Abstract
The environmental performance of a listed firm could affect its level of investment in pollution prevention and its access to financial markets. Previous studies using Tobin’s q that explore market response to environmental performance do not distinguish between the impact of performance on investment and market response, which may mislead conclusions. To overcome this problem, we simultaneously estimate the functions of the intangible asset, the replacement cost, and the toxic chemical risk. We find that the Japanese financial market does not value risk associated with toxic chemical releases. Nevertheless, even without market valuation, firms increase investment to reduce pollution. (JEL D21, Q58)
I. Introduction
Better provision and dissemination of environmental information has been recently proposed as a complement to traditional policy instruments for controlling environmental performance (Tietenberg and Wheeler 2001). Information provision comprises a quasi-regulatory mechanism: consumers, investors, the public, and other stakeholders use the information to pressure firms to change their environmental behavior (Arora and Cason 1996; Khanna and Anton 2002; Anton, Deltas, and Khanna 2004; Lyon and Maxwell 2004). For example, if consumers care about a firm’s environmental performance, the provision of more firm-specific environmental information may cause them to change their decisions on purchasing the firm’s product. If investors expect firms with better environmental performance to have lower liability costs because they are less likely to be liable for future environmental problems and thus to have higher future profitability, then publicly held firms with better performance will have a higher stock price, that is, market value (Fama 1970). If the market values environmental performance, firms have an incentive to improve their environmental performance to obtain a higher market valuation. This paper examines the efficacy of information provision policy by testing whether the Japanese market values a firm’s environmental performance as measured by its toxic releases.
Previous literature on the Pollution Release and Transfer Register (PRTR) system1 mainly analyzes the effects of the U.S. Toxic Release Inventory (TRI) on the relationship between a firm’s environmental and financial performance. Using event study methodology, studies have found that stock markets respond to the disclosure of toxic release information and published environmental news (Hamilton 1995; Konar and Cohen 1997; Khanna, Quimio, and Bojilova 1998; Klassen and McLaughlin 1996). More recently, Konar and Cohen (2001) claim that event studies cannot analyze longer-term trends or objective measures of a firm’s environmental performance that are not tied to a particular date. Instead, they employed Tobin’s q, defined as the market value of the firm divided by the replacement cost, to show that poor environmental performance had a negative effect on the value of the intangible assets of the firm in 1989.
However, it should be noted that the response of Tobin’s q to poor environmental performance does not necessarily indicate a market response, when environmental performance affects the replacement cost. In order to avoid the future environmental liability risk, firms that release and transfer more chemical substances are likely to have an incentive to have more capital stock,2 that is, greater replacement cost, to reduce the risk of pollution generated from these releases and transfers, even if the market does not respond to the risk. Although firms might potentially reduce the releases and transfers by employing an environmental management practice that is less capital intensive, they are also likely to make changes in the production process that require capital investment (see Ministry of Economy, Trade, and Industry 2002). If this were the case, even without the market valuation of the environmental performance of the firms, we would observe a lower value of Tobin’s q for the firms with more releases and transfers because of larger replacement cost: the market would “seem to” respond to the provision of toxic information. Previous studies on Tobin’s q have not considered the impact of the emission of toxic substances on the replacement cost. Thus, their conclusions might overestimate the effect of information on the firms’ market value.
To overcome this problem in Tobin’s q, we simultaneously estimate the functions of the intangible asset, the replacement cost, and the toxic chemical risk. Then we explore not only whether the Japanese financial market values the firm’s level of information on its releases and transfers of chemical substances disclosed under the PRTR system, but also whether the firm with more releases and transfers increases the capital stock, in other words, the replacement cost to respond to the market valuation and/or to reduce the future environmental liability risk.
II. Model
Japan has enforced the PRTR system since 2001; the first public release of PRTR system data was on 20 March 2003. Under this system, facilities that have more than 20 employees and produce or use chemicals on a list of 354 substances specified by the law must report annually to the central government their quantities released and transferred. The central government aggregates and sorts the reported data by industry type and geographic location and provides the information to the public. Although the central government does not specify facility-level emissions in the aggregated report, it must disclose the facility-level data when requested by a citizen. The PRTR system in Japan has an important role in reducing and managing the development of toxic chemicals.
Investors may expect that increases in the release and transfer of these chemicals will decrease future profits because of the higher expected cost of the environmental liabilities. They may also be interested in reducing the amount of chemicals released and transferred per unit of output by the firm from the perspective of social responsibility. If this is the case and the PRTR information is disclosed, firms with lower releases and transfers will have higher current stock prices or current market values (Fama 1970).3
We consider two possible roles for the PRTR system in shaping a firm’s incentives. One is the information disclosure, which is useful to the investors. If the risk information is disclosed, the investors use it to value the firm. If the investors negatively value the chemical risk, the stock price of the firm with more risk is valued relatively lower. Therefore, the information disclosure gives the firm an incentive to reduce the risk so as to obtain a higher market valuation. We define this as the indirect effect.
The other role is the information used by the firm to conduct environmental management to avoid future potential liabilities. Even if the information is not disclosed, the risk information collected under the PRTR system may help the firms to manage their environmental performance. If the firms know that their level of release and transfer is high, they will recognize that they face a potential liability. Therefore, to reduce such liability, the firms with more release and transfer have an incentive4 to reduce their chemical release and transfer levels even without the valuation of their investors. We define this as the direct effect.
In this study, we explore (1) the effect of providing information about the release and transfer of the chemical substances on the intangible assets of the firm (whether or not the market values it) and (2) the effects of releases and transfers and the market valuation on the replacement cost (whether or not the direct and/or indirect effects exist).
The market value of a firm is defined as the sum of the values of its long-term and short-term debt and its stock value. The market value can be disaggregated into tangible and intangible assets (Konar and Cohen 2001). Tangible assets consist of the replacement value of property, plant and equipment, cash, and inventory, and are measured as the replacement cost. Intangible assets are defined as market value less tangible asset value. They are based broadly on a firm’s “reputation” and the “valued added” realized from patenting, trademarks licensing, brand naming, expectation of higher future profitability, and positive public image. If the market values the environmental performance of firms, environmental information can affect the value of intangible assets. Previous studies have used the Tobin’s q function to test whether better environmental performance has an effect on the value of intangible assets. However, Tobin’s q might be problematic for the following reasons: Firms with more releases and transfers may have more capital stock (note that simple correlation between the releases and capital stock in Japan is 0.234, while it is 0.045 for the United States, using the firm-level data from the TRI of the same years [see Konar and Cohen 2001 for U.S. data sources]), and consequently higher replacement cost, to reduce future potential risk and liability. As argued above, the existence of a future liability risk for firms may lower Tobin’s q due to the higher replacement costs, regardless of whether the market values firm-level toxic releases and transfers. Thus, an analysis of Tobin’s q might result in a misleading conclusion, since it does not separate the impact of the risk of higher emissions on intangible assets from its impact on replacement cost.
In our study, considering the endogeneity of chemical risk, we estimate simultaneous equations for intangible assets, replacement cost, and the carcinogenic risk,5 rather than relying on a single equation for Tobin’s q. We formulate the system of simultaneous equations as follows:
[1]
[2]
[3]where MVi, RCi, IVi, RISK1i, RISK2i, and ENVi denote the market value, replacement cost, intangible assets, carcinogenic risk, simple sum of toxic releases and transfers (as the proxy for the other chemical risks), and vectors of measures of the environmental friendliness of the firm, such as the adoption of an environmental management system, respectively. X1i, X2i, and X3i represent vectors of firm attributes such as the firm size, financial performance, and profitability, which serve as the explanatory and control variables. Although the results using lagged variables6 are not reported in this paper, our results are robust to these alternative specifications. ε1i, ε2i, and ε3i are random error terms following a normal distribution. The detailed definitions of the variables are listed in Table 1.
Variables and their Definitions
Compared with previous studies using Tobin’s q, the above model has the following advantages:
We can distinguish the impact of information on the intangible assets from its impact on the replacement cost.
The firm is expected to increase the replacement cost to reduce the releases and transfers of the chemical substances not only to decrease its future environmental liability (a direct effect), but also to gain a positive market valuation when the market values the firm’s risk (an indirect effect). In our model, the third and fourth terms in equation [2] are risk factors that capture the direct impact, and the fifth term represents the market valuation and captures the indirect effect. Thus, we can distinguish the direct effect of the information on the replacement cost from its indirect effect, which is the impact through market valuation.
Carcinogenic risk is modeled as an endogenous variable, contrary to previous studies where it has been modeled as an exogenous variable. We believe it is endogenous because the carcinogenic risk is affected by other factors, such as the replacement cost.
III. Data
We focus on manufacturing firms that appeared in the first section of the Tokyo Stock Exchange in 2003 and 2004, and whose facilities reported PRTR information to the government. The market value is defined as the sum of the values of long- and short-term debt and the value of the stock, which is calculated using the account-end stock market price multiplied by the number of shares outstanding. The value of the stock is calculated from data obtained on March 31 (end of Japanese fiscal year) every year. The market data is obtained from the Nikkei Needs database provided by Nikkei Quick Information Technology Co. Ltd. The replacement cost of the firm is calculated as the sum of liquid assets, property, plant and equipment, and investment, and it uses accounting-based values to calculate the assets of the firm. Table 2 provides descriptive statistics of our sample.
Descriptive Statistics
This study tests two types of risk indexes using the information on toxic releases and transfers of the firms: carcinogenic risk, which is calculated using risk factor methodologies, and the mass of toxic releases and transfers. The literature analyzes the effects of toxic releases and transfers on the markets without considering the actual risk or weight of toxic data. However, firms might be willing to pay more to control highly toxic chemicals, and investors might be more concerned about highly toxic chemicals. Therefore, it is important to include both the toxicity and the mass variable.
Toxic chemical data are available from the Japanese Ministry of the Environment; we use the sum of toxic releases and transfers generated by the firm. The mass of toxic releases and transfers follows the literature (Konar and Cohen 2001) and is a proxy for the general potential toxic risk to stakeholders. The carcinogenic risk data are constructed from several sources and are used to represent the actual risk that a firm needs to take to comply with the environmental regulations.7 The Organization for Research and Communication on Environmental Risks of Chemicals summarized the literature on risk analysis related to chemical substances in the PRTR system. We utilized their methodology and constructed the index by multiplying the risk factor by the mass of each chemical substance.8
Information about the carcinogenic risks posed to humans by a variety of agents, mixtures, and exposures, and about proven and possible human carcinogens, is needed to assess the hazards posed by exposure to chemical, physical, and biological factors. Risk information is collected from the following seven sources: the Ministry of Labor in Japan (1999), the American Conference of Governmental Industrial Hygienists (2003), the International Agency for Research on Cancer (2003), the European Union (2002), the U.S. Environmental Protection Agency (2003), the U.S. National Toxicology Program (2003), and the Japan Society for Occupational Health (2003). The correlation between carcinogenic releases and aggregate releases is 0.0015. In general, the industry reduces the risk of carcinogenic emissions by end-of-pipe equipment when the releases are primarily into air.
As for the other explanatory variables, we use the current profit-sales ratio (PSR), debt ratio (DR), capital growth (CG), sales growth (SG), capital turnover ratio (CTR), number of employees (N), ISO 14001 dummy (DISO), and environmental management ranking of firms (NIKKEIIDX)9 as measures of current profitability of a firm, financial soundness, potential productivity growth, expected growth, productivity of the capital, firm size, lagged status of adoption of the ISO 14001,10 and degree of the environmental management by the firm, respectively. The market value may depend on how much of the firm’s products is exported, as more export oriented firms may be less affected by the economic recession in Japan. Therefore, we use the export sales–gross sales ratio (ER) to control for this additional level of profitability.
In addition, we use advertising expenditure (AD) to indicate the degree of product differentiation and brand value, following previous studies, and R&D (R&D) expenditure as a measure of technological capacity. Ideally, AD and R&D should be stock measures. However, because of data limitations, we measure the AD and R&D variables by taking the arithmetic means of the AD and R&D expenditures over three consecutive years. For example, the AD variable for the year 2003 is constructed from an average of years 2003, 2002, and 2001.
The industry dummies are used as proxy variables representing industrywide effects.11 Comanor and Wilson (1967) showed that firms in more-concentrated industries have higher market valuations. Because of the lack of consistent concentration data, we use the industry dummies as proxies for market conditions. Industry dummies also represent other industry-wide effects. For example, these dummies control for the possibility that a larger replacement cost in a heavily polluting (riskier) industry might result from higher investment in pollution-control equipment. On March 31, 2003, out of 673 manufacturing firms in the first section of the Tokyo Stock Exchange, 402 firms had facilities that reported PRTR information to the government. Hence, the total sample size for the analysis is 804 (for 2003 and 2004).12
IV. Estimation Results
When faced with heteroskedasticity of unknown form, the usual approach is to use the generalized method of moments (GMM), introduced by Hansen (1982). The GMM estimator is both consistent and efficient if the random error is nonpersistent. Hence, GMM is the method used to estimate the intangible asset value function, the replacement cost function, and the risk function (equations [1], [2], and [3]). Table 3 presents the estimated results. Using J statistics, we are not able to reject the hypothesis that all instruments satisfy orthogonality conditions (i.e., correct specifications). Therefore, the estimated models are adequate.
Estimation Results
In the replacement cost equation, the parameter on carcinogenicity (RISK1) is statistically significant and positive, whereas the parameter on chemical pollution (RISK2) is insignificant. This indicates that the firm pays attention to carcinogenicity, rather than total volume of chemical substances, in assessing its risk, and that a firm with larger releases and transfers of chemical substances containing carcinogens is likely to increase its capital in order to decrease its potential risk.
The parameter for intangible assets (IV) is also significant with a positive sign. This means that the firm with higher intangible assets has more capital stock. This mechanism is supported by Tobin’s q theory, since the value of intangible assets is positively correlated with Tobin’s q.
The adoption of ISO 14001 is not significant, whereas the Nikkei index is significant and negative. This indicates that firms with higher rankings, which are more likely to implement more environmental management practices, have greater replacement costs to improve their environmental performance, but that ISO 14001 does not. However, we need to note that ideally the adoption of ISO 14001 needs to be considered as an endogenous variable, since voluntarily adhering to specified environmental management standards is expected to create incentives for sustained environmental improvements. Thus, a lagged value of ISO 14001 is employed.
The number of employees and its square term are positive and statistically significant. This indicates that the replacement cost is positively associated with firm size. The squared term of the current profit-sales ratio (PSR) is statistically significant and positive, whereas the single PSR term is insignificant. This indicates that higher profitability stimulates more investment. The debt ratio (DR) and the capital turnover ratio (CTR) are significant with positive and negative signs, respectively; higher debt ratios and lower turnover ratios are associated with increased replacement costs. These results might be due to the endogeneity problem, namely, that firms with greater replacement costs will also incur greater debt to finance the costs and a lower capital turnover ratio.
As for the intangible asset equation, neither the parameter on carcinogenicity (RISK1) nor the parameter on chemical pollution (RISK2) are statistically significant, which indicates that the market does not value environmental risks as represented by the firm’s toxicity and quantity of PRTR emissions. Perhaps the carcinogenic risk is too complex for the market to calculate and comprehend. Alternatively, perhaps investors do not consider toxic volumes to be a valid measure of risk in Japan during the years immediately following the introduction of the PRTR system. From these estimation results and the evidence that the parameter for intangible assets is significant with a positive sign in equation [1], we could conclude that since the market does not value the environmental risk, the firm does not increase the replacement cost and thus the investment to reduce the risk to obtain a higher market valuation.
For comparison, we estimate Tobin’s q by the method described in the literature. As is shown in the Appendix, we find that higher risks of carcinogenicity decrease Tobin’s q. Combining the above results, we find that poor environmental performance, such as higher carcinogenic risk, increases the replacement cost but does not affect intangible assets. Hence, although poor performance clearly decreases Tobin’s q, it is not necessarily valued by the market. From this argument, results from Tobin’s q analysis, as conducted by Konar and Cohen (2001), might be misleading.
The environmental management ranking (NIKKEIIDX) parameter is significant and negative, whereas the ISO 14001 parameter is not significant. This indicates that the market values firms with higher environmental rankings. The R&D parameter is not significant, whereas the advertising parameter is positively significant. The finding shows that the market value of firms with some level of product differentiation and brand value tends to be positively associated with an increase in intangible assets, whereas the market does not value larger technological capacity.
Among the other variables, the single and the squared PSR terms are significant and of positive sign. This indicates that the higher profitability of the firm positively affects the intangible assets of the firm. Sales growth, debt ratio, and export ratio are not statistically significant, which indicates that they do not affect the intangible assets. However, capital growth is significant with a positive sign, which indicates that the market values firms with higher potential productivity growth.
Finally, in the carcinogenic risk equation, the replacement cost is significant and has a positive sign, whereas the ISO dummy is not significant. This indicates that, although the adoption of ISO 14001 does not affect carcinogenic risk, replacement cost is positively associated with risk. It should be noted that there are two types of capital stock. One is for increase in production, while the other is for pollution reduction. The available data do not distinguish the capital used to increase production from that used to reduce pollution. Therefore, the positive sign of the replacement cost parameter is likely to capture such net effects, that is, the effect of production increase on carcinogenic risk dominates the effect of pollution reduction.
V. Conclusions
The provision of information has been paid attention to in recent years as a new policy tool. In order to explore how information gives firms an incentive to improve their environmental performance, we test (1) whether financial markets value firm-level PRTR disclosure in Japan, (2) whether market valuation (the intangible assets) affects replacement cost (indirect effect), and (3) whether firms with poor environmental performance increase their replacement costs to reduce chemical releases and transfers (direct effect), and by consequence their future environmental liabilities.
In this study, we find that the Japanese financial market does not value the risk posed by toxic chemical releases and transfers, but that the intangible asset value is positively associated with replacement cost. The results suggest that if the market valued environmental performance, firms would have an incentive to increase replacement costs to improve environmental performance. Our findings on market valuation conflict with previous studies, which show that the U.S. financial markets do value the risks posed by release and transfer. We posit that the reason for the difference is related to the information disclosure protocol in Japan. The information from the Japanese PRTR system is not easy for investors to use in evaluating firms for the following reasons:
Data for firms are not available, but data for facilities are. Therefore, in order to identify the release and transfer of individual chemical substances by each firm, the data for each facility must be aggregated to the firm level. As a result, public access to firm-level data is difficult.
It is very difficult for investors to understand the degree of the risk for each firm. There are 354 types of chemical substances, and most investors do not have expert knowledge of the risks. The information disclosed is not necessarily useful for investors. Furthermore, environmental and public interest groups and media are unsuccessful in influencing public reactions to the information.13 Our findings support this conclusion; firms with positive releases and transfers are more aware of these risks than ordinary citizens and investors. Firms have incentives to reduce risk even without market valuation.
Our final finding, that, even without market valuation, firms with higher carcinogenic risk realize higher replacement costs, suggests that firms reduce their risk in order to decrease future potential environmental liabilities risk. However, we should also note that firms appear not to be concerned with the sum of the release and transfer quantities.
We believe that the financial market is able to value the environmental performance of firms, as previous studies have shown (e.g., Hamilton 1995; Khanna, Quimio, and Bojilova 1998; Konar and Cohen 1997, 2001), if suitable and easily understandable information is available and if information is readily accessible. Therefore, in order to create a more efficient market signal, it is important to consider the appropriate form and content of the information disclosure protocol so that the financial markets can effectively provide incentives to firms to reduce releases and transfers of toxic substances.
Appendix
Tobin’s q is defined as
Following previous studies, such as that by Konar and Cohen (2001), we estimate
[A1]where qi is the Tobin’s q of the firm i, ε4i a random error term drawn from a normal distribution, and X4i the vector of the attributes of the firm i.
The estimation results are shown in Table A1. The result shows that the parameter on the carcinogenic risk (RISK1) is significant at 10% level with negative sign, which indicates that higher risk decreases Tobin’s q.
Estimation Results of Tobin’s Q
Footnotes
The authors are, respectively, head, Environmental Economics and Policy Section, Social and Environmental Systems Division, National Institute for Environmental Studies, Japan; and associate professor, Graduate School of Environmental Studies, Tohoku University, Japan. The authors would like to thank two anonymous referees, Geoffrey Heal, Eric Welch, Kikuo Iwata, Yoshitami Hibiki, and participants at the 2006 AEA/AERE annual meeting in Boston for helpful comments. The authors are responsible for any remaining errors.
↵1 Important characteristics of a PRTR system are (1) facilities periodically send a mandatory report to the competent authorities on their releases to air, water, soil, and wastes, and (2) emission data of specific pollutants from individual facilities are accessible to the public.
↵2 Yamaha Corporation (producer of musical instruments and audio and video products), for example, reduced its emissions with respect to the PRTR by installing and applying regenerative exhaust-gas treatment equipment for the painting process (e.g., Yamaha Corporation 2005). The equipment renders combustible hazardous materials (such as volatile organic compounds and odor substances) harmless.
↵3 It should be noted that there are two types of market valuation on chemical risk. One is the market response to the gap between the expected risk and the realized risk. If the chemical risk turns out to be less (more) than the market expectation, the stock price will go up (down). Event studies are able to capture this type of valuation. The other type is the response to realized risk after the expectation is revised based on new information. We focus on the latter type of market valuation. This is because, as stated by Konar and Cohen (2001), event studies cannot analyze longer-term trends or objective measures of a firm’s environmental performance that are not tied to a particular date.
↵4 Lately, courts have tended to rule that the firms are liable for past emissions, when it is subsequently discovered that they are the cause of environmental damage. For example, in the Chisso Minamata disease lawsuits, the courts ruled in favor of the plaintiff. Since Chisso Co. Ltd. made an agreement on compensation with patients in 1973, it has paid 1,390 billion yen ($14.6 billion) for 2,268 certified patients (Chisso (www.chisso.co.jp/topics/minamata/index.html [in Japanese]). Additional detail about Minamata disease is available fromthe Japanese Ministry of the Environment (www.env.go.jp/en/chemi/hs/minamata2002/). These cases affect the firm’s expectation of potential liability.
↵5 As is explained in a later section, for explanatory variables of the intangible asset function and the replacement cost function, we used two types of risk indexes. These are the carcinogenic risk and simple sum of toxic releases and transfers, which are calculated using information on toxic releases and transfers of the firms. However, the simple sum of toxic releases and transfers is not significant. Hence, we exclude the risk function of the simple sum of toxic releases and transfers in our study.
↵6 The lagged variables are the firm’s attributes, such as the firm size, financial performance, and profitability, except for replacement cost and carcinogenic risk.
↵7 Bui and Mayer (2003) also constructed carcinogenic risk data in their housing market analysis. In contrast to their study of taking the sum of emissions related to carcinogenic risk, we consider the weight of the risk.
↵8 The list and weight of 178 of the 354 chemical substances specified by the PRTR law relating to the carcinogenic risk are available on request.
↵9 Nihon Keizai Shimbun Inc., a large economic newspaper company, conducts the Nikkei Environmental Management Survey every year on the listed firms in the first section of the Tokyo Stock Exchange. Using the core data based on the qualitative response obtained by the survey, Nihon Keizai Shimbun ranks the firms. Several different criteria are used, such as whether the firm has longer-term objectives for environmental management, climate change, and product management related to pollution. The rankings reflect the differences in the degree of environmental management conducted by each firm. It should be noted that there is no linkage between the index and quantitative information of emissions, and also no direct linkage to the ISO14001 adoption dummy.
↵10 ISO 14001 is an international standard for environmental management systems published by the International Organization for Standardization (ISO). Recently, the number of facilities and firms that have adopted ISO 14001 has been increasing in Japan. The adoption of an environmental management system is often expected to help firms identify their environmental goals and reduce their energy consumption, polluting waste, and the risk of future environmental liability. The decision to adopt the ISO 14001 is made primarily at the facility level. Following Nakamura, Takahashi, and Vertinsky (2001) and because of data limitations, in this model we define a firm that has at least one certified facility among all the facilities as “a certified firm.” If some facilities obtained the ISO 14001 certificate, the DISO of the firm is set to one. We obtained the information on the ISO 14001 certification status of firms from the Japanese Standards Association (2003).
↵11 We have 15 categories of industries.
↵12 On March 31, 2003, there were 1,511 firms listed in the first section of the Tokyo Stock Exchange.
↵13 For example, over a year after the release of PRTR data, only about 70 articles from major magazines and daily newspapers in Japan reported the emissions of chemicals in PRTR data, where only two firms’ emissions were mentioned. In contrast, there was much more media coverage that focused on firms with large emissions of chemicals in the U.S. TRI data (Hamilton 1995).
This article requires a subscription to view the full text. If you have a subscription you may use the login form below to view the article. Access to this article can also be purchased.
Log in using your username and password
Purchase access
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.