WPS4164 An Analysis of Livestock Choice: Adapting to Climate Change in Latin American Farms1 S. Niggol Seo University of Aberdeen Business School, UK and Robert Mendelsohn School of Forestry and Environmental Studies, Yale University, USA Abstract This paper explores how Latin American livestock farmers adapt to climate by switching species. We develop a multinomial choice model of farmer's choice of livestock species. Estimating the models across over 1200 livestock farmers in seven countries, we find that both temperature and precipitation affects the species Latin American farmers choose. We then use this model to predict how future climate scenarios would affect species choice. Global warming will cause farmers to switch to beef cattle at the expense of dairy cattle. World Bank Policy Research Working Paper 4164, March 2007 The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or the countries they represent. Policy Research Working Papers are available online at http://econ.worldbank.org. 1 This project was funded by the World Bank. We thank Emilio Ruz, Flavio Avila, Jorge Lozanoff, Luis José María Irias, Magda Aparecida de Lima, César Teherán, Jorge González, Flavio Játiva, Alfredo Albin, Liubka Valentina Trujillo, Luisa Caraballo Silva, and Ariel Dinar for their contributions to this effort. This project was funded by the Research Committee of the World Bank under the study 'Climate Change and Rural Development' that was tasked managed by Ariel Dinar. 1 1. Introduction This paper uses cross-sectional evidence to explore how farmers adapt to exogenous environmental factors such as climate and soils. By comparing choices of farmers who face different conditions, the model uncovers how farmers adapt. We specifically examine how climate affects which species Latin American farmers choose to own. We test whether climate alters species choice. Understanding species switching (adaptation) is an important goal in itself to assist planning by policy makers and private individuals. However, understanding adaptation is also important if one is interested in quantifying the impacts of climate change. The impacts of climate change not only require understanding how each species will be affected but also how farmers will switch across species. Climate impact studies have consistently predicted extensive impacts to the agricultural sector from climate change across the globe (Pearce et al. 1996; Tol 2002). The bulk of agriculture studies on the effect of climate change have focused on crops. However, a large fraction of agricultural output is from livestock. Yet there are very few economic analyses of climatic effects on livestock. Most of the livestock studies related to climate change have not explored the economic impacts (Watson et al. 1995; McCarthy et al 2001). Two important exceptions are the study of the effects of climate change on American livestock (Adams et al. 1999) and the study of climate and African livestock (Seo and Mendelsohn 2006). Individual American livestock are sensitive to climate. For example, beef catle cannot tolerate high temperatures. However, by using cool locations, protected environments (sheds, barns, etc.) and supplemental feed (e.g. hay and corn), the American livestock sector is not sensitive to warming (Adams et al 1999). In Africa, by contrast, the bulk of livestock have no protective structures and they graze off the land. They currently live in the coolest locations available. African livestock, and especially beef cattle, are sensitive to climate (Seo and Mendelsohn 2006). Warmer temperatures cause African farmers 2 to move away from beef cattle causing large damages. Interestingly, small farmers can substitute sheep and goats for beef cattle and would therefore not be vulnerable to warming. The theoretical choice model is developed in the next section. Section 3 discusses how data were collected from over 2000 farmers in seven countries across Latin America. Section 4 discusses the estimation procedure and the empirical results. Three climate change scenarios from Atmospheric Oceanic General Circulation Models (AOGCM's) are then examined in Section 5. The paper concludes with a summary of results and policy implications. 2. Theory In this paper, farmers are assumed to maximize their profits. Farmers choose the desired species to yield the highest net profit. Hence, the probability that a species is chosen depends on the profitability of that animal. We assume that farmer j's profit in choosing livestock i (i=1, 2,...,IJ) is ij =Vi(Kj,Sj)+i(Kj,Sj) (1) where K is a vector of exogenous characteristics of the farm and S is a vector of characteristics of the farmer. For example, K could include climate, soils and access variables and S could include the age of the farmer and family size. The profit function is composed of two components: the observable component V and an error term, . The error term is unknown to the researcher, but may be known to the farmer. The farmer will choose the livestock that gives him the highest profit. Defining Z = (K,S), the farmer will choose 3 animal i over all other animals k if: i (Zj) >k (Zj)fork i.[orif k(Zj)-i(Zj) 4.0). The coefficients on the quadratic terms tend to be positive when significant which implies the response function between the probability of each species being chosen and climate is U-shaped. This is especially true for precipitation. Soils can affect which species is chosen as some soils (such as Acrisols) lead to more productive grazing lands whereas others (Luvisols and Arenesols) are less productive. Households with computers are more likely to pick chickens rather than grazing animals. Female farmers are more likely to pick dairy and sheep. Finally, dairy, sheep, and pigs are more likely in the Andean region. Figure 1 graphs the relationship between the probability of choosing a species and annual temperature. Note that the mean temperature in Latin America is 18°C. The probability of choosing beef cattle and chickens decline as temperatures rise above 18°C. By contrast, the probability of choosing dairy and sheep increases. With pigs, the estimated probability first rises and then declines. The graph clearly reveals that the choice of animals in Latin America is temperature sensitive. Figure 2 displays the estimated relationship between the probability of choosing an animal and annual precipitation. The mean annual precipitation in Latin America is 118 mm/month. The probability of choosing beef cattle declines precipitously as precipitation increases above the mean. By contrast, more rain leads to more dairy cattle. The other species exhibit a U-shaped pattern where they first decline and then increase. 5. Climate scenarios In this section, we simulate the consequences of climate change using the parameter estimates in the previous section. We examine a set of climate change scenarios predicted by AOGCMs. The climate scenarios reflect the A1 SRES scenarios from the following three 7 models: the Canadian Climate Center (CCC) scenario (Boer et al. 2000), Centre for Climate System Research (CCSR) (Emori et al. 1999), and the Parallel Climate Model (PCM) scenario (Washington et al. 2000). We use country level climate change scenarios in 2020, 2060, and 2100 from each climate scenario. The change in temperature predicted by each climate model is added to the baseline temperature in each district. The percentage change in precipitation is multiplied by the baseline precipitation in each district. This gave us a new climate for every district in Latin America for each scenario. Table 2 summarizes the climate scenarios of the three models for the years 2020, 2060, and 2100. The models predict a broad set of scenarios consistent with the range of outcomes in the most recent IPCC (Intergovernmental Panel on Climate Change) report (Houghton et al. 2001). In 2100, PCM predicts a 2°C temperature increase in Latin America whereas CCC predicts a 5°C increase. Rainfall predictions are noisier: PCM predicts rainfall to increase by 8% by 2100 whereas CCC predicts rainfall to decrease by 8%. Examining the path of climate change over time reveals that temperatures are predicted to increase steadily until 2100 for all three models but precipitation will vary across time. The parameters from the estimated multinomial logit models are used to simulate the impacts of climate change on the probabilities of choosing a particular animal for each climate scenario in Table 3. The dry and hot CCC and CCSR scenarios predict that farmers would choose beef cattle and sheep more often and dairy cattle, pigs, and chickens less often. With the wetter and milder PCM scenario, farmers will pick sheep more often and beef and dairy cattle less often. 6. Conclusion This paper uses a multinomial choice model to capture the choice of species made by farmers. 8 The model is estimated across over 1200 farmers in Latin America. We observe that the choice of species varies with climate. Beef cattle are chosen more often in cooler dryer climates. Chickens are chosen in cooler places. Dairy cattle are preferred in wetter hotter climates. Sheep and pigs appear to be heat tolerant. These results are completely consistent with observations of where species are currently located. Beef cattle are currently concentrated in the relatively cool and dry regions of Argentina, Uruguay, and southern Brazil. Dairy cattle are seen more often in the hotter and wetter regions of Columbia, Brazil, and Chile. Sheep are located in relatively cool locations in Argentina and Chile. Pigs are primarily concentrated in Brazil and chickens are most common in Ecuador. The probability response functions for Latin America study are quite consistent with the response functions from Africa (Seo and Mendelsohn 2006). Beef cattle and chickens have a hill-shaped relationship with temperature. Dairy cattle are more heat tolerant. However, because Africa is hotter than Latin America, warming is more harmful to African beef cattle compared to Latin American beef cattle. Further, Latin America and especially Brazil grow pigs which are relatively rare in Africa whereas Africa owns goats which are relatively rare in Latin America. Both of these animals tend to be heat tolerant. The only species which has very different climate effects in the two continents is sheep. In Africa, sheep are often chosen in hot locations whereas in Latin America, they are far more likely in cooler locations. In interpreting these results, there are several caveats that should be kept in mind. First, this analysis does not include price effects. Large changes in animal prices may alter the results. Second, we assume that adaptations can take place as needed. For example, farmers can switch from one animal to another as temperature increases and rainfall decreases. However, this may not be the case if the adjustment requires a heavy capital investment. 9 Third, we assume that in forecasting climate change impacts, the only thing that changes in the future is climate. Many things, however, will change over the century such as population, technologies, institutional conditions, and reliance on agriculture and livestock. Future studies should address these issues and provide ever more accurate measure of climate change impacts. 10 References Adams, Richard, McCarl, Bruce, Segerson, Kathy, Rosenzweig, Cynthia, Bryant, Kelley, Dixon, Bruce, Conner, Richard, Evenson, Robert, Ojima, Dennis. "The economic effects of climate change on US agriculture." in Mendelsohn, Robert and Neumann, James, eds. The Impact of Climate Change on the United States Economy, Cambridge, UK: Cambridge University Press, 1999. Boer, G., G. Flato, and D. Ramsden (2000), "A transient climate change simulation with greenhouse gas and aerosol forcing: projected climate for the 21st century", Climate Dynamics 16, 427-450. Dubin, Jeffrey A., and McFadden, Daniel L. "An Econometric Analysis of Residential Electric Appliance Holdings and Consumption", Econometrica, 1984, Vol.52, No.2 , pp. 345-362. Emori, S. T. Nozawa, A. Abe-Ouchi, A. Namaguti, and M. Kimoto (1999), "Coupled ocean- atmospheric model experiments of future climate change with an explicit representation of sulfate aerosol scattering", J. Meteorological Society Japan 77, 1299-1307. Greene, William H. Econometric Analysis (3rd edition), New Jersey: Prentice Hall, New Jersey, 1998. Houghton, John, Yihui, Ding, Griggs, Dave, Noguer, Maria, Van der Linden, Paul, Dai, Xiaosu, Maskell, Kathy and Johnson, Cathy, (eds.) Climate Change 2001: The Scientific Basis, Intergovernmental Panel on Climate Change: Cambridge University Press, 2001. McCarthy, James, Canziani, Osvaldo F., Leary, Neil A., Dokken, David J. and White, Casey, (eds.) Climate Change 2001: Impacts, Adaptation, and Vulnerability, Cambridge: 11 Cambridge University Press, Intergovernmental Panel on Climate Change, 2001. McFadden, Daniel L. "Econometric Models of Probabilistic Choice", in Daniel. McFadden, Structural Analysis of Discrete Data and Econometric Applications, Cambridge: MIT press, 1981. McFadden, Daniel L. "Chapter 1. Discrete Response Models", University of California at Berkeley, Lecture Note, 1999. Mendelsohn, R., A. Basist, A. Dinar, F. Kogan, P. Kurukulasuriya and C. Williams. 2006. "Climate Analysis with Satellites Versus Weather Station Data" Climatic Change (forthcoming). Pearce, D. et al. 1996. "The Social Costs of Climate Change: Greenhouse Damage and Benefits of Control" in Climate Change 1995: Economic and Social Dimensions of Climate Change, J. Bruce, H. Lee, E. Haites (eds.) Cambridge Univ. Press, Cambridge, UK, pp.179-224. Seo, S. Niggol and Mendelsohn, Robert. "Climate Change Adaptation: Microeconomic Analysis of Livestock Choice" World Bank Working Paper, (forthcoming), 2006. Tol, R.S.J. (2002), `New Estimates of the Damage Costs of Climate Change, Part I: Benchmark Estimates', Environmental and Resource Economics, 21 (1), 47-73. Train, Kenneth, 2003, Discrete Choice Methods with Simulation, Cambridge, U.K.: Cambridge University Press. Washington, W., et al. (2000), "Parallel Climate Model (PCM): Control and Transient Scenarios". Climate Dynamics, 16: 755-774. 12 Table 1: Multinomial logit selection model Beef cattle Dairy cattle Variable Coefficient 2 P-value Coefficient 2 P-value Intercept 0.599 0.130 0.722 2.021 1.560 0.212 Temperature summer 0.102 0.230 0.629 -0.212 1.040 0.308 Temperature summer sq 0.001 0.010 0.906 0.003 0.170 0.683 Precipitation summer 0.016 13.020 0.000 0.011 6.950 0.008 Precipitation summer sq 0.000 15.870 <.0001 0.000 9.930 0.002 Temperature winter -0.148 2.140 0.143 0.144 2.090 0.148 Temperature winter sq 0.003 0.670 0.414 -0.001 0.120 0.733 Precipitation winter -0.008 2.310 0.128 -0.009 4.040 0.045 Precipitation winter sq 0.000 1.010 0.315 0.000 9.440 0.002 Soil Acrisols 0.055 7.580 0.006 0.044 4.820 0.028 Soil Luvisols -0.026 20.300 <.0001 -0.011 4.510 0.034 Soil Arenosols -0.022 4.380 0.036 -0.036 10.280 0.001 Computer dummy -0.566 9.760 0.002 -0.342 3.440 0.064 Female dummy 0.266 2.230 0.135 0.493 7.100 0.008 Andes dummy 0.171 0.680 0.411 0.535 6.720 0.010 13 Table 1: (continued) Sheep Pigs Variable Coefficient 2 P-value Coefficient 2 P-value Intercept -5.047 3.210 0.073 -3.015 1.200 0.273 Temperature summer 0.728 4.740 0.029 0.079 0.060 0.814 Temperature summer sq -0.016 2.660 0.103 -0.005 0.220 0.640 Precipitation summer -0.005 0.560 0.454 0.011 2.520 0.112 Precipitation summer sq 0.000 1.680 0.195 0.000 7.200 0.007 Temperature winter -0.448 8.840 0.003 0.344 4.310 0.038 Temperature winter sq 0.016 10.380 0.001 -0.008 2.130 0.144 Precipitation winter -0.028 10.080 0.002 -0.021 7.140 0.008 Precipitation winter sq 0.000 8.060 0.005 0.000 9.820 0.002 Soil Acrisols -0.138 0.000 0.000 0.046 4.460 0.035 Soil Luvisols -0.014 3.350 0.067 -0.036 5.550 0.019 Soil Arenosols -0.043 2.530 0.112 -0.016 1.290 0.257 Computer dummy -0.293 1.550 0.213 -0.628 7.360 0.007 Female dummy 0.997 6.310 0.012 0.118 0.170 0.683 Andes dummy 0.923 7.910 0.005 1.134 11.670 0.001 Note: Omitted choice is chickens. Likelihood ratio test: P<0.0001, Lagrange multiplier test: P<0.0001, Wald test: P<0.0001 14 Table 2: Latin American Average AOGCM Climate Scenarios Current 2020 2060 2100 Temperature (°C ) CCC 18.1 19.5 (+1.4) 20.8 (+2.7) 23.2 (+5.1) CCSR 18.1 19.4 (+1.3) 20.4 (+2.2) 21.3 (+3.2) PCM 18.1 18.7 (+0.6) 19.5 (+1.3) 20.1 (+2.0) Rainfall (mm/month) CCC 119 116 (-2.6%) 107 (-9.5%) 109 (-7.7%) CCSR 119 120 (+1.5%) 119 (0.0%) 114 (-3.8%) PCM 119 128 (+8.2%) 133 (+11.9% 129 (+8.4%) 15 Table 3: Predicted change in the probability of selecting each animal from AOGCM climate scenarios Beef cattle Dairy cattle Sheep Pigs Chickens Baseline 47.9% 33.4% 6.4% 4.6% 7.8% 2020 CCC +3.06% -2.04% +0.35% -0.67% -0.70% CCSR +1.79% -1.13% +0.47% -0.58% -0.55% PCM -5.02% +2.30% +1.72% +1.02% -0.02% 2060 CCC +4.63% -3.56% +1.26% -0.99% -1.34% CCSR +3.59% -2.82% +1.24% -1.22% -0.79% PCM -3.51% +1.16% +1.96% +0.32% +0.07% 2100 CCC +6.94% -5.79% +3.04% -1.94% -2.26% CCSR +5.50% -4.14% +1.09% -1.61% -0.84% PCM -1.85% -0.53% +2.61% -0.06% -0.18% 16 Figure 1: Estimated probability of selecting species given annual temperature PROB 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0 10 20 30 ANNUAL MEAN TEMPERATURE PLOT BEEF DAIRY SHEEP PIGS CHICKENS 17 Figure 2: Estimated probability of selecting species given annual precipitation PROB 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0 100 200 300 400 ANNUAL MEANPRECIPITATION PLOT BEEF DAIRY SHEEP PIGS CHICKENS 18