Is the Crop Insurance Program Effective in China? Evidence from ...

Journal of Integrative Agriculture Advance Online Publication 2014

Doi：10.1016/S2095-3119(14)60842-X

Is the Crop Insurance Program Effective in China? Evidence from Farmers Analysis in Five Provinces* 1,2*

WANG Ke

, ZHANG Qiao

1,2*

3

1

, Shingo Kimura , Suraya Akter

1

Agricultural Information Institute, Chinese Academy of Agricultural Sciences, Beijing 100081, P.R.China

2

Key Laboratory of Agri-information Service Technology, Ministry of Agriculture, Beijing 100081, P.R.China

3 Trade and Agricultural Directorate, Organization for Economic Cooperation and Development, Paris 75775, France

Abstract This paper aims to evaluate the effectiveness of the Chinese crop insurance program in terms of farmer utility and welfare. A simulation model based on the power utility function was first developed to evaluate the effectiveness of crop insurance. Then, the Monte Carlo approach was used to generate the datasets of area, price, yield, cost, and income based on the characteristics of representative farmers, which were clustered and calibrated using the farm-level data of 574 individual farmers from five Chinese provinces. Finally, the effectiveness of Chinese crop insurance was evaluated by comparing the certainty equivalence (CE) of farmer’s utility/welfare under alternative crop insurance scenarios. Government subsidy is a necessary premise for implementing the crop insurance program. The government should subsidize more than 50% of the crop insurance premium to motivate more farmers to participate in the program. The findings also show that the current crop insurance program in China has increased the welfare of farmers but still need to be improved to achieve the Pareto improvement and to make full use of the financial fund of the government. This paper is believed to not only extend academic research but also has significant implications for policymakers, especially in the context of rapid development of Chinese Crop insurance with much issues such as rate, subsidy and coverage level need to be improved. Keywords:

Crop insurance, effectiveness evaluation, expected utility model, China

Introduction Agriculture is an important industry for almost all countries, especially for developing countries with high population density such as China. However, agriculture is also considered a high-risk sector because it continually faces risks in production due to adverse weather conditions that farmers cannot control. Droughts, floods, and other natural disasters may result in serious consequences, such as crop *

Received 17 February, 2014 Accepted 15 May, 2014 WANG Ke, +86-10-82106259, E-mail: [email protected]; Correspondence ZHANG Qiao, Tel: +86-10-82109883, Fax: +86-10-82106261, E-mail: [email protected] *These authors contributed equally to this study.


Doi：10.1016/S2095-3119(14)60842-X

failure, poverty, and food insecurity. To minimize the effect of adverse weather on the income of farmers, more than 100 countries have conducted crop insurance program (Mahul and Stutley 2010). The Chinese government has also paid more attention to developing a crop insurance program. The Chinese agricultural insurance program has been very successful since its establishment in 2007, when the central government began to provide a premium subsidy. The premium of Chinese agricultural insurance rose to CNY 30.6 billion (appropriately USD 5 billion) in 2013 from CNY 5.2 billion in 2007 and CNY 24.06 billion in 2012. Until now, China is the first and the second agricultural insurance market in terms of premiums in Asia and in the world, respectively (CIRC*). Along with rapid development, the Chinese government had allotted an increasing fund to subsidize crop insurance program. Now, the Chinese government subsidizes more than 70% of crop insurance premiums. In 2012, the central government of China, who can afford to pay 40%-50% of the premium subsidy, had paid CNY 10 billion to the agricultural insurance program. At the same time, the low insured value of the Chinese crop insurance program was criticized in China. Current Chinese crop insurance only covered some of the physical costs during crop planting, which account for 25% to 40% of crop returns. Thus, government officers, agricultural economists, and farmers argue that the insured value of crop insurance is excessively low and may be of no use for farmers. Therefore, the serious question of valuing Chinese crop insurance has been proposed, and the effective evaluation of the Chinese crop insurance program in China was highlighted. In the literature from the end of the 1990s, agricultural economists began to study the effect of crop insurance, revenue insurance, hedging, loans, and other risk management tools. Certain scholars investigated the effect of crop insurance on agricultural production (Hennessy 1998, Coffey, Skees et al. 2001, O'Donoghue, Key et al. 2005), and some explored the effect of crop insurance on farmer welfare (Wang, Hanson et al. 1998, Chen, Wang et al. 2007). Other scholars studied the interaction of alternative risk management tools (Coble and Heifner 1998, Coble, Miller et al. 2004, Wang, Makus et al. 2004, Antón and Kimura 2009). Most of them had evaluated the effect of risk management tools under the expected utility framework and conducted empirical studies that adopted a robust stochastic simulation approach. However, in China, the majority of studies focused on the theoretical discussion of the necessary and potential effect of subsiding crop insurance (Tuo, 2003; Zhang et. al, 2005; Wu, 2005; Shi, 2008; Hou et al, 2010), with the exceptions of Zhang and Shi (2007) and Sun and Zhong (2008).

Zhang and Shi (2007) investigate the crop insurance effect in theory by analyzing the issues

of crop insurance subsidy and social welfare, but he fails to conduct an empirical study. Sun and Zhong (2008) estimated the net welfare of subsidized crop insurance by calculating the willingness of farmers to pay (WTP) based on survey data. However, this approach, compared to the stochastic simulation approach, lacks the feasibility to be used for hypothesis research. Agriculture insurance is initially designed to stabilize the income of farmers by helping them fight against yield loss due to adverse weather. Thus, evaluating the effect of crop insurance according to farmers is valuable and meaningful. However, no studies evaluate the effectiveness of the Chinese crop insurance program from the perspective of farmers and no studies have adopted the stochastic simulation approach, which has the advantage of analyzing the interactions among different policies, *

China Insurance Regulatory Commission: http://www.circ.gov.cn


Doi：10.1016/S2095-3119(14)60842-X

allowing the analysis of the possible consequences of modifying crucial points in each policy (Kimura and Thi 2011). This study conducts a quantitative assessment of Chinese crop insurance effectiveness based on the welfare of individual farmers. In this paper, using farm-level data of 574 individual farmers from the provinces of ShanDong (SD), HeNan (HN), JiangSu (JS), SiChuan (SC), and Shan’Xi (SX), the effectiveness of Chinese crop insurance is evaluated by comparing the certainty equivalence (CE) of the utility/welfare of representative farmers under alternative crop insurance policy scenarios. This paper is believed to not only extend academic research but will also contribute to the better design of the Chinese crop insurance program. The rest of this paper is organized as follows: the second section describes the methodology. The sample data and empirical results are shown in the third section. The fourth section presents the discussion based on empirical results, and the conclusion and policy implications are presented in the final section.

Method The basic idea to evaluate the effect of Chinese crop insurance in this paper is to compare farmer welfare in alternative scenarios. Four scenarios, including no crop insurance (NOCI), Chinese current crop insurance (CCI), modified crop insurance (MCI) and directly subsiding farmers (DSF) are hypothesized in this paper. The process to achieve this goal can be divided into four steps.

Step 1: Estimating farmer welfare In economic theory, using expected utility is the most general approach for comparing risky choices and studying risk behaviors under uncertainty. Thus, a simulation model based on expected utility function was developed in this paper to estimate farmer welfare. Similar to previous studies (Turvey 1992, Wang, Hanson et al. 1998, Coble, Heifner et al. 2000, Lin 2001), this paper also adopted the power utility function (Eq. 1) to compute for farmer utility.

U (w 0  w )  Where

1 (w  w 0 )(1   ) 1

(1)

depicts the constant relative risk aversion (CRRA) and was set to 2, 4, and 6, respectively,

to test the robustness of the analysis. Moreover, w0 depicts farmer’s initial wealth, and w is farmer’s net income, which can be expressed as Equation (2) w 

 pi

 Yi  Ai  fincother  off  Trinc  Zcinc  C  ind  pre  sub

(2)

Where Pi is the output price of crop i, Yi is the output yield of crop i, Ai is the area of land cultivated for crop i, and finc_other is the farm income from minority crops, livestock, and other agricultural production activities. Off is the income obtained from working in a city and other non-farm activities, Trinc means the transfer income that farmer received from her/his children and relative, Zcinc means farmers’ assert and other residual income, C is the crop planting costs, including physical cost and labor cost, ind and pre is the indemnity and premium* of crop insurance, respectively, and sub * Note, in theory, premium of crop insurance should be calculated based on actuarial principles. The current premium of crop insurance, however, is set based on intuitional experiences in China. Thus, when we evaluate the effectiveness of current crop

insurance, the premium is provided based on current crop insurance policies in China instead of being calculated.


Doi：10.1016/S2095-3119(14)60842-X

is the Gov. Premium subsidy for crop insurance. The indemnity and premium of crop insurance can be calculated as follows:

(1  a ) * yt  y

Ind  A * Ivalue * max(

yt

, 0)

(3)

Where A is the insured acreage, Ivalue is the insured value per unit (which equals price times yield per unit times coverage level), a is the deductible level, yt is the regular yield in a normal year, and y is the actual yield. Given that the CE is the guaranteed amount that has the same utility as farmer’s excepted utility in risk prospect but has the advantage of being easily compared, the certainty equivalent of farm income is used to compute for farmer’s welfare within a given level of risk aversion. (

CE  {(1   )E[U (w 0  w )]}

1 ) 1

 w0

(4)

Step 2: Defining representative farmers The contribution of crop insurance to farmer welfare may be different to different farmers, depending on the individual characteristics and risk exposures of farmers. Thus, the problem of which welfare should be chosen as the standard for evaluating crop insurance first needs to be addressed. The most standard method is defining the representative farm. This approach is widely used in previous studies and has the advantage of producing results that are easily interpreted and flexible structures for carrying out a large number of scenarios (Kimura and Thi 2011). When defining the representative farm, ensuring that the representative farm fully reflects the risk exposure of farmers is important. In this paper, the cluster analysis approach is adopted to group the sample farms into several clusters that are homogenous in terms of risk characteristics. Specifically, hierarchical clustering analysis is applied to group farmers in a province into several homogenous clusters according to the variance of agri. income and the proportion of agri. income to household income.

Step 3: Calibrating risk characteristics of individual farmers Given that, in a region, the good yield of one farmer may be offset by another farmer with a serious yield loss, assessing farmer production risk using aggregated data can be misleading (Coble, Dismukes et al. 2007). Micro data from agricultural planting and agribusiness are used to calculate and aggregate the characteristics of individual farmers to calibrate the characteristics of representative farmer in this paper. Due to the improvement of planting technology and crop variety, it is wildly believed that crop yield has a systemic growth over time. Thereby the yield trend should be removed when using yield data to evaluate the crop yield risk. In this paper, the crop yields of individual farmers are detrended at the county-level * to eliminate the influence of technology improvement. In addition, all monetary variables such as price and income are deflated to avoid the effects of inflation. Rural CPIs are adopted to deflate all monetary variables including price, agricultural income, off income, and covert nominate *

Given that the panel data of agricultural production is short (only six or seven years), historical data collected for each farmer may underestimate the yield risk when yield trend by farmer is removed. Therefore, the yield trend should be removed at the regional level. Taking account the huge geographic area of each province in China, the regional level is set to the county level in this paper.


Doi：10.1016/S2095-3119(14)60842-X

price into real price at the base year. Log-linear trends are estimated to get the annual percentage growth of yield at the region level.

log(Yt )  a  b * t  

(5)

Where Yt is the yield level at the county level in year t, and b can be estimated by log-linear regression to indicate the annual percentage growth of yield. Subsequently, b is subtracted from individual observations to estimate the yield risk of individual farmers, as shown in Equation (6).

Dy it  y it  y i1 * (1  (1  b )t 1 ) Where year t, and

is the detrended yield for farmer i in year t,

(6) is the yield observation of farmer i in

is the yield observation of farmer i in the base year.

After the micro data of farmers

are detrended or deflated, the statistic features (mean, standard deviation) of yield, price, cost, and returns from crop. Other farming and off-farm data are calculated by crops, by cluster, and by province. The correlation coefficient of yield and price by crops are also calculated to reflect agricultural production diversity and the relationship of yield and price.

Step 4: Calculating representative farmer’s welfare using simulated approach Although following the methods in Steps 1 to 3, representative farmer’s welfare could be calculated using historical data. Thus, we use the stochastic simulation approach in this paper to estimate farmer’s welfare because a) the time span of micro-level historical data is usually limited and unbalanced; b) the simulation approach allows us to specify a joint price and yield generating process that reflects actual conditions (Wang, Hanson et al. 1998); and c) the simulated approach has the advantage of analyzing the interactions among different policies and allows us to analyze the possible consequences of modifying crucial points in each policy (Kimura and Thi 2011). In spite of the arguments that normal distribution cannot appropriately capture crop yield and price generation process (Harri, Erdem et al. 2009), multivariable normal distribution is still primarily favored when fitting the joint distribution of crop yield and price because it is relative simple and the conclusion that normal distribution is inappropriate are doubted (Just and Weninger 1999, Wang, Makus et al. 2004). In this paper, we still assume that the joint distribution of crop yields and prices can be captured through multivariable normal distribution, and the Monte Carlo approach was adopted to generate 1,000 combinations of crop yield and price for each representative farmer. Regarding other variables, including area, household income, other farm income, cost, and off-farm income, we assume that they will be fixed and equal to the value of the previous year in the process of stochastic simulation so that the effect of crop insurance can be highlighted.

Results Data source In this paper, 574 individual farmers from 12 counties in 5 provinces (JS, SD, HN, SX, and SC) were selected as the sample for collecting farm-level data. Farm-level data were obtained from the China


Doi：10.1016/S2095-3119(14)60842-X

Rural Fixed Observation Office (CRFOO), Ministry of Agriculture of China. The time span was from 2003 to 2009. Figure 1 shows the geographic location of sample farmers. Table 1 lists the description of sample farmers. [Figure 1 here]

[Table 1 here]

Table 1 shows that the crop planting characteristics are reasonable because most farmers in the five sample provinces plant more than two crops. Moreover, the crop varieties planted by farmers in each province are consistent with local cropping practices and weather conditions. Although the observations and the number of farmers are unbalanced in Table 1, which may result in inconsistent explanatory power, we believe that the sample data are acceptable because this paper is aim to provide evidence about the effect of Chinese crop insurance on farms instead of infer the whole effect of the Chinese agricultural insurance program. Characteristics of representative farmers As mentioned above, hierarchical clustering analysis was adopted to ensure the homogeneity of representative farmers. Using hierarchical clustering analysis, the sample farmers in HN are classified into four groups. Moreover, farmers in JS, SD, SX and SC are divided into two groups, as shown in Table 2. [Table 2 here] After grouping the sample farmers, the characteristics of representative farmers in terms of crop planting and income were summarized in Table 3. The following can be observed: 1) The crop sowing area in China is relatively small. Farmers in SC and JS have the lowest sowing area, mostly less than 1.5 Mu (0.1 hectare) per crop. Wheat farmers in SD plant 6.9 Mu of wheat on average, which is the biggest in the sample but still less than 0.5 hectare; 2) Wheat and corn are the main crops for farmers in SD, HN, and SX, whereas, in JS and SC, farmers also plant rice and oilseed. 3) The yield risks of crop planting are high in China because the coefficient of variations (CVs) of crop yield in the five provinces is above 20% on average, ranging from 10% to 40%. 4) The majority of Chinese farmers may face higher risk of yield variation than price fluctuations because the CVs of yield are higher than the CVs of price for almost all sample farmers. A notable exception is the oilseed farmers in JS who were more concerned about price than yield because they have modest yield risk but may sell their oilseed with huge price fluctuations. A possible explanation for this finding is that JS is one of the best oilseed producers in China. 5) Although part-time farming is increasing at a high pace in China, agriculture remains vital for Chinese farmers because agricultural income contributes more than 50% of household income in all sample provinces, which is consistent with the values published by the National Statistics Bureau of China. 6) However, revenue from grain crops (wheat, rice, corn, and oilseed) only accounts for 3% to 8% of a farmer’s total income. This fact highlights the necessity and importance of stabilizing the grain revenue of farmers by crop insurance in China because insurance can help to increase the willingness of farmers to be involved in grain production.


Doi：10.1016/S2095-3119(14)60842-X

[Table 3 here] In practice, diversity production is a common strategy for farmers for stabilizing their crop income. The negative relationship between crop price and crop yield also contribute to the stability of the farmer’s income. Table 4 shows the correlation characteristics between yield and price by crop and by province. Notably, certain crops in Table 4 are not planted at the same time. For example, corn and wheat are usually rotated every year in SD and HN. Calculating the correlation of crops not planted at the same time is arguably problematic. However, although certain crops are planted at different times, they are planted within the same year and the crops planted in the same year experience a similar climate environment. Therefore, we believe that calculating the correlation matrix as showed in Table 4 is reasonable. [Table 4 here]

Based on the characteristics shown in Tables 3 and 4, the Monte Carlo approach was used to generate the simulated data for representative farmers following the assumption and approach shown in Step 4 in the methodology section. Effectiveness of current crop insurance program Although certain pilots of weather index insurance and crop price index insurance exist in one or two regions in China, the dominant crop insurance program in China is yield insurance, which works similar to the multi-peril crop insurance (MPCI) in the U.S. but with much lower insured value. At present, the central, provincial, and local government in China subsidizes more than 70% of crop insurance premiums. Table 5 lists the crop insurance policies in sample provinces. [Table 5 here]

Using the simulation model mentioned in the methodology section, the effectiveness of the current Chinese crop insurance program is evaluated by comparing the representative farmer’s welfare (CE) under three alternative scenarios of NOCI, CCI and DSF which means transferring the fund of government premium subsidy to farmers directly instead of subsidizing crop insurance. Figure 2 demonstrates the change in the welfare of representative farmers under the alternative scenarios. It can be seen from figure 2 that: 1) the CE of all farmers with CCI are higher than that of NOCI which implies that current crop insurance program in China has no doubt increased the welfare of Chinese farmers in spite of low risk guarantees capability of CCI; 2) but, the contribution of crop insurance to farmer seems insignificant because the welfare improvement due to CCI is only 0.4% in average, ranging from 0.1% to 0.6% in all sample provinces except HN. However the numbers are appropriated given the fact that revenue from grain crops only contributes 3% to 8% to famers’ total income (as shown in Table 3); 3) Crop insurance in China remains in need of improvement because the effect of CCI is less than DSF for 7 representative farmers out of 10 farmers and the conclusion are robust because the choose of representative farmer between CCI and DSF are constant along with different risk aversion; 4) however we cannot say that crop insurance is not needed in China because 70% farmers in this paper are prefer DSF. To our knowledge, there are two reasons for that. The first one, which has been mentioned before, is that only physical cost of crop planting are covered by current


Doi：10.1016/S2095-3119(14)60842-X

crop insurance program in China and the very low risk guarantee capability hamper the effect of crop insurance. An empirical study to test the arguments will be conducted in the following sub-section. Another possible reason is the farmers in the 7 regions have a “low” risk, lower than the charged premium which is not accurately made according to their risk exposure but in a provincial scale. The justification of this argument can be verified by comparing the actuarial premium with practical premium and that work is out of the scope of this study and will leave future research to be addressed. Although we did not conduct empirical study in this paper, we still believe the argument is logical because a study by Kang (2012) has found that crop insurance did not have advantage in the regions of low risk. [Figure 2 here]

Simulated analysis on the modification of crop insurance Figure 2 shows that farmers in four out of five provinces in China prefer DSF to WCI, which may be linked to the complaint of insufficient insured value. However, is this complaint the reason why farmers prefer DSF? To answer this question, the insured value of crop insurance are supposed to be increased to 120%, 150%, and 200% respectively with the premium, subsidy level, and other factors fixed to test whether increasing insured value will affect the results. The simulation result of increasing the insured value of crop insurance in JS, SD, SC, and SX are shown in Figure 3. Farmers in all provinces would benefit more when the insured value of the crop insurance program is increased. However, farmer’s welfare under the modified crop insurance scenario remains lower than farmers’ welfare under the DSF scenario even when the insured value of crop insurance is doubled. The finding drawn from Figure 3 implies that only increasing the insured value of crop insurance would not make farmers prefer crop insurance compared with DSF which is consistent with the study of Kang (2012). [Figure 3 here] Besides insured value, deductible is another elements of affecting the power of crop insurance. Usually deductible is set to avoid the moral hazard problem in insurance industry and will be taken into account when pricing insurance. However because the premium of CCI in China is not actuarial made, we cannot judge whether the 20% deductible in China’s CCI is rational. Therefore the scenarios of 10% deductible of CCI with other factors fixed are hypothesized to test the effect of increasing risk guarantees. The comparison of Modified crop insurance and DSF for the first group of farmers in all provinces are shown in figure 4. It can be seen clearly that farmers in SD and SX will change their choice between Crop insurance and DSF when the deductible of crop insurance are decreased to 10%, and all provinces except JS will choose purchasing the lower deductible crop insurance. Figure 4 demonstrates the advantage of crop insurance compared with DSF, and verifies the necessary of increasing risk guarantees capability of Chinese crop insurance in the future. [Figure 4 here] When the premium subsidy ratio is fixed, increasing insured value of crop insurance will increase the premium expenditures of the government. The government may wonder whether they can lower the premium subsidy ratio while increasing insured value. If so, how much can they do so? Therefore, we allow the government to change the subsidy ratio in the future to investigate whether the subsidy ratio can be decreased while insured value increases. For simplicity, the simulation results with the


Doi：10.1016/S2095-3119(14)60842-X

hypothesis that CRRA equals two are demonstrated in Figures 5. The following can be observed: 1) The government subsidy is necessary for the implementation of Chinese crop insurance because farmer’s welfare with NOCI is less than farmer’s welfare with crop insurance only when government subsidy is higher than 60% in JS, 55% in SD, 20% in SC, and 45% in SX. 2) As expected, farmers’ welfare increased as the premium subsidy ratio increased, but increasing the insured value of crop insurance decreased, rather than increased, farmer’s welfare when the subsidy ratio are lower. 3) To maintain farmer’s welfare at par, the premium subsidy ratio can be reduced by increasing the insured value of crop insurance. When increasing insured value by 50%, the government can reduce the premium subsidy ratio by 5% in JS, 5%~10% in SD, 20% in SD, and 10%~15% in SX. [Figure 5 here]

Conclusions & Discussion With the rapid development of agricultural insurance in China, the effectiveness of agricultural insurance program had elicited an increasing amount of attention and has been heavily debated by government officers, academic experts, and farmers. Unlike previous studies that evaluate the effect of Chinese crop insurance, this paper adopted the stochastic simulation model (which was originally used to compare the interaction effects of crop insurance, hedging, contract farming and other risk management tools) to evaluate the effectiveness of Chinese crop insurance. Based on the simulated results, three key conclusions can be made. The first conclusion is that the Chinese crop insurance program has increased the welfare of farms. However, the crop insurance program in China requires improvement to achieve Pareto improvement because the majority of farmers prefer DSF, in which the government subsidizes the money of farmers directly to the current crop insurance program, under the DSF scenarios. The second meaningful conclusion is that, in most cases, the government should subsidize more than 40% of premiums to make farmers participate in the crop insurance program. Although the minimum subsidy ratios of crop insurance differ across provinces, the government should provide more than 50% of crop insurance premiums to stimulate the willingness of farmers to participate in the program in China. Such finding is consistent with that of (Hazell, Pomareda et al. 1986), who found that the crop insurance for maize and beans would require a subsidy of two-thirds of the total premium to be attractive to farmers in Mexico. The conclusion in this paper provides empirical evidence to the theoretical inference that crop insurance will not work unless premium subsidy is available. Although many criticize the low insured value of the Chinese crop insurance program, Chinese crop insurance will not be improved significantly by only increasing the coverage level or insured value because these two factors would not make the majority of farmers choose crop insurance compared to DSF. However, in case of increasing the insured value of crop insurance, it is possible to reduce the premium subsidy ratio without lowering farmers’ welfare and decreasing the crop insurance participating ratio. This final conclusion has obvious policy implications for policymakers to improve Chinese crop insurance program. In spite of many meaningful conclusions had been draw in this paper, we acknowledge that the paper can be improved in several aspects. Firstly this paper assumes that the joint distribution of yield and price fits multi-normal distribution. However, alternative multivariable distribution, such as joint


Doi：10.1016/S2095-3119(14)60842-X

kernel distribution, could be used to capture the underlying joint distribution of yield and price more appropriately. Secondly we also assume that the degree of farmer’s risk aversion across sample provinces is the same, which may not be consistent with practical conditions. The two drawbacks will be improved in following works. Thirdly exploring the difference of the choice between CCI and DSF in HN and other provinces is not explained depth because it is beyond the scope of this paper, but we believe that finding the relationship between farmer’s behavior and risk is very meaningful for the policymakers. Last but not least we had to note that the evaluation method used in this paper is a static model inherently. Because farmer’s view on crop insurance may alter over time and the effect of crop insurance may appear gradually, the dynamic evaluation methodology is recommended, although it is beyond the scope of this paper, to be developed to evaluate the value of crop insurance program over time.

Reference Antón, J., & Kimura, S. (2009). Farm Level Analysis of Risk, and Risk Management Strategies and Policies: Evidence from German Crop Farms. Paper presented at the International Association of Agricultural Economists Conference, Beijing. Chen, X., Wang, H. H., & Makus, L. D. (2007). Production risk and crop insurance effectiveness: organic versus conventional apples. Paper presented at the Economics and Management Risk in Agriculture and Natural Resources, Gulf Shores Coble, K. H., Dismukes, R., & Thomas, S. (2007). Policy Implications of Crop YIeld and Revenue Variability at differing levels of Disaggregation. Paper presented at the American Agricultural Economics Association Annual Meeting 2007, Portland, Oregon. Coble, K. H., & Heifner, R. (1998). The Effect of Crop or Revenue Insurance on Optimal Hedging. Paper presented at the Proceedings of the NCR-134 Conference on Applied Commodity Price Analysis, Forecasting, and Market Risk Management. Coble, K. H., Heifner, R. G., & Zuniga, M. (2000). Implications of crop yield and revenue insurance for producer hedging. Journal of Agricultural and Resource Economics, 25(2), 432-452. Coble, K. H., Miller, J. C., Zuniga, M., & Heifner, R. (2004). The joint effect of government crop insurance and loan programmes on the demand for futures hedging. European Review of Agricultural Economics, 31(3), 309-330. doi: DOI 10.1093/erae/31.3.309 Coffey, B. K., Skees, J. R., Dillon, C. R., & Anderson, J. D. (2001). Potential Effects of Subsidized Livestock Insurance on Livestock Production. Paper presented at the 2001 Annual meeting, August 5-8, Chicago, IL. Harri, A., Erdem, C., Coble, K. H., & Knight, T. O. (2009). Crop Yield Distributions: A Reconciliation of Previous Research and Statistical Tests for Normality. Review of Agricultural Economics, 31(1), 163-182. doi: DOI 10.1111/j.1467-9353.2008.01431.x Hazell, P. B., Pomareda, C., & Valdés, A. (1986). Crop insurance for agricultural development: issues and experrience: IICA Biblioteca Venezuela. Hennessy, D. A. (1998). The Production Effects of Agricultural Income Support Policies under Uncertainty. American Journal of Agricultural Economics, 80(1), 46-57. doi: 10.2307/3180267 Hou, L.L., Y. Y., Mu and Y. Z. Zeng. 2010. Empirical Analysis on the Farmers Willing of Buying Insurance Effects and Subsidies Policy ofAgricultural Insurance, Issues in Agricultural Economy, 4, 19-25. (In Chinese) Just, R. E., & Weninger, Q. (1999). Are Crop Yields Normally Distributed? . American Journal of Agricultural Economics, 81, 287-304. Kimura, S., & Thi, C. L. (2011). Farm Level Analysis of Risk and Risk Management Strategies and Policies: TECHNICAL NOTE (Vol. OECD Food, Agriculture and Fisheries Working Paper). Paris: OECD Publishing. Kang, M (2012). Effect Evaluation of Government-supported Agricultural Insurance in Different Schemes. Master of Sciences, Chinese Academy of Agricultural Sciences Lin, C. Y. (2001). An Economic Analysis of Alternative Rice Insurance Policies in Taiwan. Taiwanness


Doi：10.1016/S2095-3119(14)60842-X

Agricultural Economic Review, 6(2), 235-253. Mahul, O., & Stutley, C. J. (2010). Government Support to Agricultural Insurance: Challenges and Options for Developing Countries. Washington D.C: The World Bank. O'Donoghue, E. J., Key, N., & Roberts, M. J. (2005). Does risk matter for farm businesses? The effect of crop insurance on production and diversification. Paper presented at the 2005 AAEA meeting, Providence, TO. Turvey, C. G. (1992). An Economic Analysis of Alternative Farm Revenue Insurance Policies. Canadian Journal of Agricultural Economics, 40, 403-426. Wang, H. H., Hanson, S. D., Myers, R. J., & Black, J. R. (1998). The Effects of Crop Yield Insurance Designs on Farmer Participation and Welfare. American Journal of Agricultural Economics, 80(4), 806-820. Wang, H. H., Makus, L. D., & Chen, X. (2004). The impact of US commodity programmes on hedging in the presence of crop insurance. European Review of Agricultural Economics, 31, 331-352. Wu, Y. 2005. Theoretical Basis of Agricultural Insurance and the Analysis of Its Effectiveness, Social Sciences, 12, 20-25 (In Chinese) Shi, H. 2008. Review on the fiscal subsidy system for Agricultural Insurance in the United States. Insurance Studies, 4, 91-94. (In Chinese) Sun, X. Y., & Zhong, F. N. (2008) The welfare economic anlysis of crop insurance subsidy. Issues in Agricultural Economy, 2, 1-11 (In Chinese) Tuo, G.Z. and G. J., Wang. 2003. Study on Chinese Agricultural Insurance and Rural Social Security, Beijing: Capital University of Economics and Business Press (In Chinese) Zhang, Y. H., & Shi, H. (2007). Subsideis, welfare and policy-oriented agricultural insurance: an intensive study based on welfare economics. Journal of Zhejiang University (Humanities and Social Sciences), 37(6), 138-146. (In Chinese) Zhang, Y. H., Q. H., Shi and H. Y., Gu. 2005. A Theoretical and Positive Study on the Demand of Crop Insurance —— Based on the questionnaires from 662 Farmers’ households of Henan province, The Journal of Quantitative & Technical Economics, 4, 65-75(In Chinese)

Annex

Figure 1 Location of sample farmers

Table 1 Description of Sample data


Doi：10.1016/S2095-3119(14)60842-X

Province

JiangSu

ShanDong

HeNan

SiChuan

Shan'Xi

Total

Farm Type

Crop Farm

Crop Farm

Crop Farm

Crop Farm

Crop Farm

Crop Farm

Major Commodity Num of Sample Length of the data

Corn,Oilseed Corn,oilseed Corn,Oilseed Corn,Oilseed ,Rice, Wheat , Wheat ,Wheat ,Rice,Wheat

Corn,Wheat

97

39

282

75

81

574

6 years (2003-2008)

6 years (2003-2008)

7 years (2003-2009)

7 years (2003-2009)

7 years (2003-2009)

6 or 7 years

Table 2 Cluster analysis for sample farmers in five provinces cluster Cluster 1

Cluster 2

Cluster 3

Cluster 4

Total

# of Individual Farmers Risk of Agri. income (Normalized) % of Agincome in Household income # of Individual Farmers Risk of Agri. income (Normalized) % of Agincome in Household income # of Individual Farmers Risk of Agri. income (Normalized) % of Agincome in Household income # of Individual Farmers Risk of Agri. income (Normalized) % of Agincome in Household income # of Individual Farmers Risk of Agri. income (Normalized) % of Agincome in Household income

JS 84 -0.389 52.90% 13 2.025 36.10%

SD 37 -0.512 63.80% 2 1.966 69.00%

97 -0.0657 50.6%

39 -0.385 64.1%

HN 235 -0.051 40.80% 37 -0.156 85.60% 9 3.304 47.50% 1 6.805 22.60% 282 0.0667 46.9%

SC 69 -0.129 48.00% 6 3.297 50.30%

SX 75 -0.351 51.40% 6 2.431 73.20%

Total

75 0.145 48.2%

81 -0.145 53.0%

574 -0.00593 49.7%


Doi：10.1016/S2095-3119(14)60842-X

Table 3 Summary of representative farmers, by crops, by cluster and by provinces JS Cluster 1 mean

Area (Mu)

Price* (CNY)

Cluster 2

sd

mean

Cluster 1

sd

corn

1.2

0.6

oilseed

1.7

0.9

1.6

0.8

rice

1.5

0.6

1.4

0.4

wheat

Yield* (kg/Mu)

SD

mean

HN Cluster 2

sd

mean

Cluster 1

sd

mean

Cluster 2

sd

mean

SC Cluster 3

sd

mean

3.1

1.9

3.2

1.1

2.9

1.5

2.9

1.6

0.3

0.1

0.2

0.1

2.0

1.4

2.2

2.2

3.1

Cluster 4

sd 1.5

mean 0.8

Cluster 1

sd

mean

0.3

SX Cluster 2

sd

mean

Cluster 1

sd

1.3

0.9

0.9

0.3

1.3

0.7

1.3

0.5

1.4

0.9

1.0

0.5

mean 4.7

Cluster 2

sd 2.0

mean

sd

5.6

2.2

1.7

1.1

6.9

2.7

6.1

3.2

4.3

2.6

4.2

3.2

3.3

1.9

0.8

0.3

1.2

0.7

0.9

0.5

6.7

3.2

5.9

1.8

corn

345.3

123.1

398.5

71.6

407.2

64.1

342.4

143.4

354.4

297.4

340.8

95.9

498.6

26.8

355.7

122.5

348.9

118.3

257.0

66.6

234.8

60.0

oilseed

181.1

29.4

172.7

8.1

168.1

68.3

186.6

59.2

56.5

43.4

62.1

33.1

163.8

58.0

139.9

38.0

rice

592.9

44.2

602.5

45.7

407.5

159.9

401.3

101.3

wheat

257.3

69.4

350.8

58.0

326.4

35.8

336.8

156.8

326.4

69.8

353.8

61.8

389.8

30.3

304.2

80.6

297.1

76.7

226.7

62.5

245.4

72.5

corn

1.4

0.3

1.1

0.1

1.1

0.1

1.0

0.1

1.0

0.1

1.0

0.1

1.0

0.1

1.3

0.2

1.3

0.2

1.0

0.1

1.0

0.1

oilseed

9.7

9.2

15.5

12.3

3.4

1.3

3.3

1.4

6.4

1.6

5.8

2.2

2.3

0.5

2.2

0.5

rice

1.5

0.1

1.5

0.1

2.4

1.3

2.3

1.2

1.1

0.1

1.3

0.1

1.3

0.1

1.2

0.1

1.2

0.1

1.2

0.1

1.2

0.1

1.3

0.1

1.3

0.1

1.3

0.1

1.3

0.1

Total income*

wheat

28353

24732

32228

11896

17265

7830

29426

34192

16806

13378

18158

14195

25727

23143

20988

18533

20862

10050

24004

17465

15888

23531

19978

20162

Agri. income*

15921

25587

12676

13372

10530

4237

22298

25562

6601

9489

15667

13580

15113

24147

10578

21606

9011

5167

13692

17022

9588

24166

16481

16812

G_Crops income*

1776

1632

1467

1509

1107

361

1183

661

735

303

706

306

774

186

907

154

1748

630

1681

608

441

195

419

203

Other Agri_income*

14145

25838

11209

13649

9423

4242

21115

25547

5866

9477

14961

13589

14339

24140

9670

21556

7263

5115

12011

17001

9147

24160

16062

16797

Cost_Farming

1184

677

1143

444

4097

1817

3936

1940

5204

88339

2032

1787

2238

1133

959

243

1996

1229

1643

1000

2756

1317

2570

1252

Non-Ag income*

1770

5682

2764

8424

1783

5174

1410

3445

3252

7776

496

2289

3147

8992

1873

4156

3820

8677

3404

7513

1371

3240

264

1195

NAInc_Local*

1234

4173

1717

5931

408.1

3937

0

0

207.5

1989

0

0

1088

7769

0

0

169.5

2107

35.56

346.6

120

1033

47.38

299.7

NAInc_Migrate*

680.4

4064

1047

5515

1375

3540

1410

3445

3045

7471

495.8

2289

2059

5002

1873

4156

3651

8481

3369

7454

1251

3094

216.7

948.2

428

1444

150

546

546

1545

3177

5904

500

1616

386

934

505

1206

27

60

890

1747

622

979

309

985

897

3038

10327

9599

16663

10143

4368

4694

2602

3989

6474

8147

1684

2874

6983

8666

8545

4881

7203

7742

6335

9158

4747

4521

2364

3129

Trincome* A&R income*

Note：The unit of all monetary variables in this table is Chinese Yuan, and * indicates deflated or detrended. G_crops income=income from grain crops (wheat, rice, corn and oilseed); Other Agri_income= other agricultural income from other crops, livestock and other farming activities; Cost_farming includes the cost of grain crops and other crops; NAinc_local and NAinc_migrate means the wage income from working in local s and outside city, respectively; Trincome means the Transfer income; A&R income means the Assert and residual income


Doi：10.1016/S2095-3119(14)60842-X

Table 4 Correlation matrix of yield and price by crops, by cluster, and by provinces


Doi：10.1016/S2095-3119(14)60842-X

JS_Cluster 1 Yield Corn Oilseed Rice Corn 1.0 -1.0 -1.0 Oilseed 1.0 1.0 yield Rice 1.0 Wheat Corn Oilseed Price Rice Wheat

JS_Cluster 2

Price Wheat Corn Oilseed Rice -0.8 -0.6 -0.8 0.8 0.8 0.6 0.8 -0.8 0.8 0.6 0.8 -0.8 1.0 1.0 1.0 -0.3 1.0 1.0 0.0 1.0 -0.3 1.0

Yield Wheat Corn Oilseed Rice 1.0 Na Na Na -1.0 1.0 0.0 -0.9 1.0 -0.9 -0.8 -0.9 0.6 1.0

SD_Cluster 1 Yield Corn Oilseed Rice Corn 1.0 -0.2 Na Oilseed 1.0 Na yield Rice Na Wheat Corn Oilseed Price Rice Wheat

Yield Wheat Corn Oilseed Rice -0.4 1.0 -0.3 Na 0.0 1.0 Na Na Na 0.2 0.4 -0.3 Na 1.0 Yield Wheat Corn Oilseed Rice 0.02 1.00 0.44 Na -0.02 1.00 Na Na Na 0.36 0.27 -0.63 Na 1.00

Price Wheat Corn Oilseed Rice 0.59 -0.04 Na Na Na Na Na Na Na Na Na Na 1.00 0.08 Na Na 1.00 Na Na Na Na Na

Yield Wheat Corn Oilseed Rice 0.08 1.00 Na Na Na Na Na Na Na 0.00 0.63 Na Na 1.00

HN_Cluster 3 Yield Corn Oilseed Rice Corn 1.00 Na Na Oilseed Na Na yield Rice Na Wheat Corn Oilseed Price Rice Wheat

Price Wheat Corn Oilseed Rice 0.3 -0.1 -0.2 Na 0.8 0.2 0.9 Na Na Na Na Na 1.0 -0.1 0.6 Na 1.0 0.7 Na 1.0 Na Na

Wheat 0.0 -0.7 Na -0.9 0.5 -0.3 Na 1.0

HN_Cluster 2 Price Oilseed Rice -0.09 Na 0.21 Na Na Na -0.50 Na -0.25 Na 1.00 Na Na

Wheat Corn 0.15 -0.01 -0.09 0.23 Na Na 1.00 -0.02 1.00

Wheat Na Na Na Na Na Na Na Na

SD_Cluster 2

Price Wheat Corn Oilseed Rice 0.0 -0.3 0.0 Na 0.0 0.0 0.0 Na Na Na Na Na 1.0 0.1 0.0 Na 1.0 0.7 Na 1.0 Na Na HN_Cluster 1

Yield Corn Oilseed Rice Corn 1.00 -0.05 Na Oilseed 1.00 Na yield Rice Na Wheat Corn Oilseed Price Rice Wheat

Price Wheat Corn Oilseed Rice Na Na Na Na Na Na 0.2 -1.0 Na Na -0.1 0.1 Na Na Na Na Na Na Na 1.0 -0.3 1.0

Price Wheat Corn Oilseed Rice 0.67 0.25 -0.50 Na 0.29 -0.39 -0.52 Na Na Na Na Na 1.00 -0.15 -0.78 Na 1.00 0.27 Na 1.00 Na Na

Wheat 0.58 0.54 Na 0.53 0.04 -0.71 Na 1.00

HN_Cluster 4

SC_Cluster 1 Yield Price Corn Oilseed Rice Wheat Corn Oilseed Rice Wheat Corn 1.00 0.36 0.32 0.28 -0.03 0.25 -0.32 0.23 Oilseed 1.00 0.43 0.25 -0.09 0.15 -0.44 0.08 yield Rice 1.00 0.47 0.05 0.16 -0.75 0.16 Wheat 1.00 0.02 0.13 -0.42 0.15 Corn 1.00 -0.47 -0.12 0.33 Oilseed 1.00 -0.15 0.05 Price Rice 1.00 -0.05 Wheat 1.00 SX_Cluster 1 Yield Price Corn Oilseed Rice Wheat Corn Oilseed Rice Wheat Corn 1.00 Na Na 0.26 -0.21 Na Na -0.02 Oilseed Na Na Na Na Na Na Na yield Rice Na Na Na Na Na Na Wheat 1.00 -0.04 Na Na -0.02 Corn 1.00 Na Na 0.52 Oilseed Na Na Na Price Rice Na Na Wheat 1.00

Price Wheat Corn Oilseed Rice 0.86 -0.22 Na Na Na Na Na Na Na Na Na Na 1.00 0.01 Na Na 1.00 Na Na Na Na Na

Wheat -0.42 Na Na -0.01 -0.17 Na Na 1.00

SC_Cluster 2 Yield Price Corn Oilseed Rice Wheat Corn Oilseed Rice Wheat 1.00 -0.17 0.34 0.79 0.99 0.15 0.37 0.77 1.00 0.38 0.30 -0.23 0.94 -0.33 0.49 1.00 0.79 0.23 0.54 0.73 0.51 1.00 0.73 0.56 0.49 0.88 1.00 0.09 0.30 0.74 1.00 -0.14 0.73 1.00 0.07 1.00 SX_Cluster 2 Yield Price Corn Oilseed Rice Wheat Corn Oilseed Rice Wheat 1.00 Na Na 0.05 0.10 Na Na 0.09 Na Na Na Na Na Na Na Na Na Na Na Na Na 1.00 -0.53 Na Na 0.13 1.00 Na Na 0.57 Na Na Na Na Na 1.00


Doi：10.1016/S2095-3119(14)60842-X

Table 5 Practical crop insurance policies in five Chinese provinces Insured Value (CNY/Mu)

Gov. subsidy ratio*

Premium Ratio (%)

Province Corn

Oilseed

Rice

Wheat

Corn

Oilseed

Rice

Wheat

JS

300

300

300

300

5

5

5

5

SD

300

300

320

3.3

5

HN

192

300

263

311

6

6

6

6

SC

300

280

300

300

7

5.5

7

7

SX

280

300

300

7

5

Crops

3.1 80%

5

CRRA=2

CRRA=4

4 3 2 1 0

JS

SD

HN

SC

SX

SC

SX

JS

SD

HN

SC

SX

CRRA=6 4 3 2 1 0

JS

SD

HN

CCI

DSF

Note: NOCI is the Baseline,=0

Figure 2 Welfare change of representative farmers under alternative scenarios

*

In practice, the crop insurance subsidy ratios range from 75% to 80% in China, depending on the region and crop. For simplicity, the subsidy ratio in this paper is set to 80% for all crops in all provinces.


Doi：10.1016/S2095-3119(14)60842-X

CRRA=2

CRRA=4

JS

JS

SD

SD

SC

SC

SX

SX -.4

-.2

0

.2

.4

.6

CRRA=6 JS SD SC SX -.4

-.2

0

.2

.4

.6

Change of Farmer's CE, % MCI120%

MCI150%

MCI200%

NOCI

Note: CCI is the Baseline equal to zero; Farmers in cluster 1 are selected; MCI120% means increaing insured value of CCI to 120%

Figure 3 Simulation results of increasing insured value of crop insurance

Note: MCI_10% means the modified crop insurance with 10% deductible; NOCI is the baseline.

Figure 4 Simulation results of decreasing deductible of crop insurance

DSF


MCI200%

Doi：10.1016/S2095-3119(14)60842-X

MCI150%

MCI100%

CCI

DSF

JS

SD

SC

SX

NOCI

1.0050 1.0000 0.9950 0.9900 0.9850

1.0050 1.0000 0.9950 0.9900 0.9850 0

10

20

30

40

50

60

70

80

90

0

10

20

30

40

50

60

70

80

90

Premium subsidy ratio by government (%) Note: CCI is the Baseline equal to one; Farmers in cluster 1 are selected; MCI150% means increaing insured value of CCI to 150%

Figure 5 Farmer’s Welfare of alternative crop insurance program in JS, SD, SC, and SX