Milton S. Boyd

Designing a neural network for forecasting financial and economic time series

Abstract Artificial neural networks are universal and highly flexible function approximators first used in the fields of cognitive science and engineering. In recent years, neural network applications in finance for such tasks as pattern recognition, classification, and time series forecasting have dramatically increased. However, the large number of parameters that must be selected to develop a neural network forecasting model have meant that the design process still involves much trial and error. The objective of this paper is to provide a practical introductory guide in the design of a neural network for forecasting economic time series data. An eight-step procedure to design a neural network forecasting model is explained including a discussion of tradeoffs in parameter selection, some common pitfalls, and points of disagreement among practitioners.

A comparison of artificial neural network and time series models for forecasting commodity prices

Abstract A feedforward neural network which can account for nonlinear relationships was used to compare ARIMA and neural network price forecasting performance. Data used was monthly live cattle and wheat prices from 1950 through 1990. The experiment was repeated seven times for successive three year periods. This involved using a walk forward or sliding window approach from 1970 through 1990 which generated out of sample results. The neural network models achieved a 27 percent and 56 percent lower mean squared error than ARIMA model. The absolute mean error and mean absolute percent error were also lower for the neural network models. The neural network models were able to capture a significant number of turning points for both wheat and cattle, while the ARIMA model was only able to do so for wheat. Since this forecasting method is not problem specific and uses only past prices, it can be applied to other forecasting problems such as stocks and other financial prices.

Commodity futures trading performance using neural network models versus ARIMA models

Neural networks trading returns are compared out‐of‐sample with traditional ARIMA returns for corn, silver, and deutsche mark. Results show that neural network and ARIMA models had positive returns, and at about the same levels. However, deutsche mark was less profitable and returns were not statistically different from zero, in contrast to corn and silver.

Crop Insurance Principles and Risk Implications for China

ABSTRACT Firms within various sectors of an economy are often faced with a number of risks. These risks can be relatively sudden and large, especially if they are weather related. In agriculture, risk often has natural causes such as weather, and therefore losses can be large and costly in particular years. Crop insurance has been commercially available in many developed countries for a number of decades, though it is only now starting to become more commercially available in a number of developing countries such as China. When facing these risks without crop insurance, farmers may use fewer inputs and invest less in crop production, resulting in lower yields and lower production. As well, lenders may be reluctant to extend credit to farmers, if farmers have not purchased crop insurance. Crop insurance has been one of the most successful risk management and longest running stabilization programs for farmers in many parts of the world. The purpose of this article is to explain the main principles underlying crop insurance, with implications for China. Challenges for crop insurance development are also pointed out, along with some possible solutions. Some North American experience with crop insurance is also discussed, including the case of Canada.

Factors affecting crop insurance purchases in China: the Inner Mongolia region

Purpose - The purpose of this paper is to explain the factors affecting crop insurance purchases by farmers in Inner Mongolia, China. Design/methodology/approach - A survey of farmers in Inner Mongolia, China, is undertaken. Selected variables are used to explain crop insurance purchases, and a probit regression model is used for the analysis. Findings - Results show that a number of variables explain crop insurance purchases by farmers in Inner Mongolia. Of the eight variables in the model, seven are statistically significant. The eight variables used to explain crop insurance purchases are: knowledge of crop insurance, previous purchases of crop insurance, trust of the crop insurance company, amount of risk taken on by the farmer, importance of low crop insurance premium, government as the main information source for crop insurance, role of head of village, and number of family members working in the city. Research limitations/implications - A possible limitation of the study is that data includes only one geographic area, Inner Mongolia, China, and so results may not always fully generalize to all regions of China, for all situations. Practical implications - Crop insurance has been recently expanded in China, and the information from this study should be useful for insurance companies and government policy makers that are attempting to increase the adoption rate of crop insurance in China. Social implications - Crop insurance may be a useful approach for stabilizing the agricultural sector, and for increasing agricultural production and food security in China. Originality/value - This is the first study to quantitatively model the factors affecting crop insurance purchases by farmers in Inner Mongolia, China.

Feedforward versus recurrent neural networks for forecasting monthly japanese yen exchange rates

Neural networks are a relatively new computer artificial intelligence method which attempt to mimic the brains problem solving process and can be used for predicting nonlinear economic time series. Neural networks are used to look for patterns in data, learn these patterns, and then classify new patterns and make forecasts. Feedforward neural networks pass the data forward from input to output, while recurrent networks have a feedback loop where data can be fed back into the input at some point before it is fed forward again for further processing and final output. Some have argued that since time series data may have autocorrelation or time dependence, the recurrent neural network models which take advantage of time dependence may be useful. Feedforward and recurrent neural networks are used for comparison in forecasting the Japanese yen/US dollar exchange rate. A traditional ARIMA model is used as a benchmark for comparison with the neural network models.Results for out of sample show that the feedforward model is relatively accurate in forecasting both price levels and price direction, despite being quite simple and easy to use. However, the recurrent network forecast performance was lower than that of the feedforward model. This may be because feed forward models must pass the data from back to forward as well as forward to back, and can sometimes become confused or unstable. Both the feedforward and recurrent models performed better than the ARIMA benchmark model.

The Effectiveness of Area-Based Yield Crop Risk Insurance in China

ABSTRACT Area-based yield insurance (AYI) is a kind of index insurance that can eliminate asymmetric information problems (adverse selection and moral hazard) and reduce transaction cost. It is important in managing crop risk for developing countries such as China, which has many small farms. However, the biggest challenge in index insurance is basis risk. In this regard, can AYI be used in China? Is AYI competitive with traditional multi-peril crop insurance (MPCI)? What are the principal factors affecting the effectiveness of AYI? Following and extending the approaches suggested by Miranda in 1991, this article compares the effectiveness of AYI and MPCI in terms of risk reduction per premium. An empirical model was established, indicating the relationship between the efficiency of AYI and the explanatory variables, using data from a survey of 108 wheat farmers from two counties in Hebei Province, China. The results provide salutary suggestions to area-based index insurance design and development in China.

Testing the effectiveness of government transfers for agricultural revenue stabilization: the case of the Western Canadian grain sector

Abstract This is the first study to document major agricultural policy transfers and their impact on agricultural stabilization. While many studies have analyzed stabilization theory, this is the first to use a complete actual data set to empirically analyze stabilization effectiveness. Transfers in the Western Canadian grain sector are identified and their effectiveness for providing revenue stability is examined. Results show that while agricultural policies from 1971 to 1990 increased the total gross grain revenue, they did not decrease the variability of the revenue. Therefore, the objective of revenue stabilization for agricultural policy may not have been adequately achieved.

Insurance Premium Calculation Using Credibility Analysis: An Example from Livestock Mortality Insurance

A major problem facing livestock producers is animal mortality risk. Livestock mortality insurance is still at the initial stages, and premium computation approaches are still relatively new and will require more research. This study seeks to provide a first step for developing a better understanding of livestock insurance as a solution to mortality risk, as it explores improved methods for livestock mortality insurance modeling procedures, and premium computation, using credibility analysis. The purpose of this study is to develop improved estimates for livestock mortality insurance premiums for Canada under a credibility framework. We illustrate our approach through one example using livestock data from 1999 to 2007.

Factors affecting farmers’ willingness to purchase weather index insurance in the Hainan Province of China

Purpose - – The purpose of this paper is to explain the factors affecting farmers’ willingness to purchase weather index insurance for crops in China, in the Province of Hainan, and to also provide additional background information on weather index insurance. Design/methodology/approach - – A survey of 134 farmers was undertaken in Hainan, China, regarding their willingness to purchase weather index insurance. A probit regression model was used, and a number of variables were included to explain willingness of farmers to purchase weather index insurance. Findings - – In total, 11 of 15 variables in the model are found to be statistically significant in explaining farmers’ willingness to purchase weather index insurance. Research limitations/implications - – First, farmers’ interest in weather index insurance may be limited due to basis risk. Second, some farmers may not sufficiently understand weather index insurance and so may not purchase it, and a considerable portion of farmers may also require a subsidy if they are to purchase weather insurance. Practical implications - – Weather index insurance may provide a lower cost alternative than traditional crop insurance, however, basis risk remains a main challenge. Originality/value - – This is the first study to quantitatively study the factors affecting the willingness of farmers to purchase weather index insurance for agriculture in the province of Hainan, China.