results of the selected log-linear model in Table 4. The R-squared value for this model is very high,
at almost 0.98, due to dummy variables. There is not much difference between the estimated
coefficients of government expenditure and tourism: their elasticity values are estimated to be
between 0.23 and 0.24. Tourism, however, has a t-value of 1.14, which is not statistically
significant. On the other hand, the t-value for government expenditure is statistically significant.
Taking into account the minimum SBIC model, which is estimated without the tourism variable,
the government expenditure in Japanese island regions has a significant positive impact on the per
capita taxable income, whereas the number of tourists does not. Table 4 also shows that the
population size has a positive and statistically significant coefficient. It indicates that population
growth also increases the productivity of the local economy in the island regions. We also note
that the adjustment coefficient is estimated to be 0.026, which indicates a relatively slow
adjustment of taxable income: it takes 88 years to achieve 90% of the “ideal income”, or 27 years
to reach 50%.
We conclude by discussing the following implications of these findings for economic
development in the remote islands in Japan. First, the government expenditure can be considered
an effective method for economic development in remote islands. Second, there is some doubt
about whether tourism development may be regarded as an effective method for economic
development in remote islands. Last, the expected population decrease in remote islands will have
a negative impact on economic development.
CONCLUSIONS
In this paper, we examine the impacts of fiscal expenditure, the number of tourists, and the
population size on the per capita taxable income of 124 island regions in Japan by conducting an
empirical analysis using panel data. A log-linear partial adjustment model that has only a time
dummy and 0 lag is selected as the optimal model by the SBIC and a nonnested test. The
estimated results indicate that both fiscal expenditure and population size have a significant
positive impact on the per capita taxable income, whereas the number of tourists has no
statistically significant impact. We discuss the political implications of these results: (1) the
taxable income in the remote islands would decrease if financial support from the central
government to local governments decreases; and (2) the tourism development expected to be an
effective measure for economic development would not work as a substitute for financial support
from the government. There is a possibility that further reduction of taxable income would result
if declines in population size and government expenditure are both taken into account, although
we have not adequately analyzed this effect in this paper.
There is a possibility that the t-value for tourism may be underestimated if the coefficient for
tourism is different in each island region. Taking into account the results for the Amami Oshima
Islands, which were shown by ISHIKAWA and FUKUSHIGE, 2006, we cannot deny the
possibility that either a linear or a log-linear model would be appropriate in particular island