Applied Econometrics and International Development Vol (2016)

Size: px
Start display at page:

Download "Applied Econometrics and International Development Vol (2016)"

Transcription

1 IMPACT OF CLIMATE CHANGE ON CORN PRODUCTION IN THE U.S.: EVIDENCE FROM PANEL STUDY Jaehyuk LEE Nazif DURMAZ Abstract This paper determines the impact of climate change on corn production using a countylevel panel data from the Corn-Belt States. The estimates of elasticity of corn production with respect to temperature and precipitation are and , respectively. The drought indices, PDSI and CMI, if the drought condition improves one unit, the corn production increases by about 2% and 10% respectively that is equivalent to about 3 and14 bushels per acre. Our simulation results indicate that such an increase will result in a decrease in corn production by about 15 to 90 bushels per acre by JEL Codes: Q54, D24 Keywords: Climate change, Corn production, Drought Index, the Palmer Drought Severity Index (PDSI), Crop Moisture Index (CMI) 1. Introduction There is consensus that green-house gas emission causes global warming and it will affect many economic aspects. Therefore, it is not surprising that global warming has been a hot issue lately all around the world. It might be the major concern to human being especially because warming will be directly related to food consumption and human health. Warming seems negatively affect agricultural production according to various reports. Because of these reasons, global warming has been receiving a lot of people s attention. As Oreskes (2004) point out, between 1993 and 2003, 928 papers that have abstracts including global climate change were published in refereed scientific journals. This is very surprising since Houghton et al. (2001) and Mendelsohn (2007) address that there is surprising absence of impact studies on climate change effects in the past partially due to the slight temperature increase on average over the globe which has warmed only by 0.5 C over the last hundred years. The impacts of global warming also have been highly controversial among scientists, scholars, and policy makers. Intergovernmental Panel on Climate Change (IPCC) and the National Academy of Sciences have reported that most of observed warming is likely due to the results of human activities such as Greenhouse gas emission while policy-makers and media argued that climate change is highly uncertain (Oreskes, 2004). However, according to the estimation by the NASA's Goddard Institute for Space Studies and the National Climatic Data Center, warming has been in increasing trend since the 1980s as it is shown in Figure 1. The National Oceanic and Contact Author. Department of Finance and Economics, Georgia Southern University, Statesboro, GA Tel: Fax: jaehyuklee@georgiasouthern.edu Department of Accounting, Economics, and Finance, University of Houston-Victoria, Sugar Land, TX Tel: Fax: durmazn@uhv.edu

2 Atmospheric Administration (NOAA) also reports that seven of the eight warmest years on record have occurred since 2001 and the 10 warmest years have all occurred since Previous studies also suggest that global warming has been in increasing trend since the1980s although the Earth's average surface temperature has increased by about 1.2 to 1.4 F in the last 100 years (Mendelsohn, 2007). Figure 1. Annual Average Global Surface Temperature Anomalies (Degrees in One Hundredths of a Celsius) Data source: Goddard Institute for Space Studies, National Aeronautics and Space Administration (NASA) There have been different forecasting and extensive debates over the concerns about the impacts of climate change. However, a broad consensus among climate scientists is that there would be drastic temperature increases and change in precipitation patterns due to greenhouse gas effect (Houghton et al., 2001). According to the NOAA s report, the recent warmth has been greatest over North America and Eurasia between 40 and 70 N although the warming has not been occurred in same fashion worldwide. That is, most of European countries and U.S. states except for the Southern states have been affected most by the recent warming. Therefore, climate change might be a major concern to humanity since it affects many economic sectors as well as different aspect of human life. Negative impacts of climate change on the agricultural sector will be especially dangerous since agriculture is directly related to food security and human health. Many believe that agricultural production will be affected most by temperature and precipitation since they are directly related to the production (Deschenes and Greenstone, 2007). This 94

3 paper mainly examines the economic impacts of climate change on the corn production in the United States. Corn is the one of major agricultural products in the U.S. which plays a crucial role in the world corn trade market. Approximately 20% of the corn traded (USDA) from the U.S. The importance of corn has been even more significant because strong demand for ethanol production has resulted in higher corn prices and has provided corn farmers incentives to increase corn acreage. Therefore, if weather negatively affects the U.S. corn production, it will cause a big problem not only for U.S. corn farmers and economy but for world food security. 2. The Impacts of climate change in literature There have been on-going debates on potential climate change and its impacts on the agricultural production. While negative impacts of climate change on U.S. agriculture are found in most previous research, some argue that warming will be beneficial or the magnitude of the impacts is not as big as it is expected to be (see, for example, Mendelsohn and Massetti 2011, and Deschenes and Greenstone 2007). In fact, it is debatable because the impacts of climate change vary by various conditions such as location, crop characteristics, soil condition, and even estimation methodology. Deschenes and Greenstone (2007) argue that previous research on climate change impacts on land value is mostly inconclusive for example. However, it is hard to ignore that many researchers find the negative impacts of warming on crop production. Schlenker and Roberts (2006) examine the relationship between weather variables (temperature and precipitation) and corn yields. They find that the growth of corn yields rapidly becomes negative for temperatures in excess of 30 and there is an inverted U-shape relationship between precipitation and corn yields with an optimal level of 26 inches. Huang and Khanna (2010) estimate the future climate change impact on U.S. crop yields using the county level panel analysis. They find that increase in temperature significantly reduces the yields of corn, soybeans, and wheat while precipitation has positive relationship with the crop yields. Deschenes and Greenstone (2007) analyze the climate impacts on agricultural yields using the fixed-effects estimation. They find negative (positive) impacts with temperature (precipitation) for corn and soybean yields although the impacts are small in magnitude. 3. Data and Methodology Conceptual Model In prior research, production function is the one of the most widely employed methods to estimate the impact of climate change along with hedonic approach (Deschenes and Greenstone, 2007). We apply a three-input production function for the analysis and the inputs are capital (K), labor (L), and fertilizer (F). The production function is also augmented by the climate variables in order to estimate their impact on the corn production. The production function takes the following form: where Y represents the corn yields per acre in bushels. The production inputs K, L, and F indicate the total number of machinery in operations per acre, the total number of 95

4 hired labor per acre, and the total expense of fertilizer purchase per acre respectively. C is the vector of climate variables. The climate variables include temperature, precipitation, and the drought indices such as the Palmer Drought Severity Index (PDSI) and the Crop Moisture Index (CMI). More details of these indices will be provided below. Econometric Model The empirical counterpart of the conceptual production function is depicted below: = is the error term. After taking natural logarithms of both sides, the model becomes: = where represents the corn yield per acre in county c at year t. We assume that the error term has three components: a time-invariant, a location-invariant, and an idiosyncratic part such that: Consequently, we included county fixed-effects and year dummies, and in the estimated equation. As mentioned above, the production input variables L, K, and F represent labor, machinery, and fertilizer, respectively. is the growing-season climate variables including temperature, precipitation, and humidity indicator indices such as PDSI and CMI. The climate variables also include both the linear and quadratic terms in order to capture the possible non-linear relationship between the corn production and climate variables. The drought indices used in the analysis are the average of a growing season in county level and weekly-level data are used for the calculation of the growing season. State Table 1. Corn for Grain: Usual Planting and Harvesting Dates, by State Production Usual planting dates Usual harvesting dates (in 1000 Bu) Begin Most active End Begin Most active End Illinois 2,130,100 Apr 22 Apr 30-May 18 May 28 Sep 24 Oct 9 - Nov 3 Nov 19 Indiana 873,600 Apr 25 May 5- May 20 June 10 Sep 20 Oct10 -Nov25 Dec 10 Iowa 2,188,800 Apr 22 May 2- May 16 June 3 Sep 17 Oct 7 - Oct 31 Nov 17 Kansas 486,420 Apr 10 Apr 25-May 15 May 25 Sep 5 Sep 20 -Oct 20 Nov 10 Minnesota 1,180,800 Apr 24 May 3- May 22 June 8 Sep 29 Oct15 - Nov12 Nov 28 Missouri 381,600 Apr 5 Apr 20-May 25 June 10 Sep 1 Sep 20 -Oct 30 Dec 1 Nebraska 1,393,650 Apr 21 May 3- May 19 June 1 Sep 21 Oct11- Nov 6 Dec 1 Ohio 421,200 Apr 22 May 1- May 30 June 12 Sep 25 Oct15- Nov14 Nov 25 S. Dakota 585,200 May 1 May 9- May 25 June 11 Sep 24 Oct10- Nov 6 Nov 30 Wisconsin 394,560 Apr 25 May 1- June 5 June 10 Oct 1 Oct15- Nov15 Nov 30 Sources: NASS, USDA 96

5 The growing season is defined as the period between the first week of June and the last week of September in the analysis. In fact, a growing season for corn slightly varies by region. However, as it is shown in Table 1, the period from June to September covers the growing season for corn in almost all sample regions (USDA-NASS, 1997). The standard errors are clustered at the county level. Data The empirical analysis utilizes cross-sectional time series (panel) data that is based on 10 major corn-producing states for years 2002 and 2007 **. The data set includes counties of Illinois, Indiana, Iowa, Kansas, Minnesota, Missouri, Nebraska, Ohio, South Dakota, and Wisconsin. These states are commonly included in the Corn Belt Region and more than 80% of the corn-for-grain acreage in the U.S. lies in the Corn Belt States according to USDA. The data are obtained from the USDA s Census of Agriculture which is conducted every 5 years census is the latest available survey. Variables of interest in the analysis are temperature, precipitation, and the drought indices. National Oceanic and Atmospheric Administration (NOAA) Climate Prediction Center is the source of these data. The Palmer Drought Severity Index (PDSI) and Crop Moisture Index (CMI) are chosen proxy for humidity of the soil. This is because, among major drought indices, the PDSI and CMI are the most widely used ones by government agencies and researchers in the United States. According to National Drought Mitigation Center, many U.S. government agencies and states rely on PDSI for evaluating drought relief programs. Therefore, PDSI is one of the most reliable measurements for drought condition. In broad terms, these indices represent degrees of dryness/ wetness in the soil. NOAA explains that total weekly precipitation, average temperature, climate division constants such as water capacity of the soil and others, and previous history of indices are included in the calculation of both PDSI and CMI for 350 climate divisions in the U.S. and Puerto Rico. Since both temperature and precipitation are included in the calculation of indices, the drought indices can be a good measurement for evaluating the effect of climate change in various fields. Table 2 shows the classifications of PDSI and CMI indices that represent the degrees of dryness/ wetness. Both PDSI and CMI indices are correlated with temperature and precipitation. As it is shown in Table 3, there is a positive relationship between precipitation and drought indices and a negative one with temperature. A one degree Fahrenheit (inch) increase in temperature (precipitation) reduces PDSI and CMI by and (5.283 and 2.460), respectively. ** 1997 census data was not used because the labor data was unavailable. The earliest census year that has the labor data available is

6 Palmer Drought Severity Index (PDSI) Table 2. Drought Index Classifications Crop Moisture Index (CMI) -4.0 or less (Extreme Drought) -3.0 or less (Severely Dry) -3.0 to -3.9 (Severe Drought) -2.0 to -2.9 (Excessively Dry) -2.0 to -2.9 (Moderate Drought) -1.0 to -1.9 (Abnormally Dry) -1.9 to 1.9 (Near Normal) -0.9 to 0.9 (Slightly Dry/ Favorably Moist) 2.0 to 2.9 (Usual Moist Spell) 1.0 to 1.9 (Abnormally Moist) 3.0 to 3.9 (Very Moist Spell) 2.0 to 2.9 (Wet) 4.0 or above (Extremely Moist) 3.0 and above (Excessively Wet) Source: NOAA-Climate Prediction Center Table 3. The Impact of Temperature and Precipitation on Drought Index Temperature Precipitation Coef. Std.Err. Coef. Std.Err. PDSI *** *** CMI *** *** Notes: *, **, and *** indicate significance at 10%, 5%, and 1% levels, respectively The definitions of the variables used in the estimation are presented in appendix. Table 4 provides the summary statistics of the variables. Table 4. Descriptive Statistics Variable Obs Mean Std. Dev. Min Max Production Harvested Acre Temp_gs Rain_gs PDSI_gs CMI_gs Capital Labor Fertilizer

7 4. Estimation and Analysis Table 5 shows the regression results of fixed effect panel analysis. Table 5. Regression Analysis Results Dependent Variable: Corn Production per acre (unit: bushels) Temp_gs *** (1) (2) (2)-A (3) (3)-A (16.922) Rain_gs *** (2.478) Temp_gs_sq *** (2.106) Rain_gs_sq *** (0.051) Temp_gs* Rain_gs *** (0.597) PDSI_gs 0.023*** *** PDSI_gs_sq 0.006*** (0.003) (0.003) (0.001) CMI_gs 0.111*** 0.101*** CMI_gs_sq 0.016* 99 (0.013) (0.012) (0.009) Capital *** * (0.051) (0.058) (0.059) (0.056) (0.056) Labor (0.025) (0.031) (0.032) (0.030) (0.030) Fertilizer 0.097** * 0.106** 0.104** (0.047) (0.053) (0.053) (0.051) (0.052) Constant *** 3.853*** 3.826*** 3.679*** 3.735*** (33.979) (0.427) (0.433) (0.417) (0.418) N R-sq No of counties Note: The outcome variable and all control variables are in natural logarithms except for the PDSI and CMI since some observations are negative in the sample.. All regressions include indicators for counties and year dummies. The standard errors are clustered at the county level and they are reported in parentheses. *, **, and *** indicate significance at 10%, 5%, and 1% levels, respectively.

8 The impact of climate variables on the corn production is estimated using the three different models which are identical except for the climate variables. Specifically, temperature and precipitation, PDSI, and CMI are included in the regressions separately. As it mentioned earlier, a quadratic term of each climate variable is also included in the estimation in order to capture the possible non-linear relationship. The elasticities of the climate variables are reported in Table 6. The first column presents results from the model in which the conventional climate variables (growing season temperature and precipitation) are included. Both the temperature and its square term are significant at 1% level. The estimated coefficients imply that one percent increase in the growing season temperature reduces the corn yield per acre by nine percent at the sample mean (Table 6). A one percent increase in the temperature corresponds to about 0.7 F. It follows that a 0.7 F increase in the growing season temperature will decrease the corn production by about 12 bushels per acre. Table 6. Elasticities of Climate Variables on Corn Production per acre (bushels) Variables Temperature Precipitation PDSI CMI Elasticities *** *** 0.018***, 0.101***, Robust S.E Note: *** significant at 1% ** significant at 5% *significant at 10% indicates semi-elasticity. The precipitation and its square term are significant at 1% level. The estimated coefficients show that there is an inverse U-shape relationship between precipitation and corn production. The elasticity of production with respect to growing season precipitation is and it is significant at conventional levels. A one percent increase is equivalent to an increase of precipitation by inches. Such an increase in precipitation will yield an additional 21.5 bushels of corn per acre. The interaction term of temperature and precipitation is negative and significant. The impact of The elasticities of climate variables (temperature and precipitation) in the estimation are evaluated at the mean and calculated using following equation for model 1; = where are linear term coefficients and are quadratic term coefficients for temperature and precipitation variables. are sample means of temperature and precipitation variables. The elasticities of the drought indices (PDSI and CMI) are not reported because of the negative signs due to the negative mean values although the coefficients of both linear and quadratic terms are positive and significant. Semi-elasticities are reported instead. 100

9 temperature on corn production per acre decreases by about 3.6% for each inch increase in precipitation. A one percent increase in capital is associated with a 0.15% decrease in corn production, while a one percent increase in fertilizer increases production by about 0.01%. Both coefficients are significant at conventional levels. The negative coefficient of capital perhaps shows the diminishing marginal product of machinery due to the sufficient amount of machinery inputs in corn production. The positive effect of fertilizer reflects the improvement of technology and its strong substitutability to other production inputs such as labor and machinery. These results are consistent with the findings of Paul and Nehring (2005) that use three types of crops including corn, soybean, and other crops, respectively. Labor is not significantly associated with corn production. Fuglie et al.(2007) explain that agricultural labor declined by 3.2 % per year since 1950s and the use of other inputs such as machinery and fertilizer have been increased due to the high labor cost and the substitutability of the inputs. Columns (2), (2)-A, (3), and (3)-A include PDSI and CMI as measures of weather effect in the estimation. All the remaining control variables are identical to those in column 1. In columns (2) and (3) both the draught index and its square term are included. In columns (2)-A and (3)-A only the main term is included. Except for the drought indicators, all variables are in natural logarithms. In both specifications, the draught indices are positive and significant at 1% level. The estimated semi-elasticity of corn production with respect to PDSI is This result implies that if the drought condition improves from near normal to the next level usual moist, the production increases by1.8% (an increase equivalent to about 3 bushels per acre). CMI and its square term are also positive and significant at conventional levels. The coefficient of CMI in column (3)-A is A one unit increase in CMI increases the corn production by about 10%. In other words, if the drought condition improves from near normal to usual moist, the corn production increases by about 10% that is equivalent to about 14 bushels per acre. In comparison, the magnitude of CMI is much bigger than that of PDSI. According to NOAA-NCDC, CMI measures short-term drought and is used to quantify drought s impacts on agriculture during the growing season. On the other hand, PDSI measures the duration and intensity of the long-term drought-inducing circulation patterns. Therefore, CMI is more proper measure to estimate the effect of weather in our analysis. The coefficients of other input variables are similar to those reported in the first column of Table Conclusion This paper examines the impact of climate change on corn production using a county-level panel data from the Corn-Belt States from years 2002 and This is the first paper that utilizes drought indices (PDSI and CMI) in addition to conventional weather variables such as temperature and precipitation. The estimates of elasticity of corn production with respect to temperature and precipitation are and , respectively. An increase in the humidity of the soil from normal to moist level is 101

10 associated with a raise in corn production by about 2-10% (about 3-14 bushels per acre). The Intergovernmental Panel on Climate Change (IPCC) forecasts that global temperature will increase between 1-6 C by Karl et al. (2009) report that the average U.S. temperature is projected to increase by about 4 F and 11 F by Based on IPCC forecasting, our results suggest that when temperature increases by 1 C in growing season (minimum increase), corn production will decrease by 15 bushels per acre. On the other extreme, (a 6 C increase), the corn production will decrease by 91 bushels per acre. This is similar to the findings of previous researchers. For example, Huang and Khanna (2010) predict that U.S. corn production will decrease by 55 bushels per acre if temperature increases by 6 C. Schlenker and Roberts (2009) forecast that corn production will decrease by 55% by 2100 when temperature increases by 6 C. As many studies indicate, agriculture will be very vulnerable to global warming. This especially holds for crops in most parts of the world. Since the U.S. is the major exporter of corn, soybean, and wheat, climate change might be a major concern to global economy and peoples well-being. Therefore, technologies that help with the agricultural sector s adaptation to climate change will be necessary. Examples of such technologies could be developing new varieties that are more tolerant to higher temperature or less humidity. References Deschenes, O. and M. Greenstone (2007): The Economic Impacts of Climate Change: Evidence from Agricultural Output and Random Fluctuation in Weather. The American Economic Review, Vol 97. No.1. Fugile, K.O., J.M. MacDonald and E. Ball (2007): Productivity Growth in U.S. Agriculture. Economic Brief 9. USDA-ERS. Houghton, J., Y. Ding, D. Griggs, M. Noguer, P. van der Linden, X. Dai, K. Maskel, and C. Johnson (2001): Climate Change 2001, The Scientific Basis. Third Assessment Report of the Intergovernmental Panel on Climate Change (IPCC). Cambridge University Press. Huang, H., and M. Khanna (2010): An Econometric Analysis of U.S. Crop Yields and Cropland Acreages: Implications for the Impact of Climate Change. Paper presented at AAEA annual meeting, Denver, Colorado, 25-27July. Intergovernmental Panel on Climate Change (1990): Climate Change: The IPCC Scientific Assessment (J. T. Houghton, G. J. Jenkins, and J. J. Ephraums, eds.), Cambridge: Cambridge University Press. Mendelshon, R. (2007): Past Climate Change Impacts on Agriculture. Handbook of Agricultural Economics, Vol 3. Ch.60 pp

11 National Climatic Data Center (NCDC). NASA- Goddard Institute for Space Studies and the National Climatic Data Center Oreskes, N. (2004): The Scientific Consensus on Climate Change. Science 3, Vol.306. No.5702: Paul,C.M. and R. Nehring (2005): Product Diversification, Production Systems, and Economic Performance in U.S. Agricultural Production. Journal of Econometrics, 126(2): Schlenker, W., W.M. Hanemann and A.C. Fisher (2006): The Impact of Global Warming on U.S. Agriculture: An Econometric Analysis of Optimal Growing Conditions. Review of Economics and Statistics, 88(1): Schlenker, W. and M.J. Roberts (2006): Nonlinear Effects of Weather on Corn Yields. Review of Agricultural Economics, 28(3): Schlenker, W. and M.J. Roberts (2009): Nonlinear Temperature Effects Indicate Severe Damages to U.S. Crop Yields under Climate Change. PNAS, 106(37): United States Department of Agriculture, NASS (1997): Usual Planting and Harvesting Dates for U.S. Field Crops. Agricultural Handbook, No United State Department of Agriculture (USDA), 103

12 Appendix Variable Definitions Production Temp_gs Rain_gs Corn yields per acre (unit: bushels) Growing season mean temperature (unit: degrees in Fahrenheit) Growing season mean precipitation (unit: inch) Temp_gs_sq Square term of growing season mean temperature Rain_gs_sq Temp*Rain PDSI_gs CMI_gs PDSI_gs_sq CMI_gs_sq Labor Capital Fertilizer Square term of growing season mean precipitation Interaction variable of growing season mean temperature and precipitation Growing season Palmer Drought Severity Index Growing season Crop Moisture Index Square term of growing season Palmer Drought Severity Index Square term of growing season Crop Moisture Index Total number of hired labor per acre Total number of machinery in operation per acre Total expense of fertilizer purchase per acre (unit: dollar) Journal published by the EAAEDS: 104