日本戰後農地價格變動因素之分析－－資訊元分析法的應用

Factors Influencing Farmland Prices of Post-War Japan-- An Approach of Bits of Information Measure

本文利用資本化模式和資訊元分析法，應用在1962～1996年日本各都道府縣的資料上，以研究農業淨所得、利率、通貨膨脹率和農業依存度、經營規模面積、稻作比例、區位等外生變數對農地價格的影響。資本化模式是利用預期報酬之折現值以評估地價，而資訊元則是根據各變數的偏相關係數編製指標，以反映各自變數所提供之資訊程度大小。根據資訊元與自變數資訊不等數分析，日本全國的農地訂價模式，大部份都道府縣以農業淨所得為最重要的因素，而通貨膨脹率次之，利率更次之。但都會區如東京都、大阪府的訂價模式則以通貨膨脹率為最重要的因素。而各地區農業淨所得所提供的資訊元比例以東海地區最高，東北地區最低；利率方面以東北地區最高，關東•東山地區最低；通貨膨脹率方面以北海道地區最高，東海地區最低。此外，由各地區的自變數資訊不等數得知，關東•東山地區、近畿地區是資訊分布差異較大的地區，東海地區、九州地區是資訊分布差異較小的地區。而由迴歸資訊不等數可知，關東•東山地區和近畿地區的迴歸資訊不等數較高，九州地區、東海地區的迴歸資訊不等數較低，所以農地訂價模式較適於分析後者的資料。最後，根據Pooling Data迴歸分析，農業淨所得與通貨膨脹率對農地價格的變動率，有正的影響，利率則有負的影響。至於農業依存度對農地價格變動率的影響為負向，亦即顯示農業依存度越小、工商業化和都市化的程度越深之都道府縣，農地價格的變動率越大。而稻作比例的係數為負，表示稻作比例較高之都道府縣，農地價格的變動率較小。同時結果亦顯示，經營規模面積較大之都道府縣，農地價格變動率較大。且東日本地區或西日本地區，農地價格的變動率皆比北海道地區為大。目前由於在WTO體制下，自由進口的壓力，將使國內農產品價格下跌，隨而農業資產報酬亦將減少，農地價格亦將下降。因此只要控制利率和通貨膨脹率之穩定，農地價格下降將有助於經營規模面積之擴大，以提高競爭力，減少未來所面臨的各種衝擊。

A Capitalization Model and Bits of Information Measure are used to study the land price behavior of Japan prefectures and districts in the period of 1962~96. Net agricultural returns, inflation, interest rate, degree of agricultural dependence, farm land size, percentage of rice production, dummy variables for various period are employed to explain the land price variation. Capitalization models use discounted flow of expected returns to evaluate farmland. The Bits of Information Measure composed of the partial correlation coefficients, weighs independent variables' explanation power for farmland price changes. The analysis of Bits of Information and Information Inequality concluded that the variable of net agricultural returns was the most important factor in most of the models. And inflation and interest rates were the second and third important variables, respectively. However, the models of urban areas such as Tokyo and Osaka had inflation as the most critical variable. In the analysis of districts, net agricultural returns offered the highest ratio of Bits of Information in the model of Tokai, while the least ratio in that of Tohoku. Interest rate offered the highest ratio of Bits of Information in the model of Tohoku, while the least ratio in that of Kanto-Tosan. Inflation offered the highest ratio of bits of information on land price variation of Hokkaido, while the least ratio on land price variation of Tokai, The analysis of Information Inequality across Regressors showed that the distributions of Information Bits in Kanto-Tosan and Kirnki were more dispersed, while the distributions in Tokai and Kyushu were less diffuse. The analysis of Inequality of Information in Regressions, which revealed the variations of Bits of Information offered by models across prefectures, were high in Kanto-Tosan and Kinki and low in Kyushu and Tokai. This means that the capitalization model is appropriate for the latter data. The regressive analysis of pooling data showed that net agricultural returns and inflation had a positive influence on land price, and that interest rate had a negative influence. Moreover, the smaller the degree of agricultural dependence was or the deeper the industrialization and urbanization was, the more the land price fluctuated. The prefectures with higher percentage of rice production or those with smaller farm size would have lower land price fluctuation. Both land prices of Eastern Japan and Western Japan were more volatile than that of Hokkaido. Under the framework of WTO, the free import has pushed down the prices of domestic sgricultursl products and the returns of agricultural assets. This has resulted in the decrease in farmland prices. Therefore, controlling interest rate and inflation as stable levels, the decrease in land prices may stimulate the expansion of farm sizes such that Japanese agriculture may become more competitive in the circumstances of globalization and liberalization.