37260 D I R E C T I O N S I N D E V E LO P M E N T Environment and Sustainable Development Overcoming Drought Adaptation Strategies for Andhra Pradesh, India Overcoming Drought Overcoming Drought Adaptation Strategies for Andhra Pradesh, India ©2006 The International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org E-mail: feedback@worldbank.org All rights reserved 1 2 3 4 5 09 08 07 06 This volume is a product of the staff of the International Bank for Reconstruction and Development / The World Bank. The findings, interpretations, and conclusions expressed in this volume do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgement on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development / The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; telephone: 978-750-8400; fax: 978-750-4470; Internet: www.copyright.com. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2422; e-mail: pubrights@worldbank.org. ISBN-10: 0-8213-6664-5 ISBN-13: 978-0-8213-6664-6 eISBN: 0-8213-6665-3 DOI: 10.1596/978-0-8213-6665-6 Library of Congress Cataloging-in-Publication data has been applied for. The World Bank has formally agreed to follow the recommended standards for paper usage set by the Green Press Initiative, a nonprofit program supporting publishers in using fiber that is not sourced from endangered forests. By printing this book on 30% post-consumer chlorine-free recycled fiber paper, the following were saved: 4 trees, 185 pounds of solid waste, 1,437 gallons of water, 346 pounds of greenhouse gases, and 3 million BTUs of total energy. For more information, visit www.greenpressinitiative.org. Contents Foreword xi Acknowledgments xiii Acronyms and Abbreviations xvii Executive Summary xix Chapter 1 Introduction 1 Drought in Andhra Pradesh 1 Andhra Pradesh's Drought-Related Initiatives 4 Objectives of the Study 6 Broader Context of Adaptation to Climate Variability and Changes 7 Structure of the Report 8 Chapter 2 Drought and the Andhra Pradesh Economy: Historical Perspective 9 Chapter 3 Methodological Framework 15 Selection of Districts and Crops for Analysis 15 Probabilistic Drought Risk Assessment Model 17 v vi Contents Chapter 4 Reducing Vulnerability of Agriculture to Drought in Eight Drought-Prone Districts 31 Crop Yield Variability: Benchmark Case 31 Crop Production Losses 35 Adaptation Strategies at the Farm Level 38 Impact of Climate Change 43 Implications for Agriculture Financing and Risk Insurance 46 Chapter 5 Managing Economic Impact of Drought at the State Level 51 Assessment of Direct and Indirect Loss Potentials: Benchmark Case 51 Simulating the Impact of Structural Changes in the Andhra Pradesh Economy 56 Socioeconomic Strategies to Decrease Vulnerability to Drought Risk: Issues and Options 58 Chapter 6 Conclusions and Recommendations 63 Methodology Development 63 Findings and Observations from the Quantitative Analysis 65 Areas for Future Action 69 Annex 1 Programs Addressing Drought in Andhra Pradesh 73 Annex 2 Definitions of Drought 79 Annex 3 Stochastic Crop Production Loss Model 83 Annex 4 WXGEN in EPIC 87 Annex 5 EPIC: Crop Growth Model 91 Annex 6 Planting Area Model 95 Annex 7 Input-Output Analysis for Andhra Pradesh 101 Annex 8 Structure of the Andhra Pradesh Economy (1980­1 to 2002­3) 113 Glossary 121 Bibliography 127 Index 131 Contents vii Tables S.1 Simulated Rice Yield Losses in Drought Years (Percent Normal Year) xxiii 1.1 Government Programs and Initiatives Addressing Drought in Andhra Pradesh: Illustrative List 5 2.1 Sector-Wise VOP, Constant 1998­9 Prices (Rs. Lakhs) 11 3.1 Drought Events and Standard Precipitation Index Values 19 3.2 Simulated Return Periods (in Years) of Droughts in Drought-Prone Districts 19 3.3 Livestock in Andhra Pradesh 20 3.4 Employment Coefficients and Output Multipliers 1998­9 26 4.1 Rice Yields in Normal Years and Rice Yield Losses in Drought Years 35 4.2 Crop Yield Changes Under Climate Change Scenarios: Average Results for Anantapur and Mahbubnagar 45 5.1 Assistance Provided to Drought- Affected States from National Calamity Contingency Fund 53 5.2 Scenarios on the Structure of the Andhra Pradesh Economy, Percent 57 A1.1 Calamity Relief Fund for Andhra Pradesh, 2000­5 (Rs in Lakhs) 74 A3.1 Drought Categories 84 A7.1 Employment Coefficients and Output Multipliers 109 A7.2 Sector-Wise VOP at Current 1998­9 and 2002­3 at Constant 1998­9 Prices 111 Figures S.1 Probabilistic Drought Risk Assessment Model xxi S.2 Conditional Average Loss in GVA Owing to Drought (Percent), by Sector and Drought Category (Minor, Moderate, Severe) xxv 1.1 Rainfall Levels in the Most Drought-Prone Districts in Andhra Pradesh 3 1.2 Groundwater Assessment for Irrigation in Andhra Pradesh (Million Cubic Meters) 4 2.1 Rainfall and Economic Performance in Andhra Pradesh 10 viii Contents 2.2 2002 Drought Effect on Agriculture 10 2.3 Percentage Changes in Agriculture and LGVA, 1993­4 Constant Prices 12 2.4 Sector-Wise GVA Time Series, 1980­2003, 1993­4 Constant Prices 12 3.1 Seasonal Rainfall Deviation in Eight Selected Districts of Andhra Pradesh 16 3.2 Crop Area in the Eight Selected Districts of Andhra Pradesh 16 3.3 Probabilistic Drought Risk Assessment Model 18 3.4 Estimated and Observed AGVA 28 4.1 Average Normal Yield by Crop (Metric Tons per Hectare) 33 4.2 Impact of Severe Drought on Yield (Percent Decrease With Respect to Normal Yield) 34 4.3 Crop Production Losses Caused by Drought in the Eight Districts: EPC 36 4.4 Average Annual Loss of Value of Production Output, by District 36 4.5 Reducing Cultivable Rice Area in Anantapur: Value of Production LEC 40 4.6 Reducing Cultivable Rice Area in Mahbubnagar: Value of Production LEC 41 5.1 Average Sectoral GVA For Normal Years 52 5.2 Average Annual Loss as Percent of GVA Owing to Droughts 53 5.3 Conditional Average Loss in GVA, by Sector and Drought Category 54 5.4 Economic Losses, in Sectoral GVA, Caused by Droughts, Andhra Pradesh: EPC 55 5.5 Loss in Total GVA Under Different Economic Scenarios, by Drought Category 57 5.6 Losses in Total GVA Caused by Droughts Under Different Economic Scenarios: EPC 58 A2.1 Definitions of Droughts 80 A3.1 Validation of SPI for Anantapur 84 A6.1 Log of Lag Rain Versus Log of Lag Gross Rain-Fed Areas for Anantapur 97 A6.2 Previous Year June Rain Deviations Versus Current Year Gross Rain-Fed Area Deviation for Anantapur 97 Contents ix A6.3 Log Transformed Current Year Rain Versus Gross Rain-Fed Area for Anantapur 98 A6.4 Gross Cropped Area and Monsoon Strike Date Variation Over the Years for Anantapur 98 A6.5 Change in Gross Irrigated Area Versus Change in Annual Rain for Anantapur 99 A6.6 Change in Gross Cropped Area Versus Change in Annual Rain for Anantapur 99 A6.7 Change in Gross Rain-Fed Area Versus Change in Annual Rain for Anantapur 100 A6.8 Structure of the Planting Area Model 100 A8.1 Agriculture GVA, 1993­4 to 2002­3, Estimated and Observed 116 A8.2 Livestock GVA, 1993­4 to 2002­3, Estimated and Observed 117 A8.3 Secondary Sector GVA, 1993­4 to 2002­3, Estimated and Observed 117 A8.4 Tertiary Sector GVA, 1993­4 to 2002­3, Estimated and Observed 117 Boxes 3.1 Risk Metrics 24 4.1 Coping With Drought: Findings From Mahbubnagar 37 4.2 Reducing Cultivable Rice Area: Assumptions 39 4.3 Farm-Level Adaptation Strategies: Expert Recommendations Regarding Cropping Patterns 42 4.4 Climate Change Scenarios 45 4.5 Weather Insurance in India: Advantages and Caveats 47 5.1 Financing Rapid Onset Natural Disaster Losses in India 56 Foreword Climate change is both an environment and a development issue. Any human-induced change in climate conditions will affect not just ecosys- tems and biodiversity, but will also affect agricultural productivity, water quantity and quality, human health, and human settlement patterns. All this has direct and potentially damaging implications for developing countries as it can, ultimately, undermine their aspirations of sustainable development. Nowhere is this more critical than in India, where the poor are the most at risk from the increased variability and volatility in weather pat- terns. One of the key climate-related vulnerabilities of India's economy is its heavy dependence on the monsoons. Monsoon analysis reveals that some part or the other of the Indian subcontinent has been hit by drought almost every two years. The disastrous effect of feeble or failed monsoons has been particular- ly acute in the state of Andhra Pradesh, where more than 70 percent of the people depend on agriculture for their livelihood. The potential for devastation has been tragically brought home to us by the suicides of thousands of farmers in recent drought years. And droughts will only get more frequent and more severe in the future as the climate changes in response to human activities. xi xii Foreword The Government of Andhra Pradesh had requested World Bank assis- tance in helping the state adapt to climate change by integrating issues of climate variability into economic planning. This report is the product of collective research by the Bank and national experts to understand and quantify the drought risk that Andhra Pradesh faces. It seeks to help the state government develop anticipatory strategies for adapting to drought not just at the macro level, but also at the level of village communities and individual farms. Developed after detailed consultations with a range of stakeholders, this report is a ground-breaking attempt to draw up a comprehensive drought management framework. It has been based on a probability model that brings together risk-mitigation and risk-financing techniques, and also considers the long-term impact of climate change. It has also led to a practical application on the ground, with the Government of Andhra Pradesh piloting an innovative Drought Adaptation Initiative with Bank support. Michael Carter Country Director, India World Bank Acknowledgments The authors would like to acknowledge and sincerely thank the Government of Andhra Pradesh, the Confederation of Indian Farmers Association, and various organizations and researchers that guided the direction of this study at different stages and contributed in a substan- tive way toward its successful completion. Special gratitude extends to Dr. Mohan Kanda, former Chief Secretary, Government of Andhra Pradesh; Mr. Tishya Chatterjee, Principal Secretary, Environment, Forests, Science and Technology Department; and Mr. Shashank Goel, Commissioner, and Asutosh Mishra, former Commissioner, Disaster Management Department, for supporting and guiding the study. We would also like to thank for their support and contribution Mr.A. K. Goel, Principal Secretary, Planning Department, Mr.A. K. Parida, former Secretary of Planning, and Dr. S. N. Jadhav, Additional Secretary and Chief Vigilance Officer, Environment, Forests, Science, and Technology Department; and Mr. W. G. Prasanna Kumar, Director, National Green Corps, Andhra Pradesh. We are grateful for the cooperation and help extended to the study by officials from several other Andhra Pradesh Government departments and agencies, including Departments of Rural Development, Agriculture, xiii xiv Acknowledgments Rain-Shadow Areas Development, and Irrigation, as well as officials from the Government of India's Departments of Agriculture and Rural Development.The team would like to specially acknowledge Mr. K. Raju, Principal Secretary, Department of Rural Development; Mr. Manohara Prasad; Secretary, Rain-Shadow Areas Development Department; Mr. Vijay Kumar, Chief Executive Officer, Society for Elimination of Rural Poverty; and Mr. M. C. Pargaien,Additional Secretary, Rain-Shadow Areas Development Department, for taking the recommendations of this study forward and supporting the pilot Andhra Pradesh Drought Adaptation Initiative. We would like to recognize the insights and comments provided by experts from research organizations in Andhra Pradesh, particularly those from the International Crop Research Institute for Semi-Arid Tropics, the Central Research Institute for Dryland Agriculture, Agriculture University, the Administrative Staff College of India, and the National Institute for Rural Development, who participated in several consultation meetings. Finally, we thank all participants of the review meetings held in Hyderabad in October 2004 and February 2005 and a stakeholder con- sultation workshop held in Hyderabad on July 19, 2005, to discuss the draft final report. The study team involved in preparing this report was led by Kseniya Lvovsky, lead environmental economist, the South Asia Environment and Social Development Department of the World Bank, and included the following World Bank staff and consultants: · World Bank: Olivier Mahul, senior insurance specialist, Financial Sector Operations and Policy Department; Ian Noble, senior climate change specialist, Environment Department; and, from the South Asia Environ- ment and Social Development Department, Yuka Makino, natural re- sources management specialist, Richard Damania, senior environment economist, Sameer Akbar, senior environment specialist, Uma Balasub- ramanian, program assistant, and Jack Williams, program assistant. · RMSI: Adityam Krovvidi, general manager and project director for this study; Dr. Satya Priya, general manager; Simon Francis, senior manager; Dr. Murthy Bachu, project manager; Dr. Rakesh Mohindra, project manager; Dr. Annes Hassankunju, project leader; and Arindam Dutta, analyst. · Dr. A. L. Nagar, professor, National Institute of Public Finance and Policy; Mr. M. R. Saluja, professor, India Development Foundation; Acknowledgments xv Dr. S. Madduri, professor, University of Hyderabad; Dr. G. S. Mandal, former Deputy Director General, Indian Meteorological Depart- ment; and Dr. D. Raji Reddy, senior scientist, ANGR Agricultural University. The report greatly benefited from comments and suggestions from peer reviewers Sofia Bettencourt, Eugene Gurenko, and Dina Umali- Deininger, all of the World Bank. Bilal Rahill from the World Bank and Dr. Ian Burton from the Meteorological Services of Canada also provided useful comments. The study was undertaken with support and guidance from Jeffrey S. Racki, acting sector director, South Asia Environment and Social Development Department; Michael Carter, country director for India; Fayez Omar, senior country manager for India; and Rachid Benmessaoud, operations advisor for India. Financial assistance from the Government of Norway through the Trust Fund for Environmentally and Socially Sustainable Development is gratefully acknowledged. The principal authors of this report are Kseniya Lvovsky, Olivier Mahul, Yuka Makino, Ian Noble, Adityam Krovvidi, Simon Francis, and Satya Priya. Acronyms and Abbreviations AAL average annual loss ACFC agricultural consumption of fixed capital AGVA agricultural sector gross value added ANGRAU Acharya NG Ranga Agricultural University ELT event loss table EPC exceedance probability curve EPIC Environment Policy Integrated Climate model GCA gross cultivated area GCM global climate model GDP gross domestic product GFCE Government final consumption expenditure GFCF gross fixed capital formation GIA gross irrigated area GoAP Government of Andhra Pradesh GoI Government of India GrfA gross rain-fed area GSDP gross state domestic product GVA gross value added ha hectare I-O Input-output xvii xviii Acronyms and Abbreviations LCFC livestock consumption of fixed capital LEC loss exceedance curve LGVA livestock sector gross value added MT metric ton PML probable maximum loss RCM regional climate models SCFC secondary sector consumption of fixed capital SGVA secondary sector gross value added SPI Standardized Precipitation Index SUR Seemingly Unrelated Regressions TCFC tertiary sector consumption of fixed capital TGVA tertiary sector gross value added VOI value of inputs VOP value of production WXGEN weather generator Executive Summary Drought sets off a vicious cycle of socioeconomic impacts beginning with crop-yield failure, unemployment, erosion of assets, decrease in income, worsening of living conditions, poor nutrition, and, subsequently, decreased risk absorptive capacity, and thus increasing vulnerability of the poor to another drought and other shocks. The mitigation of the impacts of drought has been a key area of focus of India since the 1950s, as evident through programs such as the Drought Prone Areas Programme, Desert Development Programme, National Watershed Management Programme for Rain-Fed Areas, National Calamity Contingency Fund, and the National Agricultural Crop Insurance Scheme. However, the human and social costs of droughts remain devastating. Andhra Pradesh is one of the states in India that has historically been most severely affected by drought. The lack of monsoons has had a disas- trous effect on the state's sizable agriculture sector and on a large share of the population dependent on agriculture for livelihood. This study focuses on 8 (out of a total of 23) districts in Andhra Pradesh that are particularly vulnerable to drought: Anantapur, Chittoor, Cuddapah, and Kurnool in the Rayalaseema region; Rangareddi, xix xx Executive Summary Mahbubnagar, and Nalgonda in the Teleangana region; and Prakasam in coastal Andhra. Together, these districts are home to about 30 million people and account for about 70 percent of the state's drought-related crop production loss. These districts also include some of the poorest areas and communities in the state. The Government of Andhra Pradesh (GoAP) accords high priority to uplifting these areas, as demonstrated by the creation of a dedicated Department for Rain-Shadow Areas Development. While the GoAP continues to explore possibilities to increase areas under surface water irrigation and/or further develop groundwater resources, there are seri- ous technical and economic constraints to increasing the volume of irri- gation water for the majority of areas within these districts. Thus, there is a wide recognition in Andhra Pradesh of the need to complement these efforts with an adaptation process of a gradual shift to agricultural and economic practices that are more sustainable. Study Objectives The objectives of the study were agreed through extensive consultation with several concerned GoAP departments (Environment, Disaster Management, Planning, Agriculture, Rural Development, Irrigation, and Rain-Shadow Areas Development) and other stakeholders so as to complement the existing state and Government of India (GoI) programs by enhancing the state's capacity to assess the long-term effects of drought and increase resilience at different spatial levels to drought risks. The study intended to (a) develop a robust analytical framework for simulating the long-term impacts of drought at the micro (drought- prone areas) and macro (state) levels, (b) conduct a quantitative prob- abilistic risk assessment of the impacts under different scenarios, and (c) assist the GoAP in the development of a forward-looking and antic- ipatory strategy for adapting to frequent drought events and water deficit conditions. In addition to the macroeconomic and drought management scenar- ios, the development of the modeling framework aimed to account for the possible increase in frequency and severity of droughts that may occur owing to human-induced climate change. In this context, this study is linked to a larger program of work by the World Bank in a new strategic area on adaptation to climate variability and longer-term changes. Executive Summary xxi Methodology The probabilistic drought risk assessment model developed for this study consisted of four modules, as described in figure S.1. Figure S.1. Probabilistic Drought Risk Assessment Model hazard module historical weather simulated weather generator stochastic normal and drought events vulnerability module crops soil management crop yield planting area model model production losses direct loss module commodity prices translation of production losses into monetary losses direct monetary losses indirect loss module macroeconomic data input-output macroeconomic model model indirect monetary losses xxii Executive Summary The model developed for this study is a powerful tool for thorough drought risk assessment (with statistical outputs, such as average annual loss [AAL] and loss exceedance curve [LEC]) and to investi- gate the impact of risk coping strategies and climate scenarios on crop yield and production in each block of the eight drought-prone districts. This model was calibrated using local experience in manage- ment practices and crop phenology in the eight selected districts. Its validation was very successful for the five major crops grown in these districts (that is, rice, maize, jowar (sorghum), sunflower, and groundnut). The results presented in this report are aggregated to the district level, which is the smallest scale for which comprehensive validation data were available. Thus, the validation results do not provide a fair and full illustration of this model's capability to quantify the effect of drought and coping strategies on crop yield and production at the block level. One of the particular challenges was in estimating the economic impacts of slow onset events, such as drought. Contrary to rapid onset disasters, droughts normally lack highly visible impacts; instead, their impacts are generally nonstructural and spread over long periods and large areas.Thus, though the approach broadly followed the general catastrophe risk-mod- eling framework used for assessing the impacts of rapid onset disasters (such as cyclones, floods, and earthquakes), it was customized to be appli- cable for slow onset events. Droughts generate significant indirect losses as compared to direct losses in crop production. Indirect losses were estimated through a macro- econometric analysis and an input-output analysis. A critical task was to link drought risk analysis at the block level for the eight districts to the statewide macroeconomic analysis.A prototype macroeconometric model was developed to explain how the variability of the value of crop pro- duction in the eight selected districts affects the variability of the state gross value added (GVA) in the main economic sectors of Andhra Pradesh. The validation of the macroeconometric model was satisfacto- ry, as the estimated agricultural GVA mirrored the observed agricultural GVA over the period of 1993­2002, especially during the drought years. The input-output model, the first ever developed for Andhra Pradesh, was used to provide details of the linkages between the different sectors and subsectors of the economy, the flow of goods and services, and employment. Executive Summary xxiii Key Findings · Despite a variety of antidrought programs, the human and social costs of drought have been and remain devastating for the millions of people in Andhra Pradesh. Under the "business as usual" long- term scenario, the agricultural sector of these districts faces some loss in the value of crop production output for the five major crops combined, compared to a "normal"rainfall year, every second or third year (in other words, in 40 percent of all years). The AAL of output owing to the drought-prone climate is 5 percent for the eight dis- tricts. Individual farmers may suffer greater losses, and for many small and marginal farmers in these districts, a loss of output value of 10 or even 5 percentage points would mean falling below the poverty line. This suggests the need for enhancing an existing strategy by innovative, futuristic approaches and tools to help these people adapt to frequent droughts. The study findings highlight the impor- tance of intensifying efforts to support economic and social develop- ment of drought-prone areas that are sustainable and resilient to water-scarce conditions in the long term. · Impacts of droughts vary greatly across locations and crops depending on drought severity. For example, severe (once in 30 years) drought is likely to decrease rice yields from 29 percent in Nalgonda to 62 percent in Kurnool (table S.1).Yield losses of maize, a rain-fed crop, appear par- ticularly staggering in Anantapur, Kurnool, and Mahbubnagar, which are the driest districts with less than 600 millimeters of rainfall every year. Importantly, different crops can be particularly vulnerable in dif- ferent districts. · In the situation of acute water deficit caused by a major drought, farmers often "rationalize" the use of available water by reducing an area under water-intensive rice in favor of less water-intensive crops. However, this is practiced as a temporary measure with the area of rice typically restored once the drought is over. Modeling results showed losses borne by farmers because of drought can be significantly decreased by adjustments in farming practices that reduce water Table S.1. Simulated RiceYield Losses in DroughtYears (Percent NormalYear) Anantapur Mahbubnagar Kurnool Cuddapah Chittoor Prakasham Rangareddy Nalgonda Minor 14 10 13 11 10 10 19 8 Moderate 27 19 32 21 18 19 24 16 Severe 45 26 62 31 35 33 31 29 Source: Simulations by the model developed under this study. xxiv Executive Summary demand, such as a permanent shift to a larger share of less water- intensive crops in the cropping mix. For example, in Anantapur this strategy can reduce by half the AAL of the overall crop production output during the drought years and increase the all-year average annual crop production output by one-third. · The impacts of measures that can be adopted by farmers are highly location-specific. Reallocation of irrigation water was found to be much less effective in Mahbubnagar, where further change in the cropping mix is apparently needed.This suggests that there is a significant scope for increasing the effectiveness of advice to farmers about undertaking location-specific drought coping measures, such as switching to alter- native crops in response to a poor monsoon. · Global climate change is likely to further increase the benefits of shifting from rice to less water-intensive crops. Modeling of the longer-term impact of human-induced climate change undertaken by the study reinforces the need for shifting to less water-intensive activities in the drought-prone districts. · Against the significant losses caused by drought to farmers and communities in drought-prone areas, the statewide macroeconomic im- pact is rather modest. The long-term AAL in GVA for the state caused by all drought events (averaged over a larger number of drought and nondrought years) is estimated at only 0.2 percent, even in the bench- mark (business as usual) case (figure S.2). During the years of severe drought, the loss in total GVA increases to 1.6 percent. For sectors, the macroeconometric model shows a significant negative impact of drought on the agricultural sector, a much more limited impact on the livestock sector and the secondary sector (manufacturing), and a posi- tive impact (with one year lag) on the tertiary sector (services).The live- stock sector appears less affected by drought than the secondary sector, which is quite dependent on performance of the agriculture sector. · The macroeconomic impact of drought is decreasing further because of the underlying structural changes in the Andhra Pradesh economy. The trend of the Andhra Pradesh economy since the 1980s has been a decrease in contribution by the most vulnerable agriculture sector against increasing contributions from the secondary and tertiary sec- tors. Because this trend is most likely to continue, the macroeconomic impact of drought will further decrease. The future impact of drought on the rural economy can be moderated because of the increasing role of the livestock sector. This is consistent with the analysis of historical data on past droughts, which revealed a decreasing impact trend on Executive Summary xxv Figure S.2. Conditional Average Loss in GVA Owing to Drought (Percent), by Sector and Drought Category (Minor, Moderate, Severe) 8.0 7.0 minor 6.0 (%) moderate 5.0 severe loss 4.0 3.0 GVA 2.0 1.0 sectoral 0.0 ­1.0 agriculture livestock secondary tertiary total ­2.0 Source: Study model simulations. Note: Positive numbers/percentages represent loss while negative numbers mean gains. both the overall economy and the primary sector. The AAL caused by droughts on the Andhra Pradesh economy appears lower than that due to cyclones or floods, as was estimated by another World Bank study, although any comparison should be made with caution because the losses were not measured in the same units. · Accelerating the structural shift in the economy from the agriculture sector to the secondary and particularly tertiary sectors can be inter- preted as a powerful macroeconomic drought adaptation strategy. The impact of such a shift would be an 80 percentage point decrease in total GVA loss attributed to drought events when the shares of the agriculture, secondary, and tertiary sectors roughly approximate the structure of the economy of Brazil, as compared to the impact when maintaining the current structure of the Andhra Pradesh economy.This means that the loss in total GVA can be decreased from 1.6 to 0.2 per- cent under severe drought in the eight districts. · The findings of a relatively small and declining macroeconomic impact of drought are consistent with a growing body of evidence on the macroeconomic impact of climate-related disaster. A recent study, which used worldwide historical data, showed that the maximum an- nual impact of drought is 0.8 percent of GDP for developing countries as a group. · However, droughts continue to have a negative impact on the perform- ance of the agriculture sector and, thus, the lives of the millions who depend on agriculture for livelihood, income, and employment. This highlights the need for strategies that specifically target those most xxvi Executive Summary affected by drought economic indicators: output and employment in the agriculture sector and, particularly in the most vulnerable districts, mandals and communities. · Moderating loss of employment during droughts remains a key chal- lenge. The agricultural sector is the major employment generator for the state, with the total employment loss of more than 44 lakhs for 2002­3 linked to the estimated decrease in the agricultural output caused by drought. The analysis identified several opportunities out- side the agriculture sector that can mitigate the impacts of drought on employment and income in the short to medium term, including (a) trade and transport (except railways), (b) the construction sector and related (cement, bricks, and steel) industry, (c) the mining and quarrying sectors, and (d) the poultry sector. · Responses to drought depend on the situation of a particular house- hold and may range from a change in farming decisions to migration to extreme cases of starvation, loss of health, and even life itself (including cases of suicide). These responses reinforce the findings of the analysis that tailored assistance is required for those in need. Areas for Future Action The analysis revealed stark contrasts through which drought manifests itself at different geographic levels, on different economic indicators, on different crops and sectors, on different population groups, and on different measures of human well-being. The growing gap between the encourag- ing macroeconomic trends and the impacts on farmers and communities in drought-prone areas is particularly disconcerting. Thus, an effective strategy needs to combine statewide economic and sectoral policies with intensified, well-targeted efforts at the micro level: · Continue and accelerate the ongoing changes in the economic structure at the macro level. This can significantly contribute to increasing the resilience of the state economy and/or its people to drought in the long term. These include facilitating growth of the tertiary sec- tor; supporting the development of the livestock sector, particularly poultry; and encouraging a shift in cropping pattern from rice to less water-intensive crops to decrease vulnerability to drought im- pact (including reviewing and addressing unfavorable incentives associated with current agricultural input subsides and rice procure- ment prices). Executive Summary xxvii · Encourage investments in sectors with significant employment potential for the labor displaced from the agriculture sector, such as certain services (trade and transport), construction, mining, and quarrying subsectors, to moderate the impact of drought on affected communities in the short to medium term. · Initiate development and implementation of drought adaptation plans for the most affected areas to deliver better targeted, coordinated, and packaged assistance to those in need. These plans would include (a) measures that promote a gradual shift to more sustainable agricultural practices (for example, changing cropping pattern in favor of less water-intensive crops), (b) other economic activities that are less vulnerable to drought (for example, livestock, agro-industry), (c) water conservation and watershed management activities, and (d) short-term relief and safety net measures to help protect the nutritional health and education attainments of affected communities. This initiative should use a participatory approach and build on the existing successful experiences with the community-based watershed man- agement in Andhra Pradesh, as well as integrate relevant schemes by different departments. · Consider special support programs for marginal farmers, landless and poorest. A particular challenge, as always, is to effectively reach out to those poorest and most vulnerable. The reason is that affluent farmers and households are typically better able to use alternative opportunities, including temporarily changing farming practices or migrating to other sectors, whereas the poorer farmers and landless laborers are least resilient to shocks. · Create a supporting institutional and policy framework with the involvement of all levels of the Government (local, district, and state) to provide extensive technical assistance and other support mecha- nisms to farmers and communities for drought adaptation planning and action. It would need to be supported by adequate institutional arrangements to deliver assistance to communities, strengthened poli- cies and incentives for encouraging diversification of rural economy and water conservation, an aggressive awareness campaign, massive capacity building efforts for all key stakeholders, and innovative finan- cial schemes that mitigate the risks and startup costs of transition to different crops, technologies, and economic activities. · Explore and introduce innovative microfinancing and insurance schemes for farmers that promote a shift to more sustainable practices. The probabilistic drought risk model developed in this study opens new, xxviii Executive Summary and until today virtually nonexistent, growth opportunities for com- mercial agricultural insurance and credit. Furthermore, the analysis highlights the importance of ensuring that risk-sharing arrangements to help farmers finance their losses do not perpetuate the current sit- uation of heavy dependency of farmers on rainfall. Helping farmers in drought-prone areas to maintain a business-as-usual scenario for some time will make the insurance product unviable and the transition harder in the longer term. This consideration is relevant to both the Government crop insurance schemes and innovative insurance products by the private sector, such as rainfall or weather insurance, which were recently offered to farmers in Andhra Pradesh. Rather, new financing products should be able to provide an incentive to per- manently switch to alternative, more sustainable agricultural and eco- nomic practices, such as less water-intensive crops (particularly high value cash crops), livestock, or some agro-processing activities. The study suggests that one particular promising area of innovation in fu- ture risk financing products is the development of contingent financing schemes that could facilitate the transitional "drought adaptation" process. Two possible innovative financing products are proposed by the study: (a) drought adaptation insurance, which could provide coverage against risks of transition from nonviable in the long term but familiar farming business to a viable (agricultural and nona- gricultural) business, and (b) drought adaptation credit, which could provide initial capital to shift to a long-term viable business. · Develop a Decision Support Toolkit to provide a good scientific basis for supporting drought management and adaptation planning at different levels. Among other tools, the innovative modeling framework that emerged from this study, and which expands previous work on catas- trophe modeling to include drought, offers significant opportunities for strengthening drought risk analysis, as well as planning and financing models, that can be used in various states of India and other drought- prone countries. C H A P T E R 1 Introduction Drought sets off a vicious cycle of socioeconomic impacts beginning with crop-yield failure, unemployment, erosion of assets, decrease in income, worsening of socioeconomic conditions, poor nutrition, and, subsequently, decreased risk absorptive capacity, and thus increasing vulnerability of the poor to another drought and other shocks. The mitigation of the impacts of drought has been a key area of focus of the Government of India (GoI) since the 1950s, as evidenced through programs, such as the Desert Development Programme, Drought Prone Areas Programme, National Watershed Management Programme for Rain-Fed Areas, National Calamity Contingency Fund, and National Agricul- tural Crop Insurance Scheme. However, the human and social costs of droughts remain devastating. Following a major drought in the summer of 2002, the worst since 1987, a significant slowdown in India's economic growth occurred in the fiscal year 2002­3. Drought in Andhra Pradesh Andhra Pradesh is the fifth largest state in India, with a population of 76 million,1 of which more than 70 percent lives in rural areas. Agriculture has historically been of key importance to the economy of the state. Irrigated 1 2 Overcoming Drought by three major rivers, Krishna, Godavari, and Pennar,Andhra Pradesh ranks among the top five states in terms of cultivable land and is one of the top producers of rice and fruit. It also leads in the poultry sector. Andhra Pradesh is also one of the three states in India with the largest drought-prone land area.2 It falls under the semiarid region of peninsular India and is broadly divided into three regions: coastal Andhra (compris- ing 9 districts), Teleangana (10 districts), and Rayalaseema (4 districts). During the major drought of 2002, 22 of the total 23 districts of Andhra Pradesh reportedly had less than 75 percent of the normal rainfall during the monsoon season. The strain on water resources has always been acute during low rain- fall years in the state. This has further worsened in the past decades as the demand for water increased sharply owing to growth in agricultural production, population, and the industrial and urban sectors. The over- exploitation of groundwater for irrigation and the gradual decrease in the groundwater levels in certain pockets causing wells to dry up in the dry season is a further cause for worry. The impact is felt most by the farmers, agricultural laborers, and the rural community in dry-land rain- fed areas. Recently, increasing problems in water supply on a larger scale have affected urban centers, too. Since the 1970s, the number of groundwater wells in Andhra Pradesh increased from 8 to 22 lakhs, and the area under groundwater irrigation increased from 10 lakh hectares (ha) to 26 lakh ha. This increased the overall level of groundwater use in the state from 16 to 43 percent. The use of groundwater varies across the state with some areas still having significant unused groundwater potential while in other areas ground- water use has exceeded 100 percent. Most of the groundwater is being used in areas not covered by surface irrigation. Groundwater use in these areas is 56 percent as against 16 percent in areas with surface irrigation, which covers only 5 percent of the state's geographical area. The eight rain-shadow districts of the 23 districts in Andhra Pradesh, with an annual average rainfall well below the state average, are the worst affected by drought.3 Four districts are situated in the Rayalaseema region (Anantapur, Chittoor, Cuddapah, and Kurnool), three in the Telengana region (Rangareddi, Mahbubnagar, and Nalgonda), and one in coastal Andhra region (Prakasam) (figure 1.1). In the eight rain-shadow districts, surface water appears to be fully uti- lized with only a modest scope for further increase. In the medium term, irrigation projects can increase the area under surface irrigation by about 8 percent. Groundwater extraction is also quite high in these districts, Introduction 3 Figure 1.1. Rainfall Levels in the Most Drought-Prone Districts in Andhra Pradesh Source: Ministry of Agriculture, Government of India. compared to the state average amounting to about half of the total groundwater extracted in the state (figure 1.2). A large quantum of unuti- lized groundwater is estimated to be available in surface-water irrigated areas, and the Government of Andhra Pradesh (GoAP) plans to increase areas under the surface irrigation. However, there are serious technical and economic constraints to increasing the volume of irrigation water for these districts or at least for those areas within these districts that have become drought-prone hot spots. There is a wide recognition in the state of the need to start an adaptation process for a gradual shift to less water-intensive agricultural and other economic practices (such as livestock and agro- industry) that are more sustainable in water-deficit areas. These eight districts are home to 35 percent4 of the population of Andhra Pradesh (that is, about 30 million people). A large proportion is involved in agriculture (the economic sector most vulnerable to rainfall variability occurring in these districts) compared to the other 15 districts (31 percent versus 27 percent). The eight districts account for 43 per- cent of the cultivators and 36 percent of the agricultural laborers of the 4 Overcoming Drought Figure 1.2. Groundwater Assessment for Irrigation in Andhra Pradesh (Million Cubic Meters) 20,000 balance potential current draft 15,000 10,000 5,000 0 8 districts others Source: Prasad 2002. entire Andhra Pradesh population. While variations in income are signif- icant, the average per capita income for these eight districts is below the state average (90 percent), and particularly low in Mahbubnagar (75 percent). Andhra Pradesh's Drought-Related Initiatives A large number of drought-related initiatives are currently in place in Andhra Pradesh with support from the GoI, GoAP, and several donors (table 1.1 and annex 1). Examples of major programs include irrigation schemes by the Irrigation Department, the Calamity Relief Fund by the Revenue Department (Revenue [Relief] Department 1981 and 1995) and the Rural Guaranteed Employment Scheme by the Rural Development Department. Other important initiatives include a pro- gram to provide alternative crop seeds by the Department of Agriculture and groundwater monitoring undertaken by the Groundwater Department (Groundwater Department 2003). In 2004, the GoAP created a new Department for Rain-Shadow Areas Development to support the eco- nomic and social development in the most drought-affected communities. The vast majority of these areas are in the eight districts covered by the study. There is considerable experience in drought management at the com- munity level through watersheds programs, sponsored by GoI and/or GoAP, such as the Drought Prone Area Programme, Hariyali Watershed Development Programme, Indira Prabha, Neeru Meeru, and Joint Forest Management/Community Forestry Programme. In April 2002, the Water, Land and Tree Act, promoting water conservation and tree cover, was Introduction 5 Table 1.1. Government Programs and Initiatives Addressing Drought in Andhra Pradesh: Illustrative List Type of programs Name of programs Risk financing ·Calamity Relief Fund ·National Calamity Contingency Fund ·Crop insurance Drought proofing ·Irrigation schemes ·Drought Prone Areas Programme ·Joint Forest Management/Community Forest Management ·Water harvesting schemes ·Microirrigation projects ·Statewide irrigation development ·Andhra Pradesh Rural Livelihood Project ·Watershed Development Programme ·Integrated Wastelands Development Programme ·Rural Infrastructure Development Programme ·Jawahar Gram Samridhi Yojna ·Sampoorna Grameen Rozgar Yojana Employment generation ·Self-employment programs on income generation ·Employment Generation Mission ·Women self-help groups ·Food for work programme ·National Rural Guaranteed Employment Scheme ratified by the Andhra Pradesh Legislative Assembly (The Andhra Pradesh Gazette Part IV-B 2002). Drought-related issues have also been addressed, to varying degrees, by donor-funded rural development programs such as the U.K. Department for International Development­funded Rural Livelihood Program and the World Bank­funded Andhra Pradesh Rural District Poverty Initiatives Project and Andhra Pradesh Rural Poverty Reduction Project. The national Hydrology Project financed by the World Bank helped to organize and maintain the database relevant to water resource and drought management. The Department of Planning with the technical help of the Central Research Institute for Dryland Agriculture has also developed, as part of preparing a drought management plan, a real-time decision support sys- tem to forecast and warn the farmers about the likely upcoming drought and suggest actions, such as changes in cropping patterns to mitigate these impacts. Research institutions, including the International Crop Research Institute for Semi-Arid Tropics and the Acharya NG Ranga 6 Overcoming Drought Agricultural University (ANGRAU), have been conducting extensive research on drought-resistant crops, appropriate agricultural strategies for drought-prone regions, and the socioeconomic impact of drought in select rural communities. Objectives of the Study Since both the GoAP and GoI have numerous programs on drought and watershed management, this study was designed to complement these efforts by enhancing the long-term dimension of drought management planning through the assessment of economic implications of drought and the effectiveness of various policy measures to moderate its impacts at the state and drought-prone area levels.The scope and objectives of the study were agreed through extensive consultations with various concerned Government departments in Andhra Pradesh (Environment, Disaster Management, Agriculture, Rural Development, Rain-Shadow Areas Development, and Planning) and other stakeholders. The objectives of this study were to: · Develop a robust analytical framework for simulating the long-term impacts of drought at the micro (drought-prone areas) and macro (state) levels · Conduct a quantitative assessment of the impacts under different scenarios · Assist the GoAP in the development of a future-looking anticipatory strategy for adapting to drought risks and conditions of chronic water deficit In addition to the macroeconomic and drought-management scenarios, the development of the modeling framework aims to account for the pos- sible increase in frequency and severity of drought risks that may occur as the result of human-induced climate change. The study developed and used a probabilistic risk assessment model that can simulate long-term agricultural and economic impacts of droughts under different climate change and risk mitigation scenarios. The specific steps involved in developing this model were: · Analyzing historical data and developing an agro-meteorological model that can determine the impact of droughts on agricultural assets in Andhra Pradesh Introduction 7 · Developing a probabilistic drought risk model to assess the long-term direct impacts of droughts on losses in production outputs including risk metrics, such as probable maximum loss (PML) and average annual loss (AAL) · Developing a macroeconomic model to capture indirect loss on various sectors of the economy, based on the direct loss given by the probabilis- tic drought risk model The model is a powerful tool to undertake a thorough drought-risk assessment (with statistical outputs, such as AAL, conditional average loss by drought category, loss exceedance curve [LEC]) and to investigate the impact of alternative farming practices and climate change scenarios in each block of the eight districts. The results presented in this report are aggregated to the (eight) dis- trict level and thus do not provide a full illustration of this model's capability to quantify the effect of drought and risk-coping strategies on crop yield and production at the block level.As a follow up to the study, the decision support tool can be applied in Andhra Pradesh to analyze the drought impacts and various adaptation options in detail, as per the specific needs of the responsible Government departments. Broader Context of Adaptation to Climate Variability and Changes This study is linked to a larger program of research undertaken jointly by the World Bank and the GoI in a new strategic area of adaptation to climate variability and long-term changes. The importance of these issues is attributed to the magnitude of losses from climate variability, manifested by droughts, heat waves, floods, and cyclones, which has increased in India over the past decades. Furthermore, as global climate changes, the frequency and severity of these events are expected to increase. The linkages between this study and a larger national study, entitled "Addressing Vulnerability to Climate Variability and Climate Change Through an Assessment of Adaptation Issues and Options," are twofold. On the one hand, the modeling methodology, results, and recommenda- tions that are derived from the Andhra Pradesh study inform and feed directly into the national adaptation study. On the other hand, the national study, which has a component in Andhra Pradesh, offers an opportunity to provide further technical assistance to Andhra Pradesh with respect to 8 Overcoming Drought reducing vulnerability of agriculture and rural communities to climate variability and drought. Structure of the Report The report consists of six chapters, starting with this Introduction. Chapter 2 provides a historical overview of the impact of drought on the Andhra Pradesh economy. Chapter 3 describes the methodology for analyzing the long-term impacts of future droughts under different scenarios. Chapter 4 presents the results of the analysis of crop produc- tion losses caused by drought for the eight selected districts. Chapter 5 discusses the results of the analysis of direct and indirect economic losses at the state level. Chapter 6 contains the conclusions and recom- mendations. The report ends with eight technical annexes. Notes 1. Based on 2001 census of India. 2. The top three states with the most drought-prone land area are Rajasthan (21.9 million ha), Karnataka (15.2 million ha), and Andhra Pradesh (12.5 mil- lion ha).The Central Water Commission defines drought as a situation occur- ring when the annual rainfall is less than 75 percent of the normal (defined as 30 years average) in 20 percent of the years examined and less than 30 per- cent of the cultivated area is irrigated. 3. The definitions of droughts are discussed in annex 2. 4. Based on 2001 census of India. C H A P T E R 2 Drought and the Andhra Pradesh Economy: Historical Perspective Drought undoubtedly causes loss of livelihood and human suffering at the individual and community level.Yet, its macroeconomic impact is less apparent. Figure 2.1 demonstrates the relationship between rainfall and the performance of the Andhra Pradesh economy during 1993­2002. Rainfall is represented as percentage deviation from the normal and the economy by two indicators: (a) agriculture gross state domestic product (GSDP) and (b) overall GSDP. In particular, the graph shows that the growth of both agricultural gross domestic product and GSDP slowed down dur- ing the drought years of 1997, 1999, 2001, and 2002. It is important to consider the impacts at a more disaggregated level. In the 2002 drought year there was a decrease in the contribution of agriculture to GSDP (figure 2.2).While agriculture contributed to about 21 percent of GSDP during the 2000­1 normal year, it decreased to about 15 percent in the following severe drought year. In particular, the contribution of water-intensive crops, such as rice, decreased from about 7 percent in 2000­1 to about 4 percent in 2002­3. The economic impact of past droughts in Andhra Pradesh can also be captured through a comparative study of value of production (VOP) 9 10 Overcoming Drought Figure 2.1. Rainfall and Economic Performance in Andhra Pradesh 40 % growth of AGDP 35 % growth of GSDP 30 % dev of rainfall from normal wtho 25 20 15 10 5 viation/gr 0 de eg ­5 ­10 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 ­15 centa ­20 ­25 per ­30 ­35 ­40 Note: AGDP, agricultural gross domestic product; GSDP, gross state domestic product. Figure 2.2. 2002 Drought Effect on Agriculture 2000­2001 2002­2003 2.2 1.91 3.79 6.96 3.35 3.69 rice groundnut sugarcane 0.82 cotton chilies 0.83 3.04 horticulture dairy 2.07 0.81 food products 0.92 0.67 1.37 1.11 2.82 total contribution of 21.02% in gross total contribution of 15.34% in gross state domestic product state domestic product output tables. Sector-wise VOP tables were estimated for normal year 1998­9 and drought year 2002­3, at constant prices for 1998­9 (table 2.1). The effect of 2002­3 droughts is apparent in the changes in VOP of different sectors. The agriculture sector was worst hit by the 2002 drought, and its VOP decreased by almost one-third. The production of rice decreased to such an extent that the state needed to import rice. Similarly, decrease in out- put of other food grains and food crops resulted in imports from other states. Drought and the Andhra Pradesh Economy 11 Table 2.1. Sector-Wise VOP, Constant 1998­9 Prices (Rs. Lakhs) Sectors VOP 1998­9 VOP 2002­3 Percent change Primary sector 4,883,329 5,244,715 7 Agriculture 3,073,906 2,251,601 ­ 27 Rice 1,203,027 741,465 ­ 38 Jowar (sorghum) 35,957 41,443 15 Maize 68,442 73,502 7 Other food grains 159,296 173,853 9 Groundnut 298,189 128,745 ­ 57 Other crops 1,308,995 1,092,593 ­ 17 Livestock 948,749 1,677,690 77 Forestry and logging 167,625 170,715 2 Fishing 351,600 583,779 66 Mining and quarrying 341,449 560,930 64 Secondary sector 8,077,322 9,949,970 23 Construction 1,022,581 1,524,684 49 Tertiary sector 7,426,711 9,844,996 33 Contrary to the economic impact on agriculture, the livestock sector experienced an increase of 77 percentage points in production despite the drought. While this could be due to some Government interventions in the poultry sector, which performed especially well, it highlights a poten- tial for greater resilience to drought in this sector. Figure 2.3 compares the annual changes with respect to the previous year in gross valued added (GVA) of the agricultural sector and the livestock sector during 1994­2003. The GVA is calculated as the difference between the value of output and the value of inputs excluding consumption of fixed capital. Drought years 1997­8 and 2002­3 clearly affected the agriculture sector, with a loss in GVA of higher than 20 percentage points, while these drought events did not significantly affect the livestock sector. These data indicate that a structural change in the primary sector activ- ities, such as diversifying into livestock production, is likely to make the economy (as well as the primary sector itself) less vulnerable to drought. Indeed, during the drought of 2002, the primary sector as a whole experi- enced an increase in VOP of 7 percent despite a decrease in agricultural sector VOP. The structure of the Andhra Pradesh economy and the impact of drought are shown in terms of changes in the GVA in various sectors of the economy and interrelations between them over the period 1980­2003 (figure 2.4). The key drought years are marked for quick identification. Each of these years can be compared with the preceding normal/drought 12 Overcoming Drought Figure 2.3. Percentage Changes in Agriculture and LGVA, 1993­94 Constant Prices 40 agriculture GVA 30 livestock GVA 20 eg 10 centa 0 per ­10 ­20 1994­95 1995­96 1996­97 1997­98 1998­99 1999­00 2000­01 2001­02 2002­03 ­30 Figure 2.4. Sector-Wise GVATime Series, 1980­2003, 1993­94 Constant Prices 10,000,000 9,000,000 primary sector secondary sector tertiary sector 8,000,000 7,000,000 lakhs 6,000,000 Rs. 5,000,000 in 4,000,000 A 3,000,000 GV 2,000,000 1,000,000 0 1980­81 1981­821982­831983­841984­851985­861986­871987­881988­891989­901990­911991­921992­931993­941994­951995­961996­971997­981998­991999­002000­012001­022002­03 : drought years year to assess the impacts on the overall economy and across the three aggregate sectors (primary, secondary, and tertiary). Importantly, while the agriculture GVA was lower for every drought year as compared to the previous year, the two latest drought periods (1999­2000 and 2002­3) did not cause an absolute reduction in the overall state GVA. An apparently increasing resilience of the overall state economy to drought, as observed from figure 2.4, is because the share of agriculture has gradually decreased while that of the secondary sector and, particularly, that of the tertiary sector, has increased significantly.This trend suggests that the impact of drought on the overall economy of the state is decreasing over Drought and the Andhra Pradesh Economy 13 time, as the impact on agriculture is kept under check by other sectors of the economy. To summarize, droughts continue to have a negative impact on the economy of Andhra Pradesh, particularly on the performance of the agri- culture sector and on the lives of the millions of rural poor. However, the impact of drought on the overall economic indicators has lately been decreasing owing to structural changes, such as the growth of the sec- ondary and tertiary sectors. Furthermore, the impact of drought on the rural economy is showing, on average, some signs of moderation due to the increasing role of the less vulnerable livestock sector. C H A P T E R 3 Methodological Framework The methodological framework developed for this study intended to (a) conduct a detailed risk analysis of impacts of drought events on yield and production at the block and district level and (b) assess the direct and indirect economic impacts at the state level. Selection of Districts and Crops for Analysis The major part of the modeling framework focuses on drought risk analysis in the eight most drought-prone districts: Anantapur, Chittoor, Cuddapah, and Kurnool in the Rayalaseema region; Rangareddi, Mahbubnagar, and Nalgonda in the Telengana region; and Prakasam in the coastal Andhra region. Rainfall in these districts is well below the Andhra Pradesh average of 938 millimeters (with the southwest mon- soon in June­September months contributing 66 percent and northeast monsoon in October­December months contributing 24 percent), and is particularly low in Anantapur and Mahbubnagar (figure 3.1). Four dryland crops (jowar [sorghum], maize, groundnut, and sunflower) and one water-intensive crop (rice) are mainly affected owing to drought in these eight districts. These crops together account for the largest land area used for agriculture in most of the eight districts (figure 3.2) and 15 16 Overcoming Drought Figure 3.1. Seasonal Rainfall Deviation in Eight Selected Districts of Andhra Pradesh seasonal normal rainfall 900 kharif rabi 800 700 600 (mm) 500 400 rainfall300 200 100 0 Chittoor akasam Kurnool ubnagar Pr Nalgonda Cuddapah Rangareddi Ananthapur Mahb Figure 3.2. Crop Area in the Eight Selected Districts of Andhra Pradesh 1,200 other crops sunflower 1,000 groundnut maize 800 jowar ha) rice 600 ('1000 area 400 200 0 ubnagar Kurnool akasam Chittor Anantapur Pr Nalgonda Cuddapah Rangareddi Mahb district name account for more than 80 percent of the all-crop production variability in these districts. The analysis of drought events in the eight districts can be extended to assess the economic impact of these events at the state level. The eight districts contributed to about 70 percent of the decrease in agricultural production at the state level during the past eight drought events (1980­1, Methodological Framework 17 1984­5, 1985­6, 1986­7, 1992­3, 1994­5, 1999­2000, 2002­3). The eight districts account for 88 percent of the total variability in crop losses at the state level. Probabilistic Drought Risk Assessment Model A probabilistic drought risk assessment model was developed to estimate the economic impact of droughts and to assess the effects of different drought mitigation strategies and climate change scenarios. Such models are well established to deal with rapid onset disasters (for example, earth- quakes, cyclones, and floods).1 The economic impact of drought is more complex than that of rapid onset disasters because the impact of rainfall shortage on agricultural assets (for example, crops) is a complex hydrologic and agronomic phenomenon, and drought normally lacks the highly visible direct impacts associated with rapid onset disasters, making indirect economic losses difficult to quantify. Because slow onset disasters, such as drought, have different characteristics and are more difficult to quantify than rapid onset events, it required an innovative risk assessment model using a different risk management paradigm than the one applied for rapid onset disasters.To the best of our knowledge, this is the first time ever that latest catastrophe modeling techniques have been used to address the impact of drought. The drought risk assessment model comprises four main modules (figure 3.3): hazard, vulnerability, direct loss, and indirect loss. Hazard module In the hazard module, daily weather data (precipitations, air temperature, solar radiation, and wind speed) are simulated over a period of 500 years, based on historical data at a location. Normal and drought events (for example, minor, moderate, severe drought) are captured from this time series and their frequencies calculated. The model is capable of simulat- ing different climate change scenarios. The hazard module defines the frequency and severity of a drought event at a specific location. This is done by analyzing historical data on the severity and frequencies of drought in Andhra Pradesh. The first step is to define precisely what a drought event is. Several def- initions of drought have been proposed in the literature (see annex 2). The Standardized Precipitation Index (SPI) based on the precipitation deficit over a specified period of time was selected for this study (McKee et al. 1993; and annex 3).The SPI quantifies the impact of drought on the 18 Overcoming Drought Figure 3.3. Probabilistic Drought Risk Assessment Model historical hazard module weather simulated weather generator stochastic normal and drought events crops vulnerability module soil manage- crop yield planting area ment model model production losses commodity direct loss module prices translation of production losses into monetary losses direct monetary losses macroeconomic indirect loss module data input-output macroeconomic model model indirect monetary losses availability of different water resources. Soil moisture conditions respond to precipitation anomalies on a relative short time scale (days), while groundwater, stream flow, and reservoir storage reflect the long-term pre- cipitation anomalies. In this study, the intensity of drought events is defined with respect to SPI values as shown in table 3.1. Each drought event, therefore, has a duration defined by its beginning and end and intensity for each month during which it occurs. SPI was computed at both district and block levels. The seasonal rainfall at block level was aggregated to the district level to compute the district level index and to the eight districts study areas.This allowed simulation of crop yields at different stages of drought through the vulnerability module. Methodological Framework 19 Table 3.1. Drought Events and Standard Precipitation Index Values SPI values Event ­0.5 to 0.5 Normal ­1.0 to ­0.5 Minor drought ­2.0 to ­1.0 Moderate drought ­3.0 to ­2.0 Severe drought Lower than ­3.0 Extreme drought The simulation in this study was based on 30-year data. The frequency of drought over periods much longer than the period of observation was calculated by using a stochastic weather generator (WXGEN), which is embedded in the agro-meteorological model (see annex 4). The WXGEN was first parameterized based on historical data for the study region: daily rainfall data at the block level and other meteorological data at the district level. Daily weather data were then simulated for 500 years to generate the long-term drought frequencies (see annex 3). Table 3.2 shows the simulated frequency of droughts, by category as defined in table 3.1 in each of the eight drought-prone districts of Andhra Pradesh, as well as the entire study region. The results were validated by comparing the historical and estimated exceedance proba- bility curves (EPC) that have shown a good match (see annex 3). Review of rainfall aggregations (from block to district to the region of eight dis- tricts) showed that the drought category (minor, moderate, and so on) at the block level does not necessarily translate to the same category at the district and regional level. Similarly, return period of a drought category for a block need not be the same for the district as well as the region. Table 3.2. Simulated Return Periods (in Years) of Droughts in Drought-Prone Districts District Minor Moderate Severe Extreme Any Anantapur 6.1 7.8 41.7 -- 3.2 Prakasam 6.8 8.9 29.4 -- 3.4 Rangareddy 7.5 7.7 35.7 500.0 3.4 Nalgonda 7.4 6.8 41.7 -- 3.3 Chittoor 6.5 9.6 38.5 500.0 3.5 Cuddapah 6.3 9.1 35.7 250.0 3.3 Kurnool 6.8 7.9 38.5 500.0 3.3 Mahabubnagar 6.8 7.5 41.7 500.0 3.3 Eight districts 6.8 8.2 38.5 -- 3.3 Source: Model simulations based on historical data. Note: --, not available. 20 Overcoming Drought Vulnerability module The vulnerability of the agricultural assets (for example, crops) at risk to simulated weather events is estimated from a crop yield model (annex 5) and a planting area model (annex 6). The crop yield model simulates crop yields for different drought events. Farmers' crop plant- ing decisions are estimated through the planting area model. Production is the product of crop yield and cropped area, while production loss for a given drought event and crop is calculated as the difference between production during a normal year and production in a given drought year. This module quantifies the damage caused to each asset by the inten- sity and duration of a given drought at a site. Drought mainly affects flow items, such as crops, while rapid onset disasters cause main losses among stock items. Asset classification for drought is based on a combination of crops and sensitivity to water. In addition to the five selected crops, livestock was also considered, because drought directly affects the productivity of livestock by affecting the availability of drinking water, fodder, and so on. Area, yield, and pro- duction data were available from the Directorate of Economics and Statistics of Andhra Pradesh for the past 3 years at block level and for the past 10 years at the district level. However, these data are available only on an annual basis. Livestock data were obtained from the four-yearly livestock census of Andhra Pradesh of 1993 and 1999 at district level.An analysis of the live- stock census of Andhra Pradesh since 1951 did not show the affects of droughts conclusively (see table 3.3). The lack of annual livestock data precluded any direct quantitative assessment of the impact of drought on Table 3.3. Livestock in Andhra Pradesh Census year Total livestock (in thousands) Percent change 1951 34,287 -- 1956 29,513 ­14 1961 32,643 11 1966 31,594 ­3 1972 33,064 5 1977 31,472 ­5 1983 35,756 14 1987 33,667 ­6 1993 32,911 ­2 1999 36,010 9 Note: --, not available. Methodological Framework 21 livestock. It was captured indirectly through the macro model developed in the economic module. Crop yield model In this study, damage was measured as the loss in yield of the selected crops. Loss in yield could be estimated from simple statistical relation- ships between yield and drought/nondrought categories. However, since the study aimed to analyze a wide range of response options and eventually the effects of climate change, statistical relationships would not suffice, as they cannot allow estimates of yield changes in circum- stances not yet experienced. Thus, a simulation model of crop growth and, if possible, water availability and livestock production was sought. A number of models were considered depending on whether they were well established, tested in practice, likely to be maintained over the next 5­10 years, and suitable for application in drought-prone agricultur- al systems such as those in Andhra Pradesh. These included the Decision Support System for Agro-Technology Transfer suite of models maintained by the International Consortium for Agricultural Systems Applications and the Environment Policy Integrated Climate (EPIC) model developed by scientists from the U.S. Department ofAgricultureAgricultural Research Service, Soil Conservation Service, and Economic Research Service. The EPIC (Sharpley and Williams 1990; Izaurralde et al. 2003) model was selected because it provided a more coherent modeling environment and there was relevant experience with its application in India (Priya and Shibasaki 1998a, b). The EPIC model was originally designed to assess the effect of soil ero- sion on productivity; however, it helps answer a larger number of research questions. The model simulates the effects of management decisions on soil, water, nutrient, and pesticide movements and their combined impact on soil loss, water quality, and crop yields for areas with homogeneous soils and management. Some of the important components of EPIC are WXGEN; hydrology, erosion and sedimentation, and nutrient cycling; crop growth; tillage; economics; and plant environment control. The five crops--rice,groundnut,sunflower,maize,and jowar (sorghum)-- selected for the analysis had already been included in EPIC, but needed to be modified to reflect conditions in Andhra Pradesh. About 47 param- eters relating to crop phenology, its environment, and crop growth in a stressed environment are used in EPIC. Some parameters that are used to estimate outputs (nutrient levels at various times in the growing season, and so on) were not used in this study. Parameter values for the selected 22 Overcoming Drought crops and the management practices associated with them were based on previous modeling exercises with EPIC and on advice from experts at ANGRAU, Hyderabad. Annex 5 presents detailed technical information on the EPIC model and its application in this study. An important decision during model development was the level of hydrological modeling required. EPIC calculates soil moisture based on rainfall and irrigation data. Rainfall data were available at the block level. The availability of irrigation water depends on both local rainfall, which recharges surface dams and shallow wells, and water reaching a block through rivers, canals, and pipes. River flows and reservoir storages do not depend on local rainfall but on the catchments that were outside of the study area. An analysis of available data suggested that a detailed hydro- logical model was not feasible at either the block or larger scale. Irrigated and rain-fed areas were computed in the planting area model by crop, by season, and by block. EPIC was administered for two scenarios--irrigation and rain-fed--for each crop, for each block, and then superimposed on the respective areas to calculate production.This approach eliminated the need for a hydrological model. Planting area model Farmers take planting decisions at the beginning of the season based on eco- nomic parameters (for example, expected commodity price at harvest) and agro-meteorological parameters (for example, onset of monsoon, expected rainfall levels). Production flexibility is integral to the practice of dry-land farming (Jodha 1981). When crop failure is foreseen, farmers change their cropping patterns to focus their efforts on crops that have a greater chance in adverse weather conditions, as seen in semiarid tropics of India (Walker and Ryan 1990). Farmers' plans for the rainy season are contingent on rainfall, and thus the relative importance of rainy and post-rainy crops fluctuates seasonally. As a result, the area of Kharif season crops is variable. This variability is not a source of risk but a proactive response to weather risk. An example of area variability (Walker and Ryan 1990) is the substitution of sorghum by castor that is induced by the late arrival of monsoon in Aurepalle. Late-planted jowar (sorghum) is susceptible to pests and so farmers prefer to plant castor. The response to agro-climatic events is even stronger in Mahbubnagar because of the short period of about two to four days after the onset of monsoon available for planting. Low soil moisture causes farmers to reduce cultivated area, increase inter- cropping, and increase use of short-duration and low water-requiring Methodological Framework 23 crops (Gadgil et al. 1988). Therefore, production flexibility is a key fea- ture of farmers' adjustments to weather variability. A planting area model was built to capture the impact of rainfall vari- ability on planting areas at district level.The development of a behavioral model representing the farmers' planting decision was beyond the scope of this study. Instead, the model was used to estimate through statistical analysis the irrigated and rain-fed cultivated area in a rainfall scenario. Annual data of gross cultivated areas and gross irrigated areas at district levels from 1988­9 to 2002­3 were available. Unfortunately, seasonal data were not available. Several models were tested, and the selected model estimated the percentage change of gross cultivated area and gross irrigated areas with respect to percentage change of the annual cumula- tive rainfall level (annex 6). Crop production losses were then estimated under drought events. Estimated crop production is equal to estimated crop yield multiplied by estimated crop area, at the block level under a given drought event. Crop production losses are defined as the difference between crop production simulated during a normal year and crop production simulated in a drought year for each of the five crops at the block level. Losses are then aggregated to various levels of administration (district, state) as required. Direct loss module The direct loss module converts weight units to value units with respect to commodity prices. Direct monetary losses are calculated and then risk metrics (AAL, exceeding probability loss, PML, and so on) are estimated. The direct impact of drought is monetary loss to farmers caused by decreased production. Production losses were converted to monetary losses taking current market price of each crop into consideration. The direct monetary losses were then aggregated to various levels of adminis- tration. At this stage of modeling, an event loss table (ELT) was created with columns for event number, severity, frequency, and loss. Since large uncertainties are inherent in model estimates of event severity, frequency characteristics, and consequent losses caused by such events, the model is constructed using probabilistic formulations that can incorporate this uncertainty into the risk assessment. Risk metrics produced by the model using the ELT provide the policy maker with essential information necessary to manage their risks in the future (box 3.1). The stochastic crop production loss model is explained in annex 3. 24 Overcoming Drought Box 3.1 Risk Metrics AAL: This is the expected loss per year when averaged over a very long period (for example, 100 years). It is the summation of products of event losses and event probabilities of occurrence for all stochastic events in the loss model. The events are an exhaustive list affecting the block/district under consideration generated by stochastic modeling. LEC: This represents the probability that a loss of any specified (for example, mone- tary) amount would be exceeded in a given year. This is an important catastrophe risk metric since it estimates the amount of funds required to meet risk manage- ment objectives. PML: This is a subset of the LEC that represents the loss amount for a given proba- bility or return period per year. The policy maker may decide to manage for losses up to a certain return period (for example, 1 in 100 years). It is thus the 100-year loss. Indirect loss module This module aims at estimating the indirect economic losses from drought. It provides a consistent methodology that allows capturing of the complex nature of drought impacts, including direct and indirect drought losses. One of the major challenges in assessing the economic impact of drought is that contrary to rapid onset disasters, droughts normally lack highly visible impacts and generate large indirect losses compared to direct losses. Their impacts are generally nonstructural and spread out over large areas. Because of this difference, the economic impact of droughts cannot be captured only through crop production losses. In the indirect loss module, indirect monetary losses are estimated through a macroeconometric model and an input-output model. The macroeconometric model estimates the impact of crop production vari- ability on the variability of the GVA of primary, secondary, and tertiary sectors of the economy. The input-output model gives a detailed picture of the linkages between the different sectors and subsectors of the econo- my (including Government expenditure), the flow of goods and services, as well as employment. In particular, this model can track the impact of a production loss caused by a drought on the other economic sectors and Government expenditure. The indirect loss module links direct monetary Methodological Framework 25 loss estimates at the block level, as assessed in the direct loss module, with estimated indirect drought losses at the state level. Input-output model An intersectoral input-output table, the basis of the input-output model, was developed for Andhra Pradesh to measure the interactions between the economic sectors at the state level. While the Central Statistical Organization has been constructing such a table for the India-wide economy since 1973­4, it was constructed for the first time for the state of Andhra Pradesh in this study. The table included the fol- lowing sectors: agriculture (food crops, nonfood crops), mining, food processing industries, fertilizers, metal and metal product industries including capital goods, other manufacturing products, electricity, gas and water supply, construction, trade, hotels and restaurants, transport, storage and communication, financial and other business services, as well as community, social, and other services.The input-output table was pre- pared for 1998­9 and then updated for 2002­03 using recent available data (annex 7). Table 3.4 presents the employment coefficients and output multipliers calculated from the input-output table for 1998­9. The employment coefficients are high for the agricultural sector (5.4), implying that the agricultural sector is the major employment generator of the state. The output multiplier for rice showed that a 1-unit (lakh) increase in final demand of rice resulted in 1.45 (lakhs) increase in gross output in the economy. These output multipliers were the same for maize, but slightly lower for jowar (sorghum) (1.43) and groundnut (1.40). The input-output table depicts the economy of a particular year and thus cannot capture the dynamic changes in the economy over time. Such dynamic changes can be captured only through a macroeconomet- ric model, as described below. Macroeconometric model The dynamic structure of theAndhra Pradesh economy is described in terms of changes in the GVA in various sectors of the economy and interrelations between these sectors. The major sectors analyzed in this model are: · Primary sector, of which two most relevant subsectors--agriculture and livestock--were examined separately for sensitivity to drought · Secondary sector, including manufacturing (both registered and unreg- istered), electricity, gas and water supply, and construction 26 Overcoming Drought Table 3.4. Employment Coefficients and Output Multipliers 1998­9 Employment Employment Sectors coefficients Multipliers Sectors coefficients Multipliers Rice 5.40 1.45 Leather products 1.14 2.11 Jowar (sorghum) 5.40 1.43 Fertilizers 0.89 1.09 Maize 5.40 1.45 Pesticides 0.89 2.61 Other food grains 5.40 1.52 Chemicals 0.89 1.82 Groundnut 5.40 1.40 Nonmetallic 0.95 1.95 mineral products Other crops 5.40 1.22 Basic metals 0.05 2.54 and alloys Livestock 5.40 1.42 Metal products, 0.06 2.67 electric and nonelectric machinery and equipment Forestry and 1.44 1.17 Transport 0.59 2.10 logging equipment and parts Fishing 0.68 1.25 Miscellaneous 2.20 Mining 0.47 1.41 Construction 0.86 1.69 and quarrying Food products 1.01 2.23 Railway transport 0.32 2.00 services Textile products 3.15 2.08 Communication 0.41 1.27 Wood products 9.27 1.61 Ownership 0.02 1.12 of dwellings, real estate, and business services Paper products 0.37 2.17 Public 0.92 -- administration Leather products 1.14 2.11 Ownership 0.02 1.12 of dwellings, real estate, and business services Rubber, plastic, 0.45 2.12 Education, 1.09 1.79 coal, tar medical, and other services Fertilizers 0.89 1.09 Public administration 0.92 -- Note: --, not available. Methodological Framework 27 · Tertiary sector, including trade, hotels and restaurants, railways, trans- port by other means and storage, communication, real estate and business, banking and insurance, public administration, and other services The macroeconometric model was used to investigate how the direct economic impact of drought in the eight selected drought-prone districts (captured through the previous modules of the drought risk assessment model) generates indirect economic impacts in these four sectors at state level. Specification of a macroeconometric model required postulating structural equations, which describe changes in the GVA in terms of cer- tain variables and changes that directly influence the GVA. The GVA was calculated as the difference between the values of out- put and inputs (at current or constant prices). However, the inputs did not include the consumption of fixed capital.2 For example, for agricul- tural GVA the inputs are seed, chemical fertilizers, organic manure, cur- rent repairs and maintenance of fixed assets, market charges, irrigation charges, electricity, pesticides and insecticides, and diesel. Therefore, the specification of structural equations for GVA, in each of the sectors, included consumption of fixed capital as one of the explanatory variables. Several specifications of the model in terms of sector-wise GVA were tested (annex 8), and the model best fitting the observed data during 1980­2003 is given below3: where n is the natural logarithm, VOP4,8 is the value of output of the four crops (rice, maize, jowar [sorghum], and groundnut) in the eight selected districts; AGVA is the agriculture GVA, LGVA is the livestock 28 Overcoming Drought GVA, SGVA is the secondary sector GVA, TGVA is the tertiary sector GVA, and AGVA-1 is last year's agricultural GVA; agricultural consump- tion of fixed capital (ACFC), livestock consumption of fixed capital (LCFC), secondary sector consumption of fixed capital (SCFC), and ter- tiary sector consumption of fixed capital (TCFC) are the consumption of fixed capital in agriculture, livestock, secondary sector, and tertiary sector, respectively. Numbers in parenthesis are the p-values and is the co- efficient of determination adjusted for the degrees of freedom. The coefficients were calculated by the seemingly unrelated regres- sions (SUR) method and could be interpreted as partial elasticity coeffi- cients. According to the macroeconometric model, a 1 percentage point change in the production of the four selected crops in the eight selected districts will generate a 0.25 percentage point change in AGVA. Likewise, a 1 percent change in the agricultural GVA would cause a 0.24 percent- age point change in the livestock GVA. A 1 percentage point change in agricultural would cause a 0.37 percentage point change in the secondary sector GVA and ­0.12 percentage point change in the tertiary sector GVA in the next year.This macroeconometric model can explain between 73 and 98 percent of the variability of the dependent variables. The first equation estimating AGVA captured the peaks and drops observed during 1993­2002 (figure 3.4). Because of the statistical limitations of a restricted sample size, the model described above should be considered preliminary. Some alterna- tive macroeconometric models tested under this study showed a positive but statistically not significant elasticity coefficient of AGVA on SGVA, Figure 3.4. Estimated and Observed AGVA AGVA & Est. AGVA (1993­94 to 2002­03) 1,800,000 1,700,000 1,600,000 1,500,000 1,400,000 lakhs 1,300,000 Rs.1,200,000 AGVA 1,100,000 Est. AGVA 1,000,000 900,000 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 years Methodological Framework 29 and a negative but statistically not significant coefficient of AGVA on TGVA (annex 8). Therefore, the impact of a change in the agricultural GVA on the GVA of the secondary and tertiary sectors should be ana- lyzed with caution. In future applications, the model specifications could be further refined based on a larger data series. Notes 1. Gurenko and Lester (2003) developed a risk management approach for the financing of rapid onset disasters (earthquakes, cyclones, and floods) in four states of India (Andhra Pradesh, Gujarat, Maharashtra, and Orissa). 2. The net value added is defined as the difference between the GVA and consumption of fixed capital. 3. Other macroeconometric models tested included regression of detrended data using first differences and regression of detrended data using ad hoc linear trend over the period 1980­1 to 2002­3 (with a break dummy in 1993). C H A P T E R 4 Reducing Vulnerability of Agriculture to Drought in Eight Drought-Prone Districts The results of the stochastic drought risk assessment model for the eight districts, as well as the impacts of alternative drought management and climate change scenarios, will be discussed. The latter are presented for the two most severely affected districts: Anantapur and Mahbubnagar. These results are selected illustrations of the capability of this model to investigate the impact of a variety of risk-coping strategies and climate change scenarios at the farm level. Crop Yield Variability: Benchmark Case In Andhra Pradesh, 68 percent of rainfall is received during the southwest monsoon season from June to September, which is the main cropping sea- son in the rain-fed areas. Maize, jowar (sorghum), groundnut, and sun- flower are the major crops grown under rain-fed conditions during this season in the drought-prone districts. However, rice is becoming more commonly cultivated in rain-fed areas using irrigation water from wells, bore-wells, and tanks.The yields of the five major crops (rice, maize, jowar, 31 32 Overcoming Drought groundnut, and sunflower) were simulated by the probabilistic drought risk assessment model for each block of the eight districts and then aggre- gated to the district level. Variation in yields at block level The yields under normal conditions showed considerable variation across blocks and districts for the same crops.Thus, different locations appear to favor different crops (figure 4.1). For example, groundnut does better in a small region in the northwest, while jowar does better in the southwest and sunflower in the northeast. Maize does not show significant areas of high yield. Figure 4.2 shows the increasing impact of droughts on yields. During severe droughts, losses of yields were almost uniform across most districts and blocks. However, rice is particularly affected in Anantapur and Kurnool, and sunflower shows greater sensitivity in the southwest while maize shows less sensitivity than other crops. Variation in yields at district level Table 4.1 shows the average yields in normal years and yield losses in drought years for rice in each of the eight districts. Similar to block-level results, these losses vary significantly among the districts, especially for rain-fed crops. Importantly, different crops can be particularly vulnerable in different districts, implying that district (and block)-specific coping strategies are needed. It must be stressed that the model simulates yield and production (yield 3 planted area) at the block level using a relatively simple and coarse model of shifts in planting areas at a block scale (that is, the model only crudely adjusts the allocation of irrigation water among crops depending on the level of rainfall). In reality, however, farmers routinely make adjustments in farming practices, including the allocation of irriga- tion water among crops depending on immediate water needs. For example, the model assumes that, once the planting areas have been selected, irrigation water is given to rice on a priority basis irrespec- tive of rainfall and thus estimates that maize yield is more affected by drought than rice. However, during the drought of 2002, the output of crops, such as jowar, maize, and other food grains, showed an increase against the drastic fall in output of rice. This clearly indicates that, in the situation of acute water deficit caused by a major drought, farmers "rationalized" the use of available water by favoring less water-intensive Figure 4.1. Average NormalYield by Crop (MetricTons per Hectare) Reducing Vulner abilit y of Agricultur e to Dr ought 33 34 Figure 4.2. Impact of Severe Drought onYield (Percent DecreaseWith Respect to NormalYield) O ver coming Dr ought Reducing Vulnerability of Agriculture to Drought 35 Table 4.1. Rice Yields in Normal Years and Rice Yield Losses in Drought Years Ananthapur Mahbubnagar Kurnool Cuddapah Chittoor Prakasam Rangareddy Nalgonda Normal (MT/ha) 2.87 2.15 2.59 2.73 2.86 3.10 2.37 2.69 Yield losses in drought years (percent normal yields) Minor 14 10 13 11 10 10 19 8 Moderate 27 19 32 21 18 19 24 16 Severe 45 26 62 31 35 33 31 29 crops. Thus, farmers, using traditional knowledge, common sense, and guidance from local experts and extension officers, do adapt to rainfall variability. The key question is how effective their coping strategies are and whether these can be improved upon. While the model in its current form was quite successfully validated in terms of yields and production for the eight districts based on historical data, integrating more advanced farming behavior modeling techniques is a critical area for further develop- ing and applying this analytical tool. The modeling capability to simulate the behavior of single farmers could be used to assess farm level decisions and help select those that are economically viable. Crop Production Losses The EPC of the estimated loss of VOP for the region of the eight drought- prone districts is defined as the difference between the VOP of the five crops during a normal year and the VOP during a drought year. The VOP was less than that in a normal year 40 percent of the time; that is, the eight districts faced a loss in VOP because of drought every two to three years (2.5 years on average). The VOP loss is as high as over 15 percent once every 10 years on average and exceeds 25 percent once every 25 years (figure 4.3). The AAL of output owing to exposure to drought (averaged over a long series of years) was 5 percent for the eight districts in this study, assuming no changes in the current cropping pattern. This is a signifi- cant value for average loss. The AAL of output increased to 6 percent in the worst-affected Anantapur, closely followed by Mahabubnagar, and decreased to 3 percent in Prakasam district (figure 4.4.). As shown above, there were further variations within districts and across blocks. Even greater disparities in impacts of and resilience to drought can be 36 Overcoming Drought Figure 4.3. Crop Production Losses Caused by Drought in the Eight Districts: EPC 0.40 (EP) 0.35 0.30 0.25 obability 0.20 pr 0.15 0.10 0.05 ceedancexe0.00 0 5 10 15 20 25 30 35 40 percentage of VOP loss Figure 4.4. Average Annual Loss ofValue of Production Output, by District 7 6 eg 5 4 centa 3 per 2 1 0 ubnagar Kurnool icts Chittor akasam distr Anantapur Cuddapah NalgondaPr 8 Rangareddy Mahab expected at the farm and household levels. Averages always mean that individual farmers will experience greater losses than a district or block average if their particular crops are badly affected. For small and mar- ginal farmers, even a 10 percent or 5 percent decrease in output could mean falling below the poverty line. A survey of communities in one of the poorest and worst-affected dis- tricts, Mahbubnagar, showed that depending on the situation of a partic- ular household, responses ranged from a change in farming decisions to migration to extreme cases of starvation, loss of opportunities and health, and even life itself (box 4.1). These responses further highlight the find- ings of the analysis that the impacts of drought are highly localized and differentiated and require targeted assistance to those in need. Reducing Vulnerability of Agriculture to Drought 37 Box 4.1 CopingWith Drought: Findings From Mahbubnagar A survey of drought-affected communities conducted in five blocks of the Mah- bubnagar district provided insights into how farmers and villagers change their behavior during a drought season: Irrigation practices · Increasing the number of tube-wells and decreasing the number of traditional tanks and open-wells · Increasing the depth of the tube-wells to between 200 and 300 feet (approxi- mately 61­91 meters) to access lower groundwater levels Cropping practices · Decreasing the cultivable area because of lack of water and labor because family members moved out of the district · Temporarily adapting crop cycles to suit the time of rainfall · Limited examples of changing to high yield crops, horticultural (sweet orange, mango, acid lime, and so on), and mixed cropping promoted by Government programs Migration/labor · Migration of members or whole families to outside the districts for livelihood, such as construction labor · Sending children to work as laborers · Working at lower wages to generate some income Financial · Taking loans from money lenders (50­60 percent of total loans) or self-help groups, where debts of farmers vary from Rs. 30,000 to Rs. 200,000 (70 percent for agricultural inputs and 30 percent for marriages, health, house construction, or renovation) · Pawning of household items and jewelry · The poorest people reduce expenditure on basic needs, leading to malnutrition and, in extreme cases, starvation · Sale of livestock at depressed prices owing to lack of fodder or agricultural work for the livestock Extreme practices · Suicide Source: Winrock International, India (2005). 38 Overcoming Drought Adaptation Strategies at the Farm Level Farmers adapt, to some extent, their irrigation and cropping practices in response to the changing level and pattern of rainfall. However, these changes are usually short term and aimed at surviving the extreme drought rather than preparing for subsequent droughts, or to the chronic conditions of increased water deficit, occurring in some parts of the study districts. The falling groundwater table (200­250 m) is an indication that current practices are unsustainable in the long term. Energy consumption for pumping has become a costly input for rice irrigation, and, since power supply to agriculture is heavily subsidized, it drains state finances and contributes to power shortages. There is a need for a more sustained shift to water conserving prac- tices. Given the decreased groundwater resource and increased energy costs, the GoAP is taking measures to ensure that water is used more efficiently and that the demand for irrigation water is reduced. It has already taken an initiative to curb the drilling of bore-wells by bringing the Andhra Pradesh Water Land and Trees Act (The Andhra Pradesh Gazette Part IV-B 2002) into effect. Its new energy policy discourages rice cultivation by not providing free power to farmers for the second rice harvest. Rice requires 1,200 millimeters of water during its growth period with any shortfall from rainfall being made up from irrigation. In the simula- tions in this study, 600 millimeters of irrigation were provided at regular intervals over the 120-day growing season for rice. By comparison, rain- fed crops (maize, jowar, groundnut, and sunflower) require only 400­600 millimeters of water to complete their life cycle. Yield of rain-fed crops decreases if the crops suffer moisture stress during critical stages of their life cycle and particularly during early growth or at the time of grain set- ting, which might well happen during a drought year. Irrigating the crops at critical stages during their life cycle in water-scarce periods can increase their yield. Providing 50 millimeters depth of water at one or two critical stages and adding fertilizer along with the irrigation water can further enhance yield. Reallocation of irrigation water by reducing cultivable rice area Two main scenarios were investigated, under which the area planted with rice is decreased so that the water saved can be redistributed to other crops, as follows: Reducing Vulnerability of Agriculture to Drought 39 · Case 1: single irrigation of rain-fed crops at the flowering stage or its equivalent. · Case 2: first irrigation as above plus second irrigation at the time of yield (grain formation) (see also box 4.2) Box 4.2 Reducing Cultivable Rice Area: Assumptions According to local experts, one life-saving irrigation of 50 millimeters depth of water can be given to 24 ha if one hectare of rice cultivation is reduced, as rice requires 1,200 millimeters of water during its growth period. This recommendation, which holds for all rain-fed crops (groundnut, jowar [sorghum], maize, and sunflower), is based on the assumption that there is no significant rain during severe drought conditions and the entire 1,200 millimeters of water required for rice is provided by irrigation. Two cases were investigated under this risk-coping strategy: Case 1: One life-saving irrigation of 50 millimeters depth of water is given for 24 ha of rain-fed area for every hectare of rice cultivation reduced. Irrigated area for rice that needs to be reduced is calculated from the total rain-fed area of the four crops in the ratio of 1:24. The balance of irrigated rice area is left as is. Case 2: Two life-saving irrigations, each of 50 millimeters depth of water, are given for 12 ha of rain-fed area for every hectare of rice cultivation reduced. Irrigated area for rice that needs to be reduced is calculated from the total rain-fed area of the four crops in the ratio of 1:12. The balance of irrigated rice area is left as is. Cultivated areas were reallocated using a simple rule. The cultivated rice area was reduced to provide irrigation to the full area of each of the other crops with one or two irrigations according to the requirement. If there was insufficient cul- tivable rice area to yield the necessary savings in irrigation water, the area of the other crops irrigated was reduced accordingly. This strategy of changing the cul- tivable area could be made more realistic by considering the yield value of differ- ent crop combinations and farmer preferences toward risks. The study showed that under the single irrigation scenario cultivable rice area should be reduced by 54 percentage points in Anantapur and by 8 percentage points in Mahbubnagar. This difference is because cultivable rice area in Ananta- pur is less than half that in Mahbubnagar, while the reverse applies to the area un- der the four rain-fed crops (figure 3.3). 40 Overcoming Drought As stated above, farmers temporarily adopt these practices during low rainfall years; however, the scenarios analyzed here assumed that such practices are used during all years. The results were compared with the baseline where no irrigation water was available for maize, jowar, ground- nut, and sunflower (case 0), a typical real-life situation during the years of normal rainfall or minor drought. To assess the economic impacts, the changes in yield were converted as changes in product value.1 The economic impacts were measured as decrease in the loss of VOP (value impacts), where the value impacts are defined with respect to the VOP during normal (nondrought) years. Alternative uses of irrigation water are compared in figures 4.5 and 4.6. Figure 4.5 shows that in Anantapur a single life-saving irrigation sce- nario (case 1) was evidently effective, as it decreased the production loss for all drought events. Implementing single irrigation (case 1) would mean that the average loss in production value to farmers across the dis- trict would decrease from 24 percent (of value production in normal years) to 14 percent if a 1-in-10-year drought event occurs. The AAL (in value of crop production) across all drought years would be decreased from the estimated 6.7 percent under the benchmark case (case 0) to 3.7 percent under the single irrigation scenario (case 1). Implementing dou- ble irrigation (case 2) had little additional impact on crop production loss, bringing it down to 3.4 percent. These scenarios considerably increased the long-term average annual VOP across all drought and nondrought years. The average annual value Figure 4.5. Reducing Cultivable Rice Area in Anantapur:Value of Production LEC 0.40 0.35 case 0 0.30 case 1 case 2 0.25 EP0.20 0.15 0.10 0.05 0 0 5 10 15 20 25 30 35 40 value impacts (%) Reducing Vulnerability of Agriculture to Drought 41 Figure 4.6. Reducing Cultivable Rice Area in Mahbubnagar:Value of Production LEC 0.50 0.45 case 0 0.40 case 1 0.35 case 2 0.30 EP0.25 0.20 0.15 0.10 0.05 0 0 10 20 30 40 50 60 70 value impacts (%) gain was estimated at 32 percent under the single irrigation (case 1) and 47 percent under the double irrigation (case 2) scenario. Therefore, the strategy of partially reallocating water from rice cultivation to provide life-saving irrigation to less water-intensive crops in Anantapur would reduce by half the AAL of the overall crop production value during the drought years and would increase the all-year average annual crop pro- duction value by one-third for single irrigation and by almost half for double irrigation. In Mahbubnagar, this strategy is much less effective (figure 4.6). Implementing the single irrigation scenario can lead to a 3 percentage point decrease in value impacts under drought events with return periods between 4 years (25 percent frequency) and 10 years (10 percent fre- quency). As in Anantapur, adding the second irrigation scenario does not make a difference to loss reduction. The reason for a greater benefit in Anantapur than in Mahbubnagar can be attributed to the greater proportion of cultivable land for the rain-fed crops with respect to rice. In these simulation runs, a simple allocation based on current land use was used to distribute area to rain-fed crops. The results presented illustrate the model's capability (as well as its limitations) and help formulate options for further investigation. The model currently assesses the effect of switching to the four rain- fed crops only. However, research conducted by ANGRAU, Hyderabad has shown that more complex crop systems and varieties are likely to be needed (box 4.3).These and other options can be built in and assessed by the model in the future. 42 Overcoming Drought Box 4.3 Farm-Level Adaptation Strategies: Expert Recommendations Regarding Cropping Patterns In drought-prone areas, cultivation of a single crop is risky, and, hence, intercrop- ping systems need to be advocated. The following intercropping systems are recommended for drought-prone areas of Andhra Pradesh by experts from the Agricultural University. Soils Cropping system Black soils Cotton + green gram (1:2) Cotton + soybean (1:1) Red soils Groundnut + red gram (7:1 or 11:1) Groundnut + castor (7:1 or 11:1) Bajra + red gram (2:1) Setaria + red gram (5:1) Setaria + groundnut (2:1) Crop varieties also need to be adjusted by soils, as described below: Soils Crops Varieties Red soils Sorghum CSV-15, CSV-13, CSH-13, CSH-14 Palem sorghum hybrid 1. Groundnut Vemana, Tirupati-1 Green gram MGG-295, LGG-450, AMG-275 Red gram Palnadu, PRG-100 Horse gram Marukulthi-1, AK-26 Cowpea Local Pearl millet Anantha, ICMS-451 Castor Aruna Black soils Cotton Narasimha, Aravinda, L-604 Sorghum NTJ-1, NTJ-2 Black soils Cotton Narasimha, Aravinda, L-604 Sorghum NTJ-1, NTJ-2 Setaria Krishnadevaraya, Narasimharaya Lepakshi, Prasad Changes in tillage practice Another suggested water conservation practice is to minimize tillage, which in turn reduces the exposure of moist soil at planting time to drying condi- tions.The effectiveness of minimum tillage as a water conserving effect was Reducing Vulnerability of Agriculture to Drought 43 investigated via the EPIC model. Simulations comparing no-tillage with alternative tillage techniques using a moldboard plow and offset discs were done for several crops (groundnut, maize, and rice) and several districts using recorded rainfall and weather conditions from 1979 to 1998. The results showed that water retention was enhanced slightly early in the season, but these effects had little impact over the entire season.Across all treatments yields only rarely varied by more than 1 percentage point. Minimum till cropping may have other advantages such as in maintaining soil structure and nutrients and in saving energy used for plowing (although it often requires additional herbicide treatments), but it appears to have little effect as a drought mitigation method. Impact of Climate Change Greenhouse gas emissions, largely driven by human activities, are already affecting the current climate and will do more so in the future. Most parts of the earth are becoming warmer, and, overall, precipitation is increasing. However, rainfall is projected to become more variable with fewer rainy days but heavier rainfall events in most regions, consequently causing a greater risk of both droughts and floods (Watson 1998; McCarthy 2001). Thus, climate change is likely to increase the climate variability expe- rienced by farmers. Some factors, such as increased temperatures and longer drought periods, are likely to decrease production, while others, such as the higher concentrations of carbon dioxide in the atmosphere, would increase productivity. The study explored the effects of a feasible climate change scenario on climate variability and agricultural productiv- ity for the drought-prone districts of Andhra Pradesh. Climate modeling The main tool for making projections of climate change is global climate (or circulation) model(s) (GCMs), which simulate climate at a scale of about 300 km by 300 km over India. Higher resolution projections are obtained by running regional climate models (RCMs) for subregions of the globe (often about 5000 km by 5000 km).The RCMs use the output from the GCMs to provide the climate at the boundaries of the region but then simulate the cli- mate within the region at a scale of 50 by 50 km, with some coming down to 20 by 20 km. Most models fail to simulate some important aspects of the Indian climate. However, recent GCMs have improved significantly; RCMs are relatively better. In making projections of future climate, the climate models are simulated for assumed conditions (that is, greenhouse gas, 44 Overcoming Drought particulate composition, concentration in the atmosphere, and so on) for some period in the future, and these results are expressed as a dif- ference from the simulated climate for the baseline period. For one of the most commonly used scenarios of global development (IS92a scenario), the range of GCMs predict an increase in temperature of 3°­6°C and an increase in rainfall of 15­40 percent by 2100 for India as a whole with high percentage increases occurring mainly in the drier regions and thus of little impact. However, the impacts vary considerably by region. Climate projections for Andhra Pradesh In Andhra Pradesh temperatures are projected to increase by at least 3°C throughout the state by 2041­60. This increase will occur during all sea- sons of the year.While rainfall is projected to increase for India as a whole, it is projected to decrease in the drought-prone areas of Andhra Pradesh. This decrease would be 5­20 percent during the critical monsoon season with a 5 percentage point increase during the dry March­May period.The number of rainy days appears to decrease by about 5­10 percent. Rainfall intensity (millimeters rain per wet day) appears to remain roughly constant over the years, but there may be seasonal changes that do not show up in the published data as GCMs are still unreliable in predicting rainfall intensities. Hydrological modeling suggests a significant reduction in runoff (from about 150 millimeters to 110 millimeters per year) in the Pennar River basin. This implies serious problems for water supply in the southern Andhra Pradesh. The overall assessment for the drought-prone regions of Andhra Pradesh for 2041­60 is "chronic water scarcity and drought conditions." Simulation of impacts on yields In this study, a RCM for India, Hadley Regional Model 2, was used to derive projected climate change for southern Andhra Pradesh for 2050. Two simulations of climate change were generated based on these results by changing the WXGEN within EPIC. Both scenarios assumed an increase in temperature and a decrease in the number of rainy days. The second assumed a more severe reduction in rainfall during the early mon- soon months than the first (box 4.4). The results were based on 20 years of simulated weather scenarios. The impact on crop yields derived for the two most drone-prone dis- tricts, Anantapur and Mahbubnagar, is the combined effect of increased temperature, decreased rainfall, and increased carbon-dioxide (table 4.2.). Reducing Vulnerability of Agriculture to Drought 45 Box 4.4 Climate Change Scenarios The following scenarios were simulated over the next 20 years under the prob- abilistic drought risk model: Climate change scenario 1: · Maximum temperature increases by 2°C · Minimum temperature increases by 4°C · Annual rainy days decrease by 5 percentage points · Atmospheric carbon dioxide at 550 ppm Climate change scenario 2: · Maximum temperature increases by 2°C · Minimum temperature increases by 4°C · Annual rainy days decrease by 5 percentage points · Cumulative June­September (monsoon) rainfall decrease by 10 percentage points · Atmospheric carbon dioxide at 550 ppm There was a minor difference in crop yields between the two scenarios. All four rain-fed crops showed increased yields under climate change scenario 1 and, with the exception of sunflower, a little change in climate change scenario 2. Rice showed a decrease in yield by 8­9 percentage points. Previous studies (Aggarwal et al. 2001) using similar climate change scenarios have suggested a smaller decrease in rice yield of about 5 percent. The model used in this study needs to be analyzed in more detail to determine whether it properly captures the known sensitivity of rice to increases in carbon dioxide concentration in the air Table 4.2. Crop Yield Changes Under Climate Change Scenarios: Average Results for Anantapur and Mahbubnagar Average crop yield change with respect to baseline Crops Baseline scenario CCS1 CCS2 Rice 2.59 t/ha ­9% ­8% Groundnut 0.97 t/ha 2% 0% Jowar (Sorghum) 0.87 t/ha 3% 0% Sunflower 0.51 t/ha 10% 9% Maize 2.10 t/ha 3% 0% Source: Simulations by the study model. Note: CCS1, climate change scenario 1; CCS2, climate change scenario 2. 46 Overcoming Drought (yield enhancing) and increased temperatures at critical times in its growth cycle (yield depressing). While acknowledging all the uncertainties and the need for further research, the results suggest that climate change would further reinforce the benefits of shifting from rice to less water-intensive crops. Implications for Agriculture Financing and Risk Insurance Cost-effective risk mitigation measures cannot always fully protect farmers against drought risk, particularly against extreme events. Risk financing arrangements, such as insurance, can thus help farmers to transfer the residual (nonmitigated) risk. The findings of this study showed a very high variability of losses across time, locations, and crops, a potential to significantly reduce average loss through certain adaptation strategies, and useful implications for designing drought risk financing strategies at the state level, such as innovative insur- ance products. The probabilistic drought risk model developed in this study was based on sophisticated weather, soil, and crop growth information. It can be used to forecast the expected yield and loss ratio function over the crop season. The model provides a foundation for reconsidering agricultural insurance through catastrophe modeling techniques. One of the main reasons why crop insurance has so far been under- developed worldwide is the complexity of risk and the lack of adequate risk modeling technology to understand the impact of agricultural risks, particularly drought, on crop yields. As shown in this study, drought is a highly location- and crop-specific phenomenon. The probabilistic drought risk model may thus create new growth opportunities for com- mercial agricultural (crop) insurance, which until today is almost non- existent. As mentioned above, developing a capability to simulate the behavior of single farmers would be another important step in this direction. Crop insurance is a sophisticated line of business, as the impact of adverse natural events such a drought on crop yield is the result of com- plex agro-meteorological phenomena. This prototype model offers a new risk modeling technology for the design and pricing of crop insur- ance and particularly weather insurance products recently offered on a pilot basis in India (box 4.5). The probabilistic drought risk model, building on a prototype developed for this study, also offers crop insurers the opportunity to make better informed decisions, as the model can Reducing Vulnerability of Agriculture to Drought 47 identify high-risk crops and areas, to better plan reserve requirements and reinsurance needs, and to build a more diversified crop insurance portfolio. The analysis also highlights that while risk financing arrangements offer farmers a valuable opportunity to finance their losses, it is important to ensure that they do not perpetuate the current situation of farmers' Box 4.5 Weather Insurance in India: Advantages and Caveats The analysis of the Indian Crop Insurance Program between 1985 and 2002 revealed that rainfall accounted for nearly 90 percent of total claims in India; that is, 75 percent due to deficit rainfall and 15 percent due to excess rainfall. Against this background, crop insurance may be a viable risk financing solution to help farmers to absorb their potential losses. However, traditional multiperil crop insurance suffers from many shortcomings: moral hazard, leading to high claims; adverse selection of risk by taking undue advantage of the system; involvement of multiple agencies and huge administrative cost of running the programs, hidden in Government budgets; lack of reliable methodology for estimating and reporting crop yields; and lengthy process of claims settlement. Index-based insurance is an alternative form of insurance where indemni- ties are based on an index (for example, rainfall) and not on individual losses. Rainfall insurance has many advantages, particularly when dealing with small and marginal farmers heavily exposed to drought. Trigger events (such as ad- verse rainfall) can be independently verified and measured. Since India has an independent rainfall reporting system (through the Indian Meteorological Department), rainfall can be measured in the most tamper-proof environment. This would greatly neutralize moral hazard in data procurement. Rainfall insurance does not encourage potential negligence in the insured, and the cultivator's urge for a good harvest remains unaffected. Rainfall insurance is less expensive to operate because very few agencies are involved in imple- mentation. Rainfall insurance allows for speedy settlement of indemnities, as claims can be settled as early as a fortnight after the indemnity period. Rainfall insurance was launched as a pilot scheme in June 2003 in Mahbubna- gar, district by ICICI-Lombard through the Krishna Bhima Samruddhi local area bank. In 2004, three insurance companies (AIC, IFFICO-Tokyo, and ICICI-Lombard) (Continued) 48 Overcoming Drought offered rainfall-based insurance products in several states. They insured 7,181 farmers covering a sum insured of Rs. 157.0 million, earning a premium of Rs. 8.9 million. However, such a risk-financing product may have limitations in the long term, particularly if the insured crops become increasingly exposed to drought because of a falling groundwater table (or increased rainfall variability caused by climate change). An increase in the frequency and/or the severity of droughts would make rainfall insurance more expensive, as insurers will include this risk increasing effect in the pricing of their insurance products. Rainfall insurance may thus give farmers the wrong incentives to grow nonviable crops rather than providing an incentive to switch to more sustainable farming practices. These incentives may even be stronger if rainfall insurance is eligible for Government subsidies. heavy dependency on rainfall.A sizable average crop output loss owing to drought, assuming no change in the current agricultural practices, would make such insurance products unviable. Rather, new financing products should provide an incentive to permanently switch to alternative, more sustainable, agricultural and economic practices, such as less water- intensive crops (particularly high-value cash crops), livestock, or some agro-processing activities. Developing contingent financing schemes that could facilitate this transitional drought adaptation process appears an important area for further work. Two lines of possible innovative financing products are suggested by the study. · Drought adaptation insurance could provide coverage against risks owing to a shift from nonviable farming business to viable (agricul- tural and nonagricultural) businesses. This insurance product would thus protect farmers against new sources of risks resulting from a change in their familiar farming practices toward those that are more drought-resilient and less water intensive. · Drought adaptation credit could provide initial capital to shift to a long-term viable business. In the event of an unexpected loss caused by a failure in the adaptation investment, repayments may be postponed or (partially) forgiven. Reducing Vulnerability of Agriculture to Drought 49 These financial arrangements for drought adaptation would try to induce farmers to shift from farming practices that are known to be unvi- able in the long term because of increasing water scarcity in Andhra Pradesh, to be likely exacerbated by global climate change.These arrange- ments would offer farmers the opportunity to share new risks, associated with the transition, with the society, because the adaptation process would benefit both farmers and the society. Note 1. Commodity prices are the following: rice: 5,654 Rs/MT; jowar (sorghum): 3,130 Rs/MT; maize: 2,763 Rs/MT; groundnut: 9,647 Rs/MT; and sunflower: 11,900 Rs/MT. C H A P T E R 5 Managing Economic Impact of Drought at the State Level Historically, drought has caused direct and indirect economic, social, and environmental problems in Andhra Pradesh. Drought-induced economic losses include those resulting from impaired agricultural products, exces- sive demand of power for agricultural water pumping (which is heavily subsidized), decrease in agriculture-dependent industries, and increased unemployment in agriculture and other drought-affected industries. It is difficult to quantify indirect economic losses associated with droughts, as the impacts are generally nonstructural and spread over large areas. Assessment of Direct and Indirect Loss Potentials: Benchmark Case A prototype macroeconometric model was developed to capture the impact of drought at the state level through its impact on the eight selected drought-prone districts. These districts are estimated to account for 70 percent of a state-wise average loss in the agricultural production owing to drought and for 88 percent of the state-wise crop production variability (see chapter 3).This model intended to estimate the impact of drought on the main economic sectors of Andhra Pradesh: agriculture, livestock, and secondary and tertiary sectors.1 51 52 Overcoming Drought The macroeconometric model was linked to the damage assessment module that simulates crop production losses caused by droughts in the eight drought-prone districts. "Losses" mean a reduction in the simulated values that the same indicators would have under normal (nondrought) weather conditions. Under normal weather conditions, the average annual VOP of the five crops in the eight selected districts was estimated at Rs. 262,483 lakhs based on 2002­3 prices. The macroeconometric model estimated the GVA in each sector of the economy during normal years (figure 5.1). The tertiary sector GVA represented 50 percent of total GVA, and the share of the agriculture and livestock sector was 20 percent.2 These estimates are close to the current economic structure of Andhra Pradesh. In 2002­3, the agriculture and livestock GVA and the tertiary sector GVA accounted for 24 percent and 47 percent of total GVA, respectively. From the assessment of the crop production losses in the eight dis- tricts (see chapter 4), the macroeconometric model translated the impact of these crop losses on the economic sectors of Andhra Pradesh, measured in terms of loss in GVA, as described in chapter 3. The model predicts that a 1 percentage point loss in this VOP would cause 0.25 percentage point loss in the state agriculture sector GVA; a 1 per- centage point loss in agriculture GVA in turn would cause 0.24 per- centage point loss in the livestock GVA. A 1 percentage point loss in agriculture GVA of the previous year would generate 0.37 percentage point loss in GVA in the secondary sector and 0.19 percentage point increase in GVA in the tertiary sector (with a one year lag). Figure 5.1. Average Sectoral GVA For NormalYears 5,000,000 4,635,344 4,500,000 4,000,000 3,500,000 3,000,000 lakhs) 2,500,000 2,263,159 Rs.( 2,000,000 A 1,500,000 1,233,404 GV 1,000,000 576,334 500,000 0 agriculture livestock secondary tertiary sectors Managing Economic Impact of Drought at the State Level 53 Table 5.1. Assistance Provided to Drought Affected States from National Calamity Contingency Fund 2001­2 2002­3 2003­4 NCCF Foodgrains NCCF Foodgrains NCCF Foodgrains State (Rs in crores) (lakh MTs) (Rs in crores) (lakh MTs) (Rs in crores) (lakh MTs) Andhra Pradesh -- 21.50 123.51 20.00 50.58 18.20 Note: NCCF, National Calamity Contingency Fund; --, not available. The latter suggests that droughts may have a positive impact on the tertiary sector in the next year. This is consistent with the historical data on performance of the tertiary sector in drought and nondrought years.3 A number of factors may account for the drought-related boost to tertiary sector production: central government transfers, changes in con- sumption patterns caused by the drought, and an increased supply of labor. Table 5.1 shows that the GoI transferred Rs. 153.5 crores to Andhra Pradesh through the National Calamity Contingency Fund in the drought year 2002­3 (Lok Sabha Unstarred Question No. 151 2004). Figure 5.2 shows the long term AAL in GVA caused by droughts as percentage, by the main economic sectors and the contribution of each drought category (minor, moderate, and severe, as defined in chapter 3). Notably, the AAL in GVA for the overall state economy is estimated at Figure 5.2. Average Annual Loss as Percent of GVA Owing to Droughts 1.40 1.20 severe 1.00 0.29% moderate minor 0.80 eg 0.60 0.54% entac 0.40 0.11% per 0.07% 0.20 0.20% 0.05% 0.32% 0.13% 0.10% 0.12% 0.00 0.08% 0.06% ­0.04% ­0.07% ­0.04% ­0.20 agriculture livestock secondary tertiary total ­0.40 54 Overcoming Drought a very modest 0.2 percent, jumping to more than 1 percent for the agri- culture sector. The largest average damage appears to be caused by moderate droughts, which contribute almost 50 percent to the AAL in the agricultural sector. When a minor drought occurs, the conditional average loss is estimated at more than 3 percent of agriculture GVA but below 1 percent of live- stock GVA. During moderate drought the conditional average loss in the agricultural sector would be about 4 percent of agriculture GVA and that for the whole economy was 1 percent of total GVA. During severe drought, the conditional average loss approaches 8 percent in the agricul- tural sector and 2 percent for the whole economy; the tertiary sector, however, showed a gain of 2 percent (figure 5.3). The economic impact of drought events can also be captured through the EPCs (figure 5.4). A moderate drought event (occurring 1 in 10 years in the study region) would cause 4 percent GVA loss in the agricultural sector, 1.5 percent GVA loss in the secondary sector, and 1 percent GVA loss in the livestock sector. During severe drought, which is a rare event, these losses would increase to 7 percent for the agriculture sector, 3 per- cent for the secondary sector, and 2 percent for the livestock sector. Similar to the GVA analysis on figure 5.3, the secondary sector is more exposed to drought owing to its interdependence on the agriculture sec- tor than the livestock sector. The analyses presented in figures 5.3 and 5.4 show that drought events in the eight selected drought-prone districts mainly affect the agricultural sector at the state level, with modest losses in the livestock and secondary sectors. Thus, the indirect economic losses outside of the agriculture sector appear limited, or may even generate (marginal) gains in the tertiary sector. The total impact on the Andhra Pradesh Figure 5.3. Conditional Average Loss in GVA, by Sector and Drought Category 8 7 minor 6 moderate 5 severe sectoral 4 of loss eg 3 A 2 GV entac 1 0 per ­1 ­2 agriculture livestock secondary tertiary total Managing Economic Impact of Drought at the State Level 55 Figure 5.4. Economic Losses, in Sectoral GVA, Caused by Droughts, Andhra Pradesh: EPC 0.50 agriculture livestock 0.45 secondary tertiary 0.40 (EP) 0.35 0.30 obability pr 0.25 0.20 0.15 xceedancee0.10 0.05 0.00 0 2 4 6 8 10 12 14 economic loss (%) economy, as measured through the loss in total GVA, is marginal. This finding is consistent with a growing body of evidence on the macroeco- nomic impact of climate-related disasters. Based on worldwide histor- ical data, a recent study showed that the maximum impact of drought is 0.8 percent of GDP annually for a group of developing countries (Raddatz 2005). This analysis focused on the macroeconomic impact of drought in Andhra Pradesh. It did not capture the impact of drought on the Government's revenue and expenditure (that is, its fiscal impact).The state fiscal deficit (total revenue ­ revenue expenditure ­ capital outlay ­ net lending) increased by 7.6 percentage points in 2003­4 following the drought year 2002­3. It is useful to compare the statewide economic impact of drought with those resulting from other climate extremes. A World Bank study focused on cyclones and floods in Andhra Pradesh, using a modeling framework that applies to rapid onset disasters (box 5.1). Interestingly, the AAL for the Andhra Pradesh economy (measured in terms of loss in GVA) caused by droughts is lower than that caused by cyclones or floods, although any comparison has to be made with caution because losses are not measured using the same unit (loss in GVA for droughts and property loss in public infrastructure and housing for cyclones and floods). 56 Overcoming Drought Box 5.1 Financing Rapid Onset Natural Disaster Losses in India Given India's vulnerability to growing losses caused by natural disasters at the central and state levels, the World Bank undertook a detailed review of India's catastrophe exposures. The goal of this project was to examine the loss poten- tials from rapid onset natural disasters and to consider the opportunity to apply enhanced country and state level risk management techniques, with a particu- lar emphasis on the financing of post disaster reconstruction and the efficient allocation of public funds. The study analyzed and quantified the impact of historical and probable fu- ture natural catastrophes on four states (Andhra Pradesh, Gujarat, Orissa, and Maharashtra) that suffered extensively from natural disasters in the recent past. The key major objectives were to create a reasonably comprehensive exposure database for residential buildings and public infrastructure, assess the nature of hazards affecting the region, measure the exposures and vulnerability of dis- tricts/blocks in the region to catastrophic shocks, construct hazard maps based on the severity and frequency of hazards involved, and develop an "actuarially sound" flexible economic loss model that could be used for catastrophe risk-management at the state level. Owing to limited availability of data, the scope of the modeling with regard to potential losses was limited to public infrastructure (consisting of educational, medical building, roads, and bridges) and housing (residential dwellings). Gov- ernment buildings, utilities, minor irrigation systems, and commercial/industrial property were not included in the study, thus translating into lower damage estimates than would be expected in practice. The selected perils for Andhra Pradesh were cyclones and floods. The AAL was estimated at US$61.2 million for cyclones and US$21.7 million for floods. The PML for a 1 in 150 year event was US$911 million for; cyclones US$191 million. Source: Gurenko and Lester (2003). Simulating the Impact of Structural Changes in the Andhra Pradesh Economy The economic structure of Andhra Pradesh has profoundly changed over the past two decades, with a decrease of the primary sector (particularly agriculture) and an increase of the secondary and tertiary sectors. Such a Managing Economic Impact of Drought at the State Level 57 Table 5.2. Scenarios on the Structure of the Andhra Pradesh Economy, Percent Primary Secondary Tertiary Scenario Agriculture Livestock Others sector sector sector Case 0 14 6 6 26 25 49 Case 1 7 6 6 19 21 60 Case 2 4 6 6 16 17 67 structural change is likely to continue in the future, and can be interpreted as a macroeconomic drought adaptation strategy, since the secondary and tertiary sectors are only marginally affected by droughts. The impact of the economic structure of Andhra Pradesh on its resilience to drought is examined through several scenarios in the macro- econometric model (table 5.2). The baseline case 0 scenario represents the current economic structure (in terms of GVA). Alternative scenarios, cases 1 and 2, assume that the share of the agricultural sector decreases, while the share of the tertiary sector increases significantly. The impact of the structural change, as a macroeconomic risk mitiga- tion strategy, on the Andhra Pradesh economy appears to be very effec- tive and able to reduce the loss in total GVA by about three times in case 1 to eight times in case 2 (figure 5.5). Such loss reductions were similar (proportion-wise) for each drought category. Under severe drought, loss in total GVA would decrease from 1.6 percent if the Andhra Pradesh economy maintains the current structure in (case 0) to a mere 0.2 per- cent in a hypothetical case (case 2) of an economy roughly approximat- ing the current structure of the economy of Brazil. Figure 5.5. Loss inTotal GVA Under Different Economic Scenarios, by Drought Category 1.8 1.6 case 0 1.4 case 1 case 2 (%) 1.2 1.0 loss A 0.8 0.6 TGV 0.4 0.2 0.0 minor moderate severe drought category 58 Overcoming Drought Figure 5.6. Losses inTotal GVA Caused by Droughts Under Different Economic Scenarios: EPC 0.40 0.35 case 0 0.30 case 1 obability 0.25 case 2 pr 0.20 0.15 0.10 ceedancexe 0.05 0.00 0 0.5 1.0 1.5 2.0 economic loss (%) Under the current economic structure (case 0), there is a chance (of about 5 percent) that economic loss (measured in GVA) owing to drought will exceed 1 percent of total GVA in a particular year. The maximum possible impact owing to a major drought is below 1 percent of total GVA in case 1 and well below 0.5 percent in case 2 (figure 5.6). Thus, the macroeconomic impact of drought events occurring in the eight most drought-prone districts of Andhra Pradesh is limited at the state level (in terms of loss in the total GVA), and the reduction in the share of the agricultural sector in the total GVA would make this drought impact even smaller. However, while the impact of drought averaged over a series of drought and nondrought years may be marginal at the state level, its effect at the farm level in the drought-prone districts can be significant or even disastrous, as discussed in chapter 4. Furthermore, the agricultural sector is the major employment generator for the state. Therefore, any external shock to the agricultural sector has a direct impact on the state's employment scenario, with major social and political ramifications. The total employment loss for 2002­3 because of the loss in the agricultural VOP is estimated at more than 44 lakhs. This highlights the need for effective strategies that specifically target economic indicators that are most vulnerable to drought, such as output and employment in the agri- culture sector, particularly in the eight most-affected districts. Socioeconomic Strategies to Decrease Vulnerability to Drought Risk: Issues and Options Drought management strategies have been aggressively addressed by the GoAP for many years through a variety of programs (summarized in Managing Economic Impact of Drought at the State Level 59 chapter 1 and in annex 1).4 The analysis, performed in this study, provid- ed some additional insights on possible options to better adapt to the drought-prone climate and mitigate the adverse socioeconomic impacts. Encouraging alternative employment options in the secondary and tertiary sectors Loss of employment is a key concern when a sector gets affected because of loss in production. Employment coefficients, obtained from the input- output table (annex 7, table A7.1), provide a measure to account for the loss in employment corresponding to a loss in production. The agricultural employment coefficient for the state was 5.4, which is high relative to other sectors, and confirms the vital importance of the agriculture sector in securing the livelihoods of a large number of people. A one-unit fall in the agricultural output will result in a loss of 5.4 employment units.Thus, the employment profile across various sectors was examined to identify opportunities during drought years. Services. The macroeconometric model estimated that the TGVA of a given year and the AGVA of the previous year are negatively correlated (that is, a 1 percentage point change in agricultural GVA of last year would lead to ­0.12 percentage point change in the tertiary sector GVA of the current year). This may be partly due to GoI transfers through the National Calamity Contingency Fund, and movement of labor from agri- culture to services. In the service sector, significant employment potential is available in trade and transport (except railways). Construction. This sector had a 49 percent increase in VOP during the severe 2002 drought year probably because of increased Government expenditure as a result of antidrought poverty alleviation programs, hence providing alternative employment to farmers affected by drought. The construction output multiplier obtained from the input-output table was 1.69.5 Any expenditure in the construction sector would thus lead to an increase in outputs for sectors, such as cement, steel, bricks, and tiles, and additional employment opportunities, which could be of help for drought-affected people. Other selected sectors. Subsectors, such as fishing, mining, and quarry- ing, increased between 1980 and 1981 and 2002 and 2003. The increase was even greater during the drought years. For example, the share of these sectors increased from 15.9 percent in 2000­1 to 21.6 percent in 2002­3. These sectors have good employment potential, and can absorb some 60 Overcoming Drought labor displaced from the agricultural sector, thus moderating the employ- ment loss in the agricultural sector because of drought. Supporting a structural shift in the primary sector Livestock sector. During the drought years of 1993­4 and 2002­3, LGVA increased, although AGVA had decreased in these years. The livestock sector experienced a 77 percentage point increase in produc- tion for the drought year of 2002­3 over 1998­9, a normal year. This suggests that drought had no significant effect on the sector. The different components contributing to this sector have behaved differently in the drought year than in the normal year. The three major components of this sector are milk, meat, and eggs. The value of milk as a proportion of the total value of livestock decreased from 55 to 50 percent, while that of meat remained the same. Against this backdrop the value of eggs increased and its contribution almost dou- bled from 8 to 15 percent. This good performance during drought may also be due to Government interventions that were encouraging the poultry sector. Cropping pattern. The VOP of the agriculture sector decreased by 27 percentage points in 2002­3. Particularly, the VOP of rice and groundnut decreased by 38 percentage points and 57 percentage points (relative to 1998­9), respectively. This greatly affected the loss of VOP for the total agricultural sector.While the output values of crops like jowar (sorghum), maize, and other food grains showed a increase, the drastic fall in output of rice and groundnut (the major crops grown in Andhra Pradesh that are also much more water sensitive than the other crops grown here) has out- weighed the increase in other sectors. The shift from rice and groundnut, particularly rice, to other crops would increase the resilience of the agri- culture sector to drought and water scarcity. .TAy,itpu2u y.wlidagn0l tyrrzWv7anh ivgcs.i5cit csidsf rrci)g88tChapter 4 provides a quantitative assessment of decreased production et93o htssTs1yp acpdrwobe spiaoteeg osop(alsa fajetrrrc P,utao(or e)auccsg7 Oemoe)ny V9ePjlab wmhOketd dl(Vlhc nfnwhni lnf.up gd,fv uPhp itgn on h3 m losses from such a shift using the example of two drought-prone districts. The input-output analysis also points to a potential for savings in terms of the inputs required for producing these crops, for example, 30 percent more input is needed for producing 1 unit of rice than for producing 1 unit of maize. During drought the output drops but the inputs for pro- duction does not drop in the same proportion as is evident from the input proportions for different sectors under agriculture for 1998­9 and 2002­3. Managing Economic Impact of Drought at the State Level 61 Therefore, a shift in the cropping pattern from rice to less water- intensive crops, particularly in the eight study districts, is likely to result in both reduced VOP loss and savings in inputs. While this strategy would help decrease the state agricultural GVA loss, its impact would be of particular significance to farmers operating in these districts. This further emphasizes the importance of designing and providing assis- tance to farmers not only to help absorb the risk of extreme weather events but to promote sustainable agricultural and economic practices in the long term. Notes 1. As noted in chapter 3, the validation of the model on the historical data was successful; however, the specification reported here should be considered an initial test product that should be refined in future applications based on addi- tional data and econometrics techniques. 2. Other subsectors of the primary sector (forestry and lodging, fishing, and min- ing and quarrying) represented 6 percent of total GVA. 3. As mentioned in chapter 3, losses/gains in the secondary and tertiary sector must be best viewed as indicative of sectoral linkages, rather than precise estimates, as estimated coefficients are not statistically significant under alter- native econometric models. In this case, the current model may slightly over- estimate the total losses in GVA. 4. See also India Council of Agricultural Research. 2002. Drought Management Strategies. New Delhi. 5. Interpreting the multiplier, a 1 unit increase in the output of the construction sector will result in an additional 0.69 units increase in outputs of other sec- tors because of interlinkages between the sectors. C H A P T E R 6 Conclusions and Recommendations The key outcomes and conclusions of the study are presented under the following heads for the three study objectives. · Methodology development (develop a robust analytical framework for simulating the long-term impacts of drought at the micro [drought- prone areas] and macro [state] levels) · Findings and observations from the quantitative analysis (conduct a quantitative probabilistic risk assessment of the impacts under differ- ent scenarios) · Areas for future action (assist the GoAP in the development of a fu- ture-looking and anticipatory strategy for adapting to frequent drought events and conditions of water deficit) Methodology Development Catastrophe modeling is an evolving area, which aids policymakers and other stakeholders in managing risks from natural disasters. The existing models developed by international risk modeling firms focus on the impact of rapid onset disasters, such as earthquakes or hurricanes, on pub- lic and private infrastructure. These models have recently been used by the World Bank to develop risk management strategies for financing rapid 63 64 Overcoming Drought onset disasters in India and Colombia. However, slow onset disasters such as drought have characteristics different from rapid onset events and are more difficult to quantify. While they mainly directly affect agricultural output, there are several indirect impacts too. Therefore, an important contribution of this study was in modifying and testing an original model under a different risk assessment paradigm that could be applied to slow onset disasters. This probabilistic drought risk assessment model is a powerful tool for policy makers to better understand the consequences of drought in the different sectors of the economy, to quantify such impacts with respect to drought severity, and to investigate the economic impacts of risk-coping strategies, both at the farm and state levels. The stochastic dimension included in this model also allows capturing of the underlying uncertainty related to weather events, including the impact of anticipated permanent changes in the global climate. The innovative framework developed in this study, which expands on previous work on catastrophe modeling to drought, can be used to address the issue of drought in other states of India and other drought-prone countries. The model has a number of specific areas for development to increase its practical value as a planning and decision support tool. The study identified areas, such as: · Enhancing the model's capability to be applied at a farm level: This would allow the model to incorporate more realistic farmer behavior in response to the seasonal patterns of rainfall and the availability of irrigation water. In particular, a more advanced farm-level model would offer the opportunity to look much more closely at the pat- terns of demand for irrigation, energy, and fertilizers, and for labor. · Including a larger number of alternative crops: Particularly high-value drought-resistant cash crops to assess the benefits of various coping strategies available to farmers. · Refining macroeconometric specifications: Particularly on a larger dataset to increase the predictive power of the model. Another specialized area for further development and application is re-examination of agricultural insurance through catastrophe modeling techniques. One of the main reasons why crop insurance has so far been almost universally a failure worldwide is the complexity of risk and the lack of adequate risk modeling technology to understand the impact of agricultural risks, and particularly drought, on crop yields. As shown in Conclusions and Recommendations 65 this study, drought is highly location-, time- and crop-specific, with crop yield losses that can be decreased further by changing crop and irrigation patterns.The probabilistic drought risk model may thus create new growth opportunities for commercial agricultural (crop) insurance, which is almost nonexistent now. Some examples of possible future applications of the technical foun- dation created by this work in the insurance business are: · Drought risk model as a risk underwriting and pricing tool: Crop insurance is a complex line of business, as the impact of adverse natural events such as drought on crop yield is the result of complex agro-meteorolog- ical phenomena. This model offers a new risk modeling technology for the design and pricing of crop insurance, and particularly weather insur- ance products recently offered on a pilot basis in India · Drought risk model as an innovative test of economic viability of the agricul- tural business: By identifying areas exposed to drought risk and assessing the impact of drought on the crop yield variability, the model helps to determine crops that are economically viable in a particular location un- der different climate change scenarios. It thus offers a quantitative tool to target subsidies for crops viable in the long term (even if these crops are financially less attractive in the short term). Findings and Observations from the Quantitative Analysis Despite a variety of antidrought programs, the human and social costs of droughts have been and remain devastating for millions of people in Andhra Pradesh. Frequent drought is a difficult reality for farmers in the eight rain-shadow districts of Andhra Pradesh. Under the business as usual long-term scenario, the agricultural sector of these districts faces a 40 percent chance (or every two­three years) that the VOP for the five major crops (rice, maize, jowar [sorghum], sunflower, and groundnut) combined will be somewhat less than in a normal rainfall year. Loss of crop production output exceeds 5 percent of the normal year output value every 3 years, 10 percent every 5 years, 15 percent once in 10 years, and 25 percent once in 25 years. The AAL of output caused by drought was 5 percent for the eight districts, ranging from 6 percent in the worst- affected Anantapur district to 3 percent in Prakasam. Individual farmers may experience greater losses if their specific crops happen to be hard hit. Importantly, for many small and marginal farmers in 66 Overcoming Drought these districts, a loss of VOP of 10 percentage points or even 5 percentage points (which is likely to happen quite frequently) can mean falling below the poverty line. This suggests the need for enhancing the existing strategy by innovative, future-looking approaches and tools to help these people adapt to frequent droughts. The GoAP needs to intensify efforts that are sustainable and resilient to water-scarce conditions in the long term to sup- port the economic and social development of drought-prone areas. The impacts of droughts are highly variable and localized. In addition to large variations across time, impacts vary greatly across locations and crops, depending on drought severity. Modeling highlighted significant variations in impacts on a particular crop across districts and even blocks within the same district. For example, severe drought is likely to decrease rice yields from 29 percent in Nalgonda to 62 percent in Kurnool. Yield losses of maize, a rain-fed crop, were particularly staggering in Anantapur, Kurnool, and Mahbubnagar, which are the driest districts with less than 600 millimeters of rainfall every year. Farmers often rationalize the use of available water by reducing cul- tivable area under water-intensive rice in favor of less water-intensive crops when there is acute water deficit caused by a major drought. This is however practiced as a temporary measure with the cultivable area of rice typically restored once the drought is over. The model assessed some scenarios of permanently reallocating water from cultivable rice areas to provide 50 millimeters irrigation for the four rain-fed crops (included in the model) at one or two critical stages of their growth. In Anantapur, this strategy helped reduce by half the average annual loss of the overall crop production output during the drought years and increase the all-year average annual crop production output by one-third. Importantly, better water conservation practices alone (such as a change in tillage practice), without change in cropping pattern, do not appear to have a significant long-term effect on a large scale.The results of the analy- sis suggest that losses borne by farmers because of drought can be significantly decreased by adjustments in farming practices that decrease water demand, such as a permanent shift to a larger share of less water-intensive crops in the cropping mix. The impacts of measures adopted by farmers are highly location- specific. The same scenario of reallocating irrigation water was found much less effective in Mahbubnagar, where further change in the crop- ping mix is apparently needed. Even greater disparities in impact and resilience can be expected at the farm and household level. Conclusions and Recommendations 67 One of the striking findings of the analysis was that a degree of varia- tion exists for drought impacts on different crops in different locations, clearly suggesting that there is a significant scope for advising farmers about undertaking drought-coping measures, such as switching to alterna- tive crops in response to a poor monsoon. Since the focus of this study was on linking the district and state-level impacts of drought, the data used in the report was aggregated from the block to the district level (and the total data for the eight districts was mostly used). However, the prototype risk assessment model developed for this study demonstrated good capability for a more disaggregated analysis (including testing a larger number of coping measures) that could be a useful tool to support the development of such plans. The results suggest that location-specific analyses are needed for informed decision making for the development of effective drought adap- tation plans for the affected areas. The analytical capability of the model can be further strengthened as discussed in the model development section above. The long-term impact of human-induced climate change reinforces the case for shifting to less water-intensive crops. Two scenarios of human- induced climate change, based on projections by widely accepted global and regional climate models, were simulated at the district level. While further investigation is needed, preliminary results suggest that climate change would further increase the benefits of shifting from rice to less water-intensive crops. The impact of drought on the overall state economy, measured as GVA, is modest and decreasing owing to structural changes in the Andhra Pradesh economy. The long-term AAL in GVA for the state owing to all drought events is estimated at 0.2 percent, even under the benchmark (business as usual) case. During the years of severe drought, an event that happens once in about 30­40 years in the eight districts, the loss in total GVA increases to 1.6 percent. The trend of the Andhra Pradesh economy over the past two decades has been a decrease in the contribution by the most vulnerable agriculture sector against an increasing contribution of the secondary and tertiary sectors. As this trend is most likely to contin- ue, the macroeconomic impact of drought will decrease further. The above findings are consistent with a growing body of evidence on the macroeconomic impact of climate-related disasters; based on worldwide historical data, the maximum annual impact of drought is estimated at 0.8 percent of GDP for a group of developing countries. 68 Overcoming Drought Facilitating an observed structural shift in the Andhra Pradesh economy from the agriculture sector toward the secondary and tertiary sectors could be a powerful macroeconomic drought adaptation strategy. The impact of such a shift on the economy's resilience to drought was examined through several scenarios in the macroeconometric model, corresponding to dif- ferent shares of the agriculture, livestock, secondary, and tertiary sectors in total GVA. The analysis revealed that loss in total GVA because of drought events could be decreased by 80 percentage points (for a scenario when the shares of the agriculture, secondary, and tertiary sectors roughly approximate the structure of the economy of Brazil). The loss in total GVA resulting from severe drought in the eight districts could be decreased from 1.6 to 0.2 percent.These encouraging signs in the average macrolevel indicators provide an opportunity for the state to more actively and effectively provide targeted assistance to those whose lives and well-being are devastated by drought. The impact of droughts is different for different sectors and this knowledge can help in developing sound Government policies. The macroeconometric model showed a significant negative impact of drought on the agricultural sector, a much more limited impact on the livestock sector and the secondary sector, and a positive impact on the tertiary sector. Interestingly, the livestock sector is less affected by drought than the secondary sector. Thus, the future impact of drought on the rural economy can be moderated by promoting the livestock sector, particularly poultry, which performed especially well during recent droughts. Furthermore, the tertiary sector appears to gain from drought events (with a one-year lag). Several factors associated with drought, such as GoI transfers, changes in consumption patterns and an increased supply of labor displaced from agriculture, may account for the observed boost in tertiary sector production. Loss of employment during drought remains a key concern. The agricul- tural sector is the major employment generator for the state. The agri- cultural employment coefficient for the state is 5.4, relatively higher than for other sectors. This implies that a 1-unit loss in output will result in more than 5 units of employment loss. Therefore, any external shock to the agricultural sector will have a strong impact on employment. The total employment loss for 2002­3 linked to the loss in the agricultural output owing to a major drought was estimated at more than 44 lakhs. This highlights the need for strategies that specifically target the most affected drought economic indicators, such as output and employment Conclusions and Recommendations 69 in the agriculture sector, and the most vulnerable districts, mandals, and communities. A number of opportunities outside of the agriculture sector can mitigate the impacts of drought on employment and income in the short to medium term. The options that arose from the analysis included the following: (a) in the service sector, significant employment potential is available in trade and transport (except railways); (b) investment in the construction sector would increase employment in this and related (cement, bricks, steel) industries; (c) the labor displaced from the agricultural sector can find employment in the mining and quarrying sectors, thereby mitigating the employment loss in the agricultural sector; and (d) the poultry sector may have good drought risk mitigation potential under local conditions, although all the factors accounting for its strong performance during recent droughts, as well as potential risks to farmers, need to be better understood. The impacts of droughts are highly differentiated and poor farmers/ households are most vulnerable. Droughts continue to have a negative impact on the performance of the agriculture sector and thus the lives of the millions of rural poor. A range of impacts were witnessed by a study that surveyed communities in Mahbubnagar, one of the poorest and worst affected districts. While some farmers/households are able to change farming practices or migrate to other sectors, others are forced into starvation, as well as loss of health and even life. These responses reinforce the need for tailored assistance to those in need and addressing immediate problems related to vulnerability of the poor to shocks. Areas for Future Action Develop a multitiered strategy combining statewide economic and sec- toral policies with well-targeted efforts at the micro level. Though drought is a complex and challenging natural phenomenon, it is an even more complex and challenging socioeconomic phenomenon with diverse, sometimes conflicting, impacts on the micro, sectoral, and macro levels. The analysis revealed the stark contrasts through which droughts mani- fest at different geographic levels, on different economic indicators, on different crops and sectors, on different population groups, and on differ- ent measures of human well-being. Thus, an effective strategy to tackle this phenomenon would need to deal with these multiple levels and dimensions in a balanced fashion. 70 Overcoming Drought A particular challenge, as always, is to effectively reach out to those poorest and most vulnerable.The reason is that more affluent farmers and households are typically better able to use alternative opportunities, including temporarily changing farming practices or migrating to other sectors, whereas poorer farmers are least resilient to shocks. While far from being exhaustive, this study highlighted some elements of a possible strategy for increasing resilience to drought through adaptation at different levels. Continue and accelerate the ongoing changes in the economic structure at the macro level. This can significantly contribute to increasing the resilience of the state economy and/or its people to drought in the long term, such as: · Facilitating growth of the tertiary sector; · Supporting the development of the livestock sector, particularly poul- try, as an important buffer to absorb the drought impacts on rural economy; · Encouraging the shift in cropping pattern from rice to less water-inten- sive crops would decrease vulnerability to drought impacts (including reviewing and addressing unfavorable incentives associated with current agricultural input subsides and rice procurement prices) Encourage investments in sectors with significant employment potential for the labor displaced from the agriculture sector, such as certain services (trade and transport), construction, mining, and quarrying subsectors. This would moderate the impact of drought on affected communities in the short to medium term. Address the growing gap between the encouraging macroeconomic trends and impacts on farmers and communities in drought-prone areas. The state economy is well poised to become less vulnerable to rainfall variabil- ity. Yet, the same, or possibly a larger number of people, who are (and will be for many years ahead) involved in agriculture, remain at risk of loss of livelihood and opportunities because of drought. Thus, it appears critical to intensify ongoing efforts and initiatives to promote more effective, targeted, and coordinated assistance to those in greatest need. Initiate development and implementation of drought adaptation plans for the most affected areas at the micro level (gram panchayat, water- shed, mandal). The focus should be on measures that promote a gradual Conclusions and Recommendations 71 shift to more sustainable agricultural practices (for example, changing cropping pattern in favor of less water-intensive crops) and other eco- nomic activities that are less vulnerable to drought (for example, live- stock) complemented by water conservation and watershed management activities. Given that the impact of these measures is medium to long term, the plans would also include short-term relief and safety net meas- ures that would help protect the nutritional, health, and educational needs of affected communities. In addition, special measures need to be considered for the poorest segments of the community, such as landless laborers and marginal farmers, who have the lowest adaptive capacity. The planning process ought to involve participatory approaches tai- lored to the needs of specific communities and help community members agree on a common strategy for securing stable and sustainable sources of income. This initiative should build upon the existing successful experi- ences with community-based watershed management in Andhra Pradesh, as well as integrate relevant schemes by different departments. Create a supporting institutional and policy framework. This planning and implementation process would require commitment from and involvement of all levels of Government (local to state to central), to pro- vide extensive technical assistance and other support mechanisms to farmers and communities. It would need to be supported by adequate institutional arrangements to deliver assistance to communities, an enabling policy framework, an aggressive awareness campaign, massive capacity building efforts for all key stakeholders, and innovative financial schemes that mitigate the risks and startup costs of transition to different crops, technologies, and economic activities. Explore innovative microfinancing/insurance schemes for farmers that promote a shift to more sustainable practices. Cost-effective risk mitiga- tion measures cannot fully protect farmers against drought risk. Risk financing arrangements can thus help farmers to absorb this residual risk. For example, since 2003 private insurance companies are providing rainfall insurance on a pilot basis. While such innovative risk financing arrange- ments offer farmers new opportunities to finance their losses, it is important to ensure that they do not perpetuate the current situation of heavy dependency of farmers on rainfall. A sizable average crop output loss because of drought, if there were no changes in the current agricultural practices, would make such insurance products unviable. Rather, new financing products should provide an incentive to permanently switch to 72 Overcoming Drought alternative, more sustainable agricultural and economic practices, such as less water-intensive crops (particularly high value cash crops), livestock, or some agro-processing activities. Developing contingent financing schemes that could facilitate this transitional drought adaptation process appears an important area for further work. Two possible innovative financing products are proposed by the study: · Drought adaptation insurance, which could provide coverage against risks caused by a shift from nonviable farming business to a viable (agri- cultural and nonagricultural) business. This insurance product would protect farmers against new sources of risks resulting from a change toward farming practices that are more drought resilient and less water intensive. · Drought adaptation credit, which could provide initial capital to shift to a long-term viable business. In the event of an unexpected loss caused by a failure in the adaptation investment, repayments may be postponed or (partially) forgiven. Develop a Decision Support Toolkit for drought management planning. The drought risk model developed by this study, complemented by other tools and methods (such as a real-life drought forecasting system devel- oped by the Central Research Institute for Dryland Agriculture), could provide a good scientific and information basis for supporting drought adaptation and management planning at different spatial levels. Facilitate informed public debate on drought adaptation strategies by assessing and disseminating information on the impacts and options. It is possible to quantify and conduct an objective assessment of economic losses caused by drought. This information needs to be more effectively disseminated to all the concerned stakeholders to assist with developing a common vision and reaching a broad-based agreement on the program of action. A N N E X 1 Programs Addressing Drought in Andhra Pradesh Risk Financing Programs Calamity Relief Fund This fund was established separately for each state based on recommen- dations of the IX Finance Commission and has since been approved for continuation by the X and XI Finance Commissions. This fund should be used for meeting the expenditure and for providing immediate relief to the victims of cyclone, drought, earthquake, fire, flood, and hailstorm. Table A1.1 describes the financial status of this fund over the past five years. National Calamity Contingency Fund This fund came into effect in 2000­1 and continued to be in operation until the end of the financial year, 2004­5. National calamities caused by cyclone, drought, earthquake, fire, flood, and hailstorm are considered to be of severe nature requiring expenditure by the state Government in excess of the balance available in its own Calamity Relief Fund to qualify for relief assistance under the National Calamity Contingency Fund scheme. The ini- tial corpus of the National Fund is Rs. 500 crores provided by the 73 74 Overcoming Drought Table A1.1. Calamity Relief Fund for Andhra Pradesh, 2000­5 (Rs in Lakhs) Share of 2000­1 2001­2 2002­3 2003­4 2004­5 Total Center 14854 15597 16377 17196 18056 82080 State 4951 5199 5459 5732 6019 27360 Total 19806 20796 21836 22928 24074 109440 Government of India. The National Centre for Calamity Management, constituted by the Ministry of HomeAffairs,Government of India, monitors the occurrence of natural calamities relating to cyclone, drought, earth- quake, fire, flood, and hailstorm on a regular basis and assesses their impact on area and the population. Assistance from the National Calamity Contingency Fund is only for immediate relief and rehabilitation.Any recon- struction of assets or restoration of damages is financed through plan funds. The unspent balance of the fund at the end of the financial year 2004­5 will become available to the GoI to be used as a resource for the next plan. Crop Insurance The National Agriculture Insurance Scheme was implemented in Andhra Pradesh in 1999­2000. A state has the discretion to identify the areas and crops to be covered. Once the specific area-crop combinations have been notified, participation is compulsory for farmers in those areas cultivating the specific crops and taking agricultural loans. For farmers who take loans the sum insured may be at least equal to the crop loan advanced.All farmers can insure up to the value of the threshold yield of the insured crop. Eighteen crops are currently insurable under this scheme during Kharif season (for example, rice, maize, sunflower, groundnut, sugarcane, and cot- ton) and 10 crops during Rabi season (for example, rice, maize, sunflower, and groundnut). The standard area yield insurance scheme has recently been extended to include farm income insurance and rainfall insurance. The XI Finance Commission noted the need to strengthen the National Agricultural Insurance Scheme to supplement other Government meas- ures to provide relief during a natural calamity. Drought-Proofing Programs Drought-Prone Areas Program The Drought-Prone Areas Program, a centrally sponsored scheme in operation since 1973, aims at restoring the ecological balance in the drought-prone areas and mitigation of the adverse effects of drought on Programs Addressing Drought in Andhra Pradesh 75 crops and livestock through integrated development of natural resources by adoption of appropriate technologies. The program aims to develop the drought-prone area with an objec- tive of drought proofing by conserving soil-land moisture, building water harvesting structures, and setting up afforestation and horticul- ture programs on a comprehensive micro watershed basis. So far, 3518 watersheds have been taken up covering 110 blocks in 12 districts cov- ering an area of 17.6 lakh ha. Almost 30 percent of the total watersheds under the program are located in Andhra Pradesh. A total of Rs. 507.57 crores were spent toward implementation of the program from 1995­6 to 2002­3. The central and state Governments share the expenditure for this program in the ratio of 75:25. Joint Forest Management/Community Forest Management The GoAP adopted the Joint Forest Management Program in 1992, which envisages a strategy for production, improvement, and develop- ment of forests with the involvement of local communities by forming Vana Samrakshana Samithies. There are 7090 such Samithies actively involved in the protection and development of forests in Andhra Pradesh. Water harvesting structures The Forest Department has taken up large-scale water conservation struc- tures in forest areas under the Neeru-Meeru (Water and You) Program. The structures include continuous contour trenches, check dams, rockfill dams, percolation tanks, and sunken gully pits. Seven phases of Neeru- Meeru have been completed. Water storage capacity of 1566 lakh cubic meters has been created in forest areas. Microirrigation Project GoAP launched a massive Microirrigation Project in 2003­4 throughout the state, with special emphasis on water-stressed mandals. The project envisages installation of sprinklers, drip irrigation, and rain guns to use the available underground irrigation water in the most efficient manner while improving productivity. The first phase aims to cover 2.50 lakh ha, with 50 percent state Government subsidy on the unit cost given to farmers. Andhra Pradesh Rural Livelihood Project This project provides critical support to the ongoing watershed movement in five drought-prone districts in Andhra Pradesh (Anantapur, Kurnool, Mahabubnagar, Nalgonda, and Prakasam). The mandate is to position 76 Overcoming Drought livelihood concerns strategically in watershed for the inclusion of women, the poor and the landless.The project, financed by the U.K. Department for International Development, will facilitate people-centered development inputs to the watershed program of the state including 500 new innovative watersheds, sustainable rural livelihood initiatives in 2000 watersheds, capacity building of various stakeholders, research and lesson learning for policy initiatives, and infrastructure support. Watershed development The national Bank for Agriculture and Development finances a watershed development fund. Owing to the watershed development program, the proportion of area under irrigation has increased by 19­29 percentage points among all households. Total employment has gone up by 11­29 percentage points. Yield rates have gone up for irrigated as well as nonirri- gated crops. Only 50 percent of the watersheds studied are economically viable in terms of incremental returns. The equity effect is not clearly known, though the impact on rich and medium households possessing lands seems higher. The drinking water situation has improved substan- tially and groundwater levels improved to a limited extent. Migration of labor decreased during the execution period, though in the majority of cases this was not sustained after the execution period. Household's preference for education increased and the role of women in financial matters improved substantially. Integrated Wastelands Development Program Rapid depletion of green cover and vast stretches of marginal lands lying fallow are causing enormous ecological imbalance. Productivity was neg- ligent because of soil erosion and marginalization of lands. To arrest this, a massive integrated wasteland development was undertaken in 1991 with 100 percent central assistance. The program is being implemented in 17 districts in Andhra Pradesh, with 38 projects covering an area of 362,985 ha. Rural infrastructure development A fiscal package has been developed for rural infrastructure development. In Andhra Pradesh the Department of Rural Development, Forest, Panchayat Raj, Minor Irrigation have availed this scheme. In this program, each district has selected certain villages for treatment. The implementa- tion at village level is through user groups formed based on the drainage Programs Addressing Drought in Andhra Pradesh 77 line.These groups decide the treatment of drainage line or common lands. The scheme excludes private land treatment. Employment Programs There are many other self-employment programs, based on income generation, to improve the livelihood of the affected population.These pro- grams are based on people's participatory approach. The GoAP has creat- ed various employment generation programs to eradicate poverty. While considering self-employment schemes the Government has given prior- ity to mini and micro enterprises. These programs can be considered as mitigation measures at the time of drought. Sampoorna Grameen Rozgar Yojana This has the following objectives: · To provide additional wage employment in all rural areas and thereby provide food security and improve nutritional levels · To create a durable community with social and economic assets and infrastructural development in rural areas The program provides wage employment to women, scheduled casts, scheduled tribes, and parents of children withdrawn from hazardous occu- pations. The works to be taken up must be labor-intensive, leading to the creation of additional wage employment, durable assets and infrastructure, particularly those which would assist in drought-proofing, such as soil and moisture conservation works, watershed development, and afforestation. Mission-based approach to employment generation The GoAP has established an Employment Generation Mission to coordinate activities of all the concerned departments in employment generation and manpower planning. The mission will prepare a time- bound action plan for implementation. The Government will act as facilitator and would identify and prioritize key sectors with employ- ment potential and ensure successful implementation. Empowerment of poor women The Self-Help Groups of Women (thrift groups) Program has mobilized and organized 48 lakh poor women in the rural areas into 3.7 lakh groups. 78 Overcoming Drought These women's groups have built a corpus fund of Rs. 750 crores consist- ing of their savings, borrowings from banks and the Development of Women and Children in Rural Areas revolving fund from the GoI. The empowerment process has enabled the Development of Women and Children in Rural Areas and thrift group members in addressing all dimensions of poverty. The movement has contributed to the augmenta- tion of incomes, improvement of nutrition, better child-care facilities for poor women, and enhanced status of women in rural households. A sim- ilar program for urban areas has now been started under the name of Development of Women and Children in Urban Areas, and 5,523 groups have been formed and developed in urban areas. Food for work programs These programs aim at helping millions of rural poor stave off hunger and unemployment. The basic principle of these programs is to provide employment to the poor during hard times, to create community assets through labor-intensive work, and to pay the laborers in kind (food grains or other food items). In 2005, the Government of India adopted the National Rural Guaranteed Employment Scheme, which guarantees 100 days per years of employment for community works in eligible districts, with the focus on natural and water resource management activities. Agriculture diversification initiatives These include programs by the Department of Agriculture to supply alternative crop seeds to farmers in anticipation of drought, and provide knowledge and other technical assistance for alternative, less water- intensive crops, technologies, and practices (for example, through onsite farmer schools), as well as, recently, price stabilization measures for maize and some other rain-fed crops by the Marketing Department. A N N E X 2 Definitions of Drought Drought is a normal, recurrent climatic feature. It occurs in almost all climatic zones, but its characteristics vary significantly from one region to another. Drought is a temporary aberration. It differs from aridity, which is restricted to low rainfall regions and is a permanent climatic feature. Drought results from a deficiency in precipitation that persists long enough to produce a serious hydrological imbalance. It should be consid- ered relative to long-term average condition of balance between precipita- tion and evapotranspiration (that is, evaporation and transpiration) in a particular area. Drought differs in three essential characteristics: intensity, duration, and spatial coverage. Intensity is the degree of shortfall in precip- itation and/or the severity of impacts associated with the shortfalls. It is generally measured by the departure from normal of a climatic index and is closely linked to duration in the determination of impact. Impact relates to the timing (for example, delays in the start of the rainy season, occurrence of rains in relation to principal crop growth staged) and effectiveness of rainfall (for example, number of rainfall events). Other climatic factors such as temperature, wind, and humidity can significantly aggravate its severity. Droughts are categorized as meteorological, hydrological, agricultural, and socioeconomic (Nagarajan 2003). See figure A2.1. 79 80 Overcoming Drought Figure A2.1. Definitions of Droughts natural climate change high temperature high winds, precipitation deficiency drought low relative humidity, greater (amount, intensity, timing) sunshine meteorological reduce infiltration, increased evaporation runoff, deep percolation, and transpiration and groundwater soil water deficiency al iculturr drought plant water stress, ag reduced biomass, and yield reduced streamflow, inflow to reservoirs, lakes, drought and ponds ydrologicalh economic impacts social impacts environmental impacts Source: National Drought Mitigation Center 2005. Meteorological drought: Defined usually as the degree of dryness (in com- parison to some normal or average amount) and the duration of the dry period. Another definition of meteorological drought is identified periods of drought for the number of days when precipitation was less than pre- determined thresholds. The India Meteorological Department uses a meteorological definition of drought based only on rainfall deficiency from the mean annual, mean summer monsoon, mean monthly, and mean weekly rainfall. Thus, meteorological drought is said to occur when Definitions of Drought 81 the seasonal rainfall received over an area is less than 75 percent of its long-term average value. It is further classified as moderate drought if the rainfall deficit is 26­50 percent and severe drought when the deficit exceeds 50 percent of the normal. A year is considered a drought year if the area affected by the drought is more than 20 percent of the total area of the country. Hydrological drought: Associated with the impacts of precipitation short- falls on surface or subsurface water supply (for example, stream flow, reservoir and lake levels, and groundwater).The frequency and severity of hydrological drought is often defined on a watershed or river basin scale. Hydrological droughts are usually occur out of phase with meteorologi- cal droughts. Water in hydrological storage systems (for example, reser- voirs and rivers) is often used for multiple and competing purposes (for example, flood control, irrigation, recreation, and hydropower), further complicating the sequence and quantification of impacts. Although cli- mate is the primary contributor to hydrological drought, other factors such as changes in deforestation, land degradation, and the construction of dams all affect the hydrological system. Agricultural drought: Links various characteristics of meteorological and hydrological droughts to agricultural impacts. It is related to precipita- tion shortages, differences between actual and potential evapotranspira- tion, soil water deficits, and so on. Plant water requirements depend on prevailing weather conditions, biological characteristics of the specific plant, its stage of growth, and the physical and biological properties of the soil. Agricultural drought should be able to account for the variable susceptibility of crops during different stages of development, from emergence to maturity. Deficient topsoil moisture at planting may hin- der germination, leading to low plant populations per hectare and thereby decrease in final yield. Socioeconomic drought: Associated with the supply and demand of eco- nomic goods, such as water, forage, food grains, fish, and hydroelectric power. Socioeconomic drought occurs when the demand for an eco- nomic good exceeds supply due to a water-related shortfall in water supply. A N N E X 3 Stochastic Crop Production Loss Model The stochastic crop production loss model comprises three submod- ules: the stochastic hazard module, the vulnerability and exposure module, and the loss module. Stochastic Hazard Module Weather is simulated for 500 years using historical weather parameters for each block with the WXGEN weather simulator (annex 4). Each of the simulated events (years) is assigned a drought category based on the SPI derived from cumulative seasonal (June­December) rainfall (McKee et al. 1993). The SPI computation for a specific time scale and location requires an historical record of 30 or more years of monthly precipitation. The long- term record is fitted to a gamma probability distribution and is then trans- formed into a normal distribution (that is, SPI), which, by definition, has zero mean and unit variance. McKee et al. (1993) defined drought as an event that occurs if the SPI value is ­0.1 or less, while in the present study the SPI value of ­0.5 or less was used to define a drought event. The drought categories as defined in this study are presented in table A3.1. 83 84 Overcoming Drought Table A3.1. Drought Categories Seasonal SPI Drought category 0.5 to ­0.5 Normal year ­0.5 to ­1.0 Minor drought ­1.0 to ­2.0 Moderate drought ­2.0 to ­3.0 Severe drought ­3.0 and below Extreme drought Figure A3.1.Validation of SPI for Anantapur Anantapur district EP (historical vs. modeled) 0.7 0.6 0.5 historic EP 0.4 modeled EP EP 0.3 0.2 0.1 0.0 ­4 ­3 ­2 ­1 0 1 2 3 SPI An SPI value greater than 0.5 is assumed to represent a year with excess rainfall. SPI is computed for each year at the district and state (eight districts combined) levels by aggregating cumulative seasonal rain- fall for all the blocks in the district or state. For aggregation, the simple average of the rainfall of all the blocks is used. Figure A3.1 shows the results of validating the modeled SPI index with historical data for Anantapur. The rate of occurrence for any event in the 500-event set is assumed to be 1 per 500. Thus, the frequency of a given category of drought among the simulated events is computed as the number of events of the particular category divided by 500. Vulnerability and Exposure Module In the vulnerability and exposure module the average yield and planting area associated with each of the simulated events is determined for each crop at block level. Stochastic Crop Production Loss Model 85 The average yield of five crops (jowar, maize, groundnut, sunflower, and rice) associated with each block and category of drought is determined with the help of EPIC model. EPIC runs are made at block level for selected events (10 numbers) representing different categories of drought. The events are selected from the 500-year event set for every block to represent each of the drought categories based on a representative SPI value. The planting area model computes the planting area associated with each crop and for each of the 500 events at block level based on annual rainfall corresponding to the event. Loss Module In the loss module, the crop-wise loss in production associated with the four categories of drought is computed at the block, district, and state (eight districts combined) levels. For each crop, production is computed for each of the 500 events at the block level as production = planted area x average yield. Block-level production is then summed up to get the production at district and state levels. The average normal production at district/state level is computed as the simple average of production for all the events categorized as nor- mal year in the 500-year event set at the corresponding (district or state) level. Percentage loss in production for each event and crop is then calcu- lated at the required level as: % loss in production = 100 x (average normal year production ­ production for the event)/average normal year production A N N E X 4 WXGEN in EPIC WXGEN in EPIC is based on Richardson's (1981) description.The model generates daily values for precipitation, maximum and minimum temper- ature, solar radiation, wind speed, and wind direction for any number of years for a location (Sharpley and Williams 1990). The weather generator in EPIC is designed to preserve the dependence in time, internal correla- tion, and the seasonal characteristics that exist in actual weather data. Precipitation and wind are generated independent of the other variables. Maximum temperature, minimum temperature, and solar radiation are generated depending on whether the day is wet or dry. A first-order Markov chain is used to generate the occurrence of wet or dry days. For a wet day the precipitation amount is generated based on skewed normal distribution.With the first-order Markov chain model the probability of rain on a given day is conditioned on the wet or dry status of the previous day. The procedure to generate the daily values of maximum and mini- mum temperature and solar radiation is based on the weekly stationery generating process of Matalas (1967).The wind component of the model provides for generating daily values of wind speed and direction as described by Richardson (1982). 87 88 Overcoming Drought EPIC can be run using either historical weather data or it can generate some or all of the data via its weather generator. The weather generator parameters have to be based on historical data for the study area. Within the study area, historical daily rainfall data were available at the block level, but other meteorological data were available at district level. The weather generator parameters at the block level are computed using daily rainfall data at block level and other daily weather data at district level. WXGEN was evaluated for its effectiveness in simulating Indian weather conditions using 20 years of daily weather data from India Meteorological Department stations at the Anantapur and Mahbubnagar districts. The weather generator model parameters for the simulations were calculated from 20 years of daily weather data for Anantapur and Mahbubnagar. By using the weather generator model parameters, daily weather was simulated for 50, 100, and 200 years for the two stations. A chi-square test was then performed to test whether the simulated monthly means of rainfall, air temperature, and relative humidity are significantly different from those derived from historical records. It was found that simulation convergence is achieved after simulating 200 years of weather and the null hypothesis was rejected at 5 percent level of significance for rainfall and 1 percent for the other parameters. Consistency among the daily parameters is also verified by performing a t-test for regression coefficients between simulated and historical daily values. The test results showed that the model preserved the internal cor- relation among the variables with a 1 percent level of significance. The WXGEN was recommended for weather generation in this study, based on these results. There are 450 blocks in the eight selected districts. Data were available for all the blocks up to 2003. Rainfall data were available for 1988­2003 for 60 percent of the blocks. For the remaining blocks, rainfall data were available before 1988, going as far back as 1963.Wherever data were avail- able prior to 1988, the model parameters for those blocks were computed using data before 1988. However, block-level rainfall data were used only for simulation of weather, which required other parameters (temperature, humidity, and so on) as well. Other parameters were taken from district- level India Meteorological Department data available for 1973­99. Thus, the final period used in simulation of weather at block level was con- strained by rainfall at block level and other parameters at district level. WXGEN in EPIC 89 Block-level historical daily rainfall data available for the study region were subjected to a number of quality control procedures before comput- ing the block-level rainfall simulation parameters. The spatial and tempo- ral consistency were checked and suspected data points were replaced with nearby stations data. Missing years and daily gaps in the data were not used for the parameter computation. For computation of return periods of droughts at district level, rainfall data at district level supplied by Directorate of Economics and Statistics for 1973­2003 for all eight districts were used. A N N E X 5 EPIC: Crop Growth Model EPIC is a mathematical model that operates on a daily time step to simu- late evapotranspiration, soil temperature, crop potential growth, growth constraints (water stress, stress due to high or low temperature, and nitro- gen and phosphorous stress), and yield. EPIC uses a single model for simu- lating all crops. Each crop has unique values for model parameters, which can be adjusted or created by the user if needed provided they have enough knowledge of the crop and model operation. The crop growth model uses radiation-use efficiency in calculating photosynthetic production of bio- mass. The potential is adjusted daily for stress from water, temperature, nutrients (nitrogen and phosphorous), aeration, and radiation. Crop yields are estimated using the harvest index concept. Harvest index increases as a nonlinear function of heat units from zero at the planting stage to maxi- mum value at maturity. The harvest index may be decreased by high temperature, low solar radiation, or water stress during critical crop stages. Input Data Two kinds of standard data sets are required as EPIC inputs files: · Basic input about miscellaneous field information, such as climate data, soil data, and management information 91 92 Overcoming Drought · Growth and fertilizer parameter file. The parameters for most of the major crops have been established by the developers and do not need to be modified. The climate variables, the soil physical properties, and the management information required by the EPIC model are described below. EPIC uses a stochastic weather generator to generate daily weather from monthly maximum and minimum temperatures, precipitation, stan- dard deviation of precipitation, skew coefficient for daily precipitation, probability of a wet day after dry day, and probability of wet day after wet day. EPIC can accept up to 20 parameters for 10 soil layers. However, of these at least seven parameters are required: depth, percent sand, percent silt, bulk density, pH, percent organic carbon, and percent calcium car- bonate. Other related soil parameters can be estimated by EPIC itself. EPIC requires detailed descriptions of management practices. These descriptions must specify the timing of individual operations either by date or by fraction of the growth period (that is, by heat units). EPIC allows the user to simulate complex crop rotations by specifying options for irrigation and fertilizer applications in the EPIC program; the applica- tions can be made manually or automatically based on rules. Major Crop Growing Seasons Kharif (June­October) is characterized by a gradual fall in temperature, more numerous cloudy days, low light intensity, a gradual shortening of photoperiod, high relative humidity, and cyclonic weather. During Rabi (November­March), there is a gradual rise in temperature, bright sunshine, near absence of cloudy days, a gradual lengthening of the photoperiod, and lowering of relative humidity. Rabi weather is more conducive for rice, and, in general, rice yields are higher in Rabi than in Kharif season. However, this does not hold true for maize. Unlike jowar and sunflower, which are mostly Kharif crops, maize can be grown in any climate. Groundnut is raised both in Kharif and Rabi but crop acreage is more in Kharif season. Of the above, validation for five crops could not be done to compare/validate sunflower yield owing to unavailability of reported yield data at the district level. Model Testing Rigorous testing of model performance involves the comparison of reported crop yield to model outputs achieved with conditions similar to EPIC: Crop Growth Model 93 those prevailing for crop growth in the real world. Parameterization is often an iterative process of comparison of the model and observed data and parameter adjustment to achieve a better match. Once the crop model is adjusted, the next step in testing the model is validation against independent data. In this study, observed yields of crops were compared with model outputs for the same crop for the Kharif season. However, because economic data are often gathered in political reporting districts, rather than agro-climatic areas, current results of crop productivity were compared to political districts to ensure that the same can be used as input to economic models. Parameterizing EPIC Most of the crop growth parameter sets in EPIC are developed for temperate conditions. They produce high-yield estimates as the poten- tial radiation use efficiency is less in tropical conditions (that is, under Indian conditions).Accordingly, potential radiation use efficiency values (WA; also called biomass-energy ratio) have been lowered from 35 to 10, for groundnut and sunflower to match Indian data. The base temperature (TG) was changed to 8oC from 13.5oC for groundnut (Reddy et al. 2001). Water stress decreases yield in EPIC by reducing accumulated biomass and the harvest index. Water stress is only allowed to reduce the harvest index over the later portion of the growing season. Harvest index was low- ered from 0.40 to 0.30 for groundnut and sunflower to better match field trials in India (ANGRAU 2003). This is consistent with findings that many crops have developmental phases (such as pollination and grain filling) in which water stress is critical. Each of the crop-specific parameters (more than 30 in number) helps compute the growing season water stress through accumulated biomass reduction in the form of mid- and late-season water stress reductions in crop yield (Bryant et al. 1992). Validation Validations at the district level were carried out using block-level simu- lated outputs for the years 1996­8 and annual reported yields for the selected five crops (rice, maize, jowar, groundnut, and sunflower). The validation was done using only Kharif-simulated crop yield for the Kharif season and compared with annual (Kharif + Rabi) reported yields (the only data available). Crops other than rice are often grown in the Rabi season in Andhra Pradesh under drought conditions and with irrigation 94 Overcoming Drought supplements. Since the extent of irrigation for particular crops was not known, models of crop yield in the Rabi season could not be expected to match the reported data. Nevertheless, the validation test is still powerful since a predominance of annual yield is derived from the Kharif season. For instance, statistical analysis of the crop-growing region shows that in the Anantapur district of Andhra Pradesh the planted area in the Kharif versus Rabi season were 2.7 times for rice, 3 times for maize, and 41 times for groundnut. A N N E X 6 Planting Area Model To estimate crop production, both the year-to-year variation in yield and area planted are needed. A statistical model was developed to compute area of each modeled crop (and a category or "other crops") at the block- level based on the rainfall of that district. The following data sets were available for building the planting area model: · District-wise cropped areas for major crops, gross cultivated area (GCA), gross irrigated area (GIA), and cropped areas by source of irrigation were available from 1975­6 to 2002­3 · Cropped areas by season and block are available from 1999­2000 to 2003­4 · Daily, monthly, and annual rainfall for India Meteorological Department stations (1975­2002). Crop-wise rain-fed areas and gross rain-fed area (GrfA) were calcu- lated as the difference of crop-wise gross areas and irrigated areas. The planting area model aimed to make use of available data in build- ing the statistical relationship among the variables. Many combinations of variables were tested to answer the following queries. Dependent 95 96 Overcoming Drought variables were rainfall (absolute, lag rainfall, log transformed rainfall, per- cent change in rainfall) and independent variables were GCA, GIA, and individual cropped areas. One dependent variable was taken at a time to build the model. The model intended to: · Find any significant relation with previous year's rainfall among areas · Test any correlation of previous year's monsoon first month rainfall (June) with the planted areas · Test whether crop areas are related with the monsoon strike date · Check whether change in areas is related with the change in rainfall To answer the above, the following combination of variables was used in building the planting area regression models. · GCA,GIA,GrfA versus previous year annual rainfall (log transformation) · GCA, GIA, GrfA versus previous year June rainfall (absolute and log transformation) · GCA, GIA, GrfA versus current year monsoon strike date · Change in GCA, GIA, and GrfA over previous year with change in rainfall over previous year Unfortunately, none of the first three versions gave satisfactory results. There was a lot of scatter and no significant correlation in the first three combinations. The fourth combination gave a good correla- tion structure among the variables. Hence, it is proposed as a candidate for the planting area model. Results of these plots for Anantapur are shown in figures A6.1­A6.8. Similar patterns have been observed for other districts. Two linear regression equations, separately for irrigated and GCAs regressed on rainfall, were developed based on district level data from 1975­6 to 2002­3. The GrfA is determined as the difference between GCA and GIA. However, the variables represent their percent changes instead of absolute values.The percent change in rainfall explains the per- cent change in area better than their absolute values would. Taking any known year/event as the base the equations can give percent changes for the simulated year/event of interest. The percent change in rainfall for each of the stochastic rainfall regression equations are used to calculate the change in GCA and GIA, then absolute areas are calculated with ref- erence to areas in 1998­9. Planting Area Model 97 Distribution factors by crop by season and by block are derived using the 2000­1 and 2003­4 block-level data separately for irrigated and rain- fed areas. The structure of the model is given in figure A6.8. The district cropped area is then disaggregated using the distribution factors. Based on the consistent cropping pattern observed in the four-year block-level data (although limited) an assumption is made that the same pattern would continue in the future. Figure A6.1. Log of Lag RainVersus Log of Lag Gross Rain-Fed Areas for Anantapur 13.80 rain-rainfed 13.75 poly. (rain-rainfed) 13.70 13.65 area 13.60 ed 13.55 rain-f 13.50 g la13.45 13.40 y = ­0.5535x2 + 7.0242x ­ 8.6647 13.35 R2 = 0.0709 13.30 5.8 5.9 6 6.1 6.2 6.3 6.4 6.5 6.6 6.7 lag rain Figure A6.2. PreviousYear June Rain DeviationsVersus CurrentYear Gross Rain-Fed Area Deviation for Anantapur 30 (%).v 20 de y = ­0.0106x + 0.0122 R2 = 0.0139 10 area ed 0 ­150 ­100 ­50 0 50 100 150 200 250 300 350 400 rain-f ­10 arey.v ­20 pre ­30 prev. year June rain dev. (%) 98 Overcoming Drought Figure A6.3. LogTransformed CurrentYear RainVersus Gross Rain-Fed Area for Anantapur 13.80 13.75 log rain linear (log rain) 13.70 13.65 area 13.60 ed 13.55 rain-f 13.50 log 13.45 13.40 y = 0.2332x + 12.116 13.35 R2 = 0.2991 13.30 5.8 5.9 6.0 6.1 6.2 6.3 6.4 6.5 6.6 6.7 log rain Figure A6.4. Gross Cropped Area and Monsoon Strike DateVariation Over theYears for Anantapur 1,200,000 17-Aug 28-Jul 1,100,000 8-Jul 1,000,000 area 18-Jun 900,000 oss 29-May gr MonsDate 800,000 9-May GCA 700,000 19-Apr MonsDate 600,000 30-Mar 1975­761977­781979­801981­821983­841985­861987­881989­901991­921993­941995­961997­981999­002001­02 year Planting Area Model 99 Figure A6.5. Change in Gross Irrigated AreaVersus Change in Annual Rain for Anantapur 25 chIA 20 linear (chIA) 15 10 (%) 5 IA in 0 hc ­70 ­50 ­30 ­10 10 30 50 70 ­5 ­10 y = 0.3129x ­ 0.0177 ­15 R2 = 0.7981 ­20 ­25 ch rain (%) Figure A6.6. Change in Gross Cropped AreaVersus Change in Annual Rain for Anantapur 20 ChGCA linear (ChGCA) 15 10 (%) 5 GCA 0 in.hc ­70 ­50 ­30 ­10 10 30 50 70 ­5 y = 0.166x ­ 0.0054 ­10 R2 = 0.7114 ­15 ch. rain (%) 100 Overcoming Drought Figure A6.7. Change in Gross Rain-Fed AreaVersus Change in Annual Rain for Anantapur 20 chGRf linear (chGRf) 15 y = 0.1334x ­ 0.0012 R2 = 0.5202 10 (%) GRfA 5 in e 0 anghc ­70 ­50 ­30 ­10 10 30 50 70 ­5 ­10 ch. rain (%) Figure A6.8. Structure of the Planting Area Model total cropped area of district irrigated rain-fed C1 C2 C3 C4 C5 C6 K R b l o C1, C2 ...crops c K - kharif k R - rabi s A N N E X 7 Input-Output Analysis for Andhra Pradesh The Central Statistical Organization has been constructing input-output (I-O) tables in India since 1973­4 at an interval of about five years. The latest (unpublished) table relates to the year 1998­9. The published tables for the years 1989­90 and 1993­4 are for 115 sectors of the econ- omy while for earlier years these relate to 60 sectors only. At the state level, however, I-O tables are not prepared on a regular basis. Some states, such as Punjab, Uttar Pradesh Assam, and Haryana have prepared tables only once. The table for Andhra Pradesh for the present analysis was constructed for the year 1998­9 using the all-India I-O table along with the available data at state level. This table has been extended for 2000­3. The first section gives details about the sources of data and the methodology. The second section provides the comparative analysis of the constructed I-O tables for the state. Data and Methodology The all-India I-O table is available for 1998­9. The I-O table for Andhra Pradesh was prepared in terms of their coefficients for all India, along 101 102 Overcoming Drought with directly available details of crop-wise VOPs for agriculture, the GVA and VOP for livestock, forestry and logging and fishing, mining and quarrying, construction, GVA estimates for registered and unregistered manufacturing (2-digit level), and directly available GVA estimates for various services. Andhra Pradesh economy has been broadly classified into the follow- ing sectors: · Agriculture · Rice · Jowar · Maize · Other food grains · Groundnut · Other crops · Livestock · Forestry and Logging · Fishing · Mining and Quarrying · Manufacturing · Food products · Textile products · Wood products · Paper products · Leather products · Rubber, plastic, coal, and tar and petroleum products · Fertilizers · Pesticide · Chemicals · Nonmetallic mineral products · Basic metals and alloys · Metal products and electrical and nonelectrical machinery and equipment · Transport equipment and parts · Miscellaneous industry · Construction · Services · Electricity, gas, and water supply · Railway transport services · Other sources of transport and storage Input-Output Analysis for Andhra Pradesh 103 · Communication · Trade hotels and restaurants · Banking and insurance · Ownership of dwellings, real estate, and business services · Education, medical, and other services · Public administration Agriculture The VOP data for all the crops (1998­9) were obtained from the Directorate of Economics and Statistics, Andhra Pradesh. The values were aggregated to the six sectors mentioned above. Certain adjustments were made in some sectors (such as rice/rice) to take into account the milling, because in the all-India tables, milling is included under agricul- ture and adjustment was done using the ratios based on the all-India table. The National Accounts Statistics figures for agriculture and from the I-O table (all-India) were considered. The ratio between I-O figures and National Accounts Statistics figures were used for adjustment. These adjusted figures (VOP data) were used along with all-India coefficients to calculate the Andhra Pradesh input vectors for the six agricultural sectors.Adjustments were made for inputs of fertilizers, elec- tricity, and petroleum products consistent with the estimates provided by the Directorate of Economics and Statistics. The VOP data for 2002­3 were available for rice, jowar, maize, other food grains, and the total VOP for all the crops from the Directorate of Economics and Statistics. The VOP of groundnut for 2002­3 was calculated by using the out- put/production data for 2002­3 and the wholesale price of groundnut for that year. The production figure multiplied with the price data determined the VOP of groundnut for 2002­3. The VOP of other crops is obtained by subtracting the values of output of the first five sectors from the total VOP of all the crops made available by the Directorate of Economics and Statistics. The sector-wise ratio of VOP for 2002­3 over 1998­9 was used over I-O 1998­9 VOP to get the I­O 2002­3 VOP. Thus, the structure of the inputs for the agricultural part of the I-O matrix for Andhra Pradesh for 2002­3 at 1998­9 prices was constructed. The adjusted GVA and VOP data for the six sectors for 2002­3 were converted to 1998­9 constant prices using the price indices. The total inputs for each of these sectors were calculated by deducting the GVA from the VOP. All the calculations were done on data of 2002­3 at 104 Overcoming Drought 1998­9 constant prices.These total inputs were distributed to sector-wise inputs by using the Andhra Pradesh 1998­9 coefficients. Livestock, forestry and logging, fishing, and mining and quarrying The GVA and VOP data for the livestock, forestry and logging, fishing, and mining and quarrying sector for Andhra Pradesh were obtained from the Directorate of Economics and Statistics. The difference between the VOP and the GVA provided the inputs to these sectors. The inputs were then distributed using the all-India coefficients.Thus, the input structure for these sectors for Andhra Pradesh for 1998­9 was obtained by using the all-India structure. The GVA and VOP for livestock, forestry and logging, fishing, and min- ing and quarrying sectors for 2002­3 were available both at current and constant 1993­4 prices. The GVA and VOP for 2002­3 (at current prices) were converted to 1998­9 constant prices using the price indices.These fig- ures were used to calculate the inputs for each of these sectors, which were distributed using the I-O 1998­9 coefficients. Thus, the input structure for livestock, forestry and logging, fishing, and mining and quarrying sectors for Andhra Pradesh for 2002­3 at constant 1998­9 prices was derived. Manufacturing sector The manufacturing sector at the 2-digit level for all-India as well as for Andhra Pradesh was classified under the previously mentioned heads. The detailed data of GVA for both registered and unregistered fac- tories were available from the Directorate of Economics and Statistics, Andhra Pradesh, for 1998­9 at current prices at 2-digit level of indus- trial classification. The Annual Survey of Industries for the registered1 manufacturing sector provides the GVA and VOP for different sectors of the economy at state level. The unregistered sector estimates of GVA and the VOP were obtained from the survey conducted by the National Sample Survey Organization with reference to 2000­01. It is assumed that the ratio of VOP to GVA for 1998­9 and 2000­1, are the same for the unregistered sector. Using the ratio of VOP to GVA from these sources on the GVA provided by the Directorate of Economics and Statistics, VOPs were obtained separately for registered and unreg- istered parts of different sectors for 1998­9 and then added. These adjusted figures were used for the calculation of the value of inputs (VOI). VOI = VOP ­ GVA Input-Output Analysis for Andhra Pradesh 105 The VOI for 1998­9 at current prices for the sectors obtained were at the purchaser's prices. The all-India I-O matrix gave the corresponding VOI at the purchaser's prices as well as their sectoral distribution. The input structure of 1998­9 for the manufacturing sector for Andhra Pradesh was obtained by assuming the all-India input structure for the dif- ferent sectors. This was done because the detailed data of inputs were not available at state level. This assumed that the structure of inputs for the different sectors is the same for the state and all of India. The detailed data of GVA for both registered and unregistered facto- ries for all the above sectors were available from the Directorate of Economics and Statistics,Andhra Pradesh, for 2002­3 both at current and constant 1993­4 prices. The GVA data thus obtained was converted to GVA at 1998­9 constant prices. The values of output as well as input structure for different sectors under manufacturing were estimated by using the 1998­99 I-O Andhra Pradesh structure on the 2002­3 GVA at 1998­9 prices. Hence the input structure of the manufacturing sector for the year 2002­3 at constant 1998­9 prices was constructed. Services The GVA data for the services sector for 1998­9 for Andhra Pradesh were obtained from the Directorate of Economics and Statistics, Andhra Pradesh, and data at the all-India level were available from the I-O 1998­9 matrix (all-India). The ratio of Andhra Pradesh to all India was computed and then used to adjust the all-India I-O structure for the serv- ices sector. The input structure for the services sector for Andhra Pradesh was thus obtained by using the all-India structure. The services sector GVA growth rate from 1998­9 to 2002­3 was obtained from the constructed I­O 1998­9 table and the Directorate of Economics and Statistics data were used on the I-O 1998­9 VOP data and the VOP for the year 2002­3 was calculated. The 2002­3 current price VOP data was converted to constant 1998-9 prices using the index. The inputs at 1998­9 constant prices were calculated using the converted GVA and VOP data. These inputs were distributed to different sectors using the 1998­9 I-O coefficients. Hence, the input structure for the serv- ices sector of I-O 2002­3 at constant 1998­9 prices was constructed. Final demand The various components of final demand are private consumption expen- diture, Government current expenditure, gross fixed capital formation, change in stocks, and exports and imports. 106 Overcoming Drought Private final consumption expenditure At all-India level the private consumption expenditure is obtained by using the commodity-flows approach. This method cannot be applied at the state level because of the openness of the economy. National Sample Survey conducts quinquennial household surveys on consumption expen- diture. Detailed item-wise estimates of consumption expenditure for 1999­2000 are available at state level and also for the country as a whole. There is a vast difference between the all-India estimates obtained from the National Sample Survey and those obtained by the commodity flow approach. For this study, the sector-wise ratios of National Sample Survey expenditure for Andhra Pradesh to that of all-India were applied to the all- India estimates given in the all-India I-O table for the year 1998­9. It is assumed that the degree of difference between National Sample Survey and the commodity flow approach estimates for Andhra Pradesh is the same as for all-India. The private final consumption expenditure is different from household consumption and includes the expenditure of nonprofit institutions serv- ing households, while the National Sample Survey estimates are only for the households. The estimates based on household survey are different from those given by the I-O table (using the commodity flow approach), and since the estimates in the present study should be consistent with the I-O table, this method was used. For 2002­3, the per capita expenditure estimates were available sep- arately for rural and urban areas of Andhra Pradesh by broad groups of items available from the National Sample Survey on consumption expenditure. These along with the corresponding population estimates were used to estimate the total expenditure for the same groups of items. Similar estimates for 1998­9 were obtained using the 1998­9 National Sample Survey. From these two sets of estimates the indices of sector-wise growth of expenditure in 2002­3 over 1998­9 was obtained. These indices were applied on the 1998­9 private final consumption expenditure estimates to get similar estimates for 2002­3. It is assumed here that the growth rate under a group remained the same. Gross fixed capital formation Directorate of Economics and Statistics provided the total gross fixed capital formation (GFCF) estimates for 1998­9. GFCF of construction was estimated by subtracting the row total of construction (that is, inter- mediate use in the form of repair and maintenance) from the VOP. Input-Output Analysis for Andhra Pradesh 107 Capital formation under livestock was obtained by using the ratio of increment in livestock in Andhra Pradesh and increment in all-India livestock to the capital formation in animal husbandry at all-India level. The estimate of the value of GFCF for rest of the sectors of Andhra Pradesh for 1998­9 was obtained by subtracting the estimates of GFCF in livestock and construction from the total GFCF for Andhra Pradesh. Similar estimates were obtained from the all India table I-O 1998­9. The ratio of these two estimates was used to obtain sector-wise estimates of GFCF for the remaining sectors. Here the assumption was that the distri- bution of GFCF of machinery sectors is the same for all-India and Andhra Pradesh. The total GFCF for the year 2002­3 at current prices was made avail- able by the Directorate of Economics and Statistics. The GFCF data for the sectors for 2002­3 was obtained by using the growth index over the 1998­9 GFCF data. The GFCF data obtained at current 2002­3 prices was then converted to constant 1998­9 prices using the sector-wise price indices. Government final consumption expenditure Government final consumption expenditure (GFCE) data for the state for the year 2001­02 to 2003­4 was available from "Economic cum Purpose Classification" of the Andhra Pradesh Government budget 2003­4. The growth rates for the above-mentioned years were used to calculate the GFCE for 1998­9 for the state. The ratio of the state to all-India GFCE for 1998­9 was used on the all-India I-OGFCE sector- wise estimates to get the estimates of GFCE for the state for the year 1998­9. The total GFCE 2002­3 for the state was available from the above- mentioned source. The growth index of GFCE was used on the 1998­9 sector-wise GFCE figures to get the GFCE for all the sectors for the year 2002­3 at current prices. These figures were then converted to 1998­9 constant prices using the sector-wise price indices. Imports and Exports The imports and exports for the sectors for the years 1998­9 and 2002­3 were obtained by subtracting GFCE, private final consumption expenditure, GFCF, and the intermediate use (row total) from the VOP of each sector. The change in stocks was not considered because of non- availability of data. 108 Overcoming Drought Employment coefficients This provides the number of workers required to produce Rs. 1 lakh VOP. Employment coefficients were used to calculate employment multipliers that will capture the total employment change scenarios. The employment coefficients for the manufacturing sector of Andhra Pradesh were calculated from employment data for the registered man- ufacturing sector provided by Annual Survey of Industries and the unregistered sector employment data from National Sample Survey. The agricultural employment figures available at the state level were used to derive the coefficients for the agricultural sector. It was assumed that the coefficients for all the crops are the same. Since data for other sectors were not available, estimates based on the all- India employment data available from the Central Statistical Organization and VOPs from the all-India I-O table were used. Multipliers This is a quantitative expression of the extent to which some initial, "exogenous" force or change is expected to generate additional effects through interdependencies associated with some assumed and/or empir- ically established, "endogenous" linkage system. Two types of multipliers were considered: output and employment. The constructed I-O matrix has been used to calculate the multipliers: A: I-O matrix expressed in terms of their coefficients I: Identity matrix of the same order as A. R = (I ­ A)­1 The R matrix is of the same order as A, and the column total for each of the sectors in the R matrix gave the output multipliers for those sectors. Table A7.1 presents the employment coefficients and output multi- pliers calculated from the I-O matrix for the year 1998­9. The employ- ment coefficients are quite high for the agricultural sector implying that this sector is the major employment generator for the state. Thus, any external shock to the agricultural sector has a direct impact on the state's employment scenario. The output multiplier for rice/rice showed that one unit (lakh) increase in the final demand of rice/rice results in increase of 1.46 units (lakhs) of gross output in the economy. The out- put multiplier is highest for metal products and electric and nonelectric Input-Output Analysis for Andhra Pradesh 109 Table A7.1. Employment Coefficients and Output Multipliers Employment Employment Sectors coefficients Multipliers Sectors coefficients Multipliers Rice 5.40 1.45 Pesticides 0.89 2.61 Jowar 5.40 1.43 Chemicals 0.89 1.82 Maize 5.40 1.45 Nonmetallic 0.95 1.95 mineral products Other food grains 5.40 1.52 Basic metals and 0.05 2.54 alloys Groundnut 5.40 1.40 Metal products, 0.06 2.67 electricity and nonelectric machinery and equipment Other crops 5.40 1.22 Transport 0.59 2.10 equipment and parts Livestock 5.40 1.42 Miscellaneous 2.20 Forestry 1.44 1.17 Construction 0.86 1.69 and logging Fishing 0.68 1.25 Electricity, gas, 0.08 2.20 and water supply Mining and quarrying 0.47 1.41 Railway transport 0.32 2.00 services Food products 1.01 2.23 Other 0.62 2.00 transportation and storage Textile products 3.15 2.08 Communication 0.41 1.27 Wood products 9.27 1.61 Trade, hotels, 1.29 1.45 and restaurants Paper products 0.37 2.17 Banking 0.16 1.36 and insurance Leather products 1.14 2.11 Ownership 0.02 1.12 of dwellings, real estate, and business services Rubber, plastic, coal, and tar 0.45 2.12 Education, 1.09 1.79 medical, and other services Fertilizers 0.89 1.09 Public 0.92 -- administration 110 Overcoming Drought machinery and equipment followed by pesticides. Sectors, such as basic metal and alloys, electricity, gas, and water supply also have very high output multipliers. Overall, the manufacturing sectors have the highest output multipliers. Analysis A comparative study of the two I-O tables reveals that that there has been a significant effect of drought on the economy. The effect of the 2002­3 drought can be seen in the production figures of the different sectors. Table A7.2 presents VOP for all the sectors classified in the I-O table for 1998­9 and for 2002­3 at 1998­9 constant prices. The VOP of the agricultural sector has gone down by as much as 27 percentage points. The output of crops, such as jowar, maize, and other food grains has shown an increase, but the drastic decrease in output of rice and ground- nut has outweighed the increase in other crops. The total VOP of rice relative to the total agricultural output is around 39 percent (1998­9), showing that rice is the major crop grown in Andhra Pradesh. Rice and groundnut are much more water sensitive than the other crops grown in the state. Hence, water scarcity will result in production loss, and the total agricultural sector will be affected by drought. The year 2002­3 was a major drought year, and the VOP of rice and groundnut decreased by 38 percentage points and 57 percentage points (relative to 1998­9), respectively, and thus loss of VOP for the total agricultural sector. The increase in the value of output for the food products, which depends on agricultural sector, was very small. The shift from rice and groundnut cultivation, particularly rice, to other crops would also result in savings in terms of inputs required for produc- ing these crops. From the I-O table it can be seen that for producing 1 unit of rice/rice 0.23 units of input is required, while for producing 1 unit of jowar and maize 0.16 and 0.20 units of inputs are required. At times of drought, output decreases but the inputs for production remain the same. This can be seen by comparing the input proportions for different sectors under agriculture for 1998­9 and 2002­3.Therefore, any shift in the crop- ping pattern will result in savings. The employment situation of a sector gets affected owing to the loss in the production. Employment coefficients provide a measure to account for loss in employment for any loss in production. The agricultural employment coefficient for the state is 5.48.This has been generalized for the agricultural sector as a whole because of the lack of detailed data on Input-Output Analysis for Andhra Pradesh 111 Table A7.2. Sector-Wise VOP at Current 1998­9 and 2002­3 at Constant 1998­9 Prices VOP (2002­3) at constant Sectors VOP (1998­9) 1998­9 prices Rice 1203027 741465 Jowar 35957 41443 Maize 68442 73502 Other food grains 159296 173853 Groundnut 298189 128745 Other crops 1308995 1092593 Livestock 948749 1677690 Forestry and logging 167625 170715 Fishing 351600 583779 Mining and quarrying 341449 560930 Food products 1771993 1979670 Textile products 228304 419022 Wood products 156431 197028 Paper products 294141 214297 Leather products 39573 44068 Rubber, plastic, coal, and tar 321623 1078885 Fertilizers 127262 155265 Pesticides 30351 36456 Chemicals 558991 570962 Nonmetallic mineral products 346845 422691 Basic metals and alloys 1616309 1338809 Metal products, electricity and nonelectric 520105 620527 machinery and equipment Transport equipments and parts 61350 97024 Miscellaneous 52298 152693 Construction 1022581 1524684 Electricity, gas, and water supply 929165 1097889 Railway transport services 313288 369824 Other transport and storage 1093358 1481011 Communication 199006 433838 Trade, hotels, and restaurants 2173966 2553891 Banking and insurance 593644 728509 Ownership of dwellings, real estate, 749610 1000480 and business services Education, medical, and other services 1794171 2577852 Public administration 509668 699591 the crop-wise employment structure for the state. The total employment loss for 2002­03 owing to loss in agricultural VOP is more than 44 lakhs. Encouraging dryland cropping would moderate the effects of drought on the employment scenario. 112 Overcoming Drought The livestock sector experienced a 77-percentage point increase in pro- duction despite the drought. This suggests that drought had no effect on this sector. The three major components contributing to this sector, milk, meat, and eggs, have behaved differently in the drought years from the normal years. The value of milk as a proportion of the total value of live- stock decreased from 55.1 percent to 50.1 percent, while that of meat remained the same (28.5 percent to 27.4 percent).The value of eggs, how- ever, increased from 8.09 percent to 15.23 percent.Thus, it is evident that the poultry sector (not meat) needs to be encouraged. Good performance of this sector during drought may have been due to Government interven- tions. Livestock feed also decreased over the years and from the I-O table it can be inferred that a substantial decrease in the cost of feed occurred from 1998­9 to 2002­3 (figure 4.1). This may be due to the substantial increase in the share of eggs because of poultry promotion. The primary sector as a whole has experienced an increase despite the drought, although agricultural sector performance was on the lower side. Construction showed a 49-percentage point increase in VOP. This may be because Government expenditure in this sector increased owing to poverty alleviation programs, and, hence, employment in this sector went up. The construction output multiplier obtained from I-O was 1.69. Any expenditure in the construction sector will lead to an increase in outputs for sectors, such as cement, steel, bricks, and tiles. Interpreting the multiplier, a 1 unit rise in the output of the construction sector will result in an additional 0.69 units rise in outputs of other sectors because of interlinkages between the sectors. Thus, any investment relating to construction by the Government will boost the increase in output in other sectors. The labor displaced from the agricultural sector may have been absorbed in this sector, moderating the employment loss in the agricultural sector due to drought. A comparative study of the demand components of the I-O tables revealed that the agriculture sector was the worst affected because of the 2002­3 drought. The production of rice was so low that the state had to import rice. Similarly, for other food grains and crops a decrease in out- put resulted in imports in these sectors for other states. Note 1. The factories that have 10 or more workers and use power or factories that have 20 or more workers and do not use power, are registered (organized) factories. The remaining enterprises are unregistered. A N N E X 8 Structure of the Andhra Pradesh Economy (1980­1 to 2002­3) The structure of the Andhra Pradesh economy may be described in terms of changes in the GVA in various sectors of the economy and interrela- tions between them. The following major sectors were included in the model and analysis: · Primary sector that includes agriculture, livestock, forestry and logging, fishing and mining and quarrying, of which two subsectors of partic- ular interest--agriculture and livestock--were examined separately for sensitivity to drought. · Secondary sector that includes manufacturing (both registered and un- registered), electricity, gas and water supply, and construction. · Tertiary sector that includes trade, hotels and restaurants, railways, transport by other means and storage, communications, banking and insurance, ownership of dwelling, real estate and business, public administration, and other services. The year 1993­94 was chosen as the benchmark year because the GVA was measured at 1993­4 prices. Fluctuations in GVA during 1980­1 and 1992­3 were compared with those during 1993­4 and 2002­3. The 113 114 Overcoming Drought postulated models were estimated for these subperiods separately, and from 1980­1 to 2002­3, to detect any structural changes. Descriptive Analysis of Changes in the GVA of Major Sectors Primary sector Agriculture and livestock were clubbed together because separate data on agriculture and livestock GVA were not available for the earlier subperi- od 1980­1 to 1992­3. Changes in the GVA of agriculture and livestock. The years 1984­5 and 1986­7 showed low productivity in agriculture and livestock. The annual percentage changed and the combined GVAs were ­8.7 and ­10.8, respectively. These years may be classified as years of severe drought. Years 1990­1 and 1992­3 had mild drought conditions, as the annual per- centage change in GVA of agriculture and livestock were ­2.9 and ­1.1, respectively. During the subperiod 1993­4 to 2002­3 we observed highly negative annual percentage change in the GVA of agriculture and livestock during the years 1997­8 and 2002­3. These were ­18.8 in 1997­8 and ­12.5 in 2002­3. Thus, 1997­8 and 2002­3 were years of severe drought. However, 1999­2000 also showed a highly negative (­4.02) percentage change in the GVA of agriculture and livestock. Thus, 1999­2000 should also be treated as the year of severe drought even though it was of rela- tively lower intensity than that of 1997­8 and 2002­3. The other years showing negative annual percentage change in agriculture and livestock were 1994­95 (with ­2.14) and 2001­02 (with ­0.95). The share of agriculture and livestock in the primary sector GVA has consistently decreased over the entire period 1980­1 to 2002­3. It was as high as 88 percent in 1982­3 and 1983­4 and decreased to 85 percent in 1992­3 and further decreased over the second subperiod (1993­4 to 2002­3). It remained in the range of 80­84 percent during 1993­4 and 2001­2 and was 75.52 percent in 2002­3. Changes in the GVA of components other than agriculture and livestock in the primary sector. Mining and quarrying, and fishing have consistently improved their share in the primary sector GVA. During 1980­1 and 1992­3 the share of mining and quarrying was 3­6 percent, and it increased to 7­9 percent during 1993­4 and 2001­2. It was 11.3 percent Structure of the Andhra Pradesh Economy (1980­1 to 2002­3) 115 in 2002­3. Similarly, the share of fishing was 4­6 percent from 1980­1 to 1992­3, and increased to 5­8 percent from 1993­4 to 2001­2. It stood at 10.32 percent in 2002­3. Forestry and logging share in the primary sector GVA (PGVA) was 3­5 percent over the entire period 1980­1 to 2002­3. Thus, it appears that while agriculture, livestock, and forestry were worst affected by drought, sectors such as mining and quarrying and fishing improved. Mining and quarrying recorded a 12-percentage point and fishing 19.6-percentage point change annually in their GVA in 2002­3. Secondary sector The share of manufacturing in the secondary sector GVA was 55­64 percent during 1980­1 and 1992­3 and 62­67 percent in 1993­4 and 2002­3.A decrease in the percentage share of manufacturing was observed from 1999­2000 to 2002­3: 62 percent and 63 percent during this period compared to more than 66 percent in 1993­4 and 1998­9. This may be attributed to drought conditions during 1997­8 and 2002­3 because the raw material for manufacturing comes from agriculture. The share of construction in the secondary sector GVA decreased consistently from 36 to 21 percent from 1980­1 to 1996­7 but started to increase slowly from 1997­8 to 2002­3 (22­26 percent). This could be due to the result of efforts made to mitigate hardships caused by drought conditions. Tertiary sector The percentage share in GVA of communications, banking and insur- ance consistently increased from 1980­1 to 2002­3. Although the share of communications was small (ranging from 2.28 in 1980­1 to 7.08 in 2002­3), it almost trebled over the years and that of banking and insur- ance doubled. The subsectors of trade, hotels and restaurants, and trans- port and storage maintained almost the same percentage share over the years. Surprisingly, the share of railways has decreased from about 5 percent (1980­1 to 1987­8) to less than 4 percent (1993­4 to 2002­3). Specification of the Macro Model in Terms of Sector-Wise GVA Specification of a macro model requires postulating structural equations, which describe changes that directly influence GVA in terms of certain 116 Overcoming Drought variables. These variables need to be identified and their relationship to GVA in each of the sectors determined. In the present study, the model was postulated in the form of a set of interdependent regressions, in log linear form, the parameters were estimated by the SUR method. The estimated coefficients may be inter- preted as partial elasticity coefficients. The GVA was calculated as the difference between the values of output and inputs (at current or constant prices). However, the inputs did not include the consumption of fixed capital.1 For example, in agri- cultural GVA the inputs are seed, chemical fertilizers, organic manure, current repairs and maintenance of fixed assets, market charges, irriga- tion charges, electricity, pesticides and insecticides, and diesel. Therefore, the specification of structural equations for GVA, in each of the sectors, should include consumption of fixed capital as one of the explanatory variables. Macro Model for the Major Components of Different Sectors of the Andhra Pradesh Economy Agriculture is the major constituent of the primary sector and manufac- turing that of the secondary sector. Agriculture and livestock were worst affected during drought years. Ideally, a model that accounted for agri- culture and livestock separately during both the subperiods (1980­1 to 1992­3 and 1993­4 to 2002­3) should have been postulated and performance of the model compared over the subperiods. Unfortunately, data on GVA from agriculture and livestock were aggregated for 1980­1 to 1992­3 and were not available separately for agriculture and livestock. However, agriculture GVA and livestock GVA were separately available for 1993­4 to 2002­3. See figures A8.1­A8.4. Figure A8.1. Agriculture GVA, 1993­94 to 2002­03, Estimated and Observed 1,900,000 1,700,000 1,500,000 lakhs 1,300,000 AGVA Rs. Est. AGVA 1,100,000 900,000 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 years Structure of the Andhra Pradesh Economy (1980­1 to 2002­3) 117 Figure A8.2. Livestock GVA, 1993­94 to 2002­03, Estimated and Observed 800,000 700,000 600,000 lakhs 500,000 LGVA Rs. 400,000 300,000 Est. LGVA 200,000 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 years Figure A8.3. Secondary Sector GVA, 1993­94 to 2002­03, Estimated and Observed 2,500,000 2,300,000 2,100,000 1,900,000 lakhs 1,700,000 1,500,000 SGVA Rs. 1,300,000 Est. SGVA 1,100,000 900,000 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 years Figure A8.4.Tertiary Sector GVA, 1993­94 to 2002­03, Estimated and Observed 5,000,000 4,500,000 4,000,000 lakhs 3,500,000 3,000,000 TGVA Rs. 2,500,000 Est. TGVA 2,000,000 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 years The estimated model for 1993­4 to 2002­3 is given below. 118 Overcoming Drought In the above equations: AGVA = agricultural GVA LGVA = livestock GVA SGVA = secondary GVA TGVA = tertiary sector GVA ACFC = consumption of fixed capital in agriculture LCFC = consumption of fixed capital in livestock SCFC = consumption of fixed capital in secondary TCFC = consumption of fixed capital in tertiary sector VOP4,8 = value of output for four crops eight districts AGVA­1 = last year's agricultural GVA The key data series exhibited strong trends and were nonstationary, which may lead to the familiar problem of "false inference." To identify the appropriate cointegrating relationships a number of detrended regres- sions were conducted. Among these, the following is a representative example. It involves a regression for the period 1981­2002, which yielded the following model: where * is statistically significant at 5 percent confidence level. In the above equations: dAGVA = AGVA ­ AGVA­1 dSGVA = SGVA ­ SGVA­1 dTGVA = TGVA ­ TGVA­1 dACFC = ACFC ­ ACFC­1 dSCFC = SCFC ­ SCFC­1 dCFC = TCFC ­ TCFC­1 dVOP4,8 = VOP4,8 ­ (VOP4,8)­1 In this model, an increase in AGVA in the previous year would create a positive impact on the secondary sector and a negative impact on the ter- tiary sector, as found under the previous model. However, these coeffi- cients were not statistically significant at 5 percent confidence level. The Structure of the Andhra Pradesh Economy (1980­1 to 2002­3) 119 small size of the sample diluted the statistical power of the usual battery of specifications and model selection tests. Hence, it was not possible to statistically identify the most appropriate model. Under the circumstances, trended specification was opted for because of its superior predictive power. However, because of the statistical limitations of a restricted sam- ple size, the choice should be viewed as tentative. Estimates obtained from most of the alternative specifications were within 3 standard errors of the chosen model. Hence, the coefficients of the chosen model were best viewed as indicative of sectoral linkages, rather than as precise estimates. Note 1. The net value added is defined as the difference between the GVA and consumption of fixed capital. Glossary Annual rate of occurrence The average number of occurrences per year. Not to be confused with the term "probability," which refers to the probability of at least one event occurring in a year. Base year The starting year for financial calculations; that is, a bench- mark with which future years are compared or calculated against. Block/mandal An administrative subdivision of the district, which in turn is a subdivision of the state. Crop simulation This can predict yield with a priori knowledge of the soil properties and management practices.The model simulated plant development and growth and soil processes to estimate yield. Crop yield The measurable produce of economic value from a crop. This may be evaluated in terms of quantity and/or quality. Yields are stated as kg/ha or t/ha. Crore 1 crore = 10,000,000. Deterministic model A model that assesses the impact of a hazard by investigating the severity of a single possible outcome. District domestic product or district income The sum of the econo- mic value of goods and services produced within the geographical boundaries of the district irrespective of the income and owned by persons living inside or outside the district. Thus, the estimates of domestic product at the district level are compiled by following the income originating approach as done for gross state domestic product 121 122 Overcoming Drought estimates. Because of the open character of the economic activities and absence of data relating to inter district flows, the income accru- ing concept is not followed as in the case of state domestic product. District per capita income estimates, when studied in relation to the total population of the district, indicate the level of per capita net output of goods and services available or the standard of living of the people in the district. Drought Drought is defined in many ways, such as a period of dry weather, a condition when precipitation is insufficient to meet established human needs, comparison of normal precipitation months and years, a prolonged dry weather causing hydrologic imbalance, a time-space duration distribution of percent of normal precipita- tion, and so on. Drought index Several indices are being used to estimate drought severity, including (a) the variable to be used (for example, rainfall, runoff aquifer level, Palmer Drought Index); (b) duration considered (for example, annual, seasonal, instantaneous minimum); (c) trunca- tion level (for example, percentage, quantile, standardized anomaly); and (d) area or region (for example, single site, river basin, country zone). Economic loss The total monetary cost incurred, whether insured or not, because of a shock. Environment Policy Integrated Climate model A computer model of crop growth and soil loss used to understand the impact of management actions and climate on agricultural productivity. The first versions of the model were referred to as the Erosion Productivity Impact Calculator but it has now been renamed as the Environment Policy Integrated Climate model to reflect its wider scope. Evaporation A process by which a liquid or a solid (sublimation) enters the gas phase. In the hydrologic context, it refers to the conversion of water and ice at the earth's surface to water vapor and its dissipa- tion into the atmosphere. Evapotranspiration The combined effect of evaporation and transpiration. Event loss table In its basic form contains columns of event identifica- tion, event loss, and event rate of occurrence. In its expanded form the columns for associated uncertainties of loss and rate are also provided. Event set A set of discrete events used in probabilistic risk modeling to simulate a range of possible outcomes. Exceedance probability See "exceeding probability." Exceeding probability Also known as "exceedance probability." The probability of exceeding specified loss thresholds. In risk analysis, this Glossary 123 probability relationship is commonly represented as a curve (the EP curve), which defines the probability of various levels of potential loss for a defined structure or portfolio of assets at risk of loss from natural hazards. Exposure The total value or replacement cost of assets (such as struc- tures) that are at risk from a loss-causing event such as a catastrophe. Final consumption expenditure The spending on goods and services used for the direct satisfaction of individual or collective needs, as distinct from purchases for use in a productive process. Fixed capital Long-term capital used for long-term investments in fixed assets (for example, land, buildings, equipment, machines). Gross cropped area The total area under all crops. Gross fixed capital formation The investment in assets that are used repeatedly or continuously over a number of years to produce goods. For example, machinery used to create a product. Gross irrigated area The total irrigated area under various crops during a year, counting the area irrigated under more than one crop during the same year as many times as the number of crops grown and irrigated. Gross state domestic product A measure of economic activity in a state. Calculated by adding the total value of the state annual output of goods and services. Gross value added The difference between output and intermediate consumption for any given sector/industry. That is, the difference between the value of goods and services produced and the cost of raw materials and other inputs, which are used up in production. Harvest index A crop parameter-based experimental data where crop stresses have been minimized to allow the crop to attain its poten- tial. The EPIC model adjusts the index as water stress occurs close to flowering time. Hazard A condition that may create or increase the chance of loss from a peril. Indirect taxes Taxes that do not come straight out of a person's pay packet or assets or out of company profit (for example, consumption tax such as value-added tax). Intensity A measure of the physical strength of a damage-causing event, such as a drought. Common scales for intensity include Standard Precipitation Index or Palmer Drought Severity Index for drought. Intermediate consumption The cost of raw materials and other inputs that are used up in the production process. Inventories Formerly called stocks. Consist of materials and supplies stored for use in production, work in progress, finished goods, and goods for resale. 124 Overcoming Drought Irrigation A method of purposely providing land with water, other than rain water, by artificial means. Kharif season Characterized by a gradual fall in temperature, more numerous cloudy days, low light intensity, a gradual shortening of photoperiod, high relative humidity and cyclonic weather. The kharif season depends entirely on the southwest monsoon receiving more than 70 percent of the annual aggregate rainfall during mon- soon months of June­September. Lakh 1 lakh = 100,000. Macroeconomic model Studies the overall aspects and workings of an economy, such as income, output, and the interrelationship among the diverse economic sectors. Mitigation A process by which adverse environmental impacts of an activity are minimized or replaced by beneficial features. Net area irrigated The total of all the areas irrigated from different sources, counting each area irrigated only once even though it was irrigated more than once in the same year. Net area sown The area sown with crops and orchards, counting the area sown more than once in the same year. Northeast monsoon The rainy season that affects southern India from October to December. Peril Includes storms (hurricane,tornado,other windstorm),earthquakes, floods, or drought. Price The value of the goods or money that must be given up to acquire goods or services. Price indices The statistical measure of average changes over time in the prices of commodities relative to a base year. Probabilistic model A model that assesses the impact of a hazard and assigns probabilities to a whole range of possible outcomes. Probability See "annual rate of occurrence." Probability of exceeding The probability that the actual loss level will exceed a particular threshold. Probability of nonexceeding The probability that the actual loss level will not exceed a particular threshold. Probable maximum loss A general concept applied in the insurance industry for defining high loss scenarios that should be considered when underwriting insurance risk. The exact probability or return period associated with a probable maximum loss can vary based on the company's policies and objectives. Rabi season In Rabi, a gradual increase in temperature, bright sunshine, near absence of cloudy days, a gradual lightening of the photoperiod, and a lower relative humidity. During Rabi season, rainfall occurs from October to December. Glossary 125 Radiation-use efficiency This is the potential (unstressed) growth rate (including roots) per unit of intercepted photosynthetically active radiation. Regression Regression analysis is the study of the dependence of one variable (the dependent variable) on one or more other variables (the explanatory variables), with a goal of estimating and/or predict- ing the mean or average value of the former in terms of the known or fixed values of the latter. Return period The expected length of time between recurrences of two events with similar characteristics. Can refer to hazard events such as hurricanes or earthquakes, or it can refer to specific levels of loss (for example, a US$100 million loss in this territory has a return period of 50 years). Risk A measure of potential financial loss, commonly encompassing two factors: exposure or elements at risk (amount of value subjected to potential hazard), and specific risk (the expected degree of loss owing to a particular natural phenomenon). This is also used more generally in insurance markets to refer to a specific property covered by an insurance or reinsurance policy. Risk management The management of the varied risks to which a busi- ness firm or corporation might be subject. It involves analyzing all exposures to gauge the likelihood of loss and determining how to minimize losses by such means as insurance, self-insurance, reduction, or elimination of risk or the practice of safety and security measures. Runoff Refers to agricultural runoff and occurs when the precipitation rate exceeds the infiltration rate of the soil. Site Same as location; defines exposure data.A site may represent mul- tiple buildings in proximity that are of similar construction and have a single deductible amount. Southwest monsoon The main rainy season in India that occurs from June to September. Stochastic drought A possible drought scenario created as part of a probabilistic model, whose probability has been assigned using prob- ability distributions from the historical record. Transpiration The process by which water vapor escapes from living plants and enters the atmosphere. Includes water that has tran- spired through leaf stomata, as well as intercepted water that has re-evaporated. When a growing crop covers the soil, transpiration greatly exceeds evaporation. t-statistic After an estimation of a coefficient, the t-statistic for that coefficient is the ratio of the coefficient to its standard error. That can be tested against a t distribution to determine how probable it is that the true value of the coefficient is really zero. 126 Overcoming Drought Validation The process by which probabilistic models and assump- tions are reviewed and compared to empirical data (such as histor- ically observed losses or insurance claims) to confirm that the model approach and assumptions generate reasonable estimates of potential loss. Value of production output Measures the total value of goods and services produced by a sector. Vulnerability The degree of loss to a system or structure resulting from exposure to a hazard of a given severity. Bibliography Aggarwal, P. K., S. Nagarajan, and Udai Kumar. 2001."Climate Change and Indian Agriculture: Current State of Understanding and Future Perspectives." Report submitted to the Indian Council of Agricultural Research, New Delhi. ANGRAU. 2003. Annual Report of Agro-Meteorological Cell, Hyderabad. Bryant, K. J., V. W. Benson, J. R. Kiniry, J. R. Williams, and R. D. Lacewell. 1992. "Simulation of Corn Yield Response to Irrigation Timing: Validation of the EPIC Model." Journal of Production Agriculture, 5: 230­2. Gadgil, S., A. K. S. Huda, N. S. Johda, R. P. Singh, and S. M. Virmani. 1988. "The Effects of Climatic Variations on Agriculture in Dry Tropical Regions of India." In The Impact of Climatic Variations on Agriculture:Volume 2: Assessments in Semi Arid Regions, eds. M. L Parry, T. R Carter, and N. T. Konijn. Dordrecht: Kluwer. Government of Andhra Pradesh. 2004. Economic cum Purpose Classification of Andhra Pradesh Government Budget 2003­2004. Directorate of Economics and Statistics, Hyderabad. Groundwater Department. 2003. "Note on Ground Water Scenario in Andhra Pradesh." New Delhi. Gurenko, E., and R. Lester. 2003. "Financing Rapid Onset Natural Disaster Losses in India: A Risk Management Approach." Report 26844-IN, World Bank, Washington, DC. 127 128 Bibliography India Council of Agricultural Research. 2002. "Drought Management Strategies." New Delhi. Izaurralde, R. C., N. J. Rosenberg, R. A. Brown, and A. M. Thomson. 2003. "Integrated Assessment of Hadley Center (HaDCM2) Climate-Change Impacts on Agricultural Productivity and Irrigation Water Supply in the Conterminous United States. Part II: Regional Agricultural Production in 2030 and 2095." Agricultural and Forest Meteorology 117 (1­2): 97­122. Jodha, N. S. 1981. "Role of Credit in Farmers' Adjustments Against Risk in Arid and Semi-Arid Tropical Areas of India." Economic and Political Weekly 16: 1696­709. Lal, M., B. Bhaskaran, and S. K. Singh. 1998."Indian Summer Monsoon Variability as Simulated by Regional Climate Model Nested in Global Climate Model." Chinese Journal of Atmospheric Science 22 (1): 93­102. Lok Sabha Unstarred Question No. 151. 2004, May 7. New Delhi. Madduri, S. 2003. "Convergence of Self Employment Schemes--Amangal Mandal Survey." Draft of a report. University of Hyderabad, India. Matalas, N. C. 1967. "Mathematical Assessment of Synthetic Hydrology." Water Resources Research 3: 937­45. McCarthy, James J., Osvaldo F. Canziani, Neil A. Leary, David J. Dokken, and Kasey S. White, eds. 2001. Climate Change 2001: Impacts, Adaptation, and Vulnerability. Working Group II, Intergovernmental Panel on Climate Change. New York: Cambridge University Press. McKee, T. B., N. J. Doesken, and J. Kleist. 1993. "The Relationship of Drought Frequency and Duration to Time Scales." Paper presented at the Eighth Conference on Applied Climatology, Anaheim, CA, January 17­22. Nagarajan, R. 2003. Drought: Assessment, Monitoring, Management and Resource Conservation. New Delhi: Capital Publishing. National Drought Mitigation Center. 2005. University of Nebraska-Lincoln. www.drought.unl.edu. National Institute for Rural Development. 2002. "Rapid Interactive Study of Management of 2002 Drought." Draft report. Centre for Disaster Management and Rural Reconstruction, Hyderabad, India. Prasad, B. N. 2002. "Report on Estimation of Groundwater Resources in Andhra Pradesh." Groundwater Department, Hyderabad, India. Priya, S., and Ryosuke Shibasaki. 1998a. "Soil Erosion and Crop Production: A Modeling Approach." Paper presented at the Global Environment Symposium, Japanese Society of Civil Engineers, Osaka, Japan, July 2­4. Priya, S., and Ryosuke Shibasaki. 1998b. "Assessing Impact of Increasing Carbon Dioxide With Climate Change on Crop Production." Paper presented at the International Conference on Modeling Geographical and Environmental Bibliography 129 Systems with Geographical Information Systems (GIS), Chinese University of Hong Kong, June 22­25. Raddatz, Claudio. 2005. "External Shocks and Output Volatility in Low Income Countries." DECRG Working Paper, World Bank, Washington, DC. Reddy, D. R., G. Sreenivas, and S. B. S. N. Rao. 2001. "Estimation of Base Temperature for Different Phenophases in Groundnut for South Telangana Agro-Climatic Zone of Andhra Pradesh." Abstract for the National Seminar on Agrometeorological Research for Sustainable Agricultural Production, Gujarat Agricultural University, Anand, September 27­8. Revenue (Relief) Department. 1981 and 1995. Drought:A Handbook for Management of Drought. Hyderabad, India: Government of Andhra Pradesh. Richardson, C.W. 1981."Stochastic Simulation of Daily Precipitation,Temperature, and Solar Radiation." Water Resources Research 17: 182­90. Richardson, C. W. 1982. "A Wind Simulation Model for Wind Erosion Estimation." ASAE Paper 82-2576,American Society ofAgricultural Engineers, St. Joseph, MI. Semenov, M. A., R. J. Brooks, E. M. Barrow, and C. W. Richardson. 1998. "Comparison of the WGEN and LARS-WG Stochastic Weather Generators for Diverse Climates." Climate Research 10: 95­107. Sharpley, A. N., and J. R. Williams, eds. 1990. EPIC-Erosion/Productivity Impact Calculator: 1. Model Documentation. Technical Bulletin 1768, U. S. Department of Agriculture, Washington, DC. The Andhra Pradesh Gazette, Part IV-B.April 19, 2002. Extraordinary No. HSE/49 Act No.10 of 2002. www.apind.gov.in/gos.html. Walker, T. S., and J. G. Ryan. 1990. Village and Household Economies in India's Semi-Arid Tropics. Baltimore: The John Hopkins University Press. Water Conservation Mission. 2002. Note on Neeru-Meeru Programme, Rome. Watson, R. T. M. C. Zinyowera, and R. H. Moss, eds. 1998. The Regional Impacts of Climate Change: An Assessment of Vulnerability. Special Report of Working Group II, Intergovernmental Panel on Climate Change. New York: Cambridge University Press. Wilhite, D. A., and M. H. Glantz. 1985. "Understanding the Drought Phenomenon: The Role of Definitions." Water International 10 (3): 111­20. Winrock International. India. 2005. "A Review of Vulnerability to Climate Change and Adaptation Strategies in India." Report for the World Bank,Washington, DC. Index A analysis, 16 findings, 65­69 Acharya NG Ranga University, 5­6 input-output, 60 adaptation strategies, 38, 68, 72, xxvii Anantapur, 16f, 66. See also drought-prone farm level, 42b areas adaptation to climate variability, 7­8 AAL of output, 35 "Addressing Vulnerability to Climate alternative uses of irrigation water, 40 Variability and Climate Change crop yields for climate scenarios, Through an Assessment of 44­45, 45t Adaptation Issues and Options," 7 agricultural drought, 81 cultivable rice area, 39b, 40f agricultural practices, 70­71. See also gross irrigated area and annual farming practices rain, 99f agricultural shift, 3 lag rain, 97f Agricultural University, 42b monsoons and gross cropped area, agriculture, 103­104 98f, 99f diversification initiatives, 78 seasonal planting, 94 financing, 46­49 shift in cropping mix, xxiv agriculture GVA (AGVA), 27­28, SPI validation, 84, 84f 28f, 116f Andhra Pradesh, 1­4, xix­xx inputs, 116 climate projections, 44­46 changes, 114 economic classifications, 102­103 agriculture sector, 12f economic sectors, 104 drought effects, 10f economy structure, 113 employment, 59 impact of structural changes on economy, impacts on people, 70 56­58 VOP, 60, 110 input-output analysis, 101­112 131 132 Index macro model for different economic crop production loss model, 83­85 sectors, 116­119 crop substitution, 22, 60, 66, 67, 110, macroeconomic impacts, 54­55 xxiii, xxviii services sector, 105 crop switching incentives, 48 Andhra Pradesh Rural Livelihood Project, crop yield model, 20, 21­22 75­76 crop yield variability, 31­32, 35 area variability, 22 crop yields, 33f, 34f assistance for drought affected areas, 53t climate change scenarios, 45, 45t assistance targeting, 36, xxvii crop-wise rain-fed areas, 95 average annual loss (AAL), 24b, 53, xxiii cropping mix, 66 cyclones and floods, 56b cropping mix shift, xxiv GVA, 53f, xxiv cropping patterns, 60, 97 of output, 35 recommendations, 42b of VOP by district, 36f cropping practices, 37b crops, 15, 38, 65, 66 B Cuddapah, 16f. See also drought-prone areas banking, 115 C D Calamity Relief Fund, 73, 74t damage assessment model, 52 capital formation, 106­107 data for EPIC input, 91­92 catastrophre exposures, 56b Decision Support System for catastrophre modeling, 63 Agro-Technology Transfer, 21 categories of drought, 19t, 54, 79­81, 84t Decision Support Toolkit, 72, xxviii Central Research Institute for Dryland Department for Rain-Shadow Areas of Agriculture, 5 Development, 4 Chittoor, 16f. See also drought-prone Department of Planning, 5 areas Development of Women and Children in climate change, xi Rural Areas, 78 human-induced, 67 direct loss module, 23 impacts, 43­46 diversification, 11, 78. See also crop scenarios, 45b substitution climate modeling, 43­44 donor programs, 5 climate variability, 7­8, 43 drought, 10, 17, xxii commodity-flows approach, 106 definitions, 8n, 79­81, 80f Community Forest Management, 75 drought adaptation insurance and credit, comparison of drought to other climate 48, 72, xxviii extremes, 55 drought initiatives and programs, construction, 112, 115 1, 4­6, 5t employment, 59 drought proofing, 5t, 74­77 coping strategies, 37b. See also strategies drought risk assessment model. See proba- and adaptation strategies bilistic drought risk assessment model crop area, 16f drought risk model. See probabilistic crop distribution factors, 97 drought risk assessment model crop growing seasons, 92 drought-prone areas, 3f, 8n, 15 crop growth model, 91­94 crop production losses, 35­36, 36f crop insurance, 46, 64­65, 74 intercropping systems, 42b crop production, 85 return periods, 19t losses, 35­36, 40­41, 65 rice yields, 35t drought-prone areas, 36f Drought-Prone Areas Program, 74­75 estimates, 23, 95 dryland crops, 15 Index 133 E fishing, GVA changes, 115 food for work programs, 78 economic classifications, 102­103 forecasting, 5, 43­44 economic impacts, 9, 10f, 67 forestry and logging, GVA changes, 115 economic losses, 24, 51 framework goals, 15 in sectoral GVA, 55f framework support, 71 economic scenarios, 57, 58f framework, institutional and policy, xxvii economic sector interaction, 25 frequency of droughts, 19 economic structural changes, 68, 70, xxiv­xxv, xxvi impacts, 56­58 G eggs, 60, 112 GCA. See gross cultivated area ELT. See event loss table GFCF. See gross fixed capital formation employment, 58, xxvii global climate model (GCM), 43 alternative options, 59­60 government final consumption coefficients, 25, 26t, 108, 109t, expenditure (GFCE), 107 110­111 greenhouse gas emissions, 43 drought adaptation strategies, 70 GrfA. See gross rain-fed area (GrfA) generation, 5t gross cultivated area (GCA), 95, 96 loss, 68, 111, xxvi gross fixed capital formation (GFCF), mitigating drought impacts, 69 106­107 programs, 77­78 gross irrigated area, 95, 96 Employment Generation Mission, 77 gross rain-fed area (GrfA), 95, 96, 97f, 98f, empowerment of poor women, 77­78 100f energy consumption, 38 gross state domestic product (GSDP), 9 Environment Policy Integrated Climate gross value added (GVA), 11, xxvf. See (EPIC) model, 21, 22, 85 also agriculture GVA; livestock GVA climate projections, 44 average loss by sector, 54f crop growth model, 91­94 average sectoral for normal years, 52f parameterizing, 93 changes in major sectors, 114­115 tillage changes, 43 economic scenarios, 57f, 58f WXGEN, 87­89 groundnut, 103 environmental imbalance, 76 groundwater, 2, 3 EPC. See exceedence probability curve assessment for irrigation, 4f EPIC. See Environment Policy Integrated Groundwater Department, 4 Climate model groundwater table, 38 estimating drought impacts, 51 GSDP. See gross state domestic product event loss table (ELT), 23 GVA time series, 12f exceedence probability curve (EPC), 19, 35 H economic impacts, 54 economic scenarios, 58f Hadley Regional Model 2, 44 export data, 107 hazard module, 17­19, 83­84 exposure module, 84­85 hotels and restaurants, 115 household responses, xxvi F hydrological drought, 81 hydrological modeling, 22, 44 factories, 112n Hydrology Project, 5 farmer coping strategies, 35 farmers, marginal, 70, xxvii I farming practices, 32, 35 final demand, 105 impacts, 70 financial strategies, 37 economic, 9, 10f, 67 134 Index location-specific, 66, xxiii livestock sector, 11, 20, 20t, 68, 112 macroeconomic, 54­55, 58, 67, xxv Andhra Pradesh, 104 marginal, 55 capital formation, 107 of severe drought location variability, xxiii crop yields, 34f location-specific, 36 positive, 53 coping measures, xxiv sector variability, 68 impacts, 66 socioeconomic, xix loss exceedence curve (LEC), 24b import data, 107 loss module, 85 index-based insurance, 47b loss potentials, 51­55 India Meteorological Department, 80 losses, 20, xxii. See also crop production Indian Crop Insurance Program, losses and production losses 47b­48b average for drought categories, 54 indirect loss module, 24­25 in GVA, xxvf input-output analysis, 60 rice, xxiiif input-output model, 24, 25 input-output tables, 101 M institutional framework, 71, xxvii insurance, 64­65, 71­72. See also crop macro model, 115­116, 116­119 insurance macroeconometric model, 24­25, 27­29 GVA changes, 115 loss potentials, 51­52 probabilistic drought risk model, 46 macroeconomic impacts, 54­55, 58 schemes, xxvii­xxviii decreasing, 67, xxv Integrated Wastelands Development macroeconomic model, 7, xxii Program, 76 Mahbubnagar, 16f, 37b. See also drought- intensity of drought, 18, 79 prone areas intercropping systems, 42b AAL of output, 35 International Consortium for Agricultural crop yields for climate scenarios, 44­45, Systems Applications, 21 45t International Research Institute for cultivable rice area, 39b, 41, 41f Semi-Arid Tropics, 5­6 maize, 33f irrigation, 2­3 management, 4, 21 coping strategies, 37b planning, 6, 72 irrigation reallocation, 38­41 strategies, 58­59 irrigation water, 22 manufacturing, 104, 115 alternative uses, 40­41 Markov chain, 87 meat, 60, 112 J meteorological drought, 80­81 methodology, xxi­xxii Joint Forest Management, 75 development, 63­65 jowar, 33f microfinancing, 71­72, xxvii­xxviii Microirrigation Project, 75 K migration, 37b Kharif, 92, 94 milk, 60, 112 Kurnool, 16f. See also drought-prone areas mining and quarrying, 114­115 mitigation, 17, 57, xix model testing, 92­93 L modeling framework, xx lag rain, 97f models. See climate model; crop growth LEC. See loss exceedence curve model; damage assessment model; livestock GVA (LGVA), 12f, 27­28, 117f drought risk model; Environment changes, 114 Policy Integrated Climate model; Index 135 Hadley Regional Model 2; input- flexibility, 22­23 output models; macro model; macro- losses, 20, 23 econometric model; planting area programs. See drought initiatives and model; probabilistic drought risk programs assessment model; risk assessment model; stochastic crop production Q loss model monetary losses, 23, 24 quantitative analysis findings, 65­69 monsoons, 98f, xi multipliers, 108, 109t, 110 R Rabi, 92, 94 N rain-fed areas, 31 Nalgonda, 16f. See also drought-prone areas rain-fed crops, 38 National Agriculture Insurance Scheme, 74 rainfall, 15 National Calamity Contingency Fund, 53, Anantapur, 97f, 98f, 100f 53t, 73­74 economic performance, 9, 10f National Centre for Calamity levels, 3f Management, 74 projections, 44 National Rural Guaranteed Employment risk model, 22 Scheme, 78 seasonal deviation, 16f National Sample Survey, 106 variables, 96 net value added, 119n rainfall insurance, 47b­48b Ranga Reddy, 16f. See also drought-prone areas O rapid onset disasters, 56b output multipliers, 25, 26t, 108, 109t, 110 reaching out to the poorest, 70 output tables, VOP, 10 regional climate model (RCM), 43 resilience, 12, 57, 60 P return periods, 19, 19t rice, 32, 38 Pennar River basin, 44 losses in drought years, xxiiit plant water requirements, 81 reducing cultivable area, 38­41, 39b, planting area model, 22­23, 95­97 40f structure, 100f yields in normal and drought years, PML. See probable maximum loss 35t policy framework, 71, xxvii risk assessment model, 6­7, 64­65, 67. poultry sector, 60, 69, 112 See also probabilistic drought risk Prakasam, 16f, 35. See also drought-prone assessment model areas risk financing, 5t, 46­47, 48b, 71, xxviii primary sector, 11f, 112, 113 programs, 73­74 diversification, 11 risk insurance, 46­49 GVA changes, 114­115 risk metrics, 23, 24b structural shift, 60­61 risk mitigation, 57 private final consumption expenditure, risk vulnerability, 58­61 106 rural infrastructure development, 76­77 probabilistic drought risk assessment model, 6­7, 17­25, 18f, 27­29, 32, S 64­65, 67, 72, xxif, xxii climate change, 45b Sampoorna Grameen Rozgar Yojana, 77 insurance applications, 46 scenarios for economic structure, 57t probable maximum loss (PML), 24b secondary sector, 11f, 113 production, 19 GVA, 115, 117f 136 Index sectors, major, 25, 59­60, 61n. See also U primary sector, secondary sector, and U.S. Department of Agriculture tertiary sector Agricultural Research Service, Soil Self-Help Groups of Women, 77­78 Conservation Service, and Economic services sector, 59, 105 Research Service, 21 severe drought impacts, 34f slow onset events, 64, xxii V socioeconomic drought, 81 socioeconomic impacts, xix value of production (VOP), 9­10 socioeconomic strategies, 58­61 agricultural sector, 110 soil erosion, 21 annual average, 52 Standardized Precipitation Index (SPI), crops for Andhra Pradesh, 103 17­18, 19t, 83, 84, 84f, 84t drought-prone areas, 35 stochastic crop production loss model, reducing cultivable rice area, 40f 83­85 sector-wise, 111t stochastic weather generator Vana Samrakshana Samithies, 75 (WXGEN), 19 vulnerability, 58­61 EPIC, 87­89 vulnerability module, 20­21, 84­85 strategies, 69­72, xxvi­xxviii adaptation, 72 W coping, 35, 37b water conservation, 38, 66 farm level, 38, 42b tillage, 42­43 socioeconomic, 58­61 water harvesting structures, 75 study objectives, 6­7, 63, xx water rationalization, 32­33, 66 surface water, 2, 3 water reallocation, 38­41 sustainable agricultural practices, water supply, 2 70­71 Water, Land and Trees Act, 4­5, 38 watersheds, 4, 75, 76 T weather generator, 87. See also stochastic tertiary sector, 11f, 68, 113 weather generator GVA, 115, 117f weather insurance, 47b­48b positive impacts of drought, 53 weather risk, 22 thrift groups, 77­78 women, empowering, 77­78 tillage practice changes, 42­43 World Bank, 63­64 trade, transport, and storage, 115 WXGEN. See stochastic weather generator Drought sets off a vicious cycle of socio-economic impacts that leave the poor more vulnerable to another drought and other shocks. Overcoming Drought applies recent advances in modeling climate-related risks and adjusts state-of-the-art catastrophic risk modeling techniques to drought.The result is an innovative long-term assessment of how to reduce the impacts of drought under several economic, drought management, and climate change scenarios.This assessment helps in identifying and recommending adaptation strategies at local and state levels. The analysis focuses on the most drought-prone districts of the state of Andhra Pradesh in India and measures the impacts of drought-related losses in these districts on state-wide indicators of economic performance. Although long-term macroeconomic effects of drought appear modest, the human and social costs remain devastating for the millions of people in the poorest rain-fed areas of Andhra Pradesh. An effective response to drought must account for a large variation in impacts from village to village. Overcoming Drought highlights the need and directions for intensifying efforts to deliver better packaged knowledge-based assistance to the affected communities, so as to enable a sustainable drought adaptation process. Overcoming Drought will be of great interest to readers working in microfinance, agriculture, rural development, natural resource management, and climate change.The study also has significant implications for agricultural and climate-related insurance. ISBN 0-8213-6664-5