The role of farm subsidies in changing India’s water footprint

The role of farm subsidies in changing India’s water footprint | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article The role of farm subsidies in changing India’s water footprint Shoumitro Chatterjee, Rohit Lamba, Esha Zaveri This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-1766947/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 05 Oct, 2024 Read the published version in Nature Communications → Version 1 posted You are reading this latest preprint version Abstract Dwindling groundwater supplies threaten food security and livelihoods. Farm subsidies which distort production decisions are an ubiquitous agricultural policy tool, yet their contribution to growing groundwater stress remains poorly quantified. Here we show how output subsidies that guarantee purchase of crops at higher than market prices have contributed substantially towards declining water tables in India. Overall, these policies have led to a 30% over-production of water intensive crops. In the northwestern state of Punjab, rice procurement accounts for 63% of the rise in groundwater declines over two decades. In the central state of Madhya Pradesh, wheat procurement adopted in the late 2000s has driven a 5.3 percentage point increase in dry wells and a consequent 3.4 percentage point increase in deep tubewells. The results demonstrate how well intentioned but poorly designed subsidies can impose harmful externalities on the environment and undermine long-term sustainable development. Figures Figure 1 Figure 2 Figure 3 Figure 4 Introduction India has witnessed a 500% increase in groundwater consumption over the past half-century (1-5) , making it one of the greatest groundwater guzzlers in the world. Unsurprisingly groundwater reserves are depleting at an alarming rate (2,6) . Since the 1980s, groundwater levels have plunged by more than 8 meters on average (7) . This trend is starkly visible in the alluvial aquifers of northwest India (2,7,8) . In central and south India, where wells deplete and replete annually due to the nature of the hard-rock aquifer systems, these trends are less apparent. Here, other metrics of over-extraction, such as increases in the number of dry or defunct wells, reinforce the alarm (9) . In this paper, we show that perverse incentives generated through the design of India’s agricultural output subsidies are a key source of this unfolding calamity. Output subsidies provided via guaranteed government procurement of water-intensive crops like rice and wheat at pre-determined prices were introduced in India during the green revolution of the 1960s. The goal then was to incentivize the adoption of high-yielding variety seeds to increase productivity and ensure food security for the nation. A second goal was to provide farmers with risk-free remunerative incomes (10) . While the policy was justifiable during the 1960s as India faced acute food scarcity, it has outlived its relevance. It is now causing unintended harm to food security, farmer incomes, and sustainability by accelerating groundwater depletion. The underlying mechanism has two parts. First, the output subsidy is implemented mainly for rice and wheat and not other crops. Thus, rice and wheat cultivation has risen unchecked, engulfing other crops, since the 1960s (Fig 2c). Today, even as production of these crops exceeds national consumption needs and emergency buffer requirements by at least 30%, a guaranteed subsidy ensures continued cultivation of rice and wheat and keeps farmers from switching to other crops. Second, since rice and wheat are water-intensive crops, their increased cultivation requires irrigation. Over time the increased irrigation needs have been met by personal irrigation systems like wells rather than surface water systems like dams or canals (3, 4) . Together these trends have contributed to rapid declines in groundwater levels. Rice and wheat provide over 50% of the calorific intake of India’s 1.4 billion people (11) and groundwater scarcity decreases crop yields ( 12, 13 ). Therefore, groundwater depletion compromises food security and farmer livelihoods (14)—the very objectives of the subsidy policy. Moreover, estimates suggest that in the long-run a complete loss in access to groundwater can reduce annual crop production by 28%, dry season crop production by 51%, and cropping intensity by 68% (15) . Furthermore, as groundwater also acts as a buffer against climatic variability, its depletion undermines adaptability to climate change (16, 17) . Even as groundwater depletes, India’s political economy contributes to the persistence of this policy (18) . Without a cap or a sunset clause, a vicious cycle has ensued. The deeper the groundwater table, the costlier the cultivation due to increasing costs for well deepening and pumps. As noted above, assured government procurement has locked in most of India’s 260 million poor farmers into growing rice and wheat. Since the farmers are now wholly dependent on this policy for their incomes, the government is forced to increase the subsidy every year to cover increasing costs. This has resulted in India currently spending about USD 4 billion annually to fund the output subsidy program (19) . While the literature has focused on the role of input subsidies like free power ( 20, 21 ), the role of output subsides in driving local water use has not been studied or quantified. This is both because the driving mechanisms are subtle, and there has been a lack of data to show them. Output subsidies affect groundwater indirectly by shaping crop choice decisions, unlike power subsidies that directly increase groundwater extraction using pumps. We fill this lacuna by collating data from several sources since 1981 and quantify the role of the output subsidy policy in eroding India’s groundwater supplies. Our analysis has three parts. First, we show that increased cultivation of a water-intensive crop like rice is closely tied to increases in groundwater stress across India. In particular, we show that across Indian districts, a 1 percentage point (pp) increase in the rate of growth in the area under rice cultivation between 1996–2015 is associated with a 1.7 pp increase in defunct wells. Second, we argue that an excess of rice and wheat production by almost 30% is sustained because of the government’s guaranteed purchase of these crops at higher than market prices. Third, we focus on case studies from two states with distinct aquifer systems to illustrate the main mechanism by which the output subsidy policy impacts groundwater. The hydrogeology and aquifer-specific properties can impact the extent to which pumping affects the water table locally (9, 22) . For the state of Punjab, we show that the output subsidy policy has contributed to 63% of the local decline in the groundwater table. Since this region overlays deep alluvial aquifers, these effects manifest gradually. Results show that the effects of policy-induced groundwater extraction are fully reflected in declining water levels after 6 years. For the central Indian state of Madhya Pradesh, we use different metrics of groundwater stress since it predominantly overlays hard rock aquifers with low storage capacity (9, 23) . Here, results show that since the inception of the output subsidy policy in 2008, incidence of dry-wells has increased by 5.3pp and demand for deep tube wells by 3.4% We make three contributions. First, we assemble a novel dataset that allows us to test the hypothesis linking output subsidies to the depletion of groundwater resources at a granular level. Second, we show that the output subsidy policy is a significant driver of groundwater stress not only in the deep alluvial aquifers of north-western India but also in the hard rock aquifers of central India. In general, showing that this is a pan-India issue affecting regions with very different hydrogeological conditions is important because agriculture policy related groundwater issues tend to be associated with north-western India. Third, we show that not paying careful attention to the nature of aquifers can lead to a gross underestimation of the effects of output subsidies on groundwater stress. This entails considering the gradual nature of adjusting groundwater tables in the alluvial aquifers of Punjab and using alternate groundwater stress metrics like dry wells and demand for deep tubewells in the hard rock aquifers of Madhya Pradesh. This study offers cautionary insights into the possible impacts of distortionary agricultural policy on freshwater resources across the globe. Governments around the world provide broad types and levels of support to agriculture (24) . Around USD 635 billion are provided annually to support agricultural production and food consumption worldwide ( 25, 26 ) . More than 70 percent of this total support, about $456 billion, consists of support for agricultural producers. A majority (82 percent) is provided through measures that are considered most distorting (24) . Given the magnitude of resources devoted to agricultural subsidies across the world, understanding their unintended consequences on environmental outcomes remains critical for policymakers to design sustainable, fair, and efficient agricultural policies. The results in this paper suggest while the most common policy objectives of such support are to provide price stability and food security as well as to support farmers’ incomes and livelihoods, they can often lead to unintended consequences that are counterproductive to policy goals. By distorting crop and irrigation choices, such support can ultimately lead to harmful spillovers on local water resources and undermine sustainable development objectives in the long-term. Cultivation Of Water-intensive Crops And Groundwater Stress In India One of the key reasons behind the groundwater stress in India is the over-production of water intensive crops like rice and wheat. For example, Fig 1a shows that across India, a 1 percentage point (pp) increase in the rate of growth in area under rice cultivation between 1996–2015 is associated with a 1.7 pp increase in active wells becoming defunct. This is not surprising given that rice is one of the most water intensive crops cultivated using flood irrigation technique ( 27, 28 ). The relationship in Fig. 1a is likely an underestimate because major increases in rice cultivation occurred in the immediate years after the introduction of the Green Revolution during the 1960s (Supplementary Information) from when groundwater data is unavailable. To address this and other potential confounding factors like cultivation of other crops and population driven demand, we estimate multivariate statistical models controlling for such factors. Specifically, we regress the fraction of wells that were active in 1996 but were defunct by 2015 on the growth in the area under rice cultivation in that district, and control for the initial area under rice cultivation, initial gross cropped area, total land area, and population (Materials and Methods). In Table 1, cols 1–2, growth in rice area is computed between 1996 and 2015, and in cols 3–4 between 1966 and 2015. Table 1, cols 1–2 show that the correlation from Fig. 1a survives the inclusion of various controls. Table 1, cols 3–4 show that districts that had a 1 pp greater rate of growth in rice area between 1966 and 2015 had 2.4 pp additional defunct wells by 2015. This is a higher magnitude than Cols 1-2 because we also include the period when rice area was rapidly expanding. Note that to compare districts across the country we do not use groundwater depth but number of defunct wells as the metric for groundwater stress. This is because the former is not a reliable measure to compare groundwater stress across different types of aquifer systems (9). Further, since wheat is grown in a few states (Supplementary Information, Table S1), lack of variation in the data precludes us from analyzing effects of wheat cultivation at an all-India level (Materials and Methods). The Role Of Output Subsidies In Over-production Of Rice & Wheat The increased groundwater stress from cultivation of rice and wheat is only a symptom of the problem. The underlying issue is that the government’s output subsidy policy meant to improve farmer incomes and ensure food security is inducing overproduction. Every year before the sowing season the central government announces a floor price (called the Minimum Support Price or MSP) for 23 crops at which it commits to procure any quantity of specified crops from farmers. The MSP has to be backed up by the actual procurement by government agencies, and this latter bureaucratic machinery acts as the true determinant of the efficacy of the policy. In practice the policy gets implemented primarily for rice and wheat—the main food crops (Supplementary Information). In 2020 alone, the production of rice and wheat exceeded consumption by 30% and government reserves were nearly 2.5 times the required norms (Figs. 1b and S2a)—an excess of 36 million tons or about 20% of India’s annual consumption. A large part will get wasted (29, 30) and hence, the associated cost on groundwater is a deadweight loss. Such quantum of excess production each year should ideally put a downward pressure on market prices of rice and wheat which, in turn, would disincentivize their production and automatically contain exploitation of groundwater. However, the government’s policy of providing a guaranteed floor price (called the Minimum Support Price or MSP) to farmers cultivating rice and wheat (see Supplementary Information) has thwarted the discipline of market forces for the last half-century. This has propagated the over cultivation of these crops and the depletion of scarce groundwater such that groundwater levels have plunged to more than 30 meters below ground level (mbgl) in some regions (31). Table 1. Effects of rice cultivation on groundwater stress: All India. Defunct Wells (as a % of Active Wells in 1996) 1996 - 2015 † 1966 - 2015 † (1) (2) (3) (4) Rice Area Growth † 1.747 *** 1.704 *** 2.344 *** 2.366 *** (0.445) (0.424) (0.419) (0.406) N 257 251 283 281 Adj. R 2 0.04 0.16 0.10 0.24 Notes: Each observation is a district. Robust standard errors clustered at the district level reported in parentheses. ∗ p < 0.10, ∗∗ p < 0.05, ∗∗∗ p < 0.01. The dependent variable in all columns is the percentage of wells that were active in 1996 but were defunct in 2015. Average Annual Rate of Growth of Rice Area over 1996–2015 in cols 1–2 and over 1966–2015 in cols 3–4. Growth is measured in percent. Cols 2 and 4 also include controls for initial gross cropped area, initial rice area, land area of the district, and population. Case Studies The main challenge in quantifying the role of MSP and public procurement in inducing the production of rice and wheat and the subsequent effect on groundwater is the lack of disaggregated and consistent data for key variables of interest—quantity of rice or wheat procured by government, cropped area, and groundwater—from a period when the policy-induced thrust toward the cultivation of rice and wheat occurred. To circumvent this issue, we focus on two states for which we were able to collect this data in different but relevant time periods, i.e., close to when the policy was instituted in the respective states—Punjab (1981–2003) and Madhya Pradesh (2002–2016) (Materials and Methods). Case Study I: Punjab Although Punjab is one of the most agriculturally productive and irrigated regions of the country, it has also witnessed one of the largest increases in groundwater stress in the world (32) . The average groundwater depth increased from 4.82 mbgl in 1973 to 14.55 mbgl in 2016. By 1999, 78.6% of all dug wells that were active in 1973 had become defunct (fig 2a). Additionally, over 75% of the area is considered overexploited according to the Central Ground Water Board (23). The historical root for this depletion lies in the adoption of high-yielding variety (HYV) wheat and rice during the Green Revolution in the 1960s. These varieties replaced local wheat varieties, cotton, maize, and oilseeds (Fig. 2c and Supplementary Information, Table S2) and required more intensive irrigation. The increased irrigation came from groundwater (dug wells and tubewells) causing its depletion (Fig. 2b). Underlying this process was the policy of assured government procurement at MSP of rice and wheat (Fig. 2d), which incentivized cultivation of these water-intensive crops over others even as India amassed a surplus of rice and wheat. The earliest year for which district-level procurement data are available is 1981. By this time, the wheat area had become stable with little change over time. On the other hand, rice area exhibits an increasing trend (Fig. 2c) and remains the focus of the analysis. Using regression models with a rich set of controls and fixed effects (Materials and Methods) we first document an output response. Farmers responded to a doubling of rice procurement by increasing area under rice cultivation by at least 54% in the following year (Supplementary Information, Table S3). Second, to show how increased rice cultivation in turn impacts groundwater levels, we regress changes in groundwater levels on the log rice area (Materials and Methods). Punjab overlies thick and deep alluvial aquifers such that declines in groundwater levels persist and adjust gradually (22). To capture these dynamics, we compute changes in groundwater level over multiple horizons and estimate a separate model for each horizon and present these results graphically in Fig. 3a For example, the estimate at T = t +4 is from a regression where the dependent variable is the proportional change in groundwater depth between an initial year t and t +4, i.e. over four years (∆= 4). The main co-variate of interest is either log rice area (Fig. 3a) or log rice procurement (Fig. 3b) in year t . Thus, the estimates show how the same “shock”—rice area or procurement in year t —effects groundwater levels over different horizons. Estimates using pre-monsoon readings are in blue and those using post-monsoon readings in red. In all models, we control for various observable factors that could bias our estimates like population, net cropped area, precipitation, and temperature. To control for time-invariant, district-specific unobserved factors like geography, we use district fixed effects. We control for year fixed effects to isolate unobserved district-invariant, time-varying factors. We also include Agro Ecological Zone (AEZ)-specific time trends that account for AEZ-specific time-varying factors like changes in socio-economic indicators that could bias our results. Our results are also robust to explicitly including wheat area as a control, which is another water-intensive crop and is grown in the following cropping season. The rich set of controls helps us estimate the effect of rice cultivation on groundwater depth. Fig. 3a depicts how the effects on groundwater table depths show up over time. A doubling of rice area (i.e. an increase by 100%) causes a fall in groundwater depth of 6 pp over a year and 24 pp over three years, as measured by pre-monsoon readings. Similar patterns are observed for post-monsoon readings. The effect keeps increasing until about six years and then stabilizes at 63 pp for pre-monsoon and 99 pp for post-monsoon readings. Fig. 3b, is the reduced form of this causal chain where we regress proportional changes in groundwater depth directly on log rice procurement with the same set of controls. The effects show a similar evolutionary pattern. By the sixth year, which is when the effects stabilize, a doubling of rice procurement results in a 46–70 pp fall in groundwater levels. The gradual adjustment in groundwater table depths is an important empirical finding and worth emphasizing. A naive analysis that ignores the adjustment process, and focuses solely on the contemporaneous relationship between changes in groundwater level declines and procurement would underestimate the effects of the output subsidy policy on groundwater level declines by 90%. Average pre-monsoon groundwater depth in Punjab fell by 65% between 1981 and 2003. During this period, rice procurement has increased at about 3.5% per year from 4.4 to 12.8 million tons. Our results predict a fall of 1.6 pp per year in groundwater depth due to this mechanism or a total fall of 41% over 22 years. In other words, increased rice cultivation that was induced by government procurement of rice explains 63% of the fall in groundwater levels in Punjab between 1981 and 2003. To confirm that our estimates are not picking up spurious correlations, we estimate placebo models in which we regress changes in groundwater depths between years k < t and t on rice area or rice procurement in year t . In line with the fact that rice cultivation or procurement today should not impact groundwater depths in previous periods, we find no association in our placebo estimates (see placebo estimates in Figs. 3a and 3b). This robustness check reinforces the validity of our analysis. Finally, it is worth reiterating that the output subsidy program keeps the wheat cultivated area in Punjab high and stable thus resulting in a continuous usage of groundwater for its irrigation and therefore its depletion. However, since there is little change in the wheat area post-1981, a lack of variation precludes us from estimating marginal effects of wheat cultivation or procurement on groundwater. Case Study II: Madhya Pradesh Historically, the government agencies in Madhya Pradesh did not procure either wheat or rice even when market prices fell below the MSP. But from 2008, the state announced a bonus on top of the national MSP and substantially expanded its procurement operations of wheat. The potency of the output subsidy policy is driven in large part by the procurement machinery of the state at the local level. With strong procurement, the policy became highly active in Madhya Pradesh. Before 2007, the largest volume purchased by state agencies in any year was 0.54 million tons (mt). In 2007 procurement was a mere 0.057 mt. In 2008, this exponentially jumped to 2.4 mt—a 40-fold increase (33) . The policy was introduced when state elections were due later in the year. Procurement operations were also more concentrated in districts that were key producers of wheat. Thus, there was widespread belief that this was an election year gift, and therefore, the farmers did not immediately increase wheat cultivation (33) . Over time, however, as wheat procurement became a fixture in subsequent seasons and spread to other districts, this belief was shed and along with procurement there was also a concomitant increase in area under wheat cultivation (Fig. 4a). Between 2000 and 2007, the area under wheat cultivation grew at 3% annually, which almost doubled to 5.8% per year between 2008 and 2015. Some of the growth in wheat area between 2000 and 2008 can be attributed to the improvements in irrigation systems (reliable power for tubewells and completion of canal irrigation projects) that were being made by the state government beginning in early 2000 (34). However, the increased wheat cultivation because of the procurement policy put additional pressure on irrigation demand. The average annual growth in wheat irrigated area was 6.5% between 1991 and 2007 but after 2008, the wheat irrigated area began to increase by 7.9% each year. As was the case in Punjab, much of the new irrigation came from groundwater (wells and tubewells) rather than surface water (Fig. 4b). This change has gradually started increasing groundwater stress, although magnitudes are comparatively low since this is a relatively new policy as compared to fifty years of procurement operations in Punjab. Madhya Pradesh is dominated by hard-rock and mixed aquifers where measurement of groundwater stress is not straightforward. Shallow hard-rock aquifers deplete and replete annually and long-term water level trends are less apparent. It is well known that in such aquifers, groundwater stress can increase even though average groundwater depth may not be increasing (9) . A robust measure of groundwater stress in this region is the need for deep tubewells (with depth >70m). Deep tubewells are expensive and risky to construct. Their failure causes a precipitous decline in the long-term income of the farmers trying to install them (35) . Farmers install deep tubewells when they do not have access to surface water or groundwater at shallower depths. Thus, an increase in the incidence of defunct dugwells, that are shallow, with a concomitant increase in deep tubewells is a reliable indicator of groundwater stress (Fig. 4c). The exogenous and phased introduction of procurement operations in this state provides a natural experiment to estimate the causal impact of this policy on groundwater stress. Given the hydrogeology, we use three different metrics to measure groundwater stress: the proportional change in groundwater depth pre-sowing and post-harvest, the fraction of wells that run dry post-wheat harvest, and the construction of deep tubewells. To estimate causal effects, we regress each of these measures on the log of wheat procured and the log of wheat procured interacted with a post-2008 indicator. The unit of observation is a district × year. All specifications always have district fixed effects that control for differences in time invariant, district-specific factors that could be correlated with the policy and bias our results. This includes differences in aquifer systems and the fact that procurement started more intensely in districts that already had wheat cultivation and then spread to other districts. We also control for district-specific time-trends that partial out any pre-existing trends in wheat procurement and groundwater stress. Further, we take into account seasonal rainfall and temperature, and for robustness, we also control for total irrigated area in the district. This battery of controls allows us to estimate the causal impact of the policy by essentially comparing groundwater stress in the districts before and after the policy change and relating it to the intensity of procurement operations. The coefficient on the interaction term estimates the causal impact of a 1% increase in wheat procurement on groundwater stress post-2008 as compared to before the policy was in place (Materials and Methods). Table 2. Effects of Wheat Procurement in Madhya Pradesh ∆GWL Dry Wells (prop) log Deep Tubewells (1) (2) (3) log Wheat Proc -0.012 -0.001 -0.000 (0.022) (0.008) (0.010) log Wheat Proc. 0.039 ∗ 0.076 ∗∗∗ 0.048 ∗∗∗ × Post-2008 (0.023) (0.007) (0.014) N 481 481 407 Clusters 37 37 37 Adj. R 2 .51 .27 .99 Notes: Each observation is a district×year. Robust standard errors clustered at the district level reported in parentheses. ∆GWL is the proportional change in groundwater depth between November in year t-1 and May in year t. Dry wells are the fraction of monitoring wells that are dry in the months after wheat harvest (June–September). All regression models include district fixed effects, district specific linear time trends, and controls for temperature and precipitation. ∗ p < 0 . 10, ∗∗ p < 0 . 05, ∗∗∗ p < 0 . 01. Table 2 shows that pre-policy, the little wheat procurement that occurred had no relationship with either measure of groundwater stress. Post-2008 however, a doubling of wheat procurement caused a 3.9 pp fall in groundwater depth (col 1), a 7.6 pp increase in the incidence of dry wells (col 2), and a 4.8% increase in construction of deep tubewells (col 3). All these results consistently show an increase in groundwater stress. The fact that we can triangulate the results across different metrics of groundwater stress increases the reliability of our results in this hard-rock aquifer region. Between 2007 and 2016, wheat procurement in Madhya Pradesh increased by almost 70% (from 0.057 mt to 4 mt). Our results therefore imply that the policy increased the incidence of dry wells by 5.3 pp and the need for deep tubewells by 3.4% during this period. These effects are significant because they have happened over a relatively short span of eight years. As wheat procurement continues, groundwater stress in Madhya Pradesh will increase exponentially as in Punjab. The statistical model in col 1 is directly comparable to the model we used for Punjab (Fig. 3b) for a one-year horizon, ∆= T − t = 1. Why should the marginal effect of wheat procurement vary before and after 2008? This is where the mechanism is crucial. Credible wheat procurement incentivizes farmers to grow more wheat over less water-intensive crops, which increases groundwater stress. In Supplementary Information, Table S4, we provide evidence for this core mechanism. We regress log wheat area, log wheat irrigated area, and log area under pulses (the other important crop) on lagged log wheat procurement and lagged log wheat procurement interacted with a post-2008 indicator. We use year fixed effects to isolate time-varying aggregate shocks like aggregate supply, price volatility, and climate that could impact procurement and wheat cultivation. We explicitly control for rainfall, temperature, irrigation, and area under other crops for robustness (Materials and Methods). Our estimates show that the policy increased the marginal effect of wheat procurement on wheat cultivation in the following year by an additional 13.5% (Supplementary Information Table S4, cols 1–2). A doubling of procurement post-2008 also resulted in a 22% increase in irrigated wheat area due to the policy. Col 4 shows that the increase in wheat area partially came at the cost of a reduction in area under pulses, a less water-intensive crop. Fig. 4a, however, shows that a greater amount of new area came under wheat cultivation increasing groundwater stress. As an intense shift toward highly irrigated wheat occurred post-2008, we see differences in the marginal effects pre- and post-2008 in table 2. Discussion Taken together, the results of this study offer a new understanding of the role that India’s four decades-long assured grain procurement program has played on groundwater stress and highlights important lessons for policymakers focused on designing environmentally sustainable agricultural policies. This is critically important as Indian policymakers today are vigorously debating the future of these policies as a part of the current government’s plan to liberalize India’s agricultural markets and in reaction to the developments in Ukraine that have affected global wheat markets. They face fierce demands from farmers for even greater support, as fertilizer prices continue to rise. Governments worldwide provide billions in subsidies to support farmers. The mechanism by which these subsidies are provided is as important as the quantum of subsidies. Many nations choose to provide subsidies as income transfers or as insurance against price crashes. These methods are less distortionary. In India, pork barrel considerations have resulted in output subsidies being limited to water-intensive rice and wheat. As compared to total agricultural subsidies that account for about 2-2.5% of GDP ( 36 ), expenditure on productivity improving public goods like agriculture research and infrastructure is negligible ( 37 ). In this article, we argue that this has had deleterious consequences for the environment, and sustainability. The thick alluvial aquifers of northern India are a key buffer against climate change-induced weather variability. As their recharge takes centuries, their depletion has exposed India’s most productive regions to the possibility of desertification. There are other consequences as well. The most important is on nutrition. In the 1960s, India faced food deficiencies. Over time, agricultural policy (output subsidies) has become coupled with food policy (consumption subsidies). The grains procured under the output subsidy program are provided as subsidized food to the poor. The latter is then used as a justification to continue with the former ( 38 ). While increased cereal productivity that followed the Green Revolution has ensured cereal availability, the diversity in food systems has suffered ( 39 – 41 ). The Green Revolution crowded out the production of other nutrient-rich crops like coarse cereals and pulses ( 42 , 43 ). This combined with consumption subsidies made nutrition-rich food relatively more expensive and crowded them out from diets as well ( 42 ). The National Food Security Act of 2013 furthered this by codifying subsidies for staples into law. As a result, the Indian population today faces “hidden hunger” in nutrition ( 43 – 47 ). Finally, there is a trade paradox. India has one of the lowest per-capita availabilities of freshwater reserves. Given such scarce water endowments, economic theory would suggest that the country import water-intensive crops and export the less water-intensive crops. On the contrary, even as taps run dry in major Indian cities, induced by agricultural policy India exports 25×10 9 m 3 in virtual water every year ( 48 ). Recent estimates suggest that India accounts for 12 percent of the global groundwater depletion that is embedded in international food trade ( 49 ). At this rate, India is likely to loose its entire available water in less than 1000 years ( 48 ). Policymakers in India thus need to rethink how they provide subsidies to farmers and consumers for securing their own future. Recent research has tried to provide optimal criteria of procurement from the point of view of maximizing nutrition and ensuring food security while saving water resources ( 50 ). However, implementing such criteria will be challenging partly due to limited state capacity. But also because political economy constraints make any change difficult. The new income transfer scheme PM-KISAN is probably a good step forward as it is independent of the farmers’ crop choice. But presently, it is provided in addition to and not instead of the output subsidy. More importantly, farmers have a deep mistrust in institutions as coverage of most farmer welfare schemes is far from perfect. In the Green Revolution states of Punjab and Haryana, farmers have been used to the MSP procurement apparatus of rice and wheat for half a century. This is also where the groundwater stress is the most severe. Successfully moving away to alternative systems will require trust-building and convincing farmers that alternative systems of support and subsidies that are not linked to water intensive-crops can work. Materials And Methods Data We compiled several different datasets on crop production, area, and irrigation (apportioned district-level data on Indian agriculture from the International Crops Research Institute for the Semi-Arid Tropics (ICRISAT)), government procurement of rice and wheat (from the Food Corporation of India and state agencies such as the State Civil Supplies Corporation), groundwater levels and stress (from the Central Groundwater Board of India and the state groundwater board of Punjab). We constructed district-level weather data by averaging gridded temperature and precipitation data from the Indian Meteorological Department over the growing season of crops. Data on crop production and area from the ICRISAT database spans the years from 1966 to 2015. Where necessary, maps corresponding to 1966 district boundaries are used so that districts and states that split after 1966 are considered together to allow comparability over time. District-level procurement data for India is generally not available. We were able to obtain this data for two states, Punjab and Madhya Pradesh. For Punjab, district-level data spans the years from 1981 to 2018, and for Madhya Pradesh from the years 2002 to 2016. Observation well data for the country from 1996 to 2016 was obtained from the Central Groundwater Board (CGWB) database of monitoring wells, which contains water level measurements recorded four times a year (January, May, August, and November) for 30,311 wells. Of these, 12,837 wells were active in 1996, representing wells with the longest possible records in the database. A separate set of observation well data for Punjab from 1973 to 2003 was obtained from the monitoring wells of the state groundwater board of Punjab, which contains water level measurements on a bi-annual basis (pre- and post-monsoon). Drawdown is measured in observation wells before (June) and after (November) the annual monsoonal recharge season (June– September). It should be noted that water tables reported by monitoring wells may not reflect exactly those reported by farmers in their irrigation wells. Therefore, the averaged water levels at the district level are indicators of water fluctuations rather than absolute value of the water tables actually experienced by farmers. Measures of Groundwater Stress In addition to averaged groundwater level data at the district level, we use two additional metrics of groundwater stress, namely defunct and dry wells. The literature ( 9 ), has shown how missing data in well records carry critical information on groundwater stress that is completely missed in analyses that routinely filter out this information. Missing data in a well record can occur in two ways: first, the well goes defunct and stops collecting data permanently during the analysis time frame, or second, the well records no data in multiple intermediate months within the time frame. The underlying cause of such missing data can be either physical, where the water level in the well falls permanently or temporarily below the well screen depth, or logistical, where operators neglect maintaining monitoring wells, or they collect or record the data inadequately. The physical change in water level is one of the key underlying reasons for missing data. To overcome the non-random occurrence of missing well data records, two alternative indicators of groundwater stress are used. The percentage of defunct wells is measured at the end of the sample period and is estimated as the proportion of monitoring wells that started collecting data at the beginning of the sample period (active wells) but then permanently stopped collecting data within the sample period. Following the literature ( 9 ), we add the condition that a well should not have recorded data for (at least) the four years preceding the last year in the sample period to be considered defunct. The percentage of dry wells is the proportion of monitoring wells that started data at the beginning of the sample period (active wells) but have missing data and did not record information in the time frame under consideration such as a given season and year. Dry wells could therefore recover in the wet years but lack of water in dry years indicates groundwater stress. Both metrics are indicative of the degree of groundwater stress in the region. Reasons for using different metrics of groundwater stress in different regions For a cross-district comparison at the national level we use defunct wells as the metric for groundwater stress rather than groundwater levels because differences in aquifer types make changes in groundwater depth incomparable across regions ( 9 ). Defunct wells or dry wells are a more consistent metric of groundwater stress for comparisons across regions. Groundwater depth as a measure of stress is the most appropriate for the deep alluvial aquifers of Punjab. However, here groundwater levels persist and adjust gradually ( 22 ). Hence, we study changes over different horizons in groundwater levels to assess stress. Madhya Pradesh predominantly has hard rock and mixed systems that deplete and replete annually. Thus, in these aquifer systems water levels fluctuate greatly but the long-term water level trends are less apparent. Hence, we use changes in groundwater levels over shorter horizons, combined with incidence of dry wells and construction of tube wells to assess groundwater stress. Data Constraints and Choice of Case Studies Studying the role of public procurement in inducing the production of water-intensive crops and thereby quantifying the effect on groundwater stress requires data to satisfy two features. First, we need spatially disaggregated data on procurement and groundwater and second, the data must be from a period when growth in the cultivation of water-intensive crops occur. The lack of the second feature precludes statistical estimation of marginal effects because of the absence of adequate variation in the data. All-India Analysis Disaggregated district-level data on procurement for most states is not generally available. Thus, for the all-India analysis we only show a cross-sectional association between groundwater stress and cultivation. Moreover, we only use rice cropped area and exclude wheat. This is because groundwater data across regions in India is available only from 1996 and by this time wheat cultivation patterns have stabilized in the country. Wheat area grew at 3.4% each year between 1966—1980 but at a meagre 0.7% each year between 1996—2010. Furthermore, wheat is grown in a few districts of only five states (Table S1). As a result, a large number of districts record near zero growth in wheat area and this substantially reduces cross-district variation. Thus, even as wheat producing regions use a lot of groundwater for irrigation, it is infeasible to estimate this statistically. Since we are restricted to cross-sectional comparisons in our all-India analysis, we view these correlations as motivational. Case Studies For the formal analysis, we focus on case studies in two states—Punjab and Madhya Pradesh—where we were able to obtain suitable data. Punjab is an important region because this is where the Green Revolution and the procurement policy were first introduced, and it is the region that has faced the highest groundwater stress. Moreover, its thick alluvial aquifers are a key reserve of groundwater and understanding the effects on it are very important from a sustainability perspective. For this state we were able to obtain data on procurement and groundwater since 1981. We have groundwater data for two periods—1973–2003 and 1996–2016—from two different sources. While the area under rice kept growing at a healthy rate for about 20 years after 1981, the wheat cropped area was high but stagnant (Fig. 2 c). Hence, for the above-mentioned reasons, we study rice in Punjab between 1981 and 2003. We do not extend the analysis beyond 2003 because we would have to splice groundwater data from two different sources and that might introduce other biases. We could separately study the period of 1996–2016 but by this time production patterns had stabilized in both cropping seasons, with mainly rice being grown in the kharif (fall) season and wheat in the rabi (spring). Rice and wheat occupied 62% and 80% of Punjab’s net cropped area in 2000 with little scope for growth. Thus, there is little change over time in either rice or wheat cultivation post-1996, and statistical analysis again becomes infeasible. For reasons already mentioned, we were not able to study the impact of wheat cultivation in either the all-India correlations or in Punjab. To address this, we use a natural experiment in the state of Madhya Pradesh. Madhya Pradesh is a state where the government suddenly decided to procure wheat in 2008. In 10 years, the state went from procuring nothing to being the largest procurer of wheat in the country. This exogenous policy-induced shift in cultivation allows us to estimate the causal effect of the wheat procurement policy on groundwater stress. Regression Models Punjab To estimate the effect of rice procurement on production (Table S3) we estimate the following model: log Q d,t = log proc d,t −1 + X′ γ + λ d + λ t + λ aez t + ε dt , where Q dt is the rice or wheat production in district d in year t . proc d,t −1 is the procurement of the respective crop in district d in year t − 1. The vector of controls X includes precipitation, seasonal temperature and temperature squared in year t and district d . λ d is a district fixed effect, λ t a year fixed effect and λ aez t is an Agro-Ecological Zone (AEZ)-specific linear time trend. In this model, the fixed effects estimator is likely to suffer from Nickell bias ( 50 ) as current procurement is mechanically correlated with current cropped area. However, the bias is likely very small because we have 23 years of data ( T is large). For transparency we show results with and without district fixed effects and AEZ-specific trends. To study the effect of rice production on groundwater in Punjab (Fig. 3 a), we estimate the following model: ∆ T GWL dt = β 0 + β 1 log rice area dt + X′ γ + λ d + λ t + λ aez t + ε dt , where ∆ T GWL dt is the proportional change in groundwater depth in district d between years t and T . rice area dt is the actual area under rice cultivation in district d in year t. The vector of controls X includes district-year level log net cropped area, log population, cumulative precipitation until year T , seasonal temperature and temperature squared. λ d is a district fixed effect, λ t a year fixed effect and λ aez t is an AEZ-specific linear time trend. To estimate the effect of rice procurement on groundwater levels (Fig. 3 b) we replace rice area in the above equation with rice procurement. In both models for statistical inference, we compute Newey-West standard errors to account for serial correlation in shocks over three periods within districts. Robust standard errors clustered at the district level would be preferred since that accounts for arbitrary correlations in error terms. However, the asymptotic properties for inference are not satisfied since Punjab has only eleven districts (small N ). Madhya Pradesh To estimate the impact of the wheat procurement policy on groundwater stress (Tables 2 and S5), we estimate the following model: GWS dt = β 0 + β 1 log wheat proc dt + β 2 log wheat proc dt ×1{ t > = 2008} + X′ γ + λ d + λ d t + ε dt , where GWS dt is a measure of groundwater stress in district d in year t . Groundwater stress is measured as either (a) the proportional change in groundwater level between November in year t − 1 and May in year t (before sowing and after harvest of the wheat crop), (b) the fraction of wells that are dry post-wheat harvest in June–September year t , or (c) the logarithm of the number of deep wells in district d in year t. wheat proc dt is the quantity of wheat procured in district d in year t . λ d and λ d t are district fixed effects and district-specific time trends respectively. The vector of controls X includes seasonal precipitation, temperature, and temperature squared. The coefficient β 2 estimates the causal impact of the policy relative to pre-policy effects of wheat procurement. Finally, to estimate the effect of the wheat procurement policy on production patterns (Table S4), we estimate: log Y dt = β 0 + β 1 log wheat proc d,t −1 + β 2 log wheat proc d,t −1 ×1{ t > = 2008}+ X′ γ + λ t + ε dt , where Y dt is either wheat area, wheat irrigated area, or pulses area in district d in year t . wheat proc d,t −1 is the quantity of wheat procured in disitrict d in year t − 1. λ t is a year fixed effect. The vector of controls X includes seasonal precipitation, temperature, temperature squared, area under other crops, and irrigated area. The coefficient β 2 estimates the causal impact of the policy relative to pre-policy effects of wheat procurement. Here, we do not include district fixed effects because after the policy we only have nine years of data and 15 years of data overall. A fixed effect model estimates the above equation by demeaning. Since contemporaneous procurement and cropped area are mechanically correlated, this would result in Nickell bias ( 51 ) and thus in inconsistent estimates. We recognize that our estimates are biased but consistent. Since the goal is to show the mechanism, we feel comfortable with this choice. We do include district effects in the equation used to estimate the effect of the policy on groundwater stress. For statistical inference in both models, we compute robust standard errors clustered at the district level accounting for an arbitrary correlation in the error terms within districts over time. References H. Garduno, S. Foster, Sustainable groundwater irrigation. approaches to reconciling demand with resources. Strategic Overview Series No. 4. Tech. Rep., The World Bank, Washington D.C. (2010). W. Aeschbach-Hertig, T. Gleeson, Regional strategies for the accelerating global problem of groundwater depletion. Nat. Geosci. 5, 853–861 (2012). T. Shah, Taming the Anarchy: Groundwater Governance in South Asia (Routledge, 2010). M. Shah, P. Vijayshankar, F. Harris, Water and agricultural transformation in India: A symbiotic relationship– I. Econ. Polit. Wkly. 56 (2021). S. S. Sayre, V. Taraz, Groundwater depletion in India: Social losses from costly well deepening. J. Environ. Econ. Manag. 93, 85–100 (2019). A. Asoka, T. Gleeson, Y. Wada, V. Mishra, Relative contribution of monsoon precipitation and pumping to changes in groundwater storage in India. Nat. Geosci. 10, 109–117 (2017). S. Sekhri, Sustaining Groundwater: Role of Policy Reforms in Promoting Conservation in India. In S. Shah, B. Bosworth, A. Panagariya, (eds.) India Policy Forum 2012-13: Volume 9 , vol. 9, 149–187 (SAGE Publishing India, 2013). M. Rodell, I. Velicogna, J.S. Famiglietti, Satellite-based estimates of groundwater depletion in India. Nature 460, 999–1002 (2009). T. Hora, V. Srinivasan, N. B. Basu, The Groundwater Recovery Paradox in South India. Geophys. Res. Lett. 46, 9602–9611 (2019). S. Chatterjee, A. Mahajan, Why Are Indian Farmers Protesting the Liberalization of Indian Agriculture? ARE Update 24, 1–4 (2021). A. Deaton, J. Drèze, Food and Nutrition in India: Facts and Interpretations. Econ. Polit. Wkly. 42–65 (2009). N. Bhattarai, A. Pollack, D. Lobell, R. Fishman, B. Singh, A. Dar, M. Jain, The impact of groundwater depletion on agricultural production in India. Environ. Res. Lett. 16 (2021). S. Sekhri, Missing water: agricultural stress and adaptation strategies in response to groundwater depletion in India. Dep. Econ. Univ. Virginia, Work. Pap. (2013). E. Zaveri, D. Grogan, K. Fisher-Vanden, S. Frolking, R. Lammers, D. Wrenn, A. Prusevich, R. Nicholas, Invisible water, visible impact: groundwater use and Indian agriculture under climate change. Environ. Res. Lett. 11, (2016). M. Jain, R. Fishman, P. Mondal, G. Galford, Groundwater depletion will reduce cropping intensity in India. Sci. Adv. 7 (2021). R. Fishman, Groundwater depletion limits the scope for adaptation to increased rainfall variability in India. Clim. change 147, 195–209 (2018). E. Zaveri, D. Lobell, The role of irrigation in changing wheat yields and heat sensitivity in India. Nat. Commun. 10, 1–7 (2019). M. Raghavan, Politics of procurement and price support. Econ. Polit. Wkly. 39, 506–508 (2004). S. Chatterjee, D. Kapur, P. Sekhsaria, A. Subramanian, Agricultural Federalism: New Facts, Constitutional Vision. Working Paper (2022) K. Jessoe, R. Badiani-Magnusson, Electricity prices, groundwater, and agriculture: The environmental and agricultural impacts of electricity subsidies in India. Agricultural Productivity and Producer Behavior , 157–184 (University of Chicago Press, 2019). B.S. Sidhu, M. Kandlikar, N. Ramankutty, Power tariffs for groundwater irrigation in india: A comparative analysis of the environmental, equity, and economic tradeoffs. World Dev. 128 (2020). R. Fishman, T. Siegfried, P. Raj, V. Modi, U. Lall, Over-extraction from shallow bedrock versus deep alluvial aquifers: Reliability versus sustainability considerations for India’s groundwater irrigation. Water Resour. Res. 47 (2011). Government of India. Dynamic groundwater resources of India 2020. Tech. Rep., Central Ground Water Board, Ministry of Jal Shakti (2021). M. Gautam, D. Laborde, A. Mamun, W. Martin, V. Piñeiro, R. Vos, Repurposing Agricultural Policies and Support: Options to Transform Agriculture and Food Systems to Better Serve the Health of People, Economies, and the Planet. Tech. Rep. The World Bank. (2022) A. Panagariya, Agricultural liberalisation and the least developed countries: six fallacies. World Econ. 28, 1277–1299 (2005). OECD. Agricultural Policy Monitoring and Evaluation 2021. Tech. Rep. (2021). K.F. Davis, D. Chiarelli, M. Rulli, A. Chhatre, B. Richter, D. Singh, R. Defries, Alternative cereals can improve water use and nutrient supply in India. Science Advances 4, (2018). K.G. Mandal, A.K. Thakur, S.K. Ambast, Current rice farming, water resources and micro-irrigation. Current Science 116(4), 568–576 (2019). V. Rawal, M. Kumar, A. Verma, J. Pais, India let 65 lakh tonnes of grain go to waste in four months, even as the poor went hungry. scroll.in (2020). https://scroll.in/article/963535/india-let-65-lakh-tonnes-of-grain-go-to-waste-in-four-months-even-as-the-poor-went-hungry R. T. Venkiteswaran, Food grains rot in India while millions live with empty stomachs. The Interpret. (2018). S. Kaur, R. Aggarwal, M. Brar, Groundwater Depletion in Punjab, India. Encylopedia of Soil Science (2017). P. Döll, H. Mueller Schmied, C. Schuh, F. Portmann, A. Eicker, Global-scale assessment of groundwater depletion and related groundwater abstractions: Combining hydrological modeling with information from well observations and grace satellites. Water Resour. Res. 50, 5698–5720 (2014). M. Krishnamurthy, States of Wheat: The Changing Dynamics of Public Procurement in Madhya Pradesh. Econ. Polit. Wkly. 72–83 (2012). A. Gulati, P. Rajkhowa, P. Sharma, Making Rapid Strides-Agriculture in Madhya Pradesh: Sources, Drivers, and Policy Lessons. Working Paper 339 Indian Council for Research on International Economic Relations (2017). D. Blakeslee, R. Fishman, V. Srinivasan, Way down in the hole: Adaptation to long-term water loss in rural India. Am. Econ. Rev. 110, 200–224 (2020). B. Ramaswami, Agricultural Subsidies – Study Prepared for XV Finance Commission. Tech. Report. Indian Statistical Institute. (2019) B. Dhar, R. Kishore, Indian Agriculture Needs a Holistic Policy Framework, Not Pro-market Reforems. Econ. Polit. Wkly. 56 (2021) A. Kotwal, M. Murugkar, B. Ramaswami, PDS forever? Econ. Polit. Wkly. 72–76 (2011). R. Radhakrishna, Food and nutrition security of the poor: emerging perspectives and policy issues. Econ. Polit. Wkly. 40, 1817–21 (2005). D. Headey, A. Chiu, S. Kadiyala, Agriculture’s role in the Indian enigma: help or hindrance to the undernutrition crisis? Gates Open Res 3, 958 (2019). A. M. Thow, S. Kadiyala, S. Khandelwal, P. Menon, S. Downs, K. S. Reddy, Toward food policy for the dual burden of malnutrition: an exploratory policy space analysis in India. Food Nutr. Bull. 37, 261–274 (2016). P. Kataki, Shifts in cropping system and its effect on human nutrition: Case study from India. J. Crop. Prod. 6, 119–144 (2002). P. Pingali, K. Ricketts, D. E. Sahn, Agriculture for nutrition. The fight against hunger malnutrition: The role food, agriculture, targeted policies 165 (2015). P. Pingali, Green revolution: impacts, limits, and the path ahead. Proc. Natl. Acad. Sci. 109, 12302–12308 (2012). P. Pingali, Agricultural policy and nutrition outcomes–getting beyond the preoccupation with staple grains. Food security 7, 583–591 (2015). P. Pingali, B. Mittra, A. Rahman, The bumpy road from food to nutrition security–slow evolution of India’s food policy. Glob. food security 15, 77–84 (2017). A. de Brauw, M. Suryanarayana, Linkages between poverty, food security and undernutrition: evidence from China and India. China Agric. Econ. Rev. (2015). P. Goswami, S. N. Nishad, Virtual water trade and time scales for loss of water sustainability: A comparative regional analysis. Sci. Reports 5, 1–11 (2015). C. Dalin, Y. Wada, T. Kastner, M. J. Puma, Groundwater depletion embedded in international food trade. Nature 543, 700–704 (2017). Devineni, N., Perveen, S. & Lall, U. Solving groundwater depletion in India while achieving food security. Nat Commun 13 , 3374 (2022) S. Nickell, Biases in dynamic models with fixed effects. Econometrica 1417–1426 (1981). S. Chatterjee, M. Krishnamurthy, D. Kapur, M. Bouton, A Study of the Agricultural Markets of Bihar, Odisha and Punjab. Final Report. Tech. Rep., Center for the Advanced Study of India, University of Pennsylvania (2020). S. Chatterjee, D. Kapur, Six Puzzles in Indian Agriculture. India Policy Forum 17, 185–229 (2017). P. Gupta, R. Khera, S. Narayanan, Minimum Support Prices in India: Distilling the facts. Working Paper (2021) Available at SSRN 3791859. Declarations Acknowledgements For helpful discussions and feedback, we are grateful to the seminar participants at Center on Food Security and the Environment, Stanford University and to Marshall Burke, Walter Falcon, Ram Fishman, Tejasvi Hora, Avinash Kishore, David Lobell, Roz Naylor, and P.S. Vijay Shankar. We thank Faiz Ahmed Kidwai and Azad Jain for facilitating access to wheat procurement data from Madhya Pradesh. Author contributions statement S.C., R.L., and E.D.Z designed research; S.C. and E.D.Z performed research; S.C. and E.D.Z analyzed data; and S.C., R.L., and E.D.Z wrote the paper. All authors reviewed the manuscript. Additional information The authors declare no competing interest. Data and materials availability: All data and codes to replicate the results in this paper are available from SC ( [email protected] ) or EDZ ( esha.d.zaver [email protected] ) Additional Declarations There is NO Competing Interest. Supplementary Files SupplementaryInformation.docx Cite Share Download PDF Status: Published Journal Publication published 05 Oct, 2024 Read the published version in Nature Communications → Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-1766947","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":119657249,"identity":"b4676fbb-3bb7-4be0-aa80-b7434791c442","order_by":0,"name":"Shoumitro Chatterjee","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAw0lEQVRIiWNgGAWjYDACCRBhACKYDxiDRQ4QpwWkhy2BFC1ga3gMmInSIj+7+ZjUjYI/+fzsZz4UF7YxyPHdSMCvxeDOsTTpHAMDy5k9uRuMZ7YxGEsS1CKRYwbSYmBwg3eDMW8bQ+IGQlrkZ8C18DwAaaknqIXhBkILA0hLggFBh91IS7bOMTA2kOxJMzDmOSdhOPPMA0IOSz54O+ePnAE/++FnxjxlNvJ8xwk5DAmwGcCiiWjATMBFo2AUjIJRMFIBAEcAPRIJ9p6IAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0003-4433-8452","institution":"Johns Hopkins University","correspondingAuthor":true,"prefix":"","firstName":"Shoumitro","middleName":"","lastName":"Chatterjee","suffix":""},{"id":119657250,"identity":"e457b761-4f93-47c2-862a-3a9a614fb6a7","order_by":1,"name":"Rohit Lamba","email":"","orcid":"https://orcid.org/0000-0002-2794-9801","institution":"The Pennsylvania State University","correspondingAuthor":false,"prefix":"","firstName":"Rohit","middleName":"","lastName":"Lamba","suffix":""},{"id":119657251,"identity":"534ff2d8-a925-4c57-a2e3-abf72e0a7c8e","order_by":2,"name":"Esha Zaveri","email":"","orcid":"","institution":"Stanford University","correspondingAuthor":false,"prefix":"","firstName":"Esha","middleName":"","lastName":"Zaveri","suffix":""}],"badges":[],"createdAt":"2022-06-17 03:30:37","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-1766947/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-1766947/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41467-024-52858-6","type":"published","date":"2024-10-05T04:00:00+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":24104670,"identity":"d7e37b08-4c25-429e-90ca-611142d3ca57","added_by":"auto","created_at":"2022-07-20 17:06:25","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":120323,"visible":true,"origin":"","legend":"\u003cp\u003eOver-production of water-intensive rice and wheat despite increased groundwater stress\u003c/p\u003e\u003cp\u003eNotes: Panel (a) plots the output from a binned regression. Each observation is a district. The average annual rate of growth in area under rice cultivation is divided into twenty equal bins. The scatter plots are the means of defunct wells and growth of rice area in each bin. The line is the OLS relationship between defunct wells and the growth of rice area. In panel (b) domestic consumption is estimated as total production - net exports - government stock. Sources: Agriculture Statistics at a Glance 2020, Ministry of Agriculture; Directorate General of Commercial Intelligence and Statistics, Ministry of Commerce and Industry; and Department of Food and Public Distribution, Govt. of India\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-1766947/v1/b280368ceb8d33c869b2afa2.png"},{"id":24104674,"identity":"67c704d9-57c4-470d-88d2-6ef22198543d","added_by":"auto","created_at":"2022-07-20 17:06:26","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":225577,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePunjab: Trends in groundwater, irrigation, crops, and procurement.\u003c/strong\u003e\u003c/p\u003e\u003cp\u003eNotes: The lines in panels (a), (b), and (c) present results from local linear regressions with bandwidth 2.5. The bars at the bottom of panels (b) and (c) plot irrigated area and gross cropped area respectively and is read off the right y-axis.\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-1766947/v1/a64d4d4dd8abd5c61e6d59a8.png"},{"id":24104870,"identity":"97f6b539-e7d8-4ca6-b00e-56f33747f948","added_by":"auto","created_at":"2022-07-20 17:11:26","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":79075,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eEffects of producing and procuring rice on groundwater depth in Punjab: 1981–2003.\u003c/strong\u003e\u003c/p\u003e\u003cp\u003eNotes: The figures plot estimates from a regression of change in groundwater depth on log rice area (panel a) or log rice procurement (panel b). The unit of observation is a district×year. Each estimate is from a different regression where the change in groundwater depth has been calculated over a different horizon. For example, the estimate at \u003cem\u003eT \u003c/em\u003e= \u003cem\u003et \u003c/em\u003e+4 is from a regression where the dependent variable is the proportional change in groundwater depth between \u003cem\u003et \u003c/em\u003eand \u003cem\u003et \u003c/em\u003e+4. All regressions include district effects, year effects, AEZ-specific trends, net cropped area, population, cumulative precipitation until date \u003cem\u003eT\u003c/em\u003e, and temperature as controls. Newey-West 95% confidence intervals accounting for three year lags is represented by the bars. The blue estimates measure groundwater changes from readings taken pre-monsoon (June) and red estimates are from readings taken post-monsoon (October).\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-1766947/v1/3637fbd60d98eeb2d3cb1162.png"},{"id":24104673,"identity":"8e3c6347-c851-4708-aef6-90aefe7370d1","added_by":"auto","created_at":"2022-07-20 17:06:26","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":167402,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMadhya Pradesh: Wheat procurement, production, irrigation, and groundwater stress.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-1766947/v1/e8f3d8ab6431013b66082fe0.png"},{"id":66001950,"identity":"bff289d7-97d7-49a0-b0e8-84cb468c630c","added_by":"auto","created_at":"2024-10-06 07:05:27","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1171999,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-1766947/v1/333cbfdb-8564-4f18-b10e-260a4ff87a54.pdf"},{"id":24104672,"identity":"8235f2a0-1845-4cf3-be0f-a52ec34efc21","added_by":"auto","created_at":"2022-07-20 17:06:26","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":2123850,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryInformation.docx","url":"https://assets-eu.researchsquare.com/files/rs-1766947/v1/e3d9f16bd5f68cbca4ed289e.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"The role of farm subsidies in changing India’s water footprint","fulltext":[{"header":"Introduction","content":"\u003cp\u003eIndia has witnessed a 500% increase in groundwater consumption over the past half-century \u003cem\u003e(1-5)\u003c/em\u003e, making it one of the greatest groundwater guzzlers in the world. Unsurprisingly groundwater reserves are depleting at an alarming rate \u003cem\u003e(2,6)\u003c/em\u003e. Since the 1980s, groundwater levels have plunged by more than 8 meters on average \u003cem\u003e(7)\u003c/em\u003e. This trend is starkly visible in the alluvial aquifers of northwest India \u003cem\u003e(2,7,8)\u003c/em\u003e. In central and south India, where wells deplete and replete annually due to the nature of the hard-rock aquifer systems, these trends are less apparent. Here, other metrics of over-extraction, such as increases in the number of dry or defunct wells, reinforce the alarm \u003cem\u003e(9)\u003c/em\u003e.\u003c/p\u003e\n\u003cp\u003eIn this paper, we show that perverse incentives generated through the design of India\u0026rsquo;s agricultural output subsidies are a key source of this unfolding calamity. Output\u0026nbsp;subsidies\u0026nbsp;provided\u0026nbsp;via\u0026nbsp;guaranteed\u0026nbsp;government\u0026nbsp;procurement\u0026nbsp;of\u0026nbsp;water-intensive\u0026nbsp;crops\u0026nbsp;like\u0026nbsp;rice\u0026nbsp;and\u0026nbsp;wheat\u0026nbsp;at pre-determined prices\u0026nbsp;were introduced\u0026nbsp;in\u0026nbsp;India\u0026nbsp;during\u0026nbsp;the\u0026nbsp;green\u0026nbsp;revolution\u0026nbsp;of\u0026nbsp;the\u0026nbsp;1960s. The goal\u0026nbsp;then\u0026nbsp;was to\u0026nbsp;incentivize\u0026nbsp;the\u0026nbsp;adoption\u0026nbsp;of\u0026nbsp;high-yielding\u0026nbsp;variety\u0026nbsp;seeds to increase productivity and ensure food security for the nation. A second goal was to provide farmers with risk-free remunerative incomes \u003cem\u003e(10)\u003c/em\u003e.\u003c/p\u003e\n\u003cp\u003eWhile the\u0026nbsp;policy was\u0026nbsp;justifiable during the 1960s as India faced acute food scarcity, it has outlived its relevance. It is now causing unintended harm to food security, farmer incomes, and sustainability by accelerating groundwater depletion. The underlying mechanism has two parts.\u0026nbsp;First, the output subsidy is implemented mainly for rice and wheat and not other crops. Thus, rice and wheat cultivation has risen unchecked, engulfing other crops, since the 1960s (Fig 2c). Today, even as production of these crops exceeds national consumption needs and emergency buffer requirements by at least 30%, a guaranteed subsidy ensures continued cultivation of rice and wheat and keeps farmers from switching to other crops.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eSecond, since rice and wheat are water-intensive crops, their increased cultivation requires irrigation. Over time the increased irrigation needs have been met by personal irrigation systems like wells rather than surface water systems like dams or canals \u003cem\u003e(3, 4)\u003c/em\u003e. Together these trends have contributed to rapid declines in groundwater levels. Rice and wheat provide over 50% of the calorific intake of India\u0026rsquo;s 1.4 billion people \u003cem\u003e(11)\u003c/em\u003e and groundwater scarcity decreases crop yields (\u003cem\u003e12, 13\u003c/em\u003e). Therefore, groundwater depletion compromises food security and farmer livelihoods (14)\u0026mdash;the very objectives of the subsidy policy. Moreover, estimates suggest that in the long-run\u0026nbsp;a complete loss in access to groundwater can reduce annual crop production by 28%, dry season crop production by 51%, and cropping intensity by 68% \u003cem\u003e(15)\u003c/em\u003e. Furthermore, as groundwater also acts as a buffer against climatic variability,\u0026nbsp;its depletion undermines adaptability to climate change \u003cem\u003e(16, 17)\u003c/em\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eEven as groundwater depletes, India\u0026rsquo;s political economy contributes to the persistence of this policy \u003cem\u003e(18)\u003c/em\u003e. Without a cap or a sunset clause, a vicious cycle has ensued. The deeper the groundwater table, the costlier the cultivation due to increasing costs for well deepening and pumps. As noted above, assured government procurement has locked in most of India\u0026rsquo;s 260 million poor farmers into growing rice and wheat. Since the farmers are now wholly dependent on this policy for their incomes, the government is forced to increase the subsidy every year to cover increasing costs. This has resulted in India currently spending about USD 4 billion annually to fund the output subsidy program \u003cem\u003e(19)\u003c/em\u003e.\u003c/p\u003e\n\u003cp\u003eWhile\u0026nbsp;the\u0026nbsp;literature\u0026nbsp;has\u0026nbsp;focused\u0026nbsp;on\u0026nbsp;the\u0026nbsp;role\u0026nbsp;of\u0026nbsp;input\u0026nbsp;subsidies\u0026nbsp;like\u0026nbsp;free\u0026nbsp;power (\u003cem\u003e20, 21\u003c/em\u003e), the\u0026nbsp;role\u0026nbsp;of\u0026nbsp;output\u0026nbsp;subsides\u0026nbsp;in\u0026nbsp;driving\u0026nbsp;local\u0026nbsp;water\u0026nbsp;use has not been studied or quantified. This is both because the driving mechanisms are subtle, and there has been a lack of data to show them. Output subsidies affect groundwater indirectly by shaping crop choice decisions, unlike power subsidies that directly increase groundwater extraction using pumps. We fill this lacuna by collating data from several sources since 1981 and quantify the role of the output subsidy policy in eroding\u0026nbsp;India\u0026rsquo;s\u0026nbsp;groundwater\u0026nbsp;supplies.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eOur analysis has three parts. First, we show that increased cultivation of a water-intensive crop like rice is closely tied to increases in groundwater stress across India. In particular, we show that across Indian districts, a 1 percentage point (pp) increase in the rate of growth in the area under rice cultivation between 1996\u0026ndash;2015 is associated with a 1.7 pp increase in defunct wells. Second, we argue that an excess of rice and wheat production by almost 30% is sustained because of the government\u0026rsquo;s guaranteed purchase of these crops at higher than market prices. Third, we focus on case studies from two states with distinct aquifer systems to illustrate the main mechanism by which the output subsidy policy impacts groundwater. The hydrogeology and aquifer-specific properties can impact the extent to which pumping affects the water table locally \u003cem\u003e(9, 22)\u003c/em\u003e. For the state of Punjab, we show that the output subsidy policy has contributed to 63% of the local decline in the groundwater table. Since this region overlays deep alluvial aquifers, these effects manifest gradually. Results show that the effects of policy-induced groundwater extraction are fully reflected in declining water levels after 6 years. For the central Indian state of Madhya Pradesh, we use different metrics of groundwater stress since it predominantly overlays hard rock aquifers with low storage capacity \u003cem\u003e(9, 23)\u003c/em\u003e. Here, results show that since the inception of the output subsidy policy in 2008, incidence of dry-wells has increased by 5.3pp and demand for deep tube wells by 3.4%\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe make three contributions. First, we assemble a novel dataset that allows us to test the hypothesis linking output subsidies to the depletion of groundwater resources at a granular level. Second, we show that the output subsidy policy is a significant driver of groundwater stress not only in the deep alluvial aquifers of north-western India but also in the hard rock aquifers of central India. In general, showing that this is a pan-India issue affecting regions with very different hydrogeological conditions is important because agriculture policy related groundwater issues tend to be associated with north-western India.\u003c/p\u003e\n\u003cp\u003eThird, we show that not paying careful attention to the nature of aquifers can lead to a gross underestimation of the effects of output subsidies on groundwater stress. This entails considering the gradual nature of adjusting groundwater tables in the alluvial aquifers of Punjab and using alternate groundwater stress metrics like dry wells and demand for deep tubewells in the hard rock aquifers of Madhya Pradesh.\u003c/p\u003e\n\u003cp\u003eThis study offers cautionary insights into the possible impacts of distortionary agricultural policy on freshwater resources across the globe. Governments around the world provide broad types and levels of support to agriculture \u003cem\u003e(24)\u003c/em\u003e. Around USD 635 billion are provided annually to support agricultural production and food consumption\u0026nbsp;worldwide (\u003cem\u003e25, 26\u003c/em\u003e)\u003cem\u003e. \u0026nbsp;\u003c/em\u003eMore than 70 percent of this total support, about $456 billion, consists of support for agricultural producers. A majority (82 percent) is provided through measures that are considered most distorting \u003cem\u003e(24)\u003c/em\u003e. \u0026nbsp;Given the magnitude of resources devoted to agricultural subsidies across the world, understanding their unintended consequences on environmental outcomes remains critical for policymakers to design sustainable, fair, and efficient agricultural policies. The results in this paper suggest while the most common policy objectives of such support are to provide price stability and food security as well as to support farmers\u0026rsquo; incomes and livelihoods, they can often lead to unintended consequences that are counterproductive to policy goals. By distorting crop and irrigation choices, such support can ultimately lead to harmful spillovers on local water resources and undermine sustainable development objectives in the long-term.\u0026nbsp;\u003c/p\u003e"},{"header":"Cultivation Of Water-intensive Crops And Groundwater Stress In India","content":"\u003cp\u003eOne of the key reasons behind the groundwater stress in India is the over-production of water intensive crops like rice and wheat. For\u0026nbsp;example,\u0026nbsp;Fig\u0026nbsp;1a\u0026nbsp;shows\u0026nbsp;that\u0026nbsp;across\u0026nbsp;India,\u0026nbsp;a\u0026nbsp;1\u0026nbsp;percentage\u0026nbsp;point\u0026nbsp;(pp)\u0026nbsp;increase\u0026nbsp;in\u0026nbsp;the\u0026nbsp;rate\u0026nbsp;of\u0026nbsp;growth\u0026nbsp;in\u0026nbsp;area\u0026nbsp;under\u0026nbsp;rice cultivation\u0026nbsp;between\u0026nbsp;1996\u0026ndash;2015\u0026nbsp;is\u0026nbsp;associated\u0026nbsp;with\u0026nbsp;a\u0026nbsp;1.7\u0026nbsp;pp\u0026nbsp;increase\u0026nbsp;in\u0026nbsp;active\u0026nbsp;wells becoming\u0026nbsp;defunct. This\u0026nbsp;is\u0026nbsp;not\u0026nbsp;surprising\u0026nbsp;given\u0026nbsp;that\u0026nbsp;rice\u0026nbsp;is\u0026nbsp;one\u0026nbsp;of\u0026nbsp;the\u0026nbsp;most\u0026nbsp;water\u0026nbsp;intensive\u0026nbsp;crops cultivated using flood irrigation technique (\u003cem\u003e27, 28\u003c/em\u003e).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe relationship in Fig. 1a is likely an underestimate because major increases in rice cultivation occurred in the immediate years after the introduction of the Green Revolution during the 1960s (Supplementary Information) from when groundwater data is unavailable. To address this and other potential confounding factors like cultivation of other crops and population driven demand, we estimate multivariate statistical models controlling for such factors. Specifically, we regress the fraction of wells that were active in 1996 but were defunct by 2015 on the growth in the area under rice cultivation in that district, and control for the initial area under rice cultivation, initial gross cropped area, total land area, and population (Materials and Methods). In Table 1, cols 1\u0026ndash;2, growth in rice area is computed between 1996 and 2015, and in cols 3\u0026ndash;4 between 1966 and 2015. Table 1, cols 1\u0026ndash;2 show that the correlation from Fig. 1a survives the inclusion of various controls. Table 1, cols 3\u0026ndash;4 show that districts that had a 1 pp greater rate of growth in rice area between 1966 and 2015 had 2.4 pp additional defunct wells by 2015. This is a higher magnitude than Cols 1-2 because we also include the period when rice area was rapidly expanding.\u003c/p\u003e\n\u003cp\u003eNote that to compare districts across the country we do not use groundwater depth but number of defunct wells as the metric for groundwater stress. This is because the former is not a reliable measure to compare groundwater stress across different types of aquifer systems (9). Further, since wheat is grown in a few states (Supplementary Information, Table S1), lack of variation in the data precludes us from analyzing effects of wheat cultivation at an all-India level (Materials and Methods).\u0026nbsp;\u003c/p\u003e"},{"header":"The Role Of Output Subsidies In Over-production Of Rice \u0026 Wheat","content":"\u003cp\u003eThe\u0026nbsp;increased\u0026nbsp;groundwater\u0026nbsp;stress\u0026nbsp;from\u0026nbsp;cultivation\u0026nbsp;of\u0026nbsp;rice\u0026nbsp;and\u0026nbsp;wheat\u0026nbsp;is\u0026nbsp;only\u0026nbsp;a\u0026nbsp;symptom\u0026nbsp;of\u0026nbsp;the\u0026nbsp;problem.\u0026nbsp;The\u0026nbsp;underlying\u0026nbsp;issue\u0026nbsp;is that\u0026nbsp;the\u0026nbsp;government\u0026rsquo;s\u0026nbsp;output subsidy\u0026nbsp;policy meant\u0026nbsp;to improve\u0026nbsp;farmer incomes\u0026nbsp;and ensure\u0026nbsp;food\u0026nbsp;security is\u0026nbsp;inducing overproduction. Every\u0026nbsp;year\u0026nbsp;before\u0026nbsp;the\u0026nbsp;sowing\u0026nbsp;season\u0026nbsp;the\u0026nbsp;central\u0026nbsp;government\u0026nbsp;announces\u0026nbsp;a floor price (called the Minimum Support Price or MSP) for 23 crops\u0026nbsp;at\u0026nbsp;which\u0026nbsp;it\u0026nbsp;commits\u0026nbsp;to\u0026nbsp;procure\u0026nbsp;any\u0026nbsp;quantity\u0026nbsp;of\u0026nbsp;specified crops from farmers. The MSP has to be backed up by the actual procurement by government agencies, and this latter bureaucratic machinery acts as the true determinant of the efficacy of the policy. In practice the policy gets implemented primarily for rice and wheat\u0026mdash;the main food crops (Supplementary Information).\u003c/p\u003e\n\u003cp\u003eIn 2020 alone, the production of rice and wheat exceeded consumption by 30% and government reserves were nearly 2.5 times the required norms (Figs. 1b and S2a)\u0026mdash;an excess of 36 million tons or about 20% of India\u0026rsquo;s annual consumption. A large part will get wasted (29, 30)\u003csup\u003e\u0026nbsp;\u003c/sup\u003eand hence, the associated cost on groundwater is a deadweight loss.\u003c/p\u003e\n\u003cp\u003eSuch quantum of excess production each year should ideally put a downward pressure on market prices of rice and wheat which, in turn, would disincentivize their production and automatically contain exploitation of groundwater. However, the government\u0026rsquo;s policy of providing a guaranteed floor price (called the Minimum Support Price or MSP) to farmers cultivating rice and wheat (see Supplementary Information) has thwarted the discipline of market forces for the last half-century. This has propagated the over cultivation of these crops and the depletion of scarce groundwater such that groundwater levels have plunged to more than 30 meters below ground level (mbgl) in some regions (31).\u003c/p\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n\u003ch2 style='margin-top:0in;margin-right:7.7pt;margin-bottom:12.0pt;margin-left:0in;font-size:13px;font-family:\"Arial\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003eTable\u0026nbsp;1.\u0026nbsp;Effects\u0026nbsp;of\u0026nbsp;rice\u0026nbsp;cultivation\u0026nbsp;on\u0026nbsp;groundwater\u0026nbsp;stress:\u0026nbsp;All\u0026nbsp;India.\u003c/span\u003e\u003c/h2\u003e\n\u003cdiv align=\"center\" style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;'\u003e\n \u003ctable style=\"width:52.84%;border-collapse:collapse;border:none;\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width:29.76%;border:none;border-top:solid windowtext 1.0pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;'\u003e\u003cspan style='font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:8.1%;border:none;border-top:solid windowtext 1.0pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"7\" style=\"width:62.14%;border-top:solid windowtext 1.0pt;border-left:none;border-bottom:solid windowtext 1.0pt;border-right:none;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-family:\"Garamond\",serif;color:black;'\u003eDefunct Wells\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-family:\"Garamond\",serif;color:black;'\u003e(as a % of Active Wells in 1996)\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width:29.76%;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;'\u003e\u003cspan style='font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:8.1%;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" style=\"width:29.84%;border:none;border-bottom:solid windowtext 1.0pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-family:\"Garamond\",serif;color:black;'\u003e1996\u003c/span\u003e\u003cspan style=\"font-family:Symbol;color:black;\"\u003e-\u003c/span\u003e\u003cspan style='font-family:\"Garamond\",serif;color:black;'\u003e2015\u003c/span\u003e\u003csup\u003e\u0026dagger;\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:2.68%;border:none;border-top:solid windowtext 1.0pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" style=\"width:29.62%;border-top:solid windowtext 1.0pt;border-left:none;border-bottom:solid windowtext 1.0pt;border-right:none;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-family:\"Garamond\",serif;color:black;'\u003e1966\u003c/span\u003e\u003cspan style=\"font-family:Symbol;color:black;\"\u003e-\u003c/span\u003e\u003cspan style='font-family:\"Garamond\",serif;color:black;'\u003e2015\u003c/span\u003e\u003csup\u003e\u0026dagger;\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width:29.76%;border:none;border-bottom:solid windowtext 1.0pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;'\u003e\u003cspan style='font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:8.1%;border:none;border-bottom:solid windowtext 1.0pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:13.56%;border:none;border-bottom:solid windowtext 1.0pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-family:\"Garamond\",serif;color:black;'\u003e(1)\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:2.68%;border-top:solid windowtext 1.0pt;border-left:none;border-bottom:solid windowtext 1.0pt;border-right:none;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:13.58%;border-top:solid windowtext 1.0pt;border-left:none;border-bottom:solid windowtext 1.0pt;border-right:none;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-family:\"Garamond\",serif;color:black;'\u003e(2)\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:2.68%;border:none;border-bottom:solid windowtext 1.0pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:13.56%;border:none;border-bottom:solid windowtext 1.0pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-family:\"Garamond\",serif;color:black;'\u003e(3)\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:2.68%;border-top:solid windowtext 1.0pt;border-left:none;border-bottom:solid windowtext 1.0pt;border-right:none;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:13.36%;border-top:solid windowtext 1.0pt;border-left:none;border-bottom:solid windowtext 1.0pt;border-right:none;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-family:\"Garamond\",serif;color:black;'\u003e(4)\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width:29.76%;border:none;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;'\u003e\u003cspan style='font-family:\"Garamond\",serif;color:black;'\u003eRice Area Growth\u003c/span\u003e\u003csup\u003e\u0026dagger;\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:8.1%;border:none;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:13.56%;border:none;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e1.747\u003csup\u003e***\u003c/sup\u003e\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:2.68%;border:none;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:13.58%;border:none;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e1.704\u003csup\u003e***\u003c/sup\u003e\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:2.68%;border:none;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:13.56%;border:none;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e2.344\u003csup\u003e***\u003c/sup\u003e\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:2.68%;border:none;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:13.36%;border:none;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e2.366\u003csup\u003e***\u003c/sup\u003e\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width:29.76%;border:none;border-bottom:solid windowtext 1.0pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;'\u003e\u003cspan style='font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:8.1%;border:none;border-bottom:solid windowtext 1.0pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:13.56%;border:none;border-bottom:solid windowtext 1.0pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e(0.445)\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:2.68%;border:none;border-bottom:solid windowtext 1.0pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:13.58%;border:none;border-bottom:solid windowtext 1.0pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e(0.424)\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:2.68%;border:none;border-bottom:solid windowtext 1.0pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:13.56%;border:none;border-bottom:solid windowtext 1.0pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e(0.419)\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:2.68%;border:none;border-bottom:solid windowtext 1.0pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:13.36%;border:none;border-bottom:solid windowtext 1.0pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e(0.406)\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width:29.76%;border:none;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;'\u003e\u003cspan style='font-family:\"Garamond\",serif;color:black;'\u003eN\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:8.1%;border:none;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:13.56%;border:none;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e257\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:2.68%;border:none;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:13.58%;border:none;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e251\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:2.68%;border:none;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:13.56%;border:none;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e283\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:2.68%;border:none;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:13.36%;border:none;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e281\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width:29.76%;border:none;border-bottom:solid windowtext 1.5pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;'\u003e\u003cspan style='font-family:\"Garamond\",serif;color:black;'\u003eAdj. \u003cem\u003eR\u003c/em\u003e\u003csup\u003e2\u003c/sup\u003e\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:8.1%;border:none;border-bottom:solid windowtext 1.5pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:13.56%;border:none;border-bottom:solid windowtext 1.5pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e0.04\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:2.68%;border:none;border-bottom:solid windowtext 1.5pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:13.58%;border:none;border-bottom:solid windowtext 1.5pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e0.16\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:2.68%;border:none;border-bottom:solid windowtext 1.5pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:13.56%;border:none;border-bottom:solid windowtext 1.5pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e0.10\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:2.68%;border:none;border-bottom:solid windowtext 1.5pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width:13.36%;border:none;border-bottom:solid windowtext 1.5pt;padding:0in 0in 0in 0in;\"\u003e\n \u003cp style='margin:0in;font-size:13px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-family: \"Garamond\",serif;color:black;'\u003e0.24\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;margin-top:0in;margin-right:121.5pt;margin-bottom:12.0pt;margin-left:121.5pt;text-align:justify;'\u003e\u003cspan style='font-size:12px;font-family:\"Garamond\",serif;color:black;'\u003eNotes: Each observation is a district. Robust standard errors clustered at the district level reported in parentheses.\u0026nbsp;\u003c/span\u003e\u003csup\u003e\u003cspan style='font-size:12px;font-family:\"Cambria Math\",serif;color:black;'\u003e\u0026lowast;\u003c/span\u003e\u003c/sup\u003e\u003cspan style='font-size:12px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;p \u0026lt; 0.10,\u0026nbsp;\u003c/span\u003e\u003csup\u003e\u003cspan style='font-size:12px;font-family:\"Cambria Math\",serif;color:black;'\u003e\u0026lowast;\u0026lowast;\u003c/span\u003e\u003c/sup\u003e\u003cspan style='font-size:12px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;p \u0026lt; 0.05,\u003csup\u003e\u0026nbsp;\u003c/sup\u003e\u003c/span\u003e\u003csup\u003e\u003cspan style='font-size:12px;font-family:\"Cambria Math\",serif;color:black;'\u003e\u0026lowast;\u0026lowast;\u0026lowast;\u003c/span\u003e\u003c/sup\u003e\u003csup\u003e\u003cspan style='font-size:12px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/sup\u003e\u003cspan style='font-size:12px;font-family:\"Garamond\",serif;color:black;'\u003ep \u0026lt; 0.01. The dependent variable in all columns is the percentage of wells that were active in 1996 but were defunct in 2015. Average Annual Rate of Growth of Rice Area over 1996\u0026ndash;2015 in cols 1\u0026ndash;2 and over 1966\u0026ndash;2015 in cols 3\u0026ndash;4. Growth is measured in percent. Cols 2 and 4 also include controls for initial gross cropped area, initial rice area, land area of the district, and population.\u003c/span\u003e\u003c/p\u003e"},{"header":"Case Studies","content":"\u003cp\u003eThe main challenge in quantifying the role of MSP and public procurement in inducing the production of rice and wheat and the subsequent effect on groundwater is the lack of disaggregated and consistent data for key variables of interest\u0026mdash;quantity of rice or wheat procured by government, cropped area, and groundwater\u0026mdash;from a period when the policy-induced thrust toward the cultivation of rice and wheat occurred. To circumvent this issue, we focus on two states for which we were able to collect this data in different but relevant time periods, i.e., close to when the policy was instituted in the respective states\u0026mdash;Punjab (1981\u0026ndash;2003) and Madhya Pradesh (2002\u0026ndash;2016) (Materials and Methods).\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCase Study I: Punjab\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAlthough Punjab is one of the most agriculturally productive and irrigated regions of the country, it has also witnessed one of the largest increases in groundwater stress in the world \u003cem\u003e(32)\u003c/em\u003e. The average groundwater depth increased from 4.82 mbgl in 1973 to 14.55 mbgl in 2016. By 1999, 78.6% of all dug wells that were active in 1973 had become defunct (fig 2a). Additionally, over 75% of the area is considered overexploited according to the Central Ground Water Board (23). The historical root for this depletion lies in the adoption of high-yielding variety (HYV) wheat and rice during the Green Revolution in the 1960s. These varieties replaced local wheat varieties, cotton, maize, and oilseeds (Fig. 2c and Supplementary Information, Table S2) and required more intensive irrigation. The increased irrigation came from groundwater (dug wells and tubewells) causing its depletion (Fig. 2b). Underlying this process was the policy of assured government procurement at MSP of rice and wheat (Fig. 2d), which incentivized cultivation of these water-intensive crops over others even as India amassed a surplus of rice and wheat.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe earliest year for which district-level procurement data are available is 1981. By this time, the wheat area had become stable with little change over time. On the other hand, rice area exhibits an increasing trend (Fig. 2c) and remains the focus of the analysis. Using regression models with a rich set of controls and fixed effects (Materials and Methods) we first document an output response. Farmers responded to a doubling of rice procurement by increasing area under rice cultivation by at least 54% in the following year (Supplementary Information, Table S3).\u003c/p\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n\u003cp\u003eSecond, to show how increased rice cultivation in turn impacts groundwater levels, we regress changes in groundwater levels on the log rice area (Materials and Methods). Punjab overlies thick and deep alluvial aquifers such that declines in groundwater levels persist and adjust gradually (22). To capture these dynamics, we compute changes in groundwater level over multiple horizons and estimate a separate model for each horizon and present these results graphically in Fig. 3a\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFor example, the estimate at \u003cem\u003eT\u0026nbsp;\u003c/em\u003e= \u003cem\u003et\u0026nbsp;\u003c/em\u003e+4 is from a regression where the dependent variable is the proportional change in groundwater depth between an initial year \u003cem\u003et\u0026nbsp;\u003c/em\u003eand \u003cem\u003et\u0026nbsp;\u003c/em\u003e+4, i.e. over four years (∆= 4). The main co-variate of interest is either log rice area (Fig. 3a) or log rice procurement (Fig. 3b) in year \u003cem\u003et\u003c/em\u003e. Thus, the estimates show how the same \u0026ldquo;shock\u0026rdquo;\u0026mdash;rice area or procurement in year \u003cem\u003et\u003c/em\u003e\u0026mdash;effects groundwater levels over different horizons. Estimates using pre-monsoon readings are in blue and those using post-monsoon readings in red. In all models, we control for various observable factors that could bias our estimates like population, net cropped area, precipitation, and temperature. To control for time-invariant, district-specific unobserved factors like geography, we use district fixed effects. We control for year fixed effects to isolate unobserved district-invariant, time-varying factors. We also include Agro Ecological Zone (AEZ)-specific time trends that account for AEZ-specific time-varying factors like changes in socio-economic indicators that could bias our results. Our results are also robust to explicitly including wheat area as a control, which is another water-intensive crop and is grown in the following cropping season. The rich set of controls helps us estimate the effect of rice cultivation on groundwater depth.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFig. 3a depicts how the effects on groundwater table depths show up over time. A doubling of rice area (i.e. an increase by 100%) causes a fall in groundwater depth of 6 pp over a year and 24 pp over three years, as measured by pre-monsoon readings. Similar patterns are observed for post-monsoon readings. The effect keeps increasing until about six years and then stabilizes at 63 pp for pre-monsoon and 99 pp for post-monsoon readings.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFig. 3b, is the reduced form of this causal chain where we regress proportional changes in groundwater depth directly on log rice procurement with the same set of controls. The effects show a similar evolutionary pattern. By the sixth year, which is when the effects stabilize, a doubling of rice procurement results in a 46\u0026ndash;70 pp fall in groundwater levels. The gradual adjustment in groundwater table depths is an important empirical finding and worth emphasizing. A naive analysis that ignores the adjustment process, and focuses solely on the contemporaneous relationship between changes in groundwater level declines and procurement would underestimate the effects of the output subsidy policy on groundwater level declines by 90%.\u003c/p\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n\u003cp\u003eAverage pre-monsoon groundwater depth in Punjab fell by 65% between 1981 and 2003. During this period, rice procurement has increased at about 3.5% per year from 4.4 to 12.8 million tons. Our results predict a fall of 1.6 pp per year in groundwater depth due to this mechanism or a total fall of 41% over 22 years. In other words, increased rice cultivation that was induced by government procurement of rice explains 63% of the fall in groundwater levels in Punjab between 1981 and 2003.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo confirm that our estimates are not picking up spurious correlations, we estimate placebo models in which we regress changes in groundwater depths between years \u003cem\u003ek \u0026lt; t\u0026nbsp;\u003c/em\u003eand \u003cem\u003et\u0026nbsp;\u003c/em\u003eon rice area or rice procurement in year \u003cem\u003et\u003c/em\u003e. In line with the fact that rice cultivation or procurement today should not impact groundwater depths in previous periods, we find no association in our placebo estimates (see placebo estimates in Figs. 3a and 3b). This robustness check reinforces the validity of our analysis. Finally, it is worth reiterating that the output subsidy program keeps the wheat cultivated area in Punjab high and stable thus resulting in a continuous usage of groundwater for its irrigation and therefore its depletion. However, since there is little change in the wheat area post-1981, a lack of variation precludes us from estimating marginal effects of wheat cultivation or procurement on groundwater.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCase Study II: Madhya Pradesh\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n\u003cp\u003eHistorically, the government agencies in Madhya Pradesh did not procure either wheat or rice even when market prices fell below the MSP. But from 2008, the state announced a bonus on top of the national MSP and substantially expanded its procurement operations of wheat. The potency of the output subsidy policy is driven in large part by the procurement machinery of the state at the local level. With strong procurement, the policy became highly active in Madhya Pradesh. Before 2007, the largest volume purchased by state agencies in any year was 0.54 million tons (mt). In 2007 procurement was a mere 0.057 mt. In 2008, this exponentially jumped to 2.4 mt\u0026mdash;a 40-fold increase \u003cem\u003e(33)\u003c/em\u003e.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe policy was introduced when state elections were due later in the year. Procurement operations were also more concentrated in districts that were key producers of wheat. Thus, there was widespread belief that this was an election year gift, and therefore, the farmers did not immediately increase wheat cultivation \u003cem\u003e(33)\u003c/em\u003e. Over time, however, as wheat procurement became a fixture in subsequent seasons and spread to other districts, this belief was shed and along with procurement there was also a concomitant increase in area under wheat cultivation (Fig. 4a). Between 2000 and 2007, the area under wheat cultivation grew at 3% annually, which almost doubled to 5.8% per year between 2008 and 2015.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eSome of the growth in wheat area between 2000 and 2008 can be attributed to the improvements in irrigation systems (reliable power for tubewells and completion of canal irrigation projects) that were being made by the state government beginning in early 2000 (34). However, the increased wheat cultivation because of the procurement policy put additional pressure on irrigation demand. The average annual growth in wheat irrigated area was 6.5% between 1991 and 2007 but after 2008, the wheat irrigated area began to increase by 7.9% each year. As was the case in Punjab, much of the new irrigation came from groundwater (wells and tubewells) rather than surface water (Fig. 4b). This change has gradually started increasing groundwater stress, although magnitudes are comparatively low since this is a relatively new policy as compared to fifty years of procurement operations in Punjab.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eMadhya Pradesh is dominated by hard-rock and mixed aquifers where measurement of groundwater stress is not straightforward. Shallow hard-rock aquifers deplete and replete annually and long-term water level trends are less apparent. It is well known that in such aquifers, groundwater stress can increase even though average groundwater depth may not be increasing \u003cem\u003e(9)\u003c/em\u003e. A robust measure of groundwater stress in this region is the need for deep tubewells (with depth \u0026gt;70m). Deep tubewells are expensive and risky to construct. Their failure causes a precipitous decline in the long-term income of the farmers trying to install them \u003cem\u003e(35)\u003c/em\u003e. Farmers install deep tubewells when they do not have access to surface water or groundwater at shallower depths. Thus, an increase in the incidence of defunct dugwells, that are shallow, with a concomitant increase in deep tubewells is a reliable indicator of groundwater stress (Fig. 4c).\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe exogenous and phased introduction of procurement operations in this state provides a natural experiment to estimate the causal impact of this policy on groundwater stress. Given the hydrogeology, we use three different metrics to measure groundwater stress: the proportional change in groundwater depth pre-sowing and post-harvest, the fraction of wells that run dry post-wheat harvest, and the construction of deep tubewells.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo estimate causal effects, we regress each of these measures on the log of wheat procured and the log of wheat procured interacted with a post-2008 indicator. The unit of observation is a district \u0026times; year. All specifications always have district fixed effects that control for differences in time invariant, district-specific factors that could be correlated with the policy and bias our results. This includes differences in aquifer systems and the fact that procurement started more intensely in districts that already had wheat cultivation and then spread to other districts. We also control for district-specific time-trends that partial out any pre-existing trends in wheat procurement and groundwater stress. Further, we take into account seasonal rainfall and temperature, and for robustness, we also control for total irrigated area in the district. This battery of controls allows us to estimate the causal impact of the policy by essentially comparing groundwater stress in the districts before and after the policy change and relating it to the intensity of procurement operations. The coefficient on the interaction term estimates the causal impact of a 1% increase in wheat procurement on groundwater stress post-2008 as compared to before the policy was in place (Materials and Methods).\u003c/p\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n\u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cstrong\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003eTable 2. Effects of Wheat Procurement in Madhya Pradesh\u003c/span\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cdiv align=\"center\" style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;'\u003e\n \u003ctable style=\"border: none;width:373.95pt;border-collapse:collapse;\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 115.25pt;border-color: windowtext currentcolor currentcolor;border-style: solid none none;border-width: 1pt medium medium;border-image: none 100% / 1 / 0 stretch;padding: 0in;height: 17.05pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57.1pt;border-color: windowtext currentcolor;border-style: solid none;border-width: 1pt medium;padding: 0in;height: 17.05pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e∆GWL\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 0.1in;border-color: windowtext currentcolor currentcolor;border-style: solid none none;border-width: 1pt medium medium;border-image: none 100% / 1 / 0 stretch;padding: 0in;height: 17.05pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1.2in;border-color: windowtext currentcolor;border-style: solid none;border-width: 1pt medium;padding: 0in;height: 17.05pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003eDry Wells (prop)\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 0.1in;border-color: windowtext currentcolor currentcolor;border-style: solid none none;border-width: 1pt medium medium;border-image: none 100% / 1 / 0 stretch;padding: 0in;height: 17.05pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1.4in;border-color: windowtext currentcolor;border-style: solid none;border-width: 1pt medium;padding: 0in;height: 17.05pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003elog Deep Tubewells\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 115.25pt;border-color: currentcolor currentcolor windowtext;border-style: none none solid;border-width: medium medium 1pt;border-image: none 100% / 1 / 0 stretch;padding: 0in;height: 17.05pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57.1pt;border-color: currentcolor currentcolor windowtext;border-style: none none solid;border-width: medium medium 1pt;border-image: none 100% / 1 / 0 stretch;padding: 0in;height: 17.05pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e(1)\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 0.1in;border-color: currentcolor currentcolor windowtext;border-style: none none solid;border-width: medium medium 1pt;border-image: none 100% / 1 / 0 stretch;padding: 0in;height: 17.05pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1.2in;border-color: currentcolor currentcolor windowtext;border-style: none none solid;border-width: medium medium 1pt;border-image: none 100% / 1 / 0 stretch;padding: 0in;height: 17.05pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e(2)\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 0.1in;border-color: currentcolor currentcolor windowtext;border-style: none none solid;border-width: medium medium 1pt;border-image: none 100% / 1 / 0 stretch;padding: 0in;height: 17.05pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1.4in;border-color: currentcolor currentcolor windowtext;border-style: none none solid;border-width: medium medium 1pt;border-image: none 100% / 1 / 0 stretch;padding: 0in;height: 17.05pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e(3)\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 115.25pt;border: medium none;padding: 0in;height: 17.05pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003elog Wheat Proc\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57.1pt;border: medium none;padding: 0in;height: 17.05pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e-0.012\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 0.1in;border: medium none;padding: 0in;height: 17.05pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1.2in;border: medium none;padding: 0in;height: 17.05pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e-0.001\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 0.1in;border: medium none;padding: 0in;height: 17.05pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1.4in;border: medium none;padding: 0in;height: 17.05pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e-0.000\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 115.25pt;padding: 0in;height: 12.3pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57.1pt;padding: 0in;height: 12.3pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e(0.022)\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 0.1in;padding: 0in;height: 12.3pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1.2in;padding: 0in;height: 12.3pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e(0.008)\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 0.1in;padding: 0in;height: 12.3pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1.4in;padding: 0in;height: 12.3pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e(0.010)\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 115.25pt;padding: 0in;height: 14.45pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003elog Wheat Proc.\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57.1pt;padding: 0in;height: 14.45pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e0.039\u003c/span\u003e\u003csup\u003e\u003cspan style='font-size:16px;font-family:\"Cambria Math\",serif;color:black;'\u003e\u0026lowast;\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 0.1in;padding: 0in;height: 14.45pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1.2in;padding: 0in;height: 14.45pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e0.076\u003c/span\u003e\u003csup\u003e\u003cspan style='font-size:16px;font-family:\"Cambria Math\",serif;color:black;'\u003e\u0026lowast;\u0026lowast;\u0026lowast;\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 0.1in;padding: 0in;height: 14.45pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1.4in;padding: 0in;height: 14.45pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e0.048\u003c/span\u003e\u003csup\u003e\u003cspan style='font-size:16px;font-family:\"Cambria Math\",serif;color:black;'\u003e\u0026lowast;\u0026lowast;\u0026lowast;\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 115.25pt;border-color: currentcolor currentcolor windowtext;border-style: none none solid;border-width: medium medium 1pt;border-image: none 100% / 1 / 0 stretch;padding: 0in;height: 15.9pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp; \u0026times; Post-2008\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57.1pt;border-color: currentcolor currentcolor windowtext;border-style: none none solid;border-width: medium medium 1pt;border-image: none 100% / 1 / 0 stretch;padding: 0in;height: 15.9pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e(0.023)\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 0.1in;border-color: currentcolor currentcolor windowtext;border-style: none none solid;border-width: medium medium 1pt;border-image: none 100% / 1 / 0 stretch;padding: 0in;height: 15.9pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1.2in;border-color: currentcolor currentcolor windowtext;border-style: none none solid;border-width: medium medium 1pt;border-image: none 100% / 1 / 0 stretch;padding: 0in;height: 15.9pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e(0.007)\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 0.1in;border-color: currentcolor currentcolor windowtext;border-style: none none solid;border-width: medium medium 1pt;border-image: none 100% / 1 / 0 stretch;padding: 0in;height: 15.9pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1.4in;border-color: currentcolor currentcolor windowtext;border-style: none none solid;border-width: medium medium 1pt;border-image: none 100% / 1 / 0 stretch;padding: 0in;height: 15.9pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e(0.014)\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 115.25pt;padding: 0in;height: 12.2pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003eN\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57.1pt;padding: 0in;height: 12.2pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e481\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 0.1in;padding: 0in;height: 12.2pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1.2in;padding: 0in;height: 12.2pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e481\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 0.1in;padding: 0in;height: 12.2pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1.4in;padding: 0in;height: 12.2pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e407\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 115.25pt;padding: 0in;height: 12.55pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003eClusters\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57.1pt;padding: 0in;height: 12.55pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e37\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 0.1in;padding: 0in;height: 12.55pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1.2in;padding: 0in;height: 12.55pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e37\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 0.1in;padding: 0in;height: 12.55pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1.4in;padding: 0in;height: 12.55pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e37\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 115.25pt;border-color: currentcolor currentcolor windowtext;border-style: none none solid;border-width: medium medium 1pt;border-image: none 100% / 1 / 0 stretch;padding: 0in;height: 16.65pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003eAdj. \u003cem\u003eR\u003c/em\u003e\u003csup\u003e2\u003c/sup\u003e\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57.1pt;border-color: currentcolor currentcolor windowtext;border-style: none none solid;border-width: medium medium 1pt;border-image: none 100% / 1 / 0 stretch;padding: 0in;height: 16.65pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e.51\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 0.1in;border-color: currentcolor currentcolor windowtext;border-style: none none solid;border-width: medium medium 1pt;border-image: none 100% / 1 / 0 stretch;padding: 0in;height: 16.65pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1.2in;border-color: currentcolor currentcolor windowtext;border-style: none none solid;border-width: medium medium 1pt;border-image: none 100% / 1 / 0 stretch;padding: 0in;height: 16.65pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e.27\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 0.1in;border-color: currentcolor currentcolor windowtext;border-style: none none solid;border-width: medium medium 1pt;border-image: none 100% / 1 / 0 stretch;padding: 0in;height: 16.65pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 1.4in;border-color: currentcolor currentcolor windowtext;border-style: none none solid;border-width: medium medium 1pt;border-image: none 100% / 1 / 0 stretch;padding: 0in;height: 16.65pt;vertical-align: top;\"\u003e\n \u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Garamond\",serif;color:black;'\u003e.99\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp style='margin:0in;font-size:15px;font-family:\"Times New Roman\",serif;text-align:justify;'\u003e\u003cspan style='font-size:13px;font-family:\"Garamond\",serif;color:black;'\u003eNotes: Each observation is a district\u0026times;year. Robust standard errors clustered at the district level reported in parentheses. ∆GWL is the proportional change in groundwater depth between November in year t-1 and May in year t. Dry wells are the fraction of monitoring wells that are dry in the months after wheat harvest (June\u0026ndash;September). All regression models include district fixed effects, district specific linear time trends, and controls for temperature and precipitation.\u0026nbsp;\u003c/span\u003e\u003csup\u003e\u003cspan style='font-size:13px;font-family:\"Cambria Math\",serif;color:black;'\u003e\u0026lowast;\u003c/span\u003e\u003c/sup\u003e\u003csup\u003e\u003cspan style='font-size:13px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/sup\u003e\u003cem\u003e\u003cspan style='font-size:13px;font-family:\"Garamond\",serif;color:black;'\u003ep \u0026lt;\u0026nbsp;\u003c/span\u003e\u003c/em\u003e\u003cspan style='font-size:13px;font-family:\"Garamond\",serif;color:black;'\u003e0\u003cem\u003e.\u003c/em\u003e10,\u0026nbsp;\u003c/span\u003e\u003csup\u003e\u003cspan style='font-size:13px;font-family:\"Cambria Math\",serif;color:black;'\u003e\u0026lowast;\u0026lowast;\u003c/span\u003e\u003c/sup\u003e\u003csup\u003e\u003cspan style='font-size:13px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/sup\u003e\u003cem\u003e\u003cspan style='font-size:13px;font-family:\"Garamond\",serif;color:black;'\u003ep \u0026lt;\u0026nbsp;\u003c/span\u003e\u003c/em\u003e\u003cspan style='font-size:13px;font-family:\"Garamond\",serif;color:black;'\u003e0\u003cem\u003e.\u003c/em\u003e05,\u0026nbsp;\u003c/span\u003e\u003csup\u003e\u003cspan style='font-size:13px;font-family:\"Cambria Math\",serif;color:black;'\u003e\u0026lowast;\u0026lowast;\u0026lowast;\u003c/span\u003e\u003c/sup\u003e\u003csup\u003e\u003cspan style='font-size:13px;font-family:\"Garamond\",serif;color:black;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/sup\u003e\u003cem\u003e\u003cspan style='font-size:13px;font-family:\"Garamond\",serif;color:black;'\u003ep \u0026lt;\u0026nbsp;\u003c/span\u003e\u003c/em\u003e\u003cspan style='font-size:13px;font-family:\"Garamond\",serif;color:black;'\u003e0\u003cem\u003e.\u003c/em\u003e01.\u003c/span\u003e\u003c/p\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n\u003cp\u003eTable 2 shows that pre-policy, the little wheat procurement that occurred had no relationship with either measure of groundwater stress. Post-2008 however, a doubling of wheat procurement caused a 3.9 pp fall in groundwater depth (col 1), a 7.6 pp increase in the incidence of dry wells (col 2), and a 4.8% increase in construction of deep tubewells (col 3). All these results consistently show an increase in groundwater stress. The fact that we can triangulate the results across different metrics of groundwater stress increases the reliability of our results in this hard-rock aquifer region. Between 2007 and 2016, wheat procurement in Madhya Pradesh increased by almost 70% (from 0.057 mt to 4 mt). Our results therefore imply that the policy increased the incidence of dry wells by 5.3 pp and the need for deep tubewells by 3.4% during this period. These effects are significant because they have happened over a relatively short span of eight years. As wheat procurement continues, groundwater stress in Madhya Pradesh will increase exponentially as in Punjab. The statistical model in col 1 is directly comparable to the model we used for Punjab (Fig. 3b) for a one-year horizon, ∆= \u003cem\u003eT\u0026nbsp;\u003c/em\u003e\u0026minus;\u003cem\u003et\u0026nbsp;\u003c/em\u003e= 1.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWhy should the marginal effect of wheat procurement vary before and after 2008? This is where the mechanism is crucial. Credible wheat procurement incentivizes farmers to grow more wheat over less water-intensive crops, which increases groundwater stress. In Supplementary Information, Table S4, we provide evidence for this core mechanism. We regress log wheat area, log wheat irrigated area, and log area under pulses (the other important crop) on lagged log wheat procurement and lagged log wheat procurement interacted with a post-2008 indicator. We use year fixed effects to isolate time-varying aggregate shocks like aggregate supply, price volatility, and climate that could impact procurement and wheat cultivation. We explicitly control for rainfall, temperature, irrigation, and area under other crops for robustness (Materials and Methods).\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eOur estimates show that the policy increased the marginal effect of wheat procurement on wheat cultivation in the following year by an additional 13.5% (Supplementary Information Table S4, cols 1\u0026ndash;2). A doubling of procurement post-2008 also resulted in a 22% increase in irrigated wheat area due to the policy. Col 4 shows that the increase in wheat area partially came at the cost of a reduction in area under pulses, a less water-intensive crop. Fig. 4a, however, shows that a greater amount of new area came under wheat cultivation increasing groundwater stress. As an intense shift toward highly irrigated wheat occurred post-2008, we see differences in the marginal effects pre- and post-2008 in table 2.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eTaken together, the results of this study offer a new understanding of the role that India\u0026rsquo;s four decades-long assured grain procurement program has played on groundwater stress and highlights important lessons for policymakers focused on designing environmentally sustainable agricultural policies. This is critically important as Indian policymakers today are vigorously debating the future of these policies as a part of the current government\u0026rsquo;s plan to liberalize India\u0026rsquo;s agricultural markets and in reaction to the developments in Ukraine that have affected global wheat markets. They face fierce demands from farmers for even greater support, as fertilizer prices continue to rise.\u003c/p\u003e \u003cp\u003eGovernments worldwide provide billions in subsidies to support farmers. The mechanism by which these subsidies are provided is as important as the quantum of subsidies. Many nations choose to provide subsidies as income transfers or as insurance against price crashes. These methods are less distortionary. In India, pork barrel considerations have resulted in output subsidies being limited to water-intensive rice and wheat. As compared to total agricultural subsidies that account for about 2-2.5% of GDP (\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e), expenditure on productivity improving public goods like agriculture research and infrastructure is negligible (\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eIn this article, we argue that this has had deleterious consequences for the environment, and sustainability. The thick alluvial aquifers of northern India are a key buffer against climate change-induced weather variability. As their recharge takes centuries, their depletion has exposed India\u0026rsquo;s most productive regions to the possibility of desertification.\u003c/p\u003e \u003cp\u003eThere are other consequences as well. The most important is on nutrition. In the 1960s, India faced food deficiencies. Over time, agricultural policy (output subsidies) has become coupled with food policy (consumption subsidies). The grains procured under the output subsidy program are provided as subsidized food to the poor. The latter is then used as a justification to continue with the former (\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e). While increased cereal productivity that followed the Green Revolution has ensured cereal availability, the diversity in food systems has suffered (\u003cspan additionalcitationids=\"CR40\" citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e). The Green Revolution crowded out the production of other nutrient-rich crops like coarse cereals and pulses (\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e, \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e). This combined with consumption subsidies made nutrition-rich food relatively more expensive and crowded them out from diets as well (\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e). The National Food Security Act of 2013 furthered this by codifying subsidies for staples into law. As a result, the Indian population today faces \u0026ldquo;hidden hunger\u0026rdquo; in nutrition (\u003cspan additionalcitationids=\"CR44 CR45 CR46\" citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eFinally, there is a trade paradox. India has one of the lowest per-capita availabilities of freshwater reserves. Given such scarce water endowments, economic theory would suggest that the country import water-intensive crops and export the less water-intensive crops. On the contrary, even as taps run dry in major Indian cities, induced by agricultural policy India exports 25\u0026times;10\u003csup\u003e9\u003c/sup\u003e m\u003csup\u003e3\u003c/sup\u003e in virtual water every year (\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e). Recent estimates suggest that India accounts for 12 percent of the global groundwater depletion that is embedded in international food trade (\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e). At this rate, India is likely to loose its entire available water in less than 1000 years (\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e).\u003c/p\u003e \u003cp\u003ePolicymakers in India thus need to rethink how they provide subsidies to farmers and consumers for securing their own future. Recent research has tried to provide optimal criteria of procurement from the point of view of maximizing nutrition and ensuring food security while saving water resources (\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e). However, implementing such criteria will be challenging partly due to limited state capacity. But also because political economy constraints make any change difficult. The new income transfer scheme PM-KISAN is probably a good step forward as it is independent of the farmers\u0026rsquo; crop choice. But presently, it is provided in addition to and not instead of the output subsidy. More importantly, farmers have a deep mistrust in institutions as coverage of most farmer welfare schemes is far from perfect. In the Green Revolution states of Punjab and Haryana, farmers have been used to the MSP procurement apparatus of rice and wheat for half a century. This is also where the groundwater stress is the most severe. Successfully moving away to alternative systems will require trust-building and convincing farmers that alternative systems of support and subsidies that are not linked to water intensive-crops can work.\u003c/p\u003e"},{"header":"Materials And Methods","content":"\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003eData\u003c/h2\u003e \u003cp\u003eWe compiled several different datasets on crop production, area, and irrigation (apportioned district-level data on Indian agriculture from the International Crops Research Institute for the Semi-Arid Tropics (ICRISAT)), government procurement of rice and wheat (from the Food Corporation of India and state agencies such as the State Civil Supplies Corporation), groundwater levels and stress (from the Central Groundwater Board of India and the state groundwater board of Punjab). We constructed district-level weather data by averaging gridded temperature and precipitation data from the Indian Meteorological Department over the growing season of crops.\u003c/p\u003e \u003cp\u003eData on crop production and area from the ICRISAT database spans the years from 1966 to 2015. Where necessary, maps corresponding to 1966 district boundaries are used so that districts and states that split after 1966 are considered together to allow comparability over time.\u003c/p\u003e \u003cp\u003eDistrict-level procurement data for India is generally not available. We were able to obtain this data for two states, Punjab and Madhya Pradesh. For Punjab, district-level data spans the years from 1981 to 2018, and for Madhya Pradesh from the years 2002 to 2016.\u003c/p\u003e \u003cp\u003eObservation well data for the country from 1996 to 2016 was obtained from the Central Groundwater Board\u003c/p\u003e \u003cp\u003e(CGWB) database of monitoring wells, which contains water level measurements recorded four times a year (January, May, August, and November) for 30,311 wells. Of these, 12,837 wells were active in 1996, representing wells with the longest possible records in the database. A separate set of observation well data for Punjab from 1973 to 2003 was obtained from the monitoring wells of the state groundwater board of Punjab, which contains water level measurements on a bi-annual basis (pre- and post-monsoon). Drawdown is measured in observation wells before (June) and after (November) the annual monsoonal recharge season (June\u0026ndash; September). It should be noted that water tables reported by monitoring wells may not reflect exactly those reported by farmers in their irrigation wells. Therefore, the averaged water levels at the district level are indicators of water fluctuations rather than absolute value of the water tables actually experienced by farmers.\u003c/p\u003e \u003cdiv id=\"Sec10\" class=\"Section3\"\u003e \u003ch2\u003eMeasures of Groundwater Stress\u003c/h2\u003e \u003cp\u003eIn addition to averaged groundwater level data at the district level, we use two additional metrics of groundwater stress, namely defunct and dry wells. The literature (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e), has shown how missing data in well records carry critical information on groundwater stress that is completely missed in analyses that routinely filter out this information. Missing data in a well record can occur in two ways: first, the well goes defunct and stops collecting data permanently during the analysis time frame, or second, the well records no data in multiple intermediate months within the time frame. The underlying cause of such missing data can be either physical, where the water level in the well falls permanently or temporarily below the well screen depth, or logistical, where operators neglect maintaining monitoring wells, or they collect or record the data inadequately. The physical change in water level is one of the key underlying reasons for missing data. To overcome the non-random occurrence of missing well data records, two alternative indicators of groundwater stress are used. The percentage of defunct wells is measured at the end of the sample period and is estimated as the proportion of monitoring wells that started collecting data at the beginning of the sample period (active wells) but then permanently stopped collecting data within the sample period. Following the literature (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e), we add the condition that a well should not have recorded data for (at least) the four years preceding the last year in the sample period to be considered defunct. The percentage of dry wells is the proportion of monitoring wells that started data at the beginning of the sample period (active wells) but have missing data and did not record information in the time frame under consideration such as a given season and year. Dry wells could therefore recover in the wet years but lack of water in dry years indicates groundwater stress. Both metrics are indicative of the degree of groundwater stress in the region.\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section4\"\u003e \u003ch2\u003eReasons for using different metrics of groundwater stress in different regions\u003c/h2\u003e \u003cp\u003eFor a cross-district comparison at the national level we use defunct wells as the metric for groundwater stress rather than groundwater levels because differences in aquifer types make changes in groundwater depth incomparable across regions (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e). Defunct wells or dry wells are a more consistent metric of groundwater stress for comparisons across regions.\u003c/p\u003e \u003cp\u003eGroundwater depth as a measure of stress is the most appropriate for the deep alluvial aquifers of Punjab. However, here groundwater levels persist and adjust gradually (\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e). Hence, we study changes over different horizons in groundwater levels to assess stress.\u003c/p\u003e \u003cp\u003eMadhya Pradesh predominantly has hard rock and mixed systems that deplete and replete annually. Thus, in these aquifer systems water levels fluctuate greatly but the long-term water level trends are less apparent. Hence, we use changes in groundwater levels over shorter horizons, combined with incidence of dry wells and construction of tube wells to assess groundwater stress.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section3\"\u003e \u003ch2\u003eData Constraints and Choice of Case Studies\u003c/h2\u003e \u003cp\u003eStudying the role of public procurement in inducing the production of water-intensive crops and thereby quantifying the effect on groundwater stress requires data to satisfy two features. First, we need spatially disaggregated data on procurement and groundwater and second, the data must be from a period when growth in the cultivation of water-intensive crops occur. The lack of the second feature precludes statistical estimation of marginal effects because of the absence of adequate variation in the data.\u003c/p\u003e \u003cdiv id=\"Sec13\" class=\"Section4\"\u003e \u003ch2\u003eAll-India Analysis\u003c/h2\u003e \u003cp\u003eDisaggregated district-level data on procurement for most states is not generally available. Thus, for the all-India analysis we only show a cross-sectional association between groundwater stress and cultivation. Moreover, we only use rice cropped area and exclude wheat. This is because groundwater data across regions in India is available only from 1996 and by this time wheat cultivation patterns have stabilized in the country. Wheat area grew at 3.4% each year between 1966\u0026mdash;1980 but at a meagre 0.7% each year between 1996\u0026mdash;2010. Furthermore, wheat is grown in a few districts of only five states (Table S1). As a result, a large number of districts record near zero growth in wheat area and this substantially reduces cross-district variation. Thus, even as wheat producing regions use a lot of groundwater for irrigation, it is infeasible to estimate this statistically.\u003c/p\u003e \u003cp\u003eSince we are restricted to cross-sectional comparisons in our all-India analysis, we view these correlations as motivational.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" type=\"CaseStudy\" class=\"Section4\"\u003e \u003ch2\u003eCase Studies\u003c/h2\u003e \u003cp\u003eFor the formal analysis, we focus on case studies in two states\u0026mdash;Punjab and Madhya Pradesh\u0026mdash;where we were able to obtain suitable data. Punjab is an important region because this is where the Green Revolution and the procurement policy were first introduced, and it is the region that has faced the highest groundwater stress. Moreover, its thick alluvial aquifers are a key reserve of groundwater and understanding the effects on it are very important from a sustainability perspective.\u003c/p\u003e \u003cp\u003eFor this state we were able to obtain data on procurement and groundwater since 1981. We have groundwater data for two periods\u0026mdash;1973\u0026ndash;2003 and 1996\u0026ndash;2016\u0026mdash;from two different sources. While the area under rice kept growing at a healthy rate for about 20 years after 1981, the wheat cropped area was high but stagnant (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec). Hence, for the above-mentioned reasons, we study rice in Punjab between 1981 and 2003. We do not extend the analysis beyond 2003 because we would have to splice groundwater data from two different sources and that might introduce other biases. We could separately study the period of 1996\u0026ndash;2016 but by this time production patterns had stabilized in both cropping seasons, with mainly rice being grown in the \u003cem\u003ekharif\u003c/em\u003e (fall) season and wheat in the \u003cem\u003erabi\u003c/em\u003e (spring). Rice and wheat occupied 62% and 80% of Punjab\u0026rsquo;s net cropped area in 2000 with little scope for growth. Thus, there is little change over time in either rice or wheat cultivation post-1996, and statistical analysis again becomes infeasible.\u003c/p\u003e \u003cp\u003eFor reasons already mentioned, we were not able to study the impact of wheat cultivation in either the all-India correlations or in Punjab. To address this, we use a natural experiment in the state of Madhya Pradesh. Madhya Pradesh is a state where the government suddenly decided to procure wheat in 2008. In 10 years, the state went from procuring nothing to being the largest procurer of wheat in the country. This exogenous policy-induced shift in cultivation allows us to estimate the causal effect of the wheat procurement policy on groundwater stress.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section3\"\u003e \u003ch2\u003eRegression Models\u003c/h2\u003e \u003cdiv id=\"Sec16\" class=\"Section4\"\u003e \u003ch2\u003ePunjab\u003c/h2\u003e \u003cp\u003eTo estimate the effect of rice procurement on production (Table S3) we estimate the following model:\u003c/p\u003e \u003cp\u003elog \u003cem\u003eQ\u003c/em\u003e\u003csub\u003e\u003cem\u003ed,t\u003c/em\u003e\u003c/sub\u003e = log proc\u003csub\u003e\u003cem\u003ed,t\u003c/em\u003e\u0026minus;1\u003c/sub\u003e + \u003cb\u003eX\u0026prime;\u003c/b\u003e\u003cem\u003eγ\u003c/em\u003e\u0026thinsp;+\u0026thinsp;\u003cem\u003eλ\u003c/em\u003e\u003csub\u003e\u003cem\u003ed\u003c/em\u003e\u003c/sub\u003e\u0026thinsp;+\u0026thinsp;\u003cem\u003eλ\u003c/em\u003e\u003csub\u003e\u003cem\u003et\u003c/em\u003e\u003c/sub\u003e\u0026thinsp;+\u0026thinsp;\u003cem\u003eλ\u003c/em\u003e\u003csub\u003eaez\u003c/sub\u003e\u003cem\u003et\u003c/em\u003e\u0026thinsp;+\u0026thinsp;\u003cem\u003eε\u003c/em\u003e\u003csub\u003e\u003cem\u003edt\u003c/em\u003e\u003c/sub\u003e,\u003c/p\u003e \u003cp\u003ewhere \u003cem\u003eQ\u003c/em\u003e\u003csub\u003e\u003cem\u003edt\u003c/em\u003e\u003c/sub\u003e is the rice or wheat production in district \u003cem\u003ed\u003c/em\u003e in year \u003cem\u003et\u003c/em\u003e. proc\u003csub\u003e\u003cem\u003ed,t\u003c/em\u003e\u0026minus;1\u003c/sub\u003e is the procurement of the respective crop in district \u003cem\u003ed\u003c/em\u003e in year \u003cem\u003et\u003c/em\u003e \u0026minus;\u0026thinsp;1. The vector of controls \u003cb\u003eX\u003c/b\u003e includes precipitation, seasonal temperature and temperature squared in year \u003cem\u003et\u003c/em\u003e and district \u003cem\u003ed\u003c/em\u003e. \u003cem\u003eλ\u003c/em\u003e\u003csub\u003e\u003cem\u003ed\u003c/em\u003e\u003c/sub\u003e is a district fixed effect, \u003cem\u003eλ\u003c/em\u003e\u003csub\u003e\u003cem\u003et\u003c/em\u003e\u003c/sub\u003e a year fixed effect and \u003cem\u003eλ\u003c/em\u003e\u003csub\u003eaez\u003c/sub\u003e\u003cem\u003et\u003c/em\u003e is an Agro-Ecological Zone (AEZ)-specific linear time trend. In this model, the fixed effects estimator is likely to suffer from Nickell bias (\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e) as current procurement is mechanically correlated with current cropped area. However, the bias is likely very small because we have 23 years of data (\u003cem\u003eT\u003c/em\u003e is large). For transparency we show results with and without district fixed effects and AEZ-specific trends.\u003c/p\u003e \u003cp\u003eTo study the effect of rice production on groundwater in Punjab (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea), we estimate the following model:\u003c/p\u003e \u003cp\u003e∆\u003csup\u003e\u003cem\u003eT\u003c/em\u003e\u003c/sup\u003e\u003cem\u003eGWL\u003c/em\u003e\u003csub\u003e\u003cem\u003edt\u003c/em\u003e\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;\u003cem\u003eβ\u003c/em\u003e\u003csub\u003e0\u003c/sub\u003e\u0026thinsp;+\u0026thinsp;\u003cem\u003eβ\u003c/em\u003e\u003csub\u003e1\u003c/sub\u003e log rice area\u003csub\u003e\u003cem\u003edt\u003c/em\u003e\u003c/sub\u003e + \u003cb\u003eX\u0026prime;\u003c/b\u003e\u003cem\u003eγ\u003c/em\u003e\u0026thinsp;+\u0026thinsp;\u003cem\u003eλ\u003c/em\u003e\u003csub\u003e\u003cem\u003ed\u003c/em\u003e\u003c/sub\u003e\u0026thinsp;+\u0026thinsp;\u003cem\u003eλ\u003c/em\u003e\u003csub\u003e\u003cem\u003et\u003c/em\u003e\u003c/sub\u003e\u0026thinsp;+\u0026thinsp;\u003cem\u003eλ\u003c/em\u003e\u003csub\u003eaez\u003c/sub\u003e\u003cem\u003et\u003c/em\u003e\u0026thinsp;+\u0026thinsp;\u003cem\u003eε\u003c/em\u003e\u003csub\u003e\u003cem\u003edt\u003c/em\u003e\u003c/sub\u003e,\u003c/p\u003e \u003cp\u003ewhere ∆\u003csup\u003e\u003cem\u003eT\u003c/em\u003e\u003c/sup\u003e\u003cem\u003eGWL\u003c/em\u003e\u003csub\u003e\u003cem\u003edt\u003c/em\u003e\u003c/sub\u003e is the proportional change in groundwater depth in district \u003cem\u003ed\u003c/em\u003e between years \u003cem\u003et\u003c/em\u003e and \u003cem\u003eT\u003c/em\u003e. rice area\u003csub\u003e\u003cem\u003edt\u003c/em\u003e\u003c/sub\u003e is the actual area under rice cultivation in district \u003cem\u003ed\u003c/em\u003e in year \u003cem\u003et.\u003c/em\u003e The vector of controls \u003cb\u003eX\u003c/b\u003e includes district-year level log net cropped area, log population, cumulative precipitation until year \u003cem\u003eT\u003c/em\u003e, seasonal temperature and temperature squared. \u003cem\u003eλ\u003c/em\u003e\u003csub\u003e\u003cem\u003ed\u003c/em\u003e\u003c/sub\u003e is a district fixed effect, \u003cem\u003eλ\u003c/em\u003e\u003csub\u003e\u003cem\u003et\u003c/em\u003e\u003c/sub\u003e a year fixed effect and \u003cem\u003eλ\u003c/em\u003e\u003csub\u003eaez\u003c/sub\u003e\u003cem\u003et\u003c/em\u003e is an AEZ-specific linear time trend. To estimate the effect of rice procurement on groundwater levels (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb) we replace rice area in the above equation with rice procurement.\u003c/p\u003e \u003cp\u003eIn both models for statistical inference, we compute Newey-West standard errors to account for serial correlation in shocks over three periods within districts. Robust standard errors clustered at the district level would be preferred since that accounts for arbitrary correlations in error terms. However, the asymptotic properties for inference are not satisfied since Punjab has only eleven districts (small \u003cem\u003eN\u003c/em\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section4\"\u003e \u003ch2\u003eMadhya Pradesh\u003c/h2\u003e \u003cp\u003eTo estimate the impact of the wheat procurement policy on groundwater stress (Tables\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e and S5), we estimate the following model:\u003c/p\u003e \u003cp\u003eGWS\u003csub\u003e\u003cem\u003edt\u003c/em\u003e\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;\u003cem\u003eβ\u003c/em\u003e\u003csub\u003e0\u003c/sub\u003e\u0026thinsp;+\u0026thinsp;\u003cem\u003eβ\u003c/em\u003e\u003csub\u003e1\u003c/sub\u003e log wheat proc\u003csub\u003e\u003cem\u003edt\u003c/em\u003e\u003c/sub\u003e\u0026thinsp;+\u0026thinsp;\u003cem\u003eβ\u003c/em\u003e\u003csub\u003e2\u003c/sub\u003e log wheat proc\u003csub\u003e\u003cem\u003edt\u003c/em\u003e\u003c/sub\u003e \u0026times;1{\u003cem\u003et\u0026thinsp;\u0026gt;\u0026thinsp;=\u003c/em\u003e\u0026thinsp;2008} + \u003cb\u003eX\u0026prime;\u003c/b\u003e\u003cem\u003eγ\u003c/em\u003e\u0026thinsp;+\u0026thinsp;\u003cem\u003eλ\u003c/em\u003e\u003csub\u003e\u003cem\u003ed\u003c/em\u003e\u003c/sub\u003e\u0026thinsp;+\u0026thinsp;\u003cem\u003eλ\u003c/em\u003e\u003csub\u003ed\u003c/sub\u003e\u003cem\u003et\u003c/em\u003e\u0026thinsp;+\u0026thinsp;\u003cem\u003eε\u003c/em\u003e\u003csub\u003e\u003cem\u003edt\u003c/em\u003e\u003c/sub\u003e,\u003c/p\u003e \u003cp\u003ewhere GWS\u003csub\u003e\u003cem\u003edt\u003c/em\u003e\u003c/sub\u003e is a measure of groundwater stress in district \u003cem\u003ed\u003c/em\u003e in year \u003cem\u003et\u003c/em\u003e. Groundwater stress is measured as either (a) the proportional change in groundwater level between November in year \u003cem\u003et\u003c/em\u003e \u0026minus;\u0026thinsp;1 and May in year \u003cem\u003et\u003c/em\u003e (before sowing and after harvest of the wheat crop), (b) the fraction of wells that are dry post-wheat harvest in June\u0026ndash;September year \u003cem\u003et\u003c/em\u003e, or (c) the logarithm of the number of deep wells in district \u003cem\u003ed\u003c/em\u003e in year \u003cem\u003et.\u003c/em\u003e wheat proc\u003csub\u003e\u003cem\u003edt\u003c/em\u003e\u003c/sub\u003e is the quantity of wheat procured in district \u003cem\u003ed\u003c/em\u003e in year \u003cem\u003et\u003c/em\u003e. \u003cem\u003eλ\u003c/em\u003e\u003csub\u003e\u003cem\u003ed\u003c/em\u003e\u003c/sub\u003e and \u003cem\u003eλ\u003c/em\u003e\u003csub\u003e\u003cem\u003ed\u003c/em\u003e\u003c/sub\u003e\u003cem\u003et\u003c/em\u003e are district fixed effects and district-specific time trends respectively. The vector of controls \u003cb\u003eX\u003c/b\u003e includes seasonal precipitation, temperature, and temperature squared. The coefficient \u003cem\u003eβ\u003c/em\u003e\u003csub\u003e2\u003c/sub\u003e estimates the causal impact of the policy relative to pre-policy effects of wheat procurement.\u003c/p\u003e \u003cp\u003eFinally, to estimate the effect of the wheat procurement policy on production patterns (Table S4), we estimate:\u003c/p\u003e \u003cp\u003elog\u003cem\u003eY\u003c/em\u003e\u003csub\u003e\u003cem\u003edt\u003c/em\u003e\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;\u003cem\u003eβ\u003c/em\u003e\u003csub\u003e0\u003c/sub\u003e\u0026thinsp;+\u0026thinsp;\u003cem\u003eβ\u003c/em\u003e\u003csub\u003e1\u003c/sub\u003e log wheat proc\u003csub\u003e\u003cem\u003ed,t\u003c/em\u003e\u0026minus;1\u003c/sub\u003e\u0026thinsp;+\u0026thinsp;\u003cem\u003eβ\u003c/em\u003e\u003csub\u003e2\u003c/sub\u003e log wheat proc\u003csub\u003e\u003cem\u003ed,t\u003c/em\u003e\u0026minus;1\u003c/sub\u003e \u0026times;1{\u003cem\u003et\u0026thinsp;\u0026gt;\u0026thinsp;=\u003c/em\u003e\u0026thinsp;2008}+ \u003cb\u003eX\u0026prime;\u003c/b\u003e\u003cem\u003eγ\u003c/em\u003e\u0026thinsp;+\u0026thinsp;\u003cem\u003eλ\u003c/em\u003e\u003csub\u003e\u003cem\u003et\u003c/em\u003e\u003c/sub\u003e\u0026thinsp;+\u0026thinsp;\u003cem\u003eε\u003c/em\u003e\u003csub\u003e\u003cem\u003edt\u003c/em\u003e\u003c/sub\u003e,\u003c/p\u003e \u003cp\u003ewhere Y\u003csub\u003e\u003cem\u003edt\u003c/em\u003e\u003c/sub\u003e is either wheat area, wheat irrigated area, or pulses area in district \u003cem\u003ed\u003c/em\u003e in year \u003cem\u003et\u003c/em\u003e. wheat proc\u003csub\u003e\u003cem\u003ed,t\u003c/em\u003e\u0026minus;1\u003c/sub\u003e is the quantity of wheat procured in disitrict \u003cem\u003ed\u003c/em\u003e in year \u003cem\u003et\u003c/em\u003e \u0026minus;\u0026thinsp;1. \u003cem\u003eλ\u003c/em\u003e\u003csub\u003e\u003cem\u003et\u003c/em\u003e\u003c/sub\u003e is a year fixed effect. The vector of controls \u003cb\u003eX\u003c/b\u003e includes seasonal precipitation, temperature, temperature squared, area under other crops, and irrigated area. The coefficient \u003cem\u003eβ\u003c/em\u003e\u003csub\u003e2\u003c/sub\u003e estimates the causal impact of the policy relative to pre-policy effects of wheat procurement.\u003c/p\u003e \u003cp\u003eHere, we do not include district fixed effects because after the policy we only have nine years of data and 15 years of data overall. A fixed effect model estimates the above equation by demeaning. Since contemporaneous procurement and cropped area are mechanically correlated, this would result in Nickell bias (\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e) and thus in inconsistent estimates. We recognize that our estimates are biased but consistent. Since the goal is to show the mechanism, we feel comfortable with this choice.\u003c/p\u003e \u003cp\u003eWe do include district effects in the equation used to estimate the effect of the policy on groundwater stress.\u003c/p\u003e \u003cp\u003eFor statistical inference in both models, we compute robust standard errors clustered at the district level accounting for an arbitrary correlation in the error terms within districts over time.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eH. Garduno, S. Foster, Sustainable groundwater irrigation. approaches to reconciling demand with resources. Strategic Overview Series No. 4. Tech. Rep., The World Bank, Washington D.C. (2010).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eW. Aeschbach-Hertig, T. Gleeson, Regional strategies for the accelerating global problem of groundwater depletion. Nat. Geosci. 5, 853\u0026ndash;861 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eT. Shah, Taming the Anarchy: Groundwater Governance in South Asia (Routledge, 2010).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eM. Shah, P. Vijayshankar, F. Harris, Water and agricultural transformation in India: A symbiotic relationship\u0026ndash; I. Econ. Polit. Wkly. 56 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eS. S. Sayre, V. Taraz, Groundwater depletion in India: Social losses from costly well deepening. J. Environ. Econ. Manag. 93, 85\u0026ndash;100 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eA. Asoka, T. Gleeson, Y. Wada, V. Mishra, Relative contribution of monsoon precipitation and pumping to changes in groundwater storage in India. Nat. Geosci. 10, 109\u0026ndash;117 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eS. Sekhri, Sustaining Groundwater: Role of Policy Reforms in Promoting Conservation in India. In S. Shah, B. Bosworth, A. Panagariya, (eds.) \u003cem\u003eIndia Policy Forum 2012-13: Volume\u0026nbsp;9\u003c/em\u003e, vol.\u0026nbsp;9, 149\u0026ndash;187 (SAGE Publishing India, 2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eM. Rodell, I. Velicogna, J.S. Famiglietti, Satellite-based estimates of groundwater depletion in India. Nature 460, 999\u0026ndash;1002 (2009).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eT. Hora, V. Srinivasan, N. B. Basu, The Groundwater Recovery Paradox in South India. Geophys. Res. Lett. 46, 9602\u0026ndash;9611 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eS. Chatterjee, A. Mahajan, Why Are Indian Farmers Protesting the Liberalization of Indian Agriculture? ARE Update 24, 1\u0026ndash;4 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eA. Deaton, J. Dr\u0026egrave;ze, Food and Nutrition in India: Facts and Interpretations. Econ. Polit. Wkly. 42\u0026ndash;65 (2009).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eN. Bhattarai, A. Pollack, D. Lobell, R. Fishman, B. Singh, A. Dar, M. Jain, The impact of groundwater depletion on agricultural production in India. Environ. Res. Lett. 16 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eS. Sekhri, Missing water: agricultural stress and adaptation strategies in response to groundwater depletion in India. Dep. Econ. Univ. Virginia, Work. Pap. (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eE. Zaveri, D. Grogan, K. Fisher-Vanden, S. Frolking, R. Lammers, D. Wrenn, A. Prusevich, R. Nicholas, Invisible water, visible impact: groundwater use and Indian agriculture under climate change. Environ. Res. Lett. 11, (2016).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eM. Jain, R. Fishman, P. Mondal, G. Galford, Groundwater depletion will reduce cropping intensity in India. Sci. Adv. 7 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eR. Fishman, Groundwater depletion limits the scope for adaptation to increased rainfall variability in India. Clim. change 147, 195\u0026ndash;209 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eE. Zaveri, D. Lobell, The role of irrigation in changing wheat yields and heat sensitivity in India. Nat. Commun. 10, 1\u0026ndash;7 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eM. Raghavan, Politics of procurement and price support. Econ. Polit. Wkly. 39, 506\u0026ndash;508 (2004).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eS. Chatterjee, D. Kapur, P. Sekhsaria, A. Subramanian, Agricultural Federalism: New Facts, Constitutional Vision. \u003cem\u003eWorking Paper\u003c/em\u003e (2022)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eK. Jessoe, R. Badiani-Magnusson, Electricity prices, groundwater, and agriculture: The environmental and agricultural impacts of electricity subsidies in India. \u003cem\u003eAgricultural Productivity and Producer Behavior\u003c/em\u003e, 157\u0026ndash;184 (University of Chicago Press, 2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eB.S. Sidhu, M. Kandlikar, N. Ramankutty, Power tariffs for groundwater irrigation in india: A comparative analysis of the environmental, equity, and economic tradeoffs. World Dev. 128 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eR. Fishman, T. Siegfried, P. Raj, V. Modi, U. Lall, Over-extraction from shallow bedrock versus deep alluvial aquifers: Reliability versus sustainability considerations for India\u0026rsquo;s groundwater irrigation. Water Resour. Res. 47 (2011).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGovernment of India. Dynamic groundwater resources of India 2020. Tech. Rep., Central Ground Water Board, Ministry of Jal Shakti (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eM. Gautam, D. Laborde, A. Mamun, W. Martin, V. Pi\u0026ntilde;eiro, R. Vos, Repurposing Agricultural Policies and Support: Options to Transform Agriculture and Food Systems to Better Serve the Health of People, Economies, and the Planet. Tech. Rep. The World Bank. (2022)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eA. Panagariya, Agricultural liberalisation and the least developed countries: six fallacies. World Econ. 28, 1277\u0026ndash;1299 (2005).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOECD. Agricultural Policy Monitoring and Evaluation 2021. Tech. Rep. (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eK.F. Davis, D. Chiarelli, M. Rulli, A. Chhatre, B. Richter, D. Singh, R. Defries, Alternative cereals can improve water use and nutrient supply in India. Science Advances 4, (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eK.G. Mandal, A.K. Thakur, S.K. Ambast, Current rice farming, water resources and micro-irrigation. Current Science 116(4), 568\u0026ndash;576 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eV. Rawal, M. Kumar, A. Verma, J. Pais, India let 65 lakh tonnes of grain go to waste in four months, even as the poor went hungry. \u003cem\u003escroll.in\u003c/em\u003e (2020). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://scroll.in/article/963535/india-let-65-lakh-tonnes-of-grain-go-to-waste-in-four-months-even-as-the-poor-went-hungry\u003c/span\u003e\u003cspan address=\"https://scroll.in/article/963535/india-let-65-lakh-tonnes-of-grain-go-to-waste-in-four-months-even-as-the-poor-went-hungry\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eR. T. Venkiteswaran, Food grains rot in India while millions live with empty stomachs. The Interpret. (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eS. Kaur, R. Aggarwal, M. Brar, Groundwater Depletion in Punjab, India. Encylopedia of Soil Science (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eP. D\u0026ouml;ll, H. Mueller Schmied, C. Schuh, F. Portmann, A. Eicker, Global-scale assessment of groundwater depletion and related groundwater abstractions: Combining hydrological modeling with information from well observations and grace satellites. Water Resour. Res. 50, 5698\u0026ndash;5720 (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eM. Krishnamurthy, States of Wheat: The Changing Dynamics of Public Procurement in Madhya Pradesh. Econ. Polit. Wkly. 72\u0026ndash;83 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eA. Gulati, P. Rajkhowa, P. Sharma, Making Rapid Strides-Agriculture in Madhya Pradesh: Sources, Drivers, and Policy Lessons. Working Paper 339 Indian Council for Research on International Economic Relations (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eD. Blakeslee, R. Fishman, V. Srinivasan, Way down in the hole: Adaptation to long-term water loss in rural India. Am. Econ. Rev. 110, 200\u0026ndash;224 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eB. Ramaswami, Agricultural Subsidies \u0026ndash; Study Prepared for XV Finance Commission. Tech. Report. Indian Statistical Institute. (2019)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eB. Dhar, R. Kishore, Indian Agriculture Needs a Holistic Policy Framework, Not Pro-market Reforems. Econ. Polit. Wkly. 56 (2021)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eA. Kotwal, M. Murugkar, B. Ramaswami, PDS forever? Econ. Polit. Wkly. 72\u0026ndash;76 (2011).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eR. Radhakrishna, Food and nutrition security of the poor: emerging perspectives and policy issues. Econ. Polit. Wkly. 40, 1817\u0026ndash;21 (2005).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eD. Headey, A. Chiu, S. Kadiyala, Agriculture\u0026rsquo;s role in the Indian enigma: help or hindrance to the undernutrition crisis? Gates Open Res 3, 958 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eA. M. Thow, S. Kadiyala, S. Khandelwal, P. Menon, S. Downs, K. S. Reddy, Toward food policy for the dual burden of malnutrition: an exploratory policy space analysis in India. Food Nutr. Bull. 37, 261\u0026ndash;274 (2016).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eP. Kataki, Shifts in cropping system and its effect on human nutrition: Case study from India. J. Crop. Prod. 6, 119\u0026ndash;144 (2002).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eP. Pingali, K. Ricketts, D. E. Sahn, Agriculture for nutrition. The fight against hunger malnutrition: The role food, agriculture, targeted policies 165 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eP. Pingali, Green revolution: impacts, limits, and the path ahead. \u003cem\u003eProc. Natl. Acad. Sci.\u003c/em\u003e 109, 12302\u0026ndash;12308 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eP. Pingali, Agricultural policy and nutrition outcomes\u0026ndash;getting beyond the preoccupation with staple grains. Food security 7, 583\u0026ndash;591 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eP. Pingali, B. Mittra, A. Rahman, The bumpy road from food to nutrition security\u0026ndash;slow evolution of India\u0026rsquo;s food policy. Glob. food security 15, 77\u0026ndash;84 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eA. de Brauw, M. Suryanarayana, Linkages between poverty, food security and undernutrition: evidence from China and India. China Agric. Econ. Rev. (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eP. Goswami, S. N. Nishad, Virtual water trade and time scales for loss of water sustainability: A comparative regional analysis. Sci. Reports 5, 1\u0026ndash;11 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eC. Dalin, Y. Wada, T. Kastner, M. J. Puma, Groundwater depletion embedded in international food trade. Nature 543, 700\u0026ndash;704 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDevineni, N., Perveen, S. \u0026amp; Lall, U. Solving groundwater depletion in India while achieving food security. Nat Commun \u003cb\u003e13\u003c/b\u003e, 3374 (2022)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eS. Nickell, Biases in dynamic models with fixed effects. Econometrica 1417\u0026ndash;1426 (1981).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eS. Chatterjee, M. Krishnamurthy, D. Kapur, M. Bouton, A Study of the Agricultural Markets of Bihar, Odisha and Punjab. Final Report. Tech. Rep., Center for the Advanced Study of India, University of Pennsylvania (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eS. Chatterjee, D. Kapur, Six Puzzles in Indian Agriculture. India Policy Forum 17, 185\u0026ndash;229 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eP. Gupta, R. Khera, S. Narayanan, Minimum Support Prices in India: Distilling the facts. Working Paper (2021) \u003cem\u003eAvailable at SSRN 3791859.\u003c/em\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u0026nbsp;\u003c/strong\u003eFor helpful discussions and feedback, we are grateful to the seminar participants at Center on Food Security and the Environment, Stanford University and to Marshall Burke, Walter Falcon, Ram Fishman, Tejasvi Hora, Avinash Kishore, David Lobell, Roz Naylor, and P.S. Vijay Shankar. We thank Faiz Ahmed Kidwai and Azad Jain for facilitating access to wheat procurement data from Madhya Pradesh.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions statement\u0026nbsp;\u003c/strong\u003eS.C., R.L., and E.D.Z designed research; S.C. and E.D.Z performed research; S.C. and E.D.Z analyzed data; and S.C., R.L., and E.D.Z wrote the paper. All authors reviewed the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAdditional information\u003c/strong\u003e The authors declare no competing interest.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData and materials availability:\u0026nbsp;\u003c/strong\u003eAll data and codes to replicate the results in this paper are available from SC (\u003ca href=\"mailto:[email protected]\"\[email protected]\u003c/a\u003e) or EDZ (\u003ca href=\"mailto:[email protected]\"\u003eesha.d.zaver\u003c/a\u003e\u003ca href=\"mailto:[email protected]\"\[email protected]\u003c/a\u003e)\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-1766947/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-1766947/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eDwindling groundwater supplies threaten food security and livelihoods. Farm subsidies which distort production decisions are an ubiquitous agricultural policy tool, yet their contribution to growing groundwater stress remains poorly quantified. Here we show how output subsidies that guarantee purchase of crops at higher than market prices have contributed substantially towards declining water tables in India. Overall, these policies have led to a 30% over-production of water intensive crops. In the northwestern state of Punjab, rice procurement accounts for 63% of the rise in groundwater declines over two decades. In the central state of Madhya Pradesh, wheat procurement adopted in the late 2000s has driven a 5.3 percentage point increase in dry wells and a consequent 3.4 percentage point increase in deep tubewells. The results demonstrate how well intentioned but poorly designed subsidies can impose harmful externalities on the environment and undermine long-term sustainable development.\u003c/p\u003e","manuscriptTitle":"The role of farm subsidies in changing India’s water footprint","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2022-07-20 17:06:23","doi":"10.21203/rs.3.rs-1766947/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-communications","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"NCOMMS","sideBox":"Learn more about [Nature Communications](http://www.nature.com/ncomms/)","snPcode":"","submissionUrl":"https://mts-ncomms.nature.com/","title":"Nature Communications","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Communications","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"709165a1-36cf-435a-b0a5-06d20ca42e62","owner":[],"postedDate":"July 20th, 2022","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2024-10-06T07:05:18+00:00","versionOfRecord":{"articleIdentity":"rs-1766947","link":"https://doi.org/10.1038/s41467-024-52858-6","journal":{"identity":"nature-communications","isVorOnly":false,"title":"Nature Communications"},"publishedOn":"2024-10-05 04:00:00","publishedOnDateReadable":"October 5th, 2024"},"versionCreatedAt":"2022-07-20 17:06:23","video":"","vorDoi":"10.1038/s41467-024-52858-6","vorDoiUrl":"https://doi.org/10.1038/s41467-024-52858-6","workflowStages":[]},"version":"v1","identity":"rs-1766947","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-1766947","identity":"rs-1766947","version":["v1"]},"buildId":"_2-kVJe1T_tPrBINL-cwx","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}