocks, mandals, or talukas. If this the total utilizable water in India (1,123 billion cubic trend continues, it will severely jeopardize physical metre), close to 80% is used for irrigation (Press and economic sustainability of agriculture (Gandhi and Information Bureau 2013). But water is also a scarce Bhamoriya 2011). Groundwater development in the resource, and with the intensification of agriculture over future should be linked with water conservation time, the demand for water has increased considerably measures. The Government of India realizes this, and and put severe pressure on groundwater resources. it has started emphasizing the importance of the Until the mid-1980s, surface water was the main source conservation and efficient utilization of water. In 2015, of irrigation; it accounted for 54% of the total irrigated it launched the Pradhan Mantri Krishi Sinchayee Yojna. area (Amarsinghe et al. 2007). The situation changed Pressurized irrigation technologies such as drip and gradually, and groundwater emerged as the main sprinkler systems can help to improve the sustainability source. In 2015–16, over 60% of the irrigated area of water resources and consequently agricultural relied on groundwater; since the mid-1980s, production systems. Several studies show that these groundwater has accounted for almost the entire technologies substantially reduce evaporation, and also increment in the irrigated area. There has been little conveyance and distribution losses, and improve investment in or maintenance of irrigation irrigation efficiency (Sivanappan, Rao, and Dikshit infrastructure, and the area irrigated by public irrigation 1994; Sivanappan 1994; Narayanamoorthy 1996, 1997, systems has almost stagnated (Amarsinghe et al. 2007; 2006; Dhawan 2002; Kumar et al. 2008; Saleth 2009; Shah 2009), leading to the over-exploitation of Narayanamoorthy and Deshpande 2005). These groundwater and a continual decline in the water table. technologies also lead to an improvement in the Several other factors, including subsidies for borewells efficiency of fertilizer and energy use (Kumar and and electricity, have contributed to the degradation of Palanisami 2010; Chandrakanth et al. 2013). Most of groundwater resources. Groundwater is over-exploited these studies prove the benefits of pressurized irrigation 240 Kishore P

Table 1 Key characteristics of Bundelkhand region

Particulars Bundelkhand Uttar Pradesh

Geographical area (million hectares) 2.962 24.170 Net sown area (million hectares) 2.058 16.598 Gross cropped area (million hectares) 2.924 26.147 Gross irrigated area (million hectares) 1.464 20.965 Surface irrigation (%) 44.49 19.32 Groundwater irrigation (%) 55.51 80.68 Cropping intensity (%) 142.07 157.53 Average rainfall (mm) in past 15 years 707.09 717.47 Average groundwater table (m) in past 15 years 9.13 7.96 technologies by comparing their important parameters and rapeseed-mustard (3% of the cropped area). against those of flood irrigation, but such a comparison can lead to biased estimates, as the adoption of Sampling pressurized irrigation technologies involves a selection The data that we use in this paper come from a farm process that is influenced by several observable and survey conducted in 2017–18 primarily to assess the unobservable factors. This paper uses data from a field impact of the Direct Benefit Transfer scheme for micro- survey of over 400 farm households to assess the irrigation launched throughout the state in 2014. The benefits of sprinkler irrigation considering selection Bundelkhand region comprises seven districts: , bias. Jalaun, Lalitpur, Hamirpur, Mahoba, Banda, and Chitrakut. For our survey we selected Jhansi and Data and descriptive statistics Mahoba because farmer registration there is higher for the Direct Benefit Transfer scheme for sprinklers. From Study area each district we selected three developmental blocks This study was conducted in the Bundelkhand region randomly; from Jhansi we selected Bamaur, Chirgaon, of the state of Uttar Pradesh. The region occupies about and and from Mahoba we selected Kabrai, 10% of the state’s geographical area and supports Jaitpur, and Panwari. For the next stage, six villages nearly 5% of the population. Of the total geographical from each block in and five villages from area, 69% is under cultivation. Agriculture is rain- each block in were randomly selected. dependent and rainfall is low (707 mm per annum) The key characteristics of the selected villages are given and erratic. Irrigation is limited to 50% of the gross in Table A1 in the appendix. cropped area, with groundwater accounting for 44% Most villages have basic amenities such as roads, (Table 1). electricity, and schools. Banking facilities are available The average depth of the groundwater table in the only in 27% of the villages, and about 33% of the Bundelkhand region is 9.13 metre, more than the state villages have some form of social organization average. The region is prone to frequent droughts that (cooperative society, gram panchayat, farmer producer affect agriculture and agriculture-based livelihoods. organization, etc.). Agricultural extension support is The cropping intensity is 142%, 16 percentage points available in about 67% of the villages. Finally, a sample less than the state average, and the cropping pattern is of 14–15 farm households was drawn randomly from dominated by pulses (black gram and green gram) and each village, thus yielding a sample of 480 farm oilseeds (sesame) in the rainy or kharif season. During households. All these households rely on groundwater this season, pulses occupy 19% of the cropped area for irrigation. and oilseeds 16% of the cropped area. In the post-rainy Our focus is on wheat, the main rabi crop in the region. or rabi season, the important crops are wheat (43% of Rainfall in this season is extremely low, only 11 % of the cropped area), chickpea (20% of the cropped area) the annual rainfall. In our sample, 84% of the Efficiency gains from micro-irrigation 241 households grew wheat. Groundwater is the main that of non-adopters. To consider this, we adopt source of irrigation for these households; only about treatment effects models from the programme 25% of them have adopted the sprinkler method of evaluation literature. In a regression framework, the irrigation and the rest follow the traditional flood treatment effects model is given by method. Ri = a + bCi + c′Xi + εi …(1)

Descriptive statistics where, Ri is an outcome variable (yield, water productivity, technical efficiency) for farmer i, C is a To assess the efficiency gains from the application of i dummy variable taking value 1 for a farmer who has sprinkler irrigation, we compare the key production adopted sprinkler method of irrigation and otherwise parameters of its adopters with those following the 0. X is a vector of control variables, and is a zero flood method of irrigation (Table 2). The yield of wheat i εi mean random variable. under the sprinkler system is 21.9% more than the traditional method of irrigation, and the difference is An ordinary least squares (OLS) estimate of Equation statistically significant. This translates into significant 1 is likely to be biased if εi contains within it random income gains, gross and net. The level of the water unobservable factors such as ability that are not table on the farms of sprinkler adopters and non- uniformly distributed within the population of adopter adopters is similar, but adopters experience and non-adopters. In such a case, the error term is likely significantly better water productivity and technical to be correlated with Ci. Thus, for instance, if adopters efficiency.1 The differences in efficiency gains could are more productive than non-adopters because of be due to differences in input use or unobservable farm unobserved ability, then a simple comparison of the and farmer characteristics of adopters and non- means as well the OLS estimates of Equation 1 would adopters. yield an overestimate of the true measure of gains from adoption. Hence, we apply the two-stage Heckman Table 2 compares the means of some important procedure to correct for the bias from the endogeneity observables. There is no significant difference between of right hand side variables. Consider the following adopters and non-adopters in the use of critical inputs adoption equation: such as fertilizers, seeds, or machines. Sprinkler irrigation adopters use significantly less human labour, Ci = αi Zi + ui …(2) irrigation water, and diesel (to pump groundwater), and Where, Ci is a binary variable (1 for adopters and 0 for they save around 15% of irrigation water, 8% of diesel, non-adopters), Zi is a vector of variables that matters and 11% of labour. Adopters of sprinkler irrigation have for adoption, variables in Zi will overlap with variables larger landholdings, and a larger proportion have their in Xi. Identification requires that there be at least one own source of irrigation, i.e., tubewells. They are also variable in Zi that is not in Xi. If this condition is met, more educated, and most own a smartphone. the predicted value from (2) c^ can be used as instrument

for Ci in regression equation (1). Thus, from Equation Empirical strategy 2, we estimate the inverse Mills ratio (IMR) and use it We compared the average yield, water productivity, as an instrument in Equation 1. This would yield a and technical efficiency for sprinkler-using wheat consistent estimate of b provided the instruments are growers with that of those using the flood irrigation uncorrelated with the error term in Equation 1. method. This is useful to demonstrate the efficiency gains from sprinkler irrigation, but a simple comparison Results and discussion of means is a biased measure of gains because sprinkler adoption involves a selection process. Farmers may Factors that influence adoption of sprinkler irrigation self-select to adopt sprinkler irrigation, or they may be Table 3 presents the estimates of the probit model selected as beneficiaries of the Direct Benefit Transfer corresponding to Equation 2. The key factors that scheme; the population of adopters likely differs from influence the adoption of sprinkler irrigation are

1 Water productivity is estimated using the total groundwater extracted for irrigation and yield obtained. Technical efficiency is estimated using the stochastic frontier production function. 242 Kishore P

Table 2 Descriptive statistics

Sprinkler irrigation Flood irrigation Difference (96) (307) (%)

Outputs Yield (quintal per acre) 14.12 11.58 19.77*** (0.14) (0.14) Gross income (INR per acre) 24495.41 20092.01 19.75*** (234.15) (234.84) Net income (INR per acre) 10673.83 5477.68 64.34*** (326.96) (260.60) Water productivity (kg per m3) 0.82 0.61 29.37*** (0.04) (0.02) Technical efficiency (%) 90.21 74.32 19.32*** (0.73) (1.12) Inputs Seed (kg per acre) 75.58 77.05 1.93 (1.27) (0.78) Irrigation (hour per acre) 50.47 53.98 6.72** (1.10) (0.84) Fertilizer (kg per acre) 94.93 94.24 0.73 (1.13) (0.72) Labour (man-day per acre) 11.49 12.80 10.79** (0.56) (0.30) Machine (hours per acre) 5.01 5.03 0.4 (0.10) (0.07) Diesel (litre per acre) 67.39 72.98 7.96** (1.86) (1.41) Groundwater draft (m3 per acre) 2134.86 2487.35 15.25** (112.24) (77.27) Household characteristics Age (year) 48.99 51.36 4.72 (1.31) (0.71) Farming experience (year) 29.75 30.79 3.44 (1.26) (0.70) Education (year) 8.33 6.34 27.13*** (0.49) (0.26) Family size (number) 5.69 5.76 1.22 (0.20) (0.15) Number of workers 3.39 3.38 0.3 (0.14) (0.10) Landholding size (acre) 6.06 4.25 35.11*** (0.44) (0.21) Water table (in feet) 56.96 54.24 4.89 (2.32) (1.42) Engine power (BHP) 7.08 7.21 1.82 (0.12) (0.12) Possess a smart mobile$ (%) 89.58 20.85 68.73** Non-farm income$ (%) 25 29.97 -4.97 Member of social organization$ (%) 16.67 9.77 6.9* Possess a Kisan Credit Card$ (%) 80.21 70.36 9.85* Have tubewell$ (%) 93.75 83.39 10.36*** Note Figures in parentheses are standard errors; *, **, and *** are significant, respectively, at 10%, 5%, and 1%; $Chi-square test Efficiency gains from micro-irrigation 243

Table 3 Probit estimates of determinants of adoption of sprinkler irrigation Dependent variable: Adoption of sprinkler (yes=1; no=0)

Coefficient Marginal effect

Ln landholding size (acre) 0.617*** 0.162*** (0.145) (0.036) Ln farming experience (years) –0.261 –0.068 (0.185) (0.048) Ln schooling (years) 0.074 0.019 (0.049) (0.013) Member of social organization (yes=1, no=0) 0.477** 0.125** (0.233) (0.060) Possess a Kisan Credit Card (yes=1, no=0) 0.134 0.035 (0.194) (0.051) Caste (SC/ST and OBC=1, others =0) 0.185 0.048 (0.266) (0.070) Own tubewell (yes=1, no=0) 0.501* 0.131* (0.288) (0.075) Own electric pump (yes=1, no=0) 0.851*** 0.223*** (0.254) (0.064) Own diesel engine (yes=1, no=0) –0.109 –0.029 (0.208) (0.054) Constant –2.192*** (0.804) Number of observations 403 LR chi2 65.84 Prob>chi2 0.0082 Note Figures in parentheses are standard errors; *, **, and *** are, respectively, significant at 10%, 5%, and 1%. landholding size, ownership of tubewell and electric significant, suggesting that the potential of sprinkler pump, and households’ association with a social irrigation towards improving crop yield. The other organization. The probability of adoption of sprinkler variables that have a positive and significant effect on irrigation is higher for larger farmers and for those who crop yield are irrigation, labour, and machines. own their own tubewells and electric engines for However, yield is negatively associated with seed rate, pumping groundwater for irrigation. Interestingly, a as most farmers apply almost double the rate household’s association with a social organization recommended. increases their chances of adopting sprinkler irrigation, As expected, technical efficiency is positively and perhaps because they learn about its benefits and the significantly associated with sprinkler irrigation. The government schemes. other variables that have a positive and significant influence on technical efficiency are education and the Efficiency gains from sprinkler irrigation availability of extension services. Sprinkler irrigation Having accounted for selection bias, the effects of leads to an improvement in water productivity, that is, sprinkler irrigation on crop yield, technical efficiency more yield with less water. The coefficient of the and water productivity are estimated (Table 4). The adopters of sprinklers is positive and highly significant. IMR is significant only in the case of crop yield. The The use of electricity aids in improving water dummy for sprinkler irrigation is positive and highly productivity. 244 Kishore P

Table 4 Results of outcome equations

Particulars Yield Technical efficiency Water productivity

Ln seed (kg per acre) –0.231*** – – (0.059) Ln fertilizer (kg per acre) 0.006 – – (0.06) Ln labour use (man-day per acre) 0.040** – – (0.019) Ln machine use (hour per acre) 0.175*** – – (0.039) Ln irrigation (hour per acre) 0.152*** – –0.609*** (0.036) (0.089) Sprinkler irrigation=1, otherwise=0 0.207*** 19.252*** 0.238*** (0.024) (2.322) (0.059) Ln farming experience (years) 0.008 0.573 –0.016 (0.02) (2.024) (0.05) Ln family size (no) – –0.1 - (0.379) Ln schooling (years) - 1.116** 0.023* (0.545) (0.014) Caste (SC/ST, OBC=1, otherwise=0) - - –0.204*** (0.069) Ownership of tubewell (yes=1, otherwise=0) - 9.301*** 0.003 (3.102) (0.079) Ownership of electric pump (yes=1, otherwise=0) - 2.623 0.282*** (3.021) (0.077) Ownership of diesel engine (yes=1, otherwise=0) – –2.057 –0.059 (2.363) (0.06) Access to extension support (yes=1, otherwise=0) - 4.910** - (2.024) District dummy (Jhansi=1, otherwise=0) 0.053*** 3.328* 0.288*** (0.021) (2.008) (0.05) Inverse Mills ratio (IMR) –0.034* 1.489 –0.058 (0.02) (2.368) (0.061) Constant 2.419*** 52.351*** 1.942*** (0.408) (8.692) (0.432) Number of observations 403 403 403 Prob > F 0.000 0.000 0.000 R-squared 0.331 0.232 0.296

Note Figures in parentheses are standard errors; *, **, and *** are, respectively, significant at 10%, 5%, and 1%.

Conclusions in the form of massive subsidies to micro-irrigate a larger area will have much better impact in preventing The Bundelkhand zone has become synonymous with the depletion of water resources in the future. Our drought. Water scarcity has led farmers to switch to findings have shown that farmers who followed micro- cultivating pulses and oilseeds in the kharif season. irrigation were significantly more educated and But in the rabi season, the area under wheat has been possessed larger landholdings. Micro-irrigation could picking up in recent times. Government interventions save farmers 15% of groundwater compared to flood Efficiency gains from micro-irrigation 245 irrigation and improve yield by about 21%. This ICRISAT, Patancheru, Hyderabad, India, 2–4 April. improvement in the yield led to an increase in the net hdl.handle.net/10568/38108 income of adopter farmers. Farmers with micro- Narayanamoorthy, A and R S Deshpande. 2005. Where water irrigation have also performed better on efficiency and seeps! towards a new phase in India’s irrigation water productivity in comparison to flood irrigation. reforms. Academic Foundation, New Delhi.

References Narayanamoorthy, A. 1996. Evaluation of drip irrigation system in Maharashtra. Gokhale Institute Mimeograph Amarasinghe, U A, T Shah, H Turral, and B K Anand. 2007. Series No 42, Agro-Economic Research Centre, India’s water future to 2025–2050: business-as-usual Gokhale Institute of Politics and Economics, Pune, scenario and deviations, Research Report 123, Maharashtra. dspace.gipe.ac.in/xmlui/bitstream/handle/ International Water Management Institute, Colombo, 10973/13655/gipem-042.pdf?sequence=2 Sri Lanka. iwmi.cgiar.org/Publications/IWMI_ Narayanamoorthy, A. 1997. Economic viability of drip Research_Reports/PDF/PUB123/RR123.pdf irrigation: an empirical analysis from Maharashtra. Chandrakanth, M G, C N Priyanka, P Mamatha, and K K Indian Journal of Agricultural Economics 52 (4): 728– Patil. 2013. Economic benefits from micro irrigation 39. ageconsearch.umn.edu/record/297570/files/ijae- for dry land crops in Karnataka. Indian Journal of 201.pdf Agricultural Economics 68 (3): 326–38. ageconsearch. Narayanamoorthy, A. 2006. Potential for drip and sprinkler umn.edu/record/206338/files/Chandrakanth68_3.pdf irrigation in India. Gokhale Institute for Politics and Dhawan, B D. 2002. Technological change in Indian Economics, Pune, Maharashtra. pdfs.semanticscholar. irrigated agriculture: a study of water saving methods. org/a9a6/b8f0299b5577e0d0b71bfc688 Commonwealth Publishers, New Delhi. 9f87a757903.pdf Gandhi, V P and V Bhamoriya. 2011. Groundwater irrigation Press Information Bureau, Government of India, Ministry in India: growth, challenges, and risks, in India of Water Resources. 2013. Withdrawal of fresh water. Infrastructure Report 2011, Water: Policy and pib.gov.in/newsite/PrintRelease.aspx?relid=101519 Performance for Sustainable Development, 90–117, Infrastructure Development Finance Company, New Saleth, R M (ed). 2009. Strategic analyses of the National Delhi. idfc.com/pdf/report/IIR-2011.pdf River Linking Project (NRLP) of India: series 3, Promoting irrigation demand management in India: Kumar, D S and K Palanisami. 2010. Impact of drip potentials, problems and prospects. International Water irrigation on farming system: evidence from southern Management Institute, Colombo, Sri Lanka. India. Agricultural Economics Research Review 23 indiaenvironmentportal.org.in/files/NRLP-Proceeding- (July-December): 265–72. core.ac.uk/download/pdf/ 3.pdf 6455751.pdf Shah, T. 2009. Taming the anarchy: groundwater Kumar, M D, H Turral, B Sharma, U Amarasinghe, and O P governance in South Asia. Resources for the Future, Singh. 2008. Water saving and yield enhancing micro- Washington, DC. irrigation technologies in India: when and where can they become best bet technologies? In Managing water Sivanappan, R K, A S Rao, and N K Dikshit. 1994. Drip in the face of growing scarcity, inequity and declining irrigation in India. Indian National Committee on returns: exploring fresh approaches, 1–36, (ed) M Irrigation and Drainage, New Delhi. Dinesh Kumar, International Water Management Sivanappan, R K. 1994. Prospects of micro irrigation in Institute (IWMI), South Asia Sub Regional Office. India. Irrigation and Drainage Systems 8 (1): 49–58. Proceeding of the 7th Annual Partners Meet of the IWMI TATA Water Policy Research Program, Received 16 July 2019 Accepted 21 November 2019 246 Kishore P

Table A1 Village characteristics of study area

Particulars Numbers (%)

Number of villages 33 (100) Road connectivity 32 (96.97) Rail connectivity 3 (9.01) Electricity connection 33 (100) Share of Social class Generals 14.39 OBCs 76.67 SCs/STs 8.93 Education Primary School 24 Middle School 21 Secondary School 5 Senior Secondary School 2 Banking facilities 9 (27.27) Health facilities 14 (42.42) Farm input availability in village 12 (36.37) Average distance of nearest input market (km) 8.54 Agriculture extension support 22 (66.67) Note Figures in parentheses are percentages