要旨：本研究は定量的空間経済モデル（quantitative spatial economics models）のパラメータを完全情報最尤推定（full information maximum likelihood, FIML）によって推定する新しい手法を提案する。内生変数の観測値から外生的な構造的残差（structural residuals）を一意に求められるモデルに対して、提案手法は適用可能である。構造的残差が確率分布に従うという仮定の下で、対数尤度関数を構築できる。提案手法は操作変数を必要としない。提案手法は、クロスセクションデータとパネルデータの双方に適用可能である。提案手法は、多くの場合において、パラメータの推定値を空間均衡が安定になる範囲に制約すると考えられる。本研究は、提案手法を二種類のデータに対して適用することにより、その有効性を確認する。一つ目は、既知のパラメータを有するモデルから発生させた実験データであり、二つ目は、1975年から2015年にかけての日本国内の観測データである。

Econometric Society World Congress 報告練習会

論文ファイル

AY2019

要旨：Land is everywhere, being the substratum of our existence. In addition, land is intimately linked to the dual concept of location in human activity. Together, land and location are essential ingredients for the lives of individuals as well as for national economies. Today, there exist two different approaches to incorporating land and location into a general equilibrium theory. Dating from the classic work of von Thünen (1826), a rich variety of land-location density models have been developed. In a density model, a continuum of agents is distributed over a continuous location space. Given that simple calculus can be used in the analysis, these density models continue even today to be the “workhorse” of urban economics and location theory. However, the behavioral meaning of each agent occupying an infinitesimal “density of land” has long been in question. Given this situation, a radically new approach, called the σ-field approach, was developed in the mid 1980s for modeling land in a general equilibrium framework. In this approach: (i) the totality of land, L, is specified as a subset of R^2, (ii) all possible land parcels in L are given by the σ-field of Lebesgue measurable subsets of L, and (iii) each of a finite number of agents is postulated to choose one such parcel . Starting with Berliant (1985), increasingly more sophisticated σ-field models of land have been developed. Given these two different approaches to modeling land within general equilibrium framework, several attempts have thus far been proposed for bridging the gap between them. But while a systematic study of the relationship between density models and σ-field models remains to be completed, the clarification of this relationship could open a new horizon toward a general equilibrium theory of land.