In this paper, social network perspective is applied to the estimation of corporate credit risk. Most of the models used for the estimation of corporate credit risk are based on the individual firm's financial conditions and its business situations. However, data shows that Japanese firms, which is thought to be embedded in large interfirm networks, are sometimes underevaluated in terms of their corporate credit risks. This study is based on the assumption that the underevaluation may be due to the lack of understanding the social network effect on corporate credit risk. To understand the social network effect on corporate credit risk, the Japanese interfirm networks, sometimes called "keiretsu," were chosen for analyses. Then, based on the discussion, the corporate credit risk model that is modified by adding the social network effects are proposed. Finally, the empirical research design is proposed to examine actual effect of social network for Japanese firms.


Assessing corporate credit risk is becoming an increasingly important element of financial management, especially for financial institutions that have a large amount of lending and corporate bond portfolios. Highly technological advancement in the area of finance has enabled them to precisely estimate the corporate credit risk, that is, the probability of bankruptcy or bond default. However, viewing from sociological perspective, it appears that there is still possibility to develop more precise estimation of corporate credit risk. The view I propose in this paper is social network perspective or the concept of social capital (e.g., Burt, 1992). That is, I propose that we should consider the social network effect on corporate credit risk. Burt (1992) defines social capital as relationships with other players and distinguishes it from other kinds of capital such as financial capital and human capital. Because most credit risk models reflect mainly financial capital, I will try to add social capital or network perspective to the previous credit risk models.

In this paper, I will focus on the credit risk of Japanese firms. I choose Japanese firms because they seem to have large amount of social capital, compared with the firms in other countries. The word "keiretsu" is widely known outside Japan as the gigantic corporate networks among Japanese firms. Keiretsu may provide good data for analyzing the network effect on corporate credit risks.


Credit ratings that reflect default rate

Moody's Investors Service states that "ratings are intended to serve as indicators or forecasts of the potential for credit loss because of failure to pay, a delay in payment, or partial payment." Standards and Poor's states that its ratings are an opinion of the general creditworthiness of an obligor, or … of an obligor with respect to a particular … obligation … based on relevant risk factors." (Treacy & Carey, 2000). Although these rating agencies do not explicitly state that their ratings clearly correspond to the default rate of the corporate debt, they announce the empirical data about the default rate of corporate debt, that indicates the relationship between their rating scale and probability of default (Treacy & Carey, 2000).

Table 1: Bond Rating scales and One-year Default Rate (%)

  S&P(81-94) Moody's(70-95) R&I(Japan)




























Sources: S&P and Moody's: Treacy et al. (2000). R&I (Japan): Financial Technology Research Institute, Inc. Japan. The periods covered by the three studies are somewhat different.

Table 2: Bond Ratings of Famous Japanese Companies

  R&I(Japan) S&P Moody's












Mitsubishi Corp




(As of March, 2000: Information from each rating agency)

Table1 compares the default rate of each rating scale reported by U.S. major rating agencies (Standard & Poor's and Moody's) and Japanese rating agency (Japan Rating & Information, Inc.). U.S. Rating agencies rate the Japanese corporate bond mostly lower than the Japanese rating agency. Table 1 suggests that the default rate for the same grade appear to be higher for the data of U. S. major rating agencies than Japanese rating agencies.

If U.S. agencies and Japanese agencies put the same grade for a same corporate bond issued by a Japanese company, it seems that U.S. agencies overestimate the credit risk of the corporate bond. In fact, this is not true. As Table 2 shows, the U.S. rating agencies tend to rate severely than Japanese rating agencies. This is not only the case for a few famous Japanese companies but also for the typical tendencies that U.S. rating agencies have for Japanese companies. Actually, Japanese rating agencies rate the corporate dept rather independently of U.S. rating agencies. Therefore, the empirical data from Japanese rating agencies reflect more precise estimation of corporate credit risk of Japanese firms.

It appears that U.S. rating agencies tend to overestimate the default risk of Japanese corporate bonds more than Table 1 implies. For example, if a U.S. rating agency rates a corporate bond of given Japanese firm as "BBB" while a Japanese rating agencies rate the same one as "A", U.S. rating agencies estimate the probability of default within one year as 0.10% while empirical data shows actual default is almost 0.00%. This is a big difference considering that the data reflects only one year period and that the amount of money treated in the financial industry is so large.

Because Standard & Poor's and Moody's are two major rating agencies, quite a large number of financial institutions and other industrial firms rely on their rating information for assessing the corporate credit risks. Large banks may have their own credit rating systems. But it is suspected that the results of estimation are rather consistent with those of the two rating agencies. If it is so, many institutions may not correctly estimate the credit risk of Japanese firms. That is, the credit risk of Japanese firms may tend to be always overestimated. It is highly possible that these overestimation is due to the fact that they are not considered the value of social capital or social networks that actually reduce the probability of default.

Credit risk models

To my knowledge, there is no credit risk modeling approach that considers the social network factor. The modern approach to credit risk and the valuation of contingent claims such as debt, starts with the work of Merton (1974). Merton (1974)'s asset pricing model basically estimates credit risk as the probability that the value of equity become less than or equal to zero. The probability may be affected by the company's financial situations and business situations. The approach used in rating agencies is based on this kind of financial model. They do not go into the actual estimation of default rate but determine the rating grade that implies the default rate based on their analyses of financial situations. As for current models, there are mainly four major methodologies for credit risk management, CreditMetrics, KMV CreditRisk+, and CreditPortfolio View (Crouhy, Galai, & Mark, 2000). CreditMetrics developed by J. P. Morgan basically relied on the information from rating agencies for each firm's credit risk and concentrate on how rating will change over the long period. KMV is rather based on the Merton's asset pricing model. CreditRsik+ developed by Credit Suisse Financial Products assume that default is assumed to follow Poisson process. CreditPortfolio View developed by McKinsey & Co. considers macro variables such as unemployment and the level of interest rate.

In any case, the current approach to the estimation of corporate credit risk does not seem to consider the effect of social network that given companies have with others.

Empirical research on social capital and firm dissolution

There is little empirical research on the relationship between social network and the bankruptcy. Among the little, Pennings, Lee, & Witteloostujin (1998) conducted an empirical study to access whether the social capital contributes to the reduction of firm dissolution. They used the data from a population of Dutch accounting firms for the period of 1980-1990. They operationalized the concept of firm-level social capital using eight variables that represents the firm members' ties with other industries, government, and client environments. Their results showed that the existence of partners who came from client environments and partners who went to the client environments significantly decreased firm dissolution. This study shows that social network has in fact effects on corporate dissolution. This is rather simple case in a sense that the existence of network ties reduces the risk of corporate failure. Using more complicated network data, I will further focus on what characteristic of network has impact on the elimination of credit risks, why it is so, and how it is so. In the following, I will discuss the social network characteristics of Japanese keiretsu and develop some propositions and hypotheses on how the social networks affect the corporate credit risks.

The characteristics of keiretsu network

Keiretsu refers to the characteristic Japanese interfirm patterns that are multiple, overlapping clusters of enterprises. Keiretsu are fundamentally socially structured relations, not legal entities. There are horizontal or intermarket clusters that include large financial institution and many kinds of industrial firms, and vertical or supplier and distributor keiretsu that structure around large manufactures such as Matsushita and Shiseido. However, boundaries among clasters are not clear and many clusters are overlapping. For example, NEC is a well-known member of the horizontal Sumitomo group but NEC also has its own vertical network of suppliers and distributors (Lincoln, Gerlach & Ahmadjian, 1998).

In terms of horizontal keiretsu, the features of these networks are (1) presidents' council as a social event among the presidents of member firms, (2) cross-shareholding, (3) interlocking directories, (4) loans from group banks, and (5) trading of goods and services within networks.

Keiretsu has a long history. During the decades leading up to Japan's wartime economy, the zaibatsu (a group of firm) played an important role in Japanese economy. The group bank helped to raise capital that was used in expansion projects; group trading firm provided international and overseas intelligence and resource support; and head office coordinated overall resource allocation through a small team of decision makers. After the World War II, zaibatsu was dissolved by U. S. occupation. Instead, the group firms' ties again became strengthened through an increase in cross-shareholding between firms. Ministry of Finance and MITI also supported the activities. Other things such as permanent employment, long-term buyer-supplier relationships created a tightly knit and internally consistent business system, which lead to the evolution of keiretsu networks (Lincoln, et al., 1998).

Gerlach (1992) analyzed Japanese keiretsu network using blockmodel analysis. One of the major findings is that financial institutions lie at the central position in Japanese interfirm networks. That is, financial institutions are similar to one another in terms of positions within the corporate network but substantially different from industrial firms in terms of roles and positions within the networks. Industrial firms are also similar to one another in terms of positions within networks, which suggests that Japanese corporate network represents a relatively well-ordered structure of relationships among member firms.

Theoretical explanation of keiretsu networks

Lincoln, Gerlach & Takahashi (1992) developed theoretical explanation of Japanese keiretsu networks. Transactional cost economics (e.g., Williamson, 1975) highlights mergers, acquisitions, and vertical integration -- the placement of boundaries around interacting economic nodes to offset the market's failure to thwart opportunism or free up information. Resource dependence theory (e.g., Pfeffer & Salancik, 1978) focuses on various interorganizational "bridging" strategies that firms use to control their environments. Both theories share core assumption that interorganizational control structures arise from economic exchange in order to offset market uncertainties and the propensity for exchange partners to bargain opportunistically.

Using transaction cost economics and social network theory, Jones, Hesterly, & Borgatti (1997) provide the conditions that foster the network governance. According to Jones, et al. (1997), exchange conditions of asset specificity, demand uncertainty, task complexity, and frequency increase the network form of governance. These conditions seem to be applied to the Japanese environment that foster the networking of Japanese firms. Jones et al. (1997) also suggest that these conditions drive firms toward structurally embedding their transactions, which enables firms to use social mechanisms for coordinating and safeguarding exchanges.

The risk buffering mechanisms in keiretsu networks

Lincoln et al. (1998) describe the risk buffering mechanisms that keiretsu networks offer to the member firms. In the 1970's, Sumitomo Bank orchestrated a rescue of ailing Mazda Motors. Sumitomo Bank provided new loans, sent skilled managers, made sure that other Sumitomo group members did not sell their Mazda shares, negotiated lower prices for steel and other inputs, and even encouraged Sumitomo executives to buy Mazda cars. Similarly, Mitsubishi group rescued Akai Electronic from bankruptcy by arranging loans, sent experts. In this way, the keiretsu networks sometimes intervene into the troubled firms for rescues. The tools of intervention are diverse. Banks roll over old loans into new financial packages. They dispatch personnel to board and operating management positions with the targeted firm. The distressed firm will sell off sable, cross-held shares in long-term keiretsu partners that are quickly repurchased by other members of the group. In sum, groups pool and diversify risk, providing an implicit insurance contact against bankruptcy and failure.

Keiretsu network and corporate performance

Although there seems to be no empirical study on how keiretsu network affects the corporate default rates, some empirical studies on keiretsu networks and corporate performance were conducted. Lincoln et al. (1996) studied how profitability was affected by firm integration in big-six horizontal keiretsu networks, using data on 197 large Japanese firms over a 24-year period. Combining measures of financial and commercial dependence on a keiretsu group with the governance ties of equity ownership, director transfers, and shacho-kai (presidents' council) membership, they showed that group firms had lower average profitability than independents. Their results suggest that weak companies benefit from group affiliation (they recover faster), while strong ones do not (they are subsequently outperformed by independent firms). Thus, there is much less variability in the performance of keiretsu firms than independents. This study suggests that there are altruistic behaviors toward the companies with highly financial risks. However, this redistribution effect decays in the second half of the 1980's during a period spanning deep structural changes in the Japanese economy. Before then the effect is evident for all five measures of firm ties to big-six keiretsu groups.

Reasons for risk buffering

Transaction cost perspective. In a highly developed interfirm network such as keiretsu, the firms are dependent of one another. That is, firms are deeply embedded within the complex corporate networks. In this situation, a firm's critical resources may span firm boundaries and may be embedded in interfirm routines and processes. For example, one company may owe a lot of business processes or resource components to other firms within the network. Without them, the firm may not be able to continue to run the business. From the viewpoint of transaction cost, the bankruptcy of a company is much costly for other companies within the corporate network. Therefore, they tend to rescue the member firms that become struggled with financial problems.

Quality of network ties. Dyer & Singh (1998) compared General Motors and Toyota in order to analyze their relationships with suppliers. GM has not cultivated a stable network of supplier companies. Rather, its relationship with supplier companies is more market-based. That is, the relationship between GM and its suppliers may not stable and changeable based on the price suppliers offer to GM and other factors. On the other hand, Toyota tried to build long-term relationships with its supplier companies by actively facilitating knowledge sharing, transferring management practices such as operations management, transferring its personnel to suppliers. As a result, Toyota's network ties with its suppliers are beyond transaction levels. Toyota does not easily change its suppliers even if other suppliers offer lower prices to it. Instead, Toyota encourages its suppliers to lower their production costs by providing consulting services, information about cost reduction technologies, and so on. As the case of Toyota, many Japanese companies within keiretsu networks made investment for establishing rich network ties. Therefore, letting members go into bankruptcy means that they are losing valuable ties that are developed by heavy investment. This may be one of the reasons that member firms try to rescue troubled firms within networks.

Country specific or cultural factors. Some researchers argue that Japanese firms appear to have been successful at generating valuable network ties in part because of a country-specific institutional environment that fosters goodwill trust and cooperation (Dyer & Singh, 1998). They contend that in other countries (e.g., the United States and Russia) may not be able to replicate that kind of network structure because of an inability to replicate the socially complex exrahybrid institutions embedded in the Japanese institutional environment. The theory of individualism-collectivism may provide a view about the risk buffering mechanism within Japanese firm networks. Some cultures (such as U.S.) develop citizens who are primarily individualistic and others (such as Japan) develop citizens who are decidedly collectivistic (Moorman & Blakely, 1995). Because collectives tend to weigh more on group benefits than individual benefits, the group of people in collectivistic culture may be easy to foster the cooperative relationship. In these relationships, altruistic behaviors such as rescue others without any return can occur quite often.

Network forms of organizations. If we consider the keiretsu network as one gigantic network organization rather than the group of individual organizations, it may be natural that the organization transfers its financial resource from one location to another as environment changes. Therefore, the risk buffering mechanism of keiretsu network can be thought as the resource re-allocation within the organization as response to the environmental changes.

Determinants of network effects on corporate credit risk

In addition to the traditional factors that are mostly studies in financial area, the credit risk of the firm that is a member of the interfirm network may be influenced by the characteristics of the interfirm network, and the position of the firm within the network. It might be difficult problem that what kind of characteristics is important for determining the effect on corporate credit risk. The basic parameters that are used in social network analysis may apply. However, the other factors should also be considered. For example, the relative strength of the network can be estimated using the amount of financial capitals that the members of the network possess. Gerlach (1992) used the concept of financial centrality in analyzing Japanese keiretsu networks. This view is based on the assumption that such capitals are jointly owned by member firms even if each capital is legally owned by individual companies. Next, the relative position of the member firms should be considered. The position of the firm affects the way the jointly owned capital can be allocated in case of the emergency financial troubles. This position is affected by which firm to have connection with, as well as the structure of ego-centered networks and the relative position within the network. Especially, the role of financial institutions is crucial in these relationships. Therefore, how each member firm relates to financial institutions within networks may have a great impact on its credit risk.



Keiretsu network structure. Data on Japanese keiretsu network can be obtained by combining several data sources that are publicly available in Japan. For example, Lincoln et al. (1992) obtained data from 50 financial corporations and 200 industrial firms. In their study, intercorporate control relations (measured by cross-shareholding and director transfer) and economic exchanges (lending and trade) are measured for each pair of organizations in their sample. The largest 200 industrial firms in Japan in terms of company sales could be identified in a list in "Nihon Kogyo Shudan Bunseki," and 50 major financial institutions based on total assets could be identified from the "Industrial Groupings in Japan." Also, "Kaisha Nenkan" would provide information on the affiliations of a firm's outside directors and the former affiliations of current directors who were transferred by other companies. "Jinji Koshin Roku" would provide biographical data on high-level managers and other prominent persons. The top ten shareholders in each industrial fims and the amount of their holdings were coded from a Japanese annual, "Keiretsu Kenkyu". These relational measures were supplemented by measures capturing dyad- and firm-level attributes.

Bond ratings and default rates. Data on bond ratings and default rates can be obtained from the major U. S. rating agencies, Standard & Poor's and Moody's Investors Service. The data is based on the empirical default statistics that correspond to each rating grade determined by the rating agencies. They also provide the each Japanese company's bond grading. In fact, they rate based on the default risk of the corporate bond. However, since bonds issued from same corporation will be obtained same grade, we can interpret that the rating represents the default risk of a company itself. We can construct data of each corporation's estimated default risk by U.S. rating agencies.


To analyze the network effect on corporate credit risk, the model can be developed that represents the network effect. The network effect or social capital for given company is considered to be a sum of the effect due to the characteristics of the network where the company belong and the effect of its relative position within the company. Thus, the formula will be as follows.

Social Capital (ij) = Effect (CNj) + Effect (PNij)


Social Capital (ij) = Network effect on credit risk of company (i) within network (j)

Effect (CNj) = Network effect due to the characteristics of network (j)

Effect (PNij) = Network effect due to company (i)'s relative position within network (j)

Positive value in Social Capital means that the variable has an negative effect on the total credit risk of a firm. Therefore, the total amount of credit risk can be estimated by subtracting network effect that is adjusted by a coefficient from traditional estimation of corporate credit risk. The result of the credit risk should be more similar to actual empirical default rate than merely traditional estimation of corporate credit risk. Parameter estimation should be conducted in order that the model is close to the actual default rate.

Total credit risk (%) = CR (TR) - k (Social Capital)


CR (TR) = Credit risk based on the traditional estimation (e.g., financial approach)

k= coefficient


Characteristics of networks. The first analysis is to identify the cohesive subgroups of actors within a network data. As described in the above sections, the keiretsu networks in Japan are overlapping with one another. Therefore, the meaningful clustering through network analysis is needed to understand the characteristics of keiretsu network. Clique analysis may be suitable for identifying subgroups within a network. The subgroup structure of networks may be multilevel. That is, there are not only one lower-level subgroups, but also middle level, or other levels of subgroups can be identified. Next, the characteristics of the identified subgroups need to be analyzed. Network density, connectivity, and reachability are some of the ways for describing network characteristics.

Each firm's characteristics. The analyses that focus on the given firm are also needed to identify each player's attribution among networks. Blockmodel analysis may be useful for identifying structural equivalence. Centrality and prestige are among the frequently used analyses of the players' position among networks. Not only the position of given player is important, but also the position of other players tied to given player are also important. Especially, the firm that has strong financial asset (e.g., financial institutions) may play an important role in supporting troubled firms.

Parameter estimation. Parameter estimation can be done to adjust each parameter of the model so that the model fit into the empirical data of default rate well. There may be several estimation methodologies.


Uzzi (1997) developed the arguments about the paradox of embeddedness in interfirm networks. One of the conclusions derived his arguments is that, while embeddedness in interfirm networks has considerable positive effect, embeddedness can also derail economic performance by making firms vulnerable to exogenous shocks or insulating them from information that exists beyond their networks. This argument provides an important message to this study. The basic assumption of this study is that interfirm networks and its characteristics such as embeddedness have a positive effect in eliminating the credit risk of member firms. However, as Uzzi suggests, it is possible that the embeddedness among member firms within the networks work negatively toward the credit risk of member firms. For example, even if there is positive effect of the interfirm networks when viewed from each firms, but the credit risk of network as a whole may become weaker when it is faced some harsh economic situations than the total credit risk in the group of companies where the members are rather independent of one another. This kind of characteristics of embeddedness should be included in developing more sophisticated credit risk models. In any wan, this study will be the first step to quantify the value of social capital in terms of its effect on corporate credit risk. Based on the empirical examination, more sophisticated model and theory development should be made in the future.


Burt, R. (1992). The Social Structure of Competition. Chapter 1 in Structural Holes. Cambridge: Harvard University Press.

Crouhy, M., Galai, D., & Mark, R. (2000). A comparative analysis of current credit risk models. Journal of Banking and Finance, 24, 59-117.

Dyer, J. H., & Singh, H. (1998). The relational view: Cooperative strategy and sources of interorganizational competitive advantage. Academy of Management Review, 23 (4), 660-679.

Gerlach, M. L. (1992). The Japanese corporate network - A blockmodel analysis. Administrative Science Quarterly, 37 (1), 105-139.

Gulati R., Gargiulo M., (1999). Where do interorganizational networks come from? American Journal of Sociology, 104 (5), 1439-1493.

Jones, C., Hesterly, W. S., & Borgatti, S. P. (1997). A general theory of network governance: Exchange conditions and social mechanisms. Academy of Management Review, 22 (4), 911-945.

Lincoln, J. R. (1990). Japanese organization and organization theory. Research in Organizational Behavior, 12, 255-294.

Lincoln, J. R., Gerlach, M., & Ahmadjian, C. (1998). Evolving patterns of keiretsu organization and action in Japan. Research in Organizational Behavior, 20, 303-345.

Lincoln J. R., Gerlach M. L., Ahmadjian C. L. (1996). Keiretsu networks and corporate performance in Japan. American Sociological Review, 61 (1) 67-88.

Lincoln, J. R., Gerlach, M, & Takahashi, P. (1992). Keiretsu networks in the Japanese economy: A dyad analysis of intercorporate ties. American Sociological Review, 57, 561-585.

Merton, R. C., (1974). On the pricing of corporate dept: The risk structure of interest rates. Journal of Finance, 449-470.

Moorman, R. H. & Blakely, G. L. (1995). Individualism-collectivism as an individual difference predictor of organizational citizenship behavior. Journal of Organizational Behavior, 16 (2), 127-142.

Pennings, J. M., Lee, K., & Witteloostujin, A. (1998). Human capital, social capital, and firm dissolution. Academy of Management Journal, 41 (4), 425-440.

Pfeffer, J., & Salancik, G. R. (1978). The External Control of Organizations: A Resource Dependence Perspective. New York: Harper and Row.

Scott, J. (1991). Social Network Analysis: A handbook. Thousand Oaks: Sage.

Treacy, W. F., & Carey, M. (2000). Credit risk rating systems at large US banks. Journal of Banking & Finance, 24, 167-201.

Uzzi, B. (1997). Social structure and competition in interfirm networks: The paradox of embeddedness. Administrative Science Quarterly, 42, 35-67.

Wasserman, S. & Faust, K. (1994). Social Network Analysis: Methods and Applications. Cambridge: Cambridge University Press.

Williamson, O. E. (1975). Markets and Hierarchies: Analysis and antitrust implications. New York: Free Press.