Method<br>We have analyzed the productivity of government capital formation in three steps.<br>Firstly, we have estimated the productivity of government capital formation in every prefecture, and compared it with the productivity of private investment in the same prefecture. To do this, we first estimated the capital stock of both government capital formation and private investment in each prefecture, and calculated the prefecture's marginal productivity on the basis of parameters gained by the following Cobb-Douglas product function:<br>Y (Gross Regional Product)=A׋Number of Workers›^α׋Private Investment›^β׋Government Capital Formation›^γ<br>Secondly, we have picked up roads, seaports and airports as important specimens of government capital formation and estimated the respective marginal productivity achieved in each region. The necessary parameters were gained by the Cobb-Douglas product function adapted for sectional analysis: Y=AE^αK^β_??_ when Y (Gross Regional Product)=A׋Number of Workers›^α׋Private Investment›^β׋Government Capital Formation›^γ.<br>Thirdly, we have analyzed the employee compensation effect of government investment in each region. To do this, we calculated both its direct and indirect effects, on the basis of “input coefficient table” and “inverse matrix table” in Inter-regional Input-output Table.<br>Results<br>1. Regional productivity achieved by government investment.<br>(1) In almost all prefectures, both MPG (marginal productivity of government investment) and MPK (marginal productivity of private investment) are gradually declining. This is most visible in non-metropolitan areas<br>(2) The trend is MPK<MPG in metropolises and areas surrounding them, whereas it is MPK>MPG in other areas.<br>(3) Speaking of regional blocks, the trend is MPK<MPG in metropolitan areas of Kanto, Tokai, Kinki and Chugoku blocks, whereas it is MPK>MPG in non-metropolitan areas of Hokkaido, Tohoku, Ko-Shin-Etsu, Hokuriku, Shikoku and Kyushu blocks.<br>2. Marginal productivity achieved by the structures.<br>(1) General trend: the marginal productivity of the above-mentioned structures gets smaller in the order of ’airports<seaports<government investment>roads, ’ and it is declining in all of them, especially in ‘roads’.<br>(2) Seaports: the marginal productivity is rather stable here, and comparatively high in the Kanto, Tokai and Kinki blocks, that is, higher in metropolitan areas.<br>(3) Airports: the marginal productivity is highest here, especially in the Tokai, Kinki and Kanto blocks, though regional fluctuation is considerable. Among non-metropolis areas it is comparatively higher in the Kyushu block.<br>(4) Roads: the marginal productivity is generally low, especially in non-metropolitan areas, lowest in the Hokkaido, Tohoku, Shikoku, Chugoku and Hokuriku blocks, highest in the three largest metropolitan areas.<br>3. Compensation of employees achieved by public construction.<br>(1) The direct employee compensation effect of public construction is 0.08_??_0.11 and the indirect effect is 0.11_??_0.16.<br>(2) The direct effect is greater in Kanto, Kinki and Hokkaido, and smaller in the Chugoku and Shikoku blocks.<br>(3) The total of direct and indirect effects are greater in metropolitan areas, partly because of capital feedback to the headquarters there. This effect is largely nullified in Hokkaido, Tohoku, Chugoku and Shikoku through the capital flow out of these blocks.