IMPROVING WATER USE IN FARMING: IMPLICATIONS DERIVED FROM FRONTIER FUNCTION STUDIES Boris E. Bravo-Ureta UCONN, USA & UTalca, Chile Roberto Jara-Rojas UTalca, Chile Daniela Martinez UTalca, Chile Susanne M. Scheierling World Bank, D.C. David O. Treguer World Bank, D.C. 12 th Annual Meeting, International Water Resource Economics Consortium (IWREC) The World Bank, Washington, DC September 11-13, 2016 1
Acknowledgements This study was partially funded by the Water Partnership Program (WPP), a multi-donor trust fund at the World Bank. 2
Outline 1. Introduction 2. Frontier Methods 3. Approach and Data 4. Results 5. Concluding Remarks 3
1. Introduction Primary Motivation: WHERE IS THE WATER IN FRONTIER STUDIES??? Scarcity of water resources a growing challenge heightened by climate change. Mitigation and adaption strategies needed. Irrigation a promising adaptation strategy (IPCC 2014). But irrigation already accounts for approx. 70% of all freshwater withdrawals worldwide. Future increase in the demand for water from income and population growth. 4
1. Introduction Growing scarcity puts pressure to improve productivity and efficiency of water use in farming so water can be diverted to other sectors (FAO 2012; World Bank 2013; Scheierling et al. 2014). Productivity & efficiency using frontier methodologies well defined area in economics. Frontiers provide measures of efficiency as a potential input reduction or output expansion, relative to a reference best practice frontier (Coelli et al. 2005). 5
1. Introduction Despite growth in Frontier literature and importance of water research bridging both is limited. Here we focus on 2 issues: (i) Role of water in enhancing farm productivity (ii) Implications from frontier studies on how to improve water use in farming Draw from our recent meta-analyses (Bravo-Ureta et al. 2016; Bravo-Ureta et al. 2007). 6
Outline 1. Introduction 2. Frontier Methods 3. Approach and Data 4. Results 5. Concluding Remarks 7
2. Frontier Methods In a production model, the frontier represents the maximum output producible given inputs, technology and the environment. TE = gap between max and observed output. Is an index between 0% and 100%, interpreted as a measure of managerial performance (Farrell 1957; Martin and Page 1983; Triebs and Kumbhakar 2013). Major typologies: Deterministic vs. Stochastic; Parametric vs. Deterministic; Stochastic Frontier Analysis vs. Data Envelopment A. 8
2. Frontier Methods. The Traditional Parametric (ALS 1977) Model Y (Output) Y i = 0 + Σ X + ΣγZ + v i - u i Stochastic Inefficiency (u) Max. Output Production Frontier Random Error (v) Deterministic Inefficiency Observed Output X (Input) 9
2. Frontier Methods SFA and DEA: key similarities & differences. Both are rigorous tools to measure efficiency relative to a frontier. Two key differences (Fried, Lovell & Schmidt, 2008): 1. SFA naturally stochastic and is parametric: makes it possible to separate noise from inefficiency, provides basis for statistical inference. 2. DEA nonparametric: avoids functional form misspecification; naturally deterministic. Narrowing the gap between SFA and DEA camps is an on-going issue!!!! 10
2. Frontier Methods Several recent innovations in frontier function methods. In SPFs: Panel data approaches (Greene 2005a and 2005b). Recent: Persistent and Transient TE, and unobserved time invariant heterogeneity (Colombi et al. 2014; Filippini and Greene 2014; Kumbhakar et al. 2014; Tsionas and Kumbhakar 2014). Panel data frontiers and TFP analysis. Correction for selectivity bias. Suitable for impact evaluation (Kumbhakar et al. 2009; Lai et al. 2009; Greene 2010; Bravo-Ureta et al. 2012). Endogeneity (Tran and Tsionas 2013;Shee and Stefanou 2014). 11
Outline 1. Introduction 2. Frontier Methods 3. Approach and Data 4. Results 5. Concluding Remarks 12
3. Approach and Data Comprehensive search and review of the frontier function literature; special focus on water studies. 426 studies, 110 with different water features. We focus on 92 farm-level water studies that report TE. Of the 92, 10 analyze Irrigation Efficiency or IE (Kopp 1981; Reinhard et al. 1999). Karagiannis et al. (2003) first to analyze IE. IE: non-radial measure of the amount of irrigation water that could be saved holding output, other inputs and technology constant. IE studies richer compared with non-ie water studies. 13
Input Oriented TE & IE (Karagiannis, Tzouvelekas & Xepapadeas, 2003) 14
Water Studies: 110 classified in 5 groups (A-E) A. Irrigation: 76 studies subdivided into A1, A2 and A3. A1. Quantity 56 studies subdivided into 5 classes: 1) Quantity of Water used = 19; 2) Hours of Irrigation = 5; 3) No. Irrigations, Index, or Irrig. Expenses = 22; 4) Percent of Irrigated Land = 7; and 5) Land Area Irrigated= 3. A2. Dummy: 9 studies dummy Yes/No Irrig. A3. Mixed: 11 papers combines A1 & A2. B. Precipitation: 6 papers, quantity or dummy. C. Both A & B. 13 irrigation and precipitation. D. Distance Functions: 5 articles. E. Aggregate: 10 papers. 15
Outline 1. Introduction 2. Frontier Methods 3. Approach and Data 4. Results 5. Concluding Remarks 16
4. Results The 92 farm water studies yielded 189 MTE observations or cases (MTE = simple average of individual reported TEs). Cases by Region: Asia 117; Africa 40; Western Europe, Australia and New Zealand 20; North America 8; Latin America 3; Eastern Europe 1. Most frequently studied countries: India, followed by China, Pakistan, and Bangladesh. Cases by income classification (WB 2015): LMICs = 64; LICs = 50; UMICS = 38; HICs = 37 17
4. Results Cases by Farming Type: Mixed crops and livestock 87; Rice 63; Wheat 23; Dairy 13; Maize 2; Other Animals 1. By REGION Highest MTE: Western Europe, Australia and New Zealand 81.8%. Lowest MTE: Latin America 55.1%; Africa 65.7%. By Farming Type: dairy is highest 84.0%. Average MTE for all 92 water studies: 73.2%. 18
4. NON-IE STUDIES, OUTPUT RESPONSES 82 Non-IE studies searched for estimates of output responses w.r.t. variable incorporating irrigation water. 33 observations listed and grouped by the water variable used. Overall, irrigation water, regardless of how it is incorporated into the analysis, has a positive and significant effect on farm output. Furthermore, the productivity gap that can be attributed to TE is close to 25-30 % points. 19
MTE & Output Response to Water: 33-Farm Non-IE Studies 20
MTE & Output Response to Water: 33-Farm Non-IE Studies continued 21
4. 10 IE STUDIES 4 use an SFA model and 6 DEA Irrigation water Variable: * 7 volume of irrigation water applied; * 3 used mixed measure (volume of water applied and a dummy variable for irrigation method) Average MTE: 68.9% IE: 46.6% ITCE: 84.4% Y Resp: 0.134 22
MTE, IE, ITCE, Out. Resp., Shadow Value, Market Price: 10 Farm IE Studies 23
4. Implications for Improved Water Use To identify policy implications on water Productivity from 43 studies: all 10 IE studies and the 33 non-ie discussed earlier. Policy implications clustered into three categories: Farming Practices Human Capital of Farmers Policy Actions 24
4. Implications for Improved Water Use FARMING PRACTICES: Choice of crops; chemical input use; land tenure (farmers who rent lower IE compared with owners) Adoption of modern irrigation technologies Improved maintenance of irrigation systems crucial role in water distribution, which helps to enhance efficiency gains Investing in water conservation and using fertigation can increase IE significantly 25
4. Implications for Improved Water Use HUMAN CAPITAL OF FARMERS: Farmer experience (measured by proxies such as a farmer s age) Participation in training programs focusing on agronomic and irrigation techniques Membership in Farmer Associations High share of hired labor to all labor higher TE levels Farmers with off-farm income tend to show lower TE levels 26
4. Implications for Improved Water Use POLICY ACTIONS: interventions beyond the farm gate, typically undertaken by public sector Financial support through credit and/or subsidies to foster adoption of modern irrigation techniques, construction canals and water storage Specialized training programs on the correct use of irrigation technologies and irrigation water Creation and strengthening water user associations and co-operatives Introducing a pricing system for irrigation water Formal water rights 27
Outline 1. Introduction 2. Frontier Methods 3. Approach and Data 4. Results 5. Concluding Remarks 28
5. Concluding Remarks Enhance farmers managerial ability related to irrigation. Reducing inefficient water use increasingly important as water scarcity rises. Free or low cost irrigation water does not help. Water pricing analysis needed (based on shadow prices as a point of departure). Evidence from the frontier literature is very limited. Improved farming practices including soil and water management need to be integrated in the analysis of irrigation productivity. 29
5. Concluding Remarks Dearth of TFP from frontier literature. Recent methods allow for comprehensive analysis that could be useful in informing policy. Rigorous impact evaluations of alternative interventions designed to improve irrigation water use in farming seem to be limited. Frontier methods allow for the separation of productivity gaps stemming from managerial performance and from technological levels. 30
5. Concluding Remarks Farm studies useful but limited. Most studies consider water applied, and assume any reduction would decrease waste and save saving. Basin level water use is interdependent. Return flows can be the source of water for downstream users. This has been ignored in the productivity literature and deserves attention. Irrigation-productivity-farm income research agenda requires interdisciplinary collaboration to define indicators, data needs, suitable methodologies provide robust farm and policy level recommendations. 31
IMPROVING WATER USE IN FARMING: IMPLICATIONS DERIVED FROM FRONTIER FUNCTION STUDIES Boris E. Bravo-Ureta UCONN, USA & UTalca, Chile Roberto Jara-Rojas UTalca, Chile Daniela Martinez UTalca, Chile Susanne M. Scheierling World Bank, D.C. David O. Treguer World Bank, D.C. THANK YOU 32