8 Level of Disaster Resilience and Migration …
141
the percentage of people with disability are not available since the Japanese government has not implemented a comprehensive survey of people with disabilities at
municipality level. Some municipalities disclose their information on the number of
people with disability certificates, but not all. Therefore, I used data on the number of
disability-related institutions per 1,000 population, including institutions for support
of the disabled, rehabilitation homes for the physically disabled, institutions for the
protection of persons with intellectual disabilities, social rehabilitation facilities for
mentally disabled persons and institutions supporting social activities for the physically disabled. I used the latest available data from 2011. I took an inverse of this value
after the z-transformation. The third indicator is out of poverty. Poverty imposes a
constraint on households and communities to respond to and recover from a disaster.
Out of poverty is measured by the inverse of per capita public assistance expenditure. The data on per capita public assistance expenditure comes from the System
of Social and Demographic Statistics. Again, I take an inverse of this value after
the z-transformation. The last socio-economic indicator is health care access. The
original RCI used health insurance coverage as a proxy for the general capacities
useful for effectively responding to and recovering from a disaster. However, this
makes little sense in the context of Japan because Japan achieved universal health
coverage through the social insurance system in 1961 and the entire population is
virtually covered by a social health insurance plan. However, there is still a disparity
in access to health care due to a shortage of physicians and their regional maldistribution. Therefore, I used the data on number of physicians per 1,000 population,
which comes from the System of Social and Demographic Statistics.
Community Connectivity Indicators reflect (1) the number of civic organizations,
(2) metropolitan stability measured by the percentage of long-term residents in the
community, (3) homeownership and (4) voter participation rates. RCI assumes the
density of civic organizations as a proxy for community engagement. Data on the
number of non-profit organizations per 10,000 residents come from the NPO Hiroba
Database. Last access to the database was on 24 December 2015. I divided the number
of NPOs by the number of total population in each municipality. The second indicator
is the measure of metropolitan stability, which assesses rootedness in community. I
used the share of population that remained resident in the municipality over a five-year
period in the Census 2015. The third indicator, homeownership, provides a measure of
attachment and commitment to place. I used data from the Housing and Land Survey
to calculate the proportion of owner-occupied housing units in total occupied housing
units. Finally, I measured civic engagement by using voter participation. Higher
voter participation indicates trust in the democratic process in local governance. The
dataset comes from the Japanese Local Elections Dataset, compiled by Horiuchi
and others (Horiuchi 2019). The share of voter-eligible population who voted in
the last election (in 2015) was calculated in each municipality. There are missing
observations because there are fewer candidates than seats available and no election
is held in 633 municipalities.
Due to the different measurement units of these 12 indicators, z-score transformation was employed to convert all the indicators to a common scale with a mean
of zero and a standard deviation of one. As noted above, I took an inverse of HHI,
141
the percentage of people with disability are not available since the Japanese government has not implemented a comprehensive survey of people with disabilities at
municipality level. Some municipalities disclose their information on the number of
people with disability certificates, but not all. Therefore, I used data on the number of
disability-related institutions per 1,000 population, including institutions for support
of the disabled, rehabilitation homes for the physically disabled, institutions for the
protection of persons with intellectual disabilities, social rehabilitation facilities for
mentally disabled persons and institutions supporting social activities for the physically disabled. I used the latest available data from 2011. I took an inverse of this value
after the z-transformation. The third indicator is out of poverty. Poverty imposes a
constraint on households and communities to respond to and recover from a disaster.
Out of poverty is measured by the inverse of per capita public assistance expenditure. The data on per capita public assistance expenditure comes from the System
of Social and Demographic Statistics. Again, I take an inverse of this value after
the z-transformation. The last socio-economic indicator is health care access. The
original RCI used health insurance coverage as a proxy for the general capacities
useful for effectively responding to and recovering from a disaster. However, this
makes little sense in the context of Japan because Japan achieved universal health
coverage through the social insurance system in 1961 and the entire population is
virtually covered by a social health insurance plan. However, there is still a disparity
in access to health care due to a shortage of physicians and their regional maldistribution. Therefore, I used the data on number of physicians per 1,000 population,
which comes from the System of Social and Demographic Statistics.
Community Connectivity Indicators reflect (1) the number of civic organizations,
(2) metropolitan stability measured by the percentage of long-term residents in the
community, (3) homeownership and (4) voter participation rates. RCI assumes the
density of civic organizations as a proxy for community engagement. Data on the
number of non-profit organizations per 10,000 residents come from the NPO Hiroba
Database. Last access to the database was on 24 December 2015. I divided the number
of NPOs by the number of total population in each municipality. The second indicator
is the measure of metropolitan stability, which assesses rootedness in community. I
used the share of population that remained resident in the municipality over a five-year
period in the Census 2015. The third indicator, homeownership, provides a measure of
attachment and commitment to place. I used data from the Housing and Land Survey
to calculate the proportion of owner-occupied housing units in total occupied housing
units. Finally, I measured civic engagement by using voter participation. Higher
voter participation indicates trust in the democratic process in local governance. The
dataset comes from the Japanese Local Elections Dataset, compiled by Horiuchi
and others (Horiuchi 2019). The share of voter-eligible population who voted in
the last election (in 2015) was calculated in each municipality. There are missing
observations because there are fewer candidates than seats available and no election
is held in 633 municipalities.
Due to the different measurement units of these 12 indicators, z-score transformation was employed to convert all the indicators to a common scale with a mean
of zero and a standard deviation of one. As noted above, I took an inverse of HHI,