Productivity Dynamics and R&D Spillovers in the Japanese Manufacturing lndustry : An Empirical Analysis Based on Micro-level Data

工場レベルのデータを用いた最近の生産性動学分析によれば，生産性の高い工場が閉鎖されるために退出効果が負であったことと，中小工場で全要素生産性（TFP）上昇が低迷したことが，1990年代以降の日本の製造業の生産性上昇低迷の主因であった．本論文では，『工業統計調査』や『科学技術研究調査』のミクロデータを接合し，地域経済学の視点からこの2つの問題を研究した．我々は第一に，生産性動学を全製造業と都道府県別に行い，どの地域で負の退出効果が生じたかを分析した．その結果，1995年以降，東京，大阪，神奈川など製造業の集積地で大きな負の退出効果が生じたことが分かった．我々は第二に，工場のTFP上昇に対する，当該工場を持つ企業の研究開発（R&D）の効果，他社や公的機関のR&Dのスピルオーバー効果，等を計測した．その結果，ある企業の工場が他企業のR&Dから受けるスピルオーバー効果は，他企業の工場との距離が遠いほど減衰すること，産業集積地におけるR&D集約的な企業の工場閉鎖が1990年代後半以降スピルオーバー効果を著しく弱めたことが分かった．産業集積地におけるR&D集約的な企業の工場閉鎖が，負の退出効果と中小工場におけるTFP上昇の低迷を同時にもたらしたことになる．この他，1990年代後半以降，大学以外の公的機関のR&Dが減少したことにより，公的R&Dから日本の製造業へのスピルオーバー効果が低下した可能性が高いことも分かった．Recent studies on productivity dynamics using plant-level data found that a major sources of the decline in aggregate productivity growth is the negative exit effect, in which the productivity level of exiting plants is higher than the industry average, and the total factor productivity (TFP) of small plants has stagnated. Using matched data of the Census of Manufactures and the Report on the Survey of Research and Development for 1987 and 2007, we examine two issues by focusing on regional economics. First, we decompose the aggregate productivity growth in Japan's manufacturing sector and prefectural level to investigate in which prefecture did negative effects occur. We found that a large negative exit effect occurred in manufacturing plants agglomerations such as Tokyo, Osaka, and Kanagawa after 1995. Second, we analyze the effect of research and development (R&D), private R&D spillovers, and public R&D spillovers on productivity growth in the Japanese manufacturing sector. Our findings are as follows. (1) The effect of R&D spillovers from other firms is attenuated by distance. (2) The effect of R&D spillovers across firms has declined remarkably since the late 1990s due to exits of plants which belong to R&D intensive firms in the agglomerations. This means that the exit by such plants simultaneously cause the negative exit effect and stagnation of TFP growth in small plants. (3) The effect of public R&D spillovers is more likely to decline. This is caused by the reduction of R&D in public research organizations since the late 1990s