This paper focuses on the links between health information related to food consumption and food consumption behavior. The impact of health risk information on salad oil consumption is investigated. Measurement of information is expressed in terms of the number of relevant articles on health risk topics in major newspapers in Japan. In recent years, Japanese salad oil consumption has not changed drastically, but rapeseed oil and safflower oil consumption have grown steadily.
　The applied model in this study is based on a discrete / continuous model which was derived by Hanemann and extended by Yoshino. This model enables the study to evaluate the effects of information implied by aggregated data without supposing a representative consumer.
　The results of cointegration analysis indicate that there is a long equilibrium relationship among health risk information, salad oil consumption and salad oil price. The relevant estimated coefficient implies that increasing health risk information has caused so-called premium oils, such as rapeseed oil or safflower oil, consumption to increase in the long term. The hypothesis that consumers never decrease health risk perception but increase it as additional health information is obtained over time is supported.

　The purpose of this paper is to evaluate the formulation and diffusion of the Japanese style of diet, one of the most important problems in Japanese food policy. It is said that the Japanese style of diet is healthy, but the long-term effects have not been evaluated. Using statistical criterion, I report on this subject. The data used in this paper is Japanese nutritional intake, which I use to formulate a time series model. I chose the ideal PFC balance to measure the formulation and diffusion of the Japanese style of diet. It is useful to show the mutual relationship of proteins, fats and carbohydrates, so I use a vector error correction model. I also consider cointegration to show that long-term relationship.
　As a result of the estimation, the hypothesis of stability was rejected when I used the data from 1955 to 1995. So I divided the data and tested again. The results of the estimation show that the ideal PFC balance has diffused from 1975 to 1995.

　This article estimated the effects of rural households' attitude toward increasing portfolio risk through changes in portfolio selection and demand for insurance using mail survey data.
　Although the majority of respondents did not change their portfolio selection, it was shown that people with stronger concerns over increasing financial risk behaved as risk averters.
　Because the ratio of respondents who canceled their insurance policies was comparatively high, demand functions for insurance were estimated for three different insurance suppliers, Seiho, Kanpo and JA, using the Tobit Type 2 model. The results indicated that, as a whole, concerns over financial risk did not have a large impact on insurance expenses, but Seiho users have stronger concerns over financial market risk than Kanpo and JA users.
　Demand functions for insurance were also estimated using a neural network model (Multi Layer Perceptron). The model improved the estimation error, and gave a more visual and detailed presentation to explain variable relationships.