Post on 10-Jul-2015
transcript
A hedonic approach toradiation contamination damages
Hatsuru Morita (Tohoku University, School of Law)
Table of contents
May 22, 2013Radiation Contamination (Morita)2
Motivation Econometric problems
Two biases Identification strategy Data
Legal problems Proprietary loss Non-proprietary loss
Concluding remarks
Motivation
May 22, 2013Radiation Contamination (Morita)3
May 22, 2013Radiation Contamination (Morita)4
May 22, 2013Radiation Contamination (Morita)5
Motivation TEPCO (and the Japanese government) has offered a
menu of compensation for radiation contamination damages It does not cover property damages nor mental damages
outside of the governmental evacuation area Within the evacuation area, TEPCO basically compensates the full
amount of property damages Many residents outside of the evacuation area are filing
lawsuits against TEPCO They are suffering a lot of inconvenience (and potential disease risk) During the 5 months after the earthquake and the accident, about
150 thousand elementary school children and junior high school children have moved out from Fukushima prefecture
May 22, 2013Radiation Contamination (Morita)6
Econometric problems
May 22, 2013Radiation Contamination (Morita)7
Hedonic method Explicit market for environmental factors usually does
not exist; however, Rosen 1974
Quality depends a set of characteristics: Q = (q1, q2, …, qn) Price of ith house: Pi = P(q1, q2, …, qn)
∂P/∂qj gives the marginal implicit price for q j
In competitive market, the marginal price is equal to an individual consumer’s ‘marginal willingness to pay’ (MWTP)
We can employ the hedonic estimates as proxies of mental damages as well as property damages
May 22, 2013Radiation Contamination (Morita)8
Econometric problems
May 22, 2013Radiation Contamination (Morita)9
Two sources of bias Unobserved omitted variables
Housing price is correlated with not only environmental factors, but also other factors E.g., air is heavily polluted in urban areas, where housing price also tends
to be high.
Self-selection Heterogeneity among individuals (both sellers and buyers) Individuals with higher valuations for environmental factor sort out to
areas with better environmental quality In this case, the structure of preference and the amount of sorting
behavior can affect the estimates.
Identification strategy
May 22, 2013Radiation Contamination (Morita)10
For radiation contamination: The omitted variable bias is not a serious problem
The degree of radiation contamination is exogenous to relevant variables excellent natural experiment The degree of contamination depends on geography, direction of wind, and
weather at the time of radiation emission, all of which are not correlated with housing price and control variables
Then simple analysis can reveal the MWTP For self-selection:
We need some measures to account for heterogeneity Sorting behavior does exist in Fukushima case (kids (and their parents)
are moving out) Change of population and age structure can mitigate the problem However, whether we need to control this factor is another problem
(discussed later)
Constructing counterfactuals How can we construct counterfactuals?
What the land price of the contaminated area would be if it had not been contaminated?
Solutions Difference-in-differences Synthetic control Structural estimation
May be a good tool for policy evaluation, but difficult to employ before the court
May 22, 2013Radiation Contamination (Morita)11
May 22, 2013Radiation Contamination (Morita)12
Matching Which cities to pick up as control?
Cities in northeastern Japan are good candidates Unobservables, such as level of urbanization/ industrialization,
demographic structures are basically similar One problem:
The coastal area of Fukushima is heavily damaged by the tsunami and many people have been moving into the inland area, which is also contaminated
In addition, those who had lived in the evacuation area have been moving into the inland area
Then the demand function for land in the inland area may have changed considerably
(solution is discussed later)
May 22, 2013Radiation Contamination (Morita)13
May 22, 2013Radiation Contamination (Morita)14
Tsunami area
Radiation contaminationarea
Data source[Under construction…]
Outcome variable Land price
Japanese government publishes land price every one year Two series of data (as of 1/1 and as of 7/1)Each city has 20-60 reference points
Treatment variable Level of radiation contamination
Other covariates Population, household income, age distribution, tax revenue,
and so on(Age structure is especially important, since younger children are more vulnerable to radiation contamination)
May 22, 2013Radiation Contamination (Morita)15
Variables
May 22, 2013Radiation Contamination (Morita)16
Do not include coastal area, but only inland area The coastal area is hit by tsumani, which makes it difficult to
separately estimate radiation contamination effect from tsunami effect
Evacuation area is also to be excluded, since we want to estimate the effect of radiation contamination only Evacuation order by the government must have put the housing price
downward strongly
Variables
May 22, 2013Radiation Contamination (Morita)17
Population change is an important variable, too Managi 2013 shows that both the level of radiation contamination and
the decrease of population have significant effect on land price However, decrease of population is not only caused by industry
structure, but also by the radiation contamination itself Whether it is appropriate to control population or not
depends on whether the change of population is within a proximate cause of the tort behavior by TEPCO Not an econometric issue, but a legal issue
A tentative result A simple DD estimate: -4.11%
Land price of Jul 1 (every one year) Fukushima city (capital of Fukushima) vs Morioka city (capital of
Iwate) Averaging 22 points from Fukushima and 46 points from Morioka
2009/2010/2011 (Assuming that each city has its own trend)
May 22, 2013Radiation Contamination (Morita)18
Change of land price
20092010 20102011
Fukushima city -3.41% -7.02%
Morioka city -7.08% -6.58%
May 22, 2013Radiation Contamination (Morita)19
Further concerns Stickiness of land market in Fukushima
Fukushima is rather rural area (compared to, say, Tokyo) and the liquidity of the land market is not so high In case of stock price, changes of fundamental firm value are rapidly
incorporated into the stock price In contrast, the low liquidity of land market in Fukushima would
cause considerable delay of incorporation of relevant information How could we account for such illiquidity?
Just observing long period after the earthquake (say, two or three years) is OK?
At the same time, estimation of longer windows leads to larger noise
May 22, 2013Radiation Contamination (Morita)20
Further concerns (cont’d)
May 22, 2013Radiation Contamination (Morita)21
Huge public funding has been provided to the disaster site and radiation contamination site (i.e., Fukushima) The inflow of public funding increases the land price We can observe the effect in Fukushima and Miyagi
Fukushima has both disaster relief and contamination relief Miyagi has only disaster relief, but Miyagi is the center of the
northeastern Japan and reconstruction activity center is located in Miyagi, which attracts lot of business
Is it necessary to account for the effect of public funding? We could argue that the effect of radiation contamination is mitigated
by the public funding and that we need not consider the effect of public funding
(similar argument with the case of insurance payment)
Legal problems
May 22, 2013Radiation Contamination (Morita)22
Legal problems
May 22, 2013Radiation Contamination (Morita)23
Let’s suppose we have found decline of real estate price: Then, there are two ways to take such decline into
account: Proprietary loss Non-proprietary loss
Dual nature of hedonic valuation Non-proprietary loss is capitalized into proprietary loss and
the latter is a good proxy for the former Double compensation needs to be avoided E.g., those who have received proprietary compensation are not
qualified for non-proprietary compensation
Proprietary loss
May 22, 2013Radiation Contamination (Morita)24
Is it possible to evaluate the ‘potential’ decrease of land price as proprietary loss?
Positive arguments Even if it is still ‘potential’, it is real loss
Negative arguments When a victim is paid for her non-proprietary loss, she gets
double compensation However, it is possible to adjust the two compensation
scheme But it may invoke additional transaction cost
Non-proprietary loss
May 22, 2013Radiation Contamination (Morita)25
If it is difficult to compensate the decline of real estate price as proprietary loss, then non-proprietary loss can be a viable alternative
In addition, those who do not own real estate and are just tenants do not suffer from the decline of housing price Rent is not so elastic (especially for downward) Coastal area residents and evacuated people have come into
inland area, which causes temporary increase of rent
Non-proprietary loss
May 22, 2013Radiation Contamination (Morita)26
Since how residents value radiation contamination is capitalized into real estate price, it is a good proxy for non-proprietary loss
A tentative proposal: Hypothesize typical size of a family house, which must be
bigger in Fukushima than in Tokyo, and calculate decrease of hypothetical decline of real estate price
Compare with TEPCO’s guideline for those from the evacuated area: \100,000 or 50,000/month for those who have evacuated \100,000 for those who have taken indoor-shelter
Non-proprietary loss
May 22, 2013Radiation Contamination (Morita)27
Need to account for heterogeneity? Families with younger children must suffer lot more than
families with no children A possible alternative
Compute the ratio of families with younger children; Families without younger children are not qualified to non-
proprietary loss; Families with younger children are qualified to non-proprietary loss
multiplied by the inverse of the above ratio; Cut-off age can be elementary school or junior high school
Non-proprietary loss
May 22, 2013Radiation Contamination (Morita)28
Underestimation? Illiquidity of Fukushima real estate market
It may take longer time for the change of utility to be capitalized into housing price
Difference of level of land price between urban area and rural area E.g., huge difference between price level of Tokyo and Fukushima However, the same problem arises in case of life and casualty loss,
where damages are determined by income level The difference is caused by capitalization of human capital and other
factors
Non-proprietary loss
May 22, 2013Radiation Contamination (Morita)29
In effect, ‘hedonic’ loss implies compensation for future loss Since future disutility is also capitalized into housing price, like
stock price Double compensation must be avoided
Selling a house after receiving ‘hedonic’ loss does not cause double compensation since the housing price is lower than before the accident
Those who come after the accident can purchase houses with lower price but are not qualified for ‘hedonic’ loss, since they ‘come to the nuisance’
Non-proprietary loss
May 22, 2013Radiation Contamination (Morita)30
Mitigation principle? If migration cost is cheaper than estimated decrease of real
estate price, then is it irrational to stay in Fukushima? No! The decision to stay in Fukushima is rational since there is
huge benefit from local ties and blood ties and moving out from Fukushima would destroy such ties Although there could be fake non-migrator, it is difficult to distinguish
them and we need to accept them
Concluding remarks
May 22, 2013Radiation Contamination (Morita)31
Concluding remarks
May 22, 2013Radiation Contamination (Morita)32
In order to achieve socially optimal deterrence, it is basically necessary to compensate for the whole externality
Traditional loss calculation method is not effective to achieve this goal and more comprehensive approach is desirable Econometric method of hedonic approach can be a useful tool
as a proxy
Concluding remarks
May 22, 2013Radiation Contamination (Morita)33
CBA of decontamination Although the cost of contamination is not included in the
intermediate guideline, it will probably included in the final guideline
However, we need a cost-benefit analysis, for example, of decontamination activity If the cost is larger than the benefit, then the decontamination
process is not justified Mitigation principle may require decreasing the amount of damages
If cost is smaller than benefit, then the decontamination is beneficial A caveat: possibility of underestimation