Association of rainfall and detrended crop yield based on piecewise regression for agricultural insurance

Agricultural insurance is one of the most useful tools for managing the financial risks associated with farming. However, traditional insurance has several drawbacks, specifically in developing countries because of high transaction costs and other challenges that may hinder the protection from risk. Index based insurance is more likely to be a superior and viable alternative to traditional insurances for many developing countries because of its independent and objective nature. Various weather related factors are one of the major uncertainties that effect crop growth and yield. In order to develop an effective weather based insurance model, crop yield needs to be correlated with the weather factor(s) such as rainfall. However, crop yield pattern may be dependent on other external factors and in general create an increasing or decreasing trend in the yield. In our analysis, we observe that both downtrend and uptrend exists in our crop yield data with a threshold in the middle and thus creates a rare trend pattern that is cyclical. In order to identify the explicit relationship between yield and rainfall, we detrended crop yield by using piecewise regression. In addition, crop yield data that are collected from the field are usually noisy and the relationship between weather factors and yield responses are in general weak. Thus, a three-period moving average smoothing technique was applied on the data to make the pattern of the trend more visible. Consequently, identification of proper trend pattern, such as, cyclical-trend rather than a simple linear trend for detrending the crop yield appeared to be significant in this research study. As a result, our study adds significant contribution in this field of research concerning the influence of unobservable factor(s) on the crop yield that creates threshold effect. The implication of these findings in this study is significant for developing an appropriate associative model for creating weather based index insurance.