Willingness to Pay for Mortality Risk Reduction from Water Accidents: Application to Recreational Beaches, Miyagi Prefecture, Japan

The purpose of this paper is to measure benefits of the risk reductions on water accidents. 763 people who recreated at beaches were selected by an internet research. The reduction rates were 10%, 50%, and 90%. The median WTPs and the mean WTPs for reduction rates with and without protest bits were calculated. As results, the median WTPs at 10% reduction rates were calculated from 6 yen to 221 yen, the mean WTPs were from 615 yen to 820 yen. At 50% reduction rates, WTPs were from 743 yen to 1287 yen, and from 1256 yen to 1695 yen. At 90% reduction rates, WTPs were from 1607 yen to 2924 yen, and were from 2411 yen to 3433 yen.


. Time Series of Mortality Risks of Water Acci dents
Generally, it would be recogn ized among peop le that the mo rtality risks of rec reat ional act iv ities are lo w. Ho wever, if a person is in a water accident, the person would dead due to the h igh rate of mo rtality. Thus, the number of d isaster protection countermeasures, such as lifeguards, emergency service, and coastal levee, are perfo rmed in recreation sites. Costs (investments) for the countermeasures are usually p rovided by the local govern ment, however, the effects of the countermeasures are vague. Sine p roject evaluat ions are required in recent years, and then it is necessary to examine the benefit analysis on the countermeasures at recreation sites. Thus, the purpose of this paper is to perform the estimat ion o f willingness to pay (hereafter WTP) for the mo rtality risk reduct ion.
Since the mo rtality risk is one o f non market goods , this study emp loys the contingent valuat ion method (hereafter CVM) fo r the benefit estimat ion (Note 2). In this study, the concept of option p rice (hereafter, OP), wh ich is presented by Weisbrod (1964), was employed as a welfare measure in the benefit analysis. Jones-Lee (1976) and Graham (1981)

Survey
The research was conducted through an Internet research company, Net Mile, Inc., fro m January to were not only recreat ional activ it ies.
An e-mail was sent to about 10,000 respondents, and 3,401 respondents answered screening questions, 1) whether they had lived in Miyagi prefecture now, 2) whether they had visited at least one of 25 beaches in the past year. 914 respondents answered "yes", and then they were invited to co mplete online questionnaires. As a result, 763 respondents answered the questionnaires.

Designing Mortality Risk
In this study, it was difficu lt to know the actual mortality risk caused from recreational activit ies at beach because of the lack of official data. Thus, the mortality risk was designed as 8 persons, which are the number of deaths in 2008 described above, per 2,340,049 persons, which is the populatio n of Miyagi prefecture. Here, the mortality risk should be adjusted to be consisted with the total number of users because the possibility involved in the risk g ive for only people who used beaches. Thus, the denominator was adjusted as 380,000, which is the total nu mber o f visitors at 25 beaches in 2009 published by Miyagi Prefectural Govern ment (2010). As a result, the mortality risk fro m water accidents designed as 1.3 persons per 380,000 persons.
Since the reasons of eight persons' death were not on ly recreat ional act ivit ies, thus the mortality risk,

. Risk Reducti on Rates Described in Contingent Scenario
This study used two figures following the manner. Figure 3 shows the relative magnitudes of mortality risks of cancer, traffic accidents, fire, and water accidents. The mortality risks (without the one of water accidents) were calculated fro m the statistics of Miyagi prefectural govern mental office. Figure 4 shows the magnitudes of the effects of risk reductions, such as the present mortality risk, and reduced mortality risks by 10%, 50%, and 90% (Note 5). The levels of mo rtality risks were showed by bars.

Contingent Scenario
Since it needs an additional found to imp lement the additional protection countermeasures, then let assume the situation that the tax used for the protection countermeasures is collected. The tax is collected fro m residents lived in M iyagi prefecture in o rder to maintenance or improve the protection countermeasures. The usages of collected tax are as follows: 1) The tax is used only for the protection countermeasures at beaches The period of the pro ject is 10 years fro m now. The ta x is used for the increase of addit ional manpower for the protection such as life savers, monitored facilit ies, breakwaters, and the increase of emergency med ical service.
2) The number of collecting the tax is once. The effect of the protection counterme asures continue during next 10 years.
3) All o f founds are only used for the protection countermeasures. The detail of the accounting is published".
The duration of the effect of the pro ject was set as 10 years not to change individuals' present situations (i.e., frequency of visitation) in respondents' images. This point is confirmed by including data on respondents' perspectives for visitations. The number o f payments was once, and the format was the tax.
The true objects of collect ing the tax were users at beaches from the view point of the benefit principle.
However, all beaches are opened for citizens (Open access), thus, an admission fee for the usage of beaches was considered as unrealistic method. Therefore, the tax for cit izen lived in Miyagi prefe cture was assumed. The effect of countermeasure is 90%

Answer Format
The effect of countermeasure is 50% The effect of countermeasure is 10% 10,000 yen ✔ 7,000 yen 300 yen ✔ 100 yen 0 yen ✔ it is difficult to construct programs of the fo rmats due to the systems of the research co mpany. Thus, this research used the payment card format (Note 6). Moreover, since three reduction rates were assumed in this study (Figure 4), then the two answer formats were made as matrix formats in Table 1 (Format A) and Table 2 (Fo rmat B). The reduction rates are showed in the first row and the WTPs (0, 100, 300, 500, 1,000, 3,000, 5,000, 7,000, and 10,000) are showed in the first colu mn. All respondents answered both questionnaires by checking (✔) the elements of matrixes; the first is Format A, then Format B. The two formats were used for checking the differences of WTPs caused from the answer formats.
In questionnaire, four categories of 0 yen were shown in Table 1 and Table 2; A) 0 yen: There is no mean ing of the effect of the project, B) 0 yen: There is no money, C) 0 yen: No possibility to meet the accident, D) 0 yen: Dislike the tax. If a respondent select A or B, the WTP were classified as 0 yen. If C or D, the WTP were classified as protest bids because the respondent objected to the payments vehicle or the mortality risk in the question. In this study, the two estimations were perfo rmed. The one is the estimation using data with protest bids and the other is without protest bids. Checkin g the differences among WTPs in each formats (internal scope test) using data with the protest bits performed in Appendix A.

Explanatory Variables
There are three categories of explanatory variables in this study, individuals ' characteristics, the experiences of injure in beaches, and issues on usage of beaches. The questionnaires are shown in Appendix B.
The first is individuals' characteristics. Data of individual i 's gender ( GND ) and age ( AGE ), wh ich were collected by the company as respondents' information (The data have updated every year), were used. The GND was a du mmy variable, 1 for male, 0 for female. The AGE was the indiv idual i 's years. Individual i 's annual household income ( M ) and educational level researched by the questionnaire of this study were used. The educational level has four categories; vocat ional school ( EDVS ), junior co llege ( EDJC ), university ( EDU ), and graduate school ( EDGS ). Each variable were emp loyed as dummy variab les. Next, irregular employ ment ( JBIE ) and a homemaker ( JBHM ) were used as the individual i 's employ ment status. The JBIE and JBHM were emp loyed as dummy variables.
The second category is the experiences of injures in beaches. A hypothetical assumption on the influence of the variable for WTP was that the WTP becomes high if the individual i had experienced to injure at a beach in a recreational activity. Here, there are some magnitudes of injures. Thus, the magnitudes of injury were categorized in the questionnaire as follows (The exp lanations were refereed by Japanese dictionary).
The minor injury : an injured person is not admitted to hospital.
The severe injury: an inju red person must be admitted to hospital more than one month.
The serious medical condition: an injured person is in mo rtal danger.
Moreover, the experiences were categorised as individual i 's experiences ( ). Basic statistics of variables are shown in Table 3.  x x x  . Figure 5 shows the response rates for WTPs, Table 4 and Table 5 shows the basic statistics after the reconstruction.

Esti mation Results
Estimations were performed using R version 2.11. The results of including protest bits are shown in Table 6, and without protest bits are shown in Table 7. Confidence intervals of parameters were calculated following Venables and Rip ley (2002), and the results are shown in Appendix C. Colu mns of Table 6 and Table 7 are categorized by 1) format A and B, 2) reduction rates (10%, 50%, and 90%). The  Table 5 were difference due to the differences of protest bits. Note A. Super script, a, in tables means p-value of a estimated parameter is less than 1%, the b means less than 5%, and the c means 10%, respectively.  is considered that the injured experiences are influenced to WTPs in the cases of low levels of reduction rates.
Next, signs and p-values of use status are discussed. First, the signs of FUSEs are positive in all categories, and the p-values are less than 10% without the cases of 10% reduction rates in all categories.
Thus, it is considered that the FUSE influences WTP in the high levels of reduction rates. In the contingent scenario, the duration of the effect of the project was set as 10 years, and assumed the numbers of visits are same with their present situations. However, it is considered that individuals did not image such future situations when they answered.
The signs of purposes to visit beaches (SW, SF, etc.) were estimated both positive and negative values.
However, the signs are positive when the p-values are less than 10%.
Finally, calcu lations of WTPs are shown in Table 8. The values in b rackets are the WTPs calculated by lower and upper bounds of parameters in Appendix C. As results, median WTPs at 10% reduction rates ranges from 6.30 yen/once to 221.36 yen/once, mean WTPs ranges from 615.25 yen/once to 820.59 yen/once. Median WTPs at 50% reduction rates ranges from 743.28 yen/once to 1287.12 yen/once, mean WTPs ranges fro m 1256.11 yen/once to 1695.90 yen/once. Median WTPs at 90% reduction ra tes ranges from 1607.40 yen/once to 2924.63 yen/once, mean WTPs ranges fro m 2411.95 yen/once to 3433.40 yen/once.

Conclusions
Water recreat ions, such as swimming and fishing, are very popular activ it ies in the wo rld. Since there are many beaches in mos t area of Japan, water recreat ions at beaches are familiar to the nations.
However, if a person is in a water accident, the person would dead due to the high rate of mortality ris k of water accident. A lthough there are many studies on benefit analyses of t he OP, litt le studies have not focused on the benefits of mo rtality risk reductions on recreat ional activ it ies. The purpose of th is paper is to perform the estimat ion of willingness to pay (hereafter WTP) for the mo rtality risk reduction researched by the contingent valuation method. three reduction rates (10%, 50%, and 90%) were assumed in th is study, then the two answer formats (Format A and Format B) were presented. the presented WTPs were 0 yen, 100 yen, 300 yen, 500 yen, 1,000 yen, 3,000 yen, 5,000 yen, 7,000 yen, and 10,000 yen, and were shown with each reduction rate.
The number of payment is at once, and the payment vehicle was tax.
Our findings are as follows. The median WTPs and the mean WTPs for each reduction rate with and without protest bits were calculated. As results, the median WTP at 10% reduction rates ranges from 6.30 yen/once to 221.36 yen/once, the mean WTP ranges fro m 615.25 yen/once to 820.59 yen/once.
The median WTP at 50% reduction rates ranges from 743.28 yen/once to 1287.12 yen/once, the mean WTP ranges from 1256.11 yen/once to 1695.90 yen/once. The med ian WTP at 90% reduction rates ranges fro m 1607.40 yen/once to 2924.63 yen/once, and the mean WTP ranges fro m 2411.95 yen/once to 3433.40 yen/once.
As for the exp lanatory variables, estimat ion results indicated that 1) the self injured experiences influence WTPs more than the other persons' injured e xperiences, 2) the inju red experiences are influenced to WTPs in the cases of low levels of reduction rates, 3) individuals' perspective for future use of beaches influences their WTPs in the high levels of reduction rates.

Appendi x
Appendix A. Scope Tests Internal scope tests were performed by testing the differences of mean values of WTPs (with protest bits). The Tukey's mult iple co mparison were employed. H0: no differences in mean values, H1: other.
The results of the statistic are shown in Table 9, p -values are in the parenthesis. Let  be a rejection  Q. How long do you use beaches in future? Please answer your perspective.