Economic Valuation of Forest Ecosystem Service in Japan

This study estimates both unit and attribute values of forest ecosystem services in Japan. In this estimation, we examined the applicable valuation techniques from the viewpoint of environmental economics, in particular regarding the economic valuation method of ecosystem services. Our results revealed that direct use, indirect use, and non-use values are important in evaluating ecosystem services. In addition, it is important to incorporate indirect use and non-use values into the ecosystem evaluation framework. In order to reflect such values, we used the payment card Contingent Valuation Method (CVM) to estimate the unit value of forest in Japan. In addition, to estimate the attribute value of each forest ecosystem service, we conducted choice experiments that share the same theoretical foundation of a random utility model with CVM. Our findings visualized the ecosystem service value of forests in Japan.

Key words: Ecosystem service, economic valuation, contingent valuation, forest resources, unit value, attribute value.

Global Analysis of Ecosystem Services from Agroforestry and Natural Capital

We enjoy tremendous value of ecosystem services from agricultural land and forest. However, it remains challenging to evaluate ecosystem services relevant to human well-being. This study illustrates global trends of the changes in values of ecosystem services from natural capital, based on the dataset in a recent United Nations report (Inclusive Wealth Report 2018), which includes 140 countries' data of ecosystem services from renewable natural capital. Among other results, we show that the value of agricultural land increased in South American countries with high levels of agricultural production, and decreased for some developed countries, e.g., in Europe, although in general natural capital has been decreasing. We also argue that in future work, other aspects of ecosystem services from natural capital should be evaluated, e.g., ecosystem disservices from agricultural land and urban ecosystem services.

Key words: Agricultural land capital, forest capital, wealth, capital valuation.

Effectiveness of Invasive Species Eradication Efforts on Biodiversity Conservation: Spatial Congruence between Conservation Priority Areas and Threat

To implement biodiversity conservation under limited resource, we should understand the effectiveness of action plans. In this study, we analyzed spatial congruence between distribution of invasive mongoose (Herpestes auropunctatus) as a major threat against biodiversity and conservation priority areas for native terrestrial vertebrate species (254 species) in the northern part of Okinawa island, Japan. Using potential species distribution maps (29, 155, 47, 23 species for mammals, birds, reptiles and amphibians, respectively) estimated at the 1km grid-cell level, we created spatial conservation priority rankings and detected top 17% priority areas for each taxonomic group within Okinawa prefecture. In the spatial prioritization, we first assumed the perfectly protected state (i.e. 100% of land is protected), and then removed least important cell sequentially following a removal rule where cells include geographically rare species (narrow distribution range) receives higher conservation values. To evaluate the dynamics of mongoose population in space and time, we used catching data obtained by gage traps in the eradication program implemented by Okinawa prefecture and the Ministry of the Environment during 2000 to 2009. Using the catching data (number of mongoose per trapping days in each trap), spatiotemporal pattern of mongoose occurrence was modelled using a hierarchical Bayesian model that estimates survival and immigration rate of mongoose and their environmental dependencies and spatial autocorrelation. The eradication program in the region successfully reduced the occurrence of mongoose within conservation priority areas, whereas the spatial congruence between the priority areas and mongoose occurrence was different between the taxa. Some of the priority areas, which mostly located on the northern part of the region, showed slight increase in the occurrence of mongoose threat, indicating that continuous eradication effort is needed to prevent the mongoose threat. Our findings demonstrate that we should use explicitly spatial biodiversity information and develop strategic conservation plans that account for synergy and conflict between conservation actions and benefits.

Key words: Hierarchical Bayesian model, mongoose (Herpestes auropunctatus), removal rules, spatial conservation prioritization.

Adjusting for Non-response Bias Using Registration Data from an Internet Research Company
—Application to Consumers' Evaluation of Japanese Grass-fed Beef—

Research on missing data is attracting growing attention as a means of improving translation of research into more reliable evidence-based decision making. In this study, we investigate the non-response bias in Contingent Valuation Method (CVM) using the Inverse Probability Weighting Estimator (IPWE). One of the major problems with application of the IPWE is related to the availability of covariates. We dealt with this problem by using individual information registered at an internet research company. Using the CVM format, we evaluated consumers' willingness to pay for a grass-fed beef produced by an environmentally friendly farming system. The results revealed that there were differences between respondents and non-respondents in individual income, age, characteristics of family members, use of SNS or video distribution websites, and hobbies. The sample mean of willingness to pay (WTP) for grass-fed beef was ¥1161.6 per 100g, and the IPWE of WTP was ¥1161.7 per 100g. In other words, there was almost no difference between the mean and IPWE of WTP. Based on this survey, we conclude that there are differences in covariates between respondents and non-respondents. However, this difference did not cause a non-response bias in estimation of consumers' WTP for grass-fed beef.

Key words: Non-response bias, contingent valuation, registered data, grass-fed beef, internet-based survey, evidence-based decision making.

The Relationship between Pollination Service by the Native Honeybee (Apis cerana) and Landscape Structure
—A Case Study of Hyuganatsu (Citrus tamurana) in Aya Town, Miyazaki Prefecture—

In this study, we examined the relationship between pollination of hyuganatsu (Citrus tamurana) (as an ecosystem service) by native honeybees (Apis cerana) and the landscape structure in Aya Town, Miyazaki Prefecture. A total of 24 hyuganatsu trees were selected in 16 orchards, and the number of honeybees visiting each tree was counted. A land use map of the study area was developed by photo interruption on the orthophoto. A stochastic model for predicting honeybee visits was developed, using the area of natural forest and the total area of agricultural field and grassland as explanatory variables. The difference in bee-finding ability among observers was also modelled as an observation model. We created an ecological model combining the relationship between the number of honeybee visits, regarded as an indicator of pollination service, landscape structure, and the observation model were combined, and estimated the parameters of these model by Bayesian inference. The estimated parameters suggested that both the area of natural forest and total area of agricultural field and grassland positively affected pollination services by native honeybees.

Key words: UNESCO Biosphere Reserve, GIS, Bayesian inference, natural forest restoration, organic agriculture.

Statistical Modeling via Functional Data Analysis

The development of technologies for measurement devices enables us to obtain high-throughput data. In agriculture in particular, more and more repeated measured data with respect to time, position, depth, etc. for individual have been obtained. However, sometimes it is difficult to analyze such data for the following three reasons. First, they usually contain observational errors, making it difficult to reveal unknown structures. Second, repeated measurements lead to the high-dimensional data, to which it is hard to apply traditional statistical analyses. Third, the observed numbers and time points may differ for individual. This leads to difficulties in applying the traditional multivariate analysis. Functional data analysis is one of the most useful tools for resolving the aforementioned problems. In this approach, longitudinal data are expressed as a smooth function for each individual, and then information is drawn from the collection of functions. In this paper, we introduce several techniques for functional data analysis that would be useful for analyzing agricultural data. In particular, we describe regression analysis, time series analysis, and spatial data analysis for functional data. For each method, we provide an empirical example in which the technique is used to analyze real data.

Key words: Functional data analysis, longitudinal data analysis, regression analysis, time series analysis, spatial data analysis.

Economic Valuation Methods for Ecosystem Services Provided by Agricultural Land and Forest

Agricultural land and forests not only supply agricultural crops and timber, but also play various roles in land conservation, cultivation of water sources, climate stabilization, prevention of global warming, conservation of biodiversity, and opportunities for recreation. To recognize the importance of these ecosystem services, it is effective to visualize their value by evaluating them on a monetary term. However, most ecosystem services are not traded on the market, and real prices do not exist; therefore, they cannot be evaluated based on price. Instead, specialized methods called environmental valuation methods are used to determine their value. In this paper, we explain the economic theory and estimation methods of the replacement cost method, hedonic price method, travel cost method, contingent valuation method (CVM), and conjoint analysis. These methods are applicable to the valuation of ecosystem services provided by agricultural land and forests.

Key words: Environmental valuation methods, replacement cost method, hedonic price method, travel cost method, contingent valuation method (CVM), conjoint analysis.

Effect of Firm Age in Credit Scoring Model for Loans to the Self-employed

A number of Japanese banks have utilized credit scoring models to reduce screening costs by automating part of the screening process. Several types of credit scoring models exist. To calculate the credit scores of small firms with twenty or fewer employees, it is common to utilize the logistic regression model, linked to the correlations between financial indicators and the occurrence of default.
Small firms include not only corporate businesses, but also self-employed businesses. The accuracy of the credit scoring model for the self-employed is not as high as might be expected. The main reason for this is that we lack adequate information regarding financial accounting variables on the balance sheet, such as assets or debts, because self-employed firms are not obligated to prepare a financial statement. We are convinced that it is effective to use explanatory variables associated with assets or debts. However, it is difficult for banks to obtain accurate data, because most self-employed households confuse the owner's private assets with business finances.
Ogi et al. (2016) suggest the use of ``firm age'' in the credit scoring model for small corporate business, and we have found this to be a statistically significant variable. There are at least three reasons for this. First, for loans to small firms, it is effective to evaluate the private assets of the owner together with the assets of the firm, but accurate information on this subject is not easily available. Second, the default rates of small firms decrease as the firms get older because the private assets of owners are expected to accumulate year by year, and this can help with management activities. Third, in the traditional screening process of loans by experienced bankers, firm age is an important factor for assessing the credit risk to debtors. In this paper, we propose to utilize ``firm age'' in credit scoring models for self-employed as well as small corporate businesses, and to analyze the correlation between firm age and default rate using a dataset for the more than 680,000 self-employed in Japan from 2007 to 2014. This dataset is owned by the Japan Finance Corporation. We will evaluate the credit scoring model using the accuracy ratio, which is commonly used as a measure.
We find that the default rate can be expressed by the density function of the Weibull distribution or a piecewise-linear function of firm ages with three ranges: up to 3 years, 4--25 years, and 26 or more years. We also show that the accuracy ratios increase by about nine percent at the maximum in both approaches, and that the model can be effectively used in practice.

Key words: Credit scoring model, credit risk, self-employed, firm age, logistic regression model, Weibull distribution.