Decomposing local fiscal multipliers: Evidence from Japan
Japan and The World Economy
A B S T R A C T Recent studies on fiscal policy use cross-sectional data and estimate local fiscal multipliers along with spillovers. This paper estimates local fiscal multipliers with spillovers using Japanese prefectural data comparable with the national accounts. We estimate the local fiscal multiplier on output to be 1.7 at the regional level. We decompose the regional fiscal multiplier into the prefectural fiscal multiplier and the region-wide effect. Converting the latter component into
... e spillover, we find that the spillover is positive and small in size. We also decompose the regional fiscal multiplier on output into multipliers on the expenditure components. Our estimates suggest that there are crowding-in effects of government spending on consumption and investment. Moreover, we find that the regional fiscal multiplier on absorption exceeds 2.0 and that the spillover to absorption is considerable in contrast to the spillover to output. (T. Tsuruga). 1 Chodorow-Reich (2019) comprehensively reviews numerous recent studies on the LFM. Ramey (2011) surveys fiscal and tax multipliers, including the time-series evidence. According to Auerbach et al. (2020) , understanding the spillover from government spending is "a fundamental and largely unresolved task in macroeconomics' ' (p. 196). In this paper, we estimate and decompose the LFM to understand the spillover in local economies. The objectives of this paper are threefold. First, we provide evidence of the LFM in Japan, which we can compare with LFMs in other countries. Second, and more importantly, we measure the spillover within the region using prefectural data. We separate the country into regions consisting of prefectures and estimate the regional fiscal multiplier (RFM) as the sum of the prefectural fiscal multiplier (PFM) and the region-wide effect. The former is a component of the RFM that is estimated from variations in prefectural government spending. The latter is also a component of the RFM but is estimated from variations in regional government spending and, thus, is related to the spillover. For consistency with the existing literature, we convert the estimated region-wide effect into the spillover within the region https://doi.