Strategic Decoupling in Korean Business Groups: Ambiguous Identity as a Strategy in Chaebol Groups

Abstract

I examine how firms affiliated with Korean business groups utilize decoupling of the stated business area and actual business activities to maintain the economic sustainability of their organizations. In examining the strategic sources of decoupling, I focus on the idiosyncratic nature of Korean business groups, otherwise known as chaebols. I suggest that decoupling of the stated and actual business areas of a chaebol affiliate is affected positively by the number of regulations in the industry, positively by the relative resource endowment of the affiliate within the chaebol, and negatively by the affiliate’s niche overlap with other affiliates. However, the negative effect of niche overlap was moderated by the affiliate’s relative resource endowment.

1. Introduction

Most of the studies on identity and classification in the organizational ecology literature generally suggest that ambiguous identities incur various disadvantages, since audiences prefer organizations with clear and sharp identities [1,2,3]. Although a few exceptions have examined the potential benefits of ambiguous identities [4], sources of such identity ambiguity have received scant attention so far. I argue that organizations often deliberately choose ambiguous identities for the purpose of maintaining the sustainability of their businesses, by decoupling how they present themselves to external stakeholders from what they actually do. More specifically, I examine a type of identity ambiguity that stems from the decoupling of an organization’s externally stated business area (announced through the Standard Industrial Classification code) and its actual business activities.

Although Meyer and Rowan [5] highlighted decoupling as an important consequence of organizational formalization in their seminal paper, this phenomenon has been seldom studied empirically. While neo-institutional theory suggests that decoupling of the formal and informal aspects of organizational phenomena occurs as organizations ceremonially adopt widespread formal practices, we still do not know much about the specific mechanisms through which the degree of such decoupling increases or decreases.

This paper aims to contribute to this gap in the literature by suggesting institutional pressures and strategic conditions that may influence firms to choose decoupling. Existing studies on decoupling have illustrated the discrepancy between formal rules and informal practices [6,7], but they did not specifically discuss whether organizations choose decoupling intentionally, and if yes, which factors drive them to choose decoupling. This study may contribute to the existing literature on decoupling by emphasizing firms’ strategic choices, arguing that organizations may deliberately choose decoupling in order to deal with institutional pressures or strategic circumstances. This paper also may contribute to the development of the decoupling literature by showing one way to empirically measure decoupling. Although decoupling has been suggested as one of key concepts in neo-institutional theory [5], it has rarely been studied empirically due to the difficulty of operationalizing the concept. This study presents one type of decoupling that may be measured by observing the discrepancy between firms’ Standard Industrial Classification codes and their actual business areas. Given that most countries adopt formal classification schemes for categorizing industries and classify firms according to these schemes, this study suggests one way of measuring decoupling that can be applied in various national contexts.

This paper may also have implications for policy makers. I observe how industrial regulations affect firms’ choices of decoupling. More specifically, I observe whether new industrial regulations lead firms to explore business areas outside their originally stated business area. Therefore, the implication for policy makers is that when designing industrial policies, they should take into account how new industrial regulations may have unintended consequences on industry incumbents.

I empirically examine sources of the decoupling between stated business areas (i.e., officially announced Standard Industrial Classification codes) and actual business activities in firms affiliated with Korean business groups, otherwise known as chaebols. I argue that firms often strategically decouple how they externally represent their activities and what they actually do. In making the argument, I focus on the idiosyncratic nature of business groups as a tightly coordinated network of multiple affiliated firms operating in diverse industries [8,9,10], since the issue of decoupling becomes particularly salient in highly diversified firms [8].

The empirical analysis of 314 affiliated firms of the thirty largest Korean business groups suggest that decoupling of the stated and actual business areas is affected positively by the number of regulations in the industry (i.e., institutional pressure), positively by the relative resource endowment of the affiliate within the business group (i.e., strategic conditions at the affiliate level), and negatively by the affiliate’s niche overlap with other affiliates in the business group (i.e., strategic conditions at the business-group level). However, the negative effect of niche overlap was found to become weaker with the increase of the focal affiliate’s relative resource endowment.

2. Decoupling in Organizations: Formal Properties as Myths and Ceremonies

The decoupling of formal structures and actual behavioral patterns has been one of the key themes in neo-institutional theory. Meyer and Rowan [5] noted decoupling of formal structures and actual operations in organizations, arguing that organizations buffer the workings of their technical cores from external disturbances and, at the same time, gain legitimacy from external stakeholders by decoupling formal structures from actual operations. Neo-institutional theorists maintain that, while profit maximizing certainly is the fundamental reason for being business firms, at the same time, internal and external stakeholders expect firms to be legitimate [11,12]. Failure to meet this expectation for legitimacy results in serious damage to resource attraction and chances of survival [11,13].

Problems may arise when there is a contradiction between these dual pressures—technical efficiency and legitimacy. One solution to this dilemma is separating the statement from the action—in other words, gaining and maintaining legitimacy by adopting formal structures for ceremonial purposes, but informally relying on routines and practices organizations believe to be profit maximizing [11,14,15]. This separation of action from statement is referred to as decoupling in neo-institutional theory [5].

Despite the importance of decoupling in the conceptual system of neo-institutional theory, only a few empirical studies exist on decoupling. The few exceptions include studies on the decoupling of incentive programs for chief executive officers from actual practices [16,17,18], the decoupling of stock repurchasing programs from actual corporate investments [19], the decoupling of local governments’ implementation of the European Foundation for Quality Management’s Business Excellence Model and actual practices [20], and the decoupling of the official announcements of price-cuts for AIDS medicines and the substantive action of actual price-cutting [21].

This paper departs from the existing studies on decoupling in the following two aspects. First, while most of the existing studies on decoupling have focused narrowly on the actual implementation of a specific practice and its inconsistency with a formally announced program, this paper examines the decoupling of a firm’s announced and actual business areas—probably the most essential factor that defines the fundamental identity of the entire organization. Among the various dimensions that collectively constitute the identity of a firm, its core business areas are arguably the single most crucial element [22].

Second, unlike in the neo-institutional theory literature where organizational responses to institutional pressures are often depicted as a preconscious-level action [23], this paper theoretically frames decoupling as a more conscious-level strategic action. I see firms as being able to deliberately utilize decoupling for strategic purposes. The original formulation of the concept by Meyer and Rowan [5] also implicitly assumed decoupling as an adaptive action that deals with potentially contradictory pressures. Oliver [24] more explicitly underlined the strategic aspects of organizational responses to institutional pressures, Hirsch and Bermiss [25] similarly suggested that strategic decoupling can contribute to the preservation of institutions, and Beverland and Luxton [26] even compared decoupling to impression management. In this regard, I view decoupling as a strategic response chosen by organizations to deal with institutional pressures and strategic situations.

3. Business Groups and Inclination to Diversify

The decoupling of the stated and actual business areas is especially salient in business groups, due to their high inclination to diversify [9,27]. A business group is broadly defined as a ‘collection of firms bound together in some formal and/or informal way’ [8]. The operation of firms affiliated with a business group is often tightly coordinated by the group headquarter. For instance, affiliates tend to submit to the strategic initiatives of the group headquarter when making major strategic decisions [8,27,28,29].

One of the most distinctive characteristics of business groups is their exceptionally strong inclination to diversify over a wide range of unrelated industries [9,27]. The emergence of business groups is often attributed to diversification under market imperfections. For instance, when the markets for production factors such as capital or labor are not relevantly developed, an economic actor with an idea for new business opportunities may find it difficult to obtain the required financial or human resources, but business-group-affiliated firms can resort to their internal capital and labor markets and therefore have better opportunities to pursue aggressive diversification [30].

In diversified business groups like Korean chaebols, affiliates are often tightly coordinated by the group headquarter in order to achieve synergy at the business-group level [27,29]. Decisions regarding diversification are probably one of the most fundamental strategic activities carried out by the group headquarter. However, diversification cannot be unilaterally decided by the headquarter because the strategic dynamics of business groups involve complex relationships between the headquarter, the affiliates, and the government [27]. When a business group needs to rearrange business areas among its affiliated firms, the group’s top management has to consider the strategic importance and bargaining power of each affiliate, and also the industrial policies imposed by the government.

4. Decoupling, Identity Classification, and the Standard Industrial Classification (SIC) Code

In this paper, I examine the decoupling of actual business activities and stated business areas officially announced through each firm’s SIC code. In the SIC system used in most countries, including Korea, a firm is generally assigned an SIC code that indicates the business area accounting for the greatest share among all of that firm’s business areas [31]. Therefore, the SIC code of a firm is the official declaration of its identity in terms of its main business area.

The SIC code is a classification system that assigns each member of the organizational population to a category in terms of its business areas. The central function of a classification system is to provide stakeholders with the foundation for the comparative differentiation of population members with a universal criterion [32,33]. The results of classification are utilized by stakeholders in their decisions related to the members of each category [32].

Organizational ecology literature suggests that the positioning of an organization in classification categories significantly affects the perception and evaluation of its identity by various audience groups [1,3]. Thus, regardless of accuracy, the classification of business firms based on SIC codes is used in decisions for industrial policy, diversification, investments, loans, regulation, taxation, stock trading, mergers and acquisitions, and so forth, by relevant stakeholders such as consumers, company executives, investors, banks, data service agencies, regulators, IRS, SEC, and the government [31,34,35,36]. Therefore, a firm has a strategic incentive to declare its SIC code in a way that represents itself more favorably in the eyes of the stakeholders, sometimes at the cost of sacrificing the accuracy of depicting its actual activities. When this is the case, the stated business areas officially declared in the SIC code of a firm may be seriously decoupled from its actual business activities.

Some studies in the organizational ecology literature suggest that organizations often have identities that span multiple categories, even within a classification scheme [1,2,3,37]. In other words, since the categories that collectively constitute a classification system are not always orthogonal with one another, an organization may have an identity that spans multiple categories, rather than one that closely matches a single category. Most of the existing studies on identity and classification suggest that identity ambiguity stemming from such spanning across multiple categories is likely to have negative effects on organizational survival and performance, since audiences tend to evaluate organizations with an ambiguous identity unfavorably [1,2,3,37]. However, Pontikes [4] argues otherwise, explaining that while market-takers such as consumers indeed tend to avoid uncertainty surrounding identity ambiguity, market-makers such as venture capitalists tend to favorably regard ambiguous identities as an indicator of future flexibility.

However, the type of identity ambiguity examined in this paper is different from ambiguity related to the multiple category spanning type reviewed in the organizational ecology literature. In the SIC system examined in this paper, each firm is classified into a single identity category without the possibility of boundary-spanning across multiple identity categories. Therefore, the identity ambiguity analyzed in this study mostly results from the firm’s diversification into new business areas outside its officially declared SIC code, or from its partial or full withdrawal from the previously declared business area. Inconsistency between SIC codes and actual activities was examined by Kahle and Walkling [36], Clarke [31], Guenther and Rosman [34], and Fan and Lang [35]. Clarke [31] especially noted that the assignment of an SIC code to diversified multi-business firms is often misleading, considering the historical fact that the SIC system was initially developed to classify individual manufacturing establishments. The problem of inconsistency between stated SIC codes and actual business areas is markedly serious in business groups like Korean chaebols because of their exceptionally high degree of diversification. In this regard, I examine the decoupling of stated and actual business areas of firms affiliated with business groups, focusing on institutional pressures and strategic conditions that lead to this decoupling.

5. Hypotheses

To examine the factors that may affect the degree of decoupling of the stated business area and actual business activities of a business-group affiliate, I focus on the effects of institutional pressures from the organizational field, strategic conditions of the affiliate and of the business groups, and their interaction.

5.1. Institutional Factor: Government Regulations on Industries

Organizations may strategically decouple actual business activities from their stated SIC to avoid certain institutional pressures from the government. Among the three mechanisms of institutional isomorphism suggested by DiMaggio and Powell [11], coercive isomorphism is the most direct and conspicuous kind of mechanism that results from pressures exerted on organizations by powerful other organizations, such as regulatory agencies. Among those ‘other organizations,’ the government is probably the most powerful player that sets the rules of the game. Choices and actions of business firms are seriously constrained by industrial policies and regulations set by the government. For instance, when the government perceives the state of a certain industry of crucial public interest to be unhealthy, such as in case of excessive competition, unethical business practices, or near monopoly, it may impose various regulations on the industry [38,39]. Additionally, if the government perceives an industry to be outdated or unhelpful for the development of the national economy, it may announce regulations on that industry in order to deter further entry into the industry and promote exits.

The abundance of regulations in certain industries restricts the industry incumbents’ autonomy to pursue self-interests. Therefore, if the government places many regulations on a certain industry, firms that declare their stated business area to be in that particular industry may try to avoid being the target of regulations by exploring other business areas, while maintaining the stated business area until the majority of its business shifts to the newly explored industry. In such a case, the government regulation of the industry to which a business-group affiliate officially belongs (in terms of its stated SIC code) is likely to result in increased decoupling of the stated and actual business areas. Therefore, decoupling of the stated and actual business areas in this case can be viewed as a strategic response made by business-group affiliates, to effectively deal with the institutional pressures from the government.

Hypothesis 1 (H1).

The more government regulations on the industry that a firm affiliated with a business group officially declares as its main business area in its SIC code, the higher the degree of decoupling of its stated business area and actual business activities in the subsequent period.

5.2. Strategic Factor at the Affiliate Level: Relative Resource Endowments of Business-Group Affiliates

The degree of decoupling may be affected by strategic conditions surrounding the affiliate or the business group. Particularly, the relative proportion of the entire business group’s strategic resources possessed by an affiliate is likely to be associated with the degree of decoupling of its stated and actual business areas, since it represents the strategic importance of the affiliate to the interests of the entire business group. The strategic resources of a business group are divided and allocated to its affiliates [40,41], with certain affiliates receiving a larger portion of strategic resources than other affiliates. This within-group disparity of resource endowment among affiliates may bring about variations in the group headquarter’s strategic control over each affiliate. In a business group, the competitiveness of an affiliate is not confined to itself, but is actively leveraged to other affiliates through various resource-sharing mechanisms arranged by the headquarter [42,43,44,45]. Since each affiliate is expected to contribute to the performance of itself at the firm level as well as to that of the entire group [44], the utilization of strategic resources possessed by an individual affiliate is not confined to their own individual performances, but widely and flexibly leveraged to generate the entire group’s performance.

Therefore, the key affiliates endowed with a large amount of critical strategic resources are often expected to play the role of a seedbed for new business creations at the entire group level [27]. They are likely to be asked by the headquarter to invest in risky new businesses for the group’s sake. Those affiliates which possess strategic resources are especially likely to face strong intervention from the headquarter to leverage their strategic resources to new business areas [46].

Pontikes [4] argues that an ambiguous identity may serve as a basis of strategic flexibility. As key affiliates with large resource endowments are expected to serve as a seedbed for the group’s new business creations, they leverage their strategic resources to new business areas which are different from the initially stated business area for group interests, and thus the degree of decoupling of its stated business area and actual business activities is likely to increase. Therefore, the degree of decoupling will be higher for key affiliates endowed with a larger amount of critical strategic resources than for others endowed with less.

Hypothesis 2 (H2).

The relative size of resource endowment possessed by a firm affiliated with a business group in comparison to those possessed by the affiliates belonging to the same business group will be positively associated with the degree of decoupling of its stated business area and actual business activities in the subsequent period.

5.3. Strategic Factor at the Business-Group Level: Within-Group Niche Overlap

The degree of decoupling may be affected by the extent to which the business areas of affiliates belonging to the same business group are redundant with one another. In other words, the degree to which an affiliate’s actual business activities overlap with those of the other affiliates in the same business group may affect the degree of decoupling of its stated and actual business areas, since it represents a need for the rationalization of business areas at the entire group level.

In a business group, diversification and other corporate strategies are often formulated and implemented by the headquarter, with the goals of realizing synergy and hedging risks at the group level [30,43,44,47]. For this purpose of pursuing synergy at the group level, the headquarter of a business group tightly aligns the strategic actions of affiliates from the macro perspective of the entire group, rather from the micro perspective of an individual affiliate. However, at the affiliate level, the advantages of specialization and focusing, such as accumulation of expertise and industry-specific knowledge and scale economy, still remain valid [48,49,50]. Therefore, the best option would be the combination of diversification at the group level and specialization at the affiliate level [51].

However, the strategic interests of affiliates do not always coincide with this strategically optimal solution at the group level. Each affiliate is compelled not only to contribute to synergy and competitiveness at the group level, but also to pursue the maximization of its own performance at the affiliate level [8]. Thus, when an affiliate detects opportunities in adjacent new business areas beyond its initially stated business area, the company may be tempted to expand its operation into the new business areas without necessarily changing its stated SIC code. However, the iterations of such expansions by all affiliates sooner or later lead to the niche overlap of business areas among the affiliates, resulting in internal niche congestion [52,53].

The existence of excessive within-group niche overlap brings about competitive conflict, rather than cooperative synergy, among business-group affiliates. When affiliates’ niches overlap, they compete not only for scarce group resources but also for markets, which is detrimental to the performance at the group level. Moreover, within-group niche overlap incurs inefficiency stemming from resource dispersion and opportunity cost arising from the lost chance for synergy generation and scale economy. Therefore, when the degree of niche overlap among affiliates is high, the group headquarter has a strategic incentive to rationalize business areas at the group level by regrouping and reassigning the overlapping businesses to the most adequate affiliate. The reduction of within-group niche overlap through the rationalization of business areas is likely to result in the decrease of the decoupling of the stated and actual business areas in each affiliate in the subsequent period.

Hypothesis 3 (H3).

The degree of niche overlap between a firm affiliated with a business group and the affiliates belonging to the same business group will be negatively associated with the degree of decoupling of its stated business area and actual business activities in the subsequent period.

5.4. Interaction Between Relative Resource Endowment and Within-Group Niche Overlap

The negative effect of within-group niche overlap on decoupling may differ depending on the relative resource endowment of each business-group affiliate. I predict that the relative resource endowment of an affiliate will reduce the negative effect of within-group niche overlap on decoupling, since it indicates bargaining power based on resource independence within the group. As I mentioned earlier, strategic interests of affiliates are not always in accordance with those of the entire business group. When within-group niche overlap increases, the headquarter has a strategic incentive to rationalize the business areas of the entire group by refocusing each affiliate’s businesses to its individual specialty area. Yet, an affiliate may have an incentive to pursue new opportunities for further growth of itself.

When these two conflicting strategic incentives collide, the result is likely to be determined by the bargaining power of the affiliate [43,54,55]. One of the most important determinants of the bargaining power of an affiliate and the headquarter is the affiliate’s resource endowment. The central tenet of resource dependence theory is that, when an organization possesses a sufficient amount of critical resources, it may enjoy autonomy in the pursuit of its own interests, free from external controls [13]. Therefore, if a business-group affiliate possesses a large amount of strategic resources, it may enjoy bargaining power in its interactions with the group headquarter, and thus may pursue its firm-level strategic goals. The firm-level strategic goals may include growth through diversification. Their bargaining power based on resource independence is likely to play a role of countervailing force against the group headquarter’s strategic incentive to rationalize business areas [56]. Thus, I predict that the negative effect of within-group niche overlap on decoupling will be weaker for affiliates with a large relative resource endowment than for ones with a small relative resource endowment.

Hypothesis 4 (H4).

The negative effect of within-group niche overlap on the decoupling of the stated business area and actual business activities of a business-group affiliate in the subsequent period will be weaker for affiliates with a large amount of relative resource endowment than for ones with a small amount of relative resource endowment.

6. Methods

6.1. Empirical Setting and Data

This study examines the decoupling of stated business areas (i.e., SIC code) and actual business activities of firms affiliated with the thirty largest Korean business groups, otherwise called chaebols. Chaebols are groups of formally independent Korean companies which are under common administrative and financial control and are owned and controlled by certain families [57]. Although business groups are not unique to Korea, chaebols have unique attributes that make them particularly suitable for the empirical setting of this study.

First, in chaebol groups, the possibility of conflict between affiliate and group headquarter interests is especially pronounced because the group headquarter of a chaebol often pursues corporate strategies that may sacrifice the profitability of each individual affiliate. In a typical chaebol system, although the affiliates are separate legal entities, the interests of the entire business group often precede the interests of individual affiliates [47,58,59,60]. Although business groups in general pursue synergy at the group level, the overarching power of the group headquarter is particularly strong in the case of Korean chaebols [29,57]. Chaebols, intentionally and unintentionally, have created a highly complicated system of corporate control and governance based on heavy reliance on cross-shareholdings among affiliates, the hidden dispersion of shares among the owner’s family members and allies, and a skillful use of formal and informal management control devices [57]. Consequently, chaebol affiliates routinely face strategic decisions that conflict with their own firm-level interests.

Second, Korean chaebols are particularly well known for their highly diversified business areas, with the largest chaebols participating in variety of related and unrelated industries including but not limited to semiconductors, consumer electronics, automobiles, construction, and financial services [61]. Since chaebol affiliates frequently enter new business areas and exit from existing ones, there is a high possibility of discrepancy between the stated business area (SIC code) and actual business activities.

The dataset contains information on firms affiliated with the thirty largest business groups in Korea. Most, but not all, of the thirty biggest Korean business-group affiliates are included in the dataset, because division-level business area data was unavailable for some of the affiliates. Further, firms operating in financial industries were excluded because they use a different form of financial statement and thus are not comparable across some variables I use in this study. The total number of companies included in the data set was 314, from the years 1988 to 2000. Since all the independent and control variables are lagged by one year, these variables are from the years 1987 to 1999, while the dependent variable, decoupling, is from the years 1988 to 2000. Of course, some companies were created in the middle of this period and some exited during this period, so not all companies have a full record of thirteen years.

To examine which factors influence decoupling of the stated and actual business areas in those firms, I assembled a dataset of three levels: affiliate level, business-group level, and industry level. Affiliate-level data constitutes each affiliate’s foundation date, SIC code, financial statements, the composition of divisions within each affiliate, and the sales figure of each division. Division data gathered was written in text form, so in order to compare the composition of divisions with the representative SIC code, matching of the division with the SIC code was necessary. To do so, I used the Korean Standard Industry Classification (SIC) system. The Korean SIC system allows people to search the relevant SIC code by entering in text-form the business area name. Sometimes, it was not easy to infer the products from the division names, in which cases I visited the company website to find out exactly which type of products each division produced. Despite this effort, some division names were too ambiguous, and no information was available to indicate what they produce. As a result, the dataset was left with 2620 firm-years in total.

Business-group level data consists of the foundation date, number of affiliates, and financial statements at the group level from the years 1987 to 1999. Industry level data consists of the SIC code, number of companies in the focal industry, number of government regulation policies announced regarding the focal industry, and financial statements for the industry for the same period of time.

Financial data required for this study was acquired from the database developed by the Korea Information Service (KIS). KIS, a database that provides comprehensive corporate and financial information, is by far the most comprehensive and reliable database available in Korea [57]. Information on government regulations at the industry level was gathered from the Official Gazette of the Korean government. To sort out relevant industry regulations from myriads of announcements on the Official Gazette, I read all announcements made in the years 1987 to 1999, counted the number of announcements that were regulative in nature, and identified which industries the announcements affected.

6.2. Measurements

The following section describes how I measured the dependent, independent, and control variables. Like the independent variables, which span the three levels (affiliate, business group, and industry), the control variables also consist of variables from these three levels. In the statistical analysis, I applied one-year lagging to all the independent and control variables.

6.2.1. Dependent and Independent Variables

• Decoupling: For most business-group affiliated firms included in the data, the affiliates are comprised of multiple divisions, and only some of these divisions produce products that match the stated SIC code of the affiliate. The dependent variable, decoupling, was operationalized by the following equation:

  $$D_{it}=1- \frac{{{\displaystyle \sum }}_{j=1}^{n_i}{S}_{jt}{I}_{jt}}{{{\displaystyle \sum }}_{j=1}^{n_i}{S}_{jt}}$$

  Let $n_i$ be the number of divisions affiliate i holds, and $D_{it}$ firm i’s degree of decoupling at time t. $S_{jt}$ is the sales of j-th division at time t, and $I_{jt}$ indicates a function that takes a value of 1 if division j’s business area coincides with the stated business area of firm i (firm i’s SIC code), and 0 otherwise. The denominator in the equation is the sum of sales from all $n_i$ divisions in firm i at time t. In other words, dividing sales from divisions that match the officially stated SIC code by the total sales from all divisions within the affiliate gives a ratio of how much the actual business areas of the affiliate coincide with the stated business areas. By subtracting this ratio from 1, the degree of decoupling was calculated.
• Industry regulations: Data necessary for measuring the extent of government regulation was collected from the Official Gazette of the Korean government published from 1987 to 1999. I identified government announcements of policies that were regulative in nature for certain industries, then I identified which SIC categories each policy announcement affected. Industry regulations were measured by counting the number of regulative announcements made in the previous year that may affect each affiliate’s SIC code industry.
• Relative resource endowments: The relative resource endowments, $R_{it}$, of affiliate i at time t is defined as follows:

  $$R_{it}= \frac{E_{ijt}}{{{\displaystyle \sum }}_{h=1}^{n_j}{E}_{hjt}}$$

  Let $E_{ijt}$ represent the sum of R&D and advertisement expenditure of affiliate i of business group j at time t, and $n_j$ represent the total number of affiliates in business group j. The denominator is the sum of R&D and advertisement expenditure of all $n_j$ affiliates in business group j. I use the sum of R&D and advertisement expenditure for measuring resource endowments because these two financial statement items are the most frequently used ones to estimate valuable and hard-to-imitate resources in the strategy literature [62,63]. This variable is mean-centered to avoid multicollinearity.
• Within-group niche overlap: Niche overlap was measured as suggested by Stuart and Podolny [64].

  $${\alpha }_{ijt}= \frac{{{\displaystyle \sum }}_{v=1}^p{a}_{ivt}{a}_{jvt}}{{{\displaystyle \sum }}_{v=1}^p{a}_{ivt}}$$

  $${\alpha }_{jit}= \frac{{{\displaystyle \sum }}_{v=1}^p{a}_{ivt}{a}_{jvt}}{{{\displaystyle \sum }}_{v=1}^p{a}_{jvt}}$$

  In Formula (1), ${\alpha }_{ijt}$ denotes the proportion of affiliate i’s niche that is occupied by another affiliate j at time t. v denotes a niche, and p indexes the total number of distinct niches that are occupied by affiliates at time t. The value $a_{ivt}$ is coded 1 if niche v is occupied by firm i at time t and 0 otherwise; $a_{jvt}$ is coded 1 if niche v is occupied by firm j at time t and 0 otherwise. Since the multiple of $a_{ivt}$ and $a_{jvt}$ becomes 1 only when firm i and firm j both occupy the same niche and is otherwise 0, the numerator gives the number of niches co-occupied by firm i and j. That divided by the sum of $a_{ivt}$ from v = 1 to v = p, or the number of niches occupied by firm i, denotes the proportion of firm i’s niche that is occupied by another affiliate firm j. Likewise, in Formula (2), ${\alpha }_{jit}$ denotes the proportion of affiliate j’s niche that is occupied by another affiliate i at time t. Therefore, both $a_{ijt}$ and $a_{jit}$ are bounded by zero and one: at zero, the two firms are completely differentiated; at one, one firm fully occupies the other’s niche. Niche overlap for affiliate i in this paper was calculated by the sum of all $a_{ijt}$ (j = 1, 2, 3,…, N, i ≠ j) divided by N-1. This variable is mean-centered to avoid multicollinearity.

6.2.2. Control Variables

• Affiliate size: I control for affiliate size because it may potentially affect decoupling, although the direction of its effect is not certain. While larger firms may have more resources to pursue diversification compared to smaller firms, they may at the same time may prefer to stay within current business areas due to structural inertia [22]. I measured affiliate size by the amount of assets, as is often done by studies in strategic management [65,66]. While firm sales or market capitalization may also be used to operationalize firm size, Dang, Li and Yang [67] suggest that total assets are the best proxy for measuring a firm’s resources. Asset data is given in Korean Won in the data source, and I divided the figure by 1 trillion to avoid making the coefficients too small to present in the results table.
• Affiliate age: Affiliate age was measured by years elapsed since the founding of each affiliate. Affiliate age is included amongst the control variables for reasons similar to affiliate size. Considering the liability of newness [68], younger firms may experience greater difficulties in pursuing diversification than older firms. Older firms generally have more resources and legitimacy to attempt new resource-consuming activities. On the other hand, considering structural inertia, which usually increases with age [22], younger firms may pursue more aggressive diversification, thus increasing the decoupling of their stated and actual business areas.
• Affiliate performance: The statistical models in this study control for the affiliate performance to account for the possibility that slack resources may facilitate firms to pursue more diversification. According to organizational learning literature, slack resources facilitate exploration through the mechanism of slack search [69,70], and slack search may increase chances of decoupling of stated and actual business areas. Affiliate performance was measured by the value of the returns on assets (ROA) for each affiliate.
• Affiliate cash: I also control for the amount of cash each affiliate holds for reasons similar to affiliate performance. Cash is the most convenient form of organizational slack that can immediately cover search costs, and thus the amount of cash may increase affiliates’ tendency to explore outside their stated business areas. The amount of cash is again divided by 1 trillion to avoid making the coefficients too small to present in the results table.
• Business-group size: Again, business-group size was controlled for the same logic applied to affiliate size. Larger business groups may have more resources and legitimacy to support diversification of their affiliates, and yet these larger groups may have more structural inertia to stay within their existing business areas. Business-group size was measured by the business group’s assets, and the values are divided by 1 quadrillion to avoid making the coefficients too small to present in the results table.
• Business-group affiliates: The population ecology literature emphasizes density as an indicator of the level of competition within a niche [71]. Higher density within an organizational niche usually implies a higher level of competition within the given niche, as firms compete for limited markets and resources. Similar logic applies to the density of affiliates within a business-group because they compete for attention and resource allocation from the business-group headquarter. The more affiliates there are to compete with each other, the less resources each affiliate may obtain to use for diversification. Therefore, I control for the number of affiliates belonging to the same business group.
• Industry size: I controlled for industry size in order to take into account the effect of industry munificence. In industries marked by high returns or a broad spectrum of activities, firms in general may have more slack resources and opportunities to pursue expansion, thus affecting the decoupling of stated and actual business areas. Industry performance was measured by the assets of each industry, and the values are divided by 1 trillion to avoid making the coefficients too small to present in the results table.
• Industry intangible resources: Some industries require more intangible resources in order to stay competitive, while others require less. Since relative resource endowment, one of the independent variables, includes R&D expenditure in its calculation, the effect of industry-wide tendency to invest in R&D and advertisement should be controlled for in order to estimate the accurate impact of relative resource endowment within the business group. Otherwise, affiliates in high-tech industries would consistently appear as if they are endowed with more resources than non-high-tech affiliates. The industry intangible resource requirement is measured by the sum of R&D expenditure and advertisement expenditure at the industry level. The values are divided by 1 trillion to avoid making the coefficients too small to present in the results table.
• Industry profitability: Similar to industry size, industry profitability is another important indicator of industry munificence. I measured industry profitability using net income of each industry, and the values are divided by 1 trillion to avoid making the coefficients too small to present in the results table.
• Industry density: The degree of competition within an industry may affect the decoupling of firms belonging to that industry, however the direction is uncertain. A high level of industry competition may inhibit firms from diversifying due to the lack of slack in these industries. On the other hand, higher competition within an industry may motivate firms in that industry to venture out of the industry by exploring other business areas. The values are divided by 1000 to avoid making the coefficients too small to present in the results table.
• Industry support policies: In contrast to industry regulation policies, governments can also announce policies that support certain industries. In order to control for the offsetting effect of support policies on the effect of regulative policies, I control for the number of industry support policies made each year in the affiliate firm’s industry, and the values are divided by 1000 to avoid making the coefficients too small to present in the results table.
• Year dummies: To take into account the unobserved heterogeneity across years, year dummies are included in the statistical models.

7. Results and Discussion

I ran a set of Ordinary Least Squares regressions for hypothesis testing. Since the data matrix contains panel data with repeated observations of given firms over 13 years, a fixed-effect treatment was employed to control for the firm-specific unobserved heterogeneity.

Table 1. Descriptive statistics.

Variable	Obs	Mean	Std. Dev.	Min	Max
Decoupling	2620	0.555	0.339	0	1
Industry regulations	2620	0.220	0.706	0	5
Relative resource endowment	2620	0.044	0.122	0.000	0.984
Within-group niche overlap	2620	0.136	0.265	0	1
Affiliate size (mil.)	2620	99,000	213,000	60.8881	2,690,000
Affiliate age	2620	25.336	12.087	0	104
Affiliate performance	2620	0.002	0.101	−1.573	1.261
Affiliate cash (mil.)	2620	30,400	69,600	3.2042	1,680,000
Business-group size (mil.)	2620	12,900,000	17,600,000	114,767	87,200,000
Business-group affiliates	2620	20.076	12.129	1	51
Industry size (mil.)	2620	14,300,000	20,100,000	5153	108,000,000
Industry intangible resources (mil.)	2620	114,577	219,427	1	1,731,952
Industry profitability (mil.)	2620	−41,594	1,163,218	−20,300,000	6,722,022
Industry density	2620	181.202	99.203	3	331
Industry support policies	2620	0.004	0.009	0	97

summarizes the descriptive statistics. The average size of a business group is 12.9 trillion Korean won, and the average size of a business group affiliate is 99 billion Korean won, showing that the firms included in this analysis are large in scale. The mean affiliate age is 25.336, which also shows that the analyzed firms tend to be established firms. We can also infer from the mean value of 0.555 for decoupling that many firms derive more than half of their sales from industries not conveyed by their officially stated SIC code.

Table 2. Correlations.

	Variable	1	2	3	4	5	6	7	8
1	Decoupling	1
2	Industry regulations	−0.029	1
3	Relative resource endowment	0.050	−0.011	1
4	Within-group niche overlap	−0.195	0.007	−0.074	1
5	Affiliate size	−0.001	0.062	0.109	−0.009	1
6	Affiliate age	0.038	0.041	0.146	−0.053	0.226	1
7	Affiliate performance	−0.042	−0.029	−0.063	0.000	−0.044	−0.066	1
8	Affiliate cash	−0.028	0.072	0.068	0.056	0.850	0.208	−0.081	1
9	Business group size	−0.028	0.021	−0.082	0.022	0.442	0.057	0.010	0.395
10	Business group affiliates	−0.044	−0.015	−0.157	0.064	0.189	−0.055	0.097	0.176
11	Industry size	−0.163	0.158	−0.029	0.216	0.223	0.154	−0.033	0.265
12	Industry intangible resources	−0.062	0.068	0.163	−0.016	0.275	0.010	−0.020	0.206
13	Industry profitability	0.069	−0.040	0.007	−0.024	−0.164	−0.088	0.180	−0.197
14	Industry density	−0.113	0.038	0.021	0.087	0.050	0.034	0.015	0.086
15	Industry support policies	0.005	−0.074	0.047	−0.018	0.082	0.009	−0.025	0.058
	Variable	9	10	11	12	13	14	15
9	Business-group size	1
10	Business-group affiliates	0.643	1
11	Industry size	0.175	−0.008	1
12	Industry intangible resources	0.203	0.122	0.340	1
13	Industry profitability	−0.136	0.002	−0.121	0.062	1
14	Industry density	−0.004	−0.013	0.201	0.137	0.010	1
15	Industry support policies	0.075	0.017	0.019	0.132	0.025	−0.008	1

shows correlations between the variables. None of the correlation values are alarmingly high, except for the correlation between affiliate size and affiliate cash. I checked the variance inflation factor (VIF) to confirm that multicollinearity is not an issue in the statistical models. The mean VIF is 1.68, with the highest value being 4.07 for the affiliate size variable. Given that VIF values less than 10 are generally accepted [72], the variables used in this study are relatively free from the multicollinearity problem.

Table 3. Ordinary Least Squaresregression results: decoupling.

	Model 1		Model 2		Model 3		Model 4		Model 5		Model 6
Independent Var.
Industry regulations			0.009	*					0.01	*	0.01	*
			(0.005)						(0.005)		(0.005)
Relative resource endowment					0.076	*			0.074	*	0.108	**
					(0.033)				(0.033)		(0.038)
Within-group niche overlap							−0.094	***	−0.093	***	−0.084	***
							(0.020)		(0.020)		(0.020)
Relative resource endowment x											0.38	†
within-group niche overlap											(0.207)
Control Var.
Affiliate size	0.003		0.004		0.008		−0.001		0.005		0.008
	(0.032)		(0.032)		(0.032)		(0.032)		(0.032)		(0.032)
Affiliate age	0.002		0.001		0.002		0.002		0.001		0.001
	(0.001)		(0.001)		(0.001)		(0.001)		(0.001)		(0.001)
Affiliate performance	−0.012		−0.011		−0.012		−0.013		−0.013		−0.014
	(0.026)		(0.026)		(0.026)		(0.026)		(0.026)		(0.026)
Affiliate cash	0.167	†	0.156	†	0.160	†	0.166	†	0.149		0.151	†
	(0.010)		(0.092)		(0.092)		(0.091)		(0.091)		(0.091)
Business group size	−0.010		0.017		0.029		−0.021		0.047		0.058
	(0.331)		(0.331)		(0.331)		(0.329)		(0.330)		(0.329)
Business group affiliates	−0.001	*	−0.001	*	−0.001	†	−0.001	†	−0.001	†	−0.001	†
	(0.001)		(0.001)		(0.001)		(0.001)		(0.001)		(0.001)
Industry size	−0.001	***	−0.001	***	−0.001	***	−0.001	***	−0.001	***	−0.001	***
	(0.000)		(0.000)		(0.000)		(0.000)		(0.000)		(0.000)
Industry intangible resources	0.055	**	0.056	**	0.051	*	0.054	**	0.050	*	0.052	**
	(0.019)		(0.019)		(0.020)		(0.019)		(0.019)		(0.019)
Industry profitability	0.003		0.003		0.003		0.003		0.004		0.004
	(0.002)		(0.002)		(0.002)		(0.002)		(0.002)		(0.002)
Industry density	−0.067	*	−0.068	*	−0.066	*	−0.065	*	−0.064	*	−0.065	*
	(0.031)		(0.031)		(0.031)		(0.031)		(0.031)		(0.031)
Industry support policies	0.345		0.389		0.367		0.360		0.426		0.441
	(0.369)		(0.369)		(0.369)		(0.367)		(0.367)		(0.367)
Constant	0.542	***	0.553	***	0.538	***	0.545	***	0.551	***	0.556	***
	(0.035)		(0.036)		(0.035)		(0.035)		(0.035)		(0.035)
R-Squared: Within	0.029		0.031		0.031		0.039		0.043		0.044
R-Squared: Between	0.012		0.012		0.013		0.022		0.022		0.021
R-Squared: Overall	0.020		0.020		0.020		0.046		0.045		0.043
Observations	2620		2620		2620		2620		2620		2620
Number of affiliates	314		314		314		314		314		314

Standard errors in parentheses. *** p < 0.001, ** p < 0.01, * p < 0.05, † p < 0.1. Year dummies included.

summarizes the results of the fixed-effect OLS regressions. Model 1 is the baseline model, including only the control variables. Some of the control variables have a statistically significant effect on decoupling. As expected, affiliate cash has a positive effect on decoupling, implying that the amount of cash increases affiliates’ tendencies to explore outside their stated business areas. The number of affiliates within the same business group has a negative effect on decoupling, probably because when many affiliates compete for resource, each affiliate would end up with less resources to use for diversification. Model 1 also shows that firms positioned in large and dense industries are less likely to pursue business opportunities outside their officially-stated area. In contrast, firms positioned in industries with a high level of intangible resources are more likely to explore other business areas, probably because the accumulation of intangible assets enables firms to exploit these assets in other markets.

Models 2, 3, and 4 show the results when the industry regulation variable, relative resource endowment, and within group niche overlap are separately entered to check their main effects. Model 5 tests the main effects of all three independent variables when they are entered together in the model. Model 6 is the fully specified model, including all independent variables and the interaction term, “relative resource endowment x within-group niche overlap”, to test the moderating effect of relative resource endowment on the relationship between within-group niche overlap and decoupling.

In all models, including industry regulations, this independent variable constantly shows a statistically significant positive effect on decoupling, supporting Hypothesis 1. As expected, affiliate firms in industries with more government interventions were more likely to have actual business areas decoupled from the stated SIC code. These firms may have chosen to explore other business areas to avoid government interventions in their stated business areas.

The independent variable relative resource endowment is also positive and significant. As hypothesized in Hypothesis 2, affiliates that hold relatively more important positions within the business groups are more likely have their actual business areas decoupled from their stated business areas. Such result supports my prediction that affiliates with higher relative resource endowment take the role of a seedbed for new business creations in the business groups.

Hypothesis 3 predicts a negative relationship between the within business-group niche overlap and decoupling, and the results support this hypothesis. The results indicate that niche overlaps between affiliates within a business group motivate the group to avoid decoupling of actual and stated business areas in the subsequent period.

Model 6 shows that the effect of the interaction between relative resource endowment and within-group niche overlap is positive, even though the statistical significance is weak (p = 0.056). Therefore, Hypothesis 4 receives weak support, showing that the negative effect of niche overlap on decoupling was found to become somewhat weaker with an increase in the relative resource endowment of the affiliate firm. Such a result implies that, while business groups wish to reduce decoupling between actual and stated business areas when there is too much niche overlap between affiliates, the affiliates with more resource endowment are less likely to be the target of the rationalization attempt.

The statistical analysis of the data supported all of the hypotheses, except for Hypothesis 4 which received only weak support (p < 0.1). General implications of empirical findings are discussed further in the following section.

8. Conclusions

In this paper, I examined sources of decoupling of the stated business areas and actual business activities in business group affiliates. The findings here illustrate that what firms actually do cannot be directly inferred from their stated SIC codes due to the complex dynamics of decoupling, which this study viewed as a strategic response to institutional and strategic imperatives. The findings from the empirical study suggest that decoupling of the stated and actual business areas of a business group affiliate is affected positively by government regulations on the industry, positively by the strategic importance of the affiliate within the business group, and negatively by the affiliate’s niche overlap with other affiliates within the group. I also found that the relative resource endowment of the affiliate reduces the negative effect of niche overlap on decoupling, although this hypothesis was only weakly supported.

Throughout this paper, I have underlined the functional and strategic aspects of identity ambiguity, a concept usually seen as a sign of inefficiency and disorder. I agree with those theoretical traditions which have viewed the modern organization as a contradictory and paradoxical entity, such as ‘organized anarchy’ [73], a ‘loosely coupled system’ [74], and a collection of ‘myths and ceremonies’ based on ‘collective rationality’ [5,11]. This study’s findings may have the following implications for the fields of organization studies and for policy makers.

First, this study suggests that decoupling is a highly important and promising concept to which organization theorists should pay more attention. Since decoupling is prevalent in organizational life, the understanding of its dynamics is crucially important when studying organizations. This study emphasizes decoupling as a strategic choice, arguing that organizations may deliberately choose decoupling in order to deal with institutional pressures or strategic circumstances. Such findings may be extended in the future by incorporating insights from various classical and contemporary sociological theories, such as impression management [75], symbolic interaction [76], and practical action [77].

Second, this paper may contribute to the development of the decoupling literature by showing one way to measure this highly abstract theoretical concept. Although the intriguing concept of decoupling was pioneered by neo-institutional theorists, it has rarely been examined through empirical studies. One of the reasons that studies on decoupling were so limited is the difficulty of measuring the concept. Although decoupling was examined by a number of theoretical and qualitative studies [6,7], the generalizability of their findings was limited due to the lack of large-sample, quantitative studies. This paper may contribute to the future development of the decoupling literature by introducing a new way of measuring decoupling in organizations through the use of SIC codes.

In addition, the use of SIC codes has practical implications for academic researchers and policy makers. Findings from the analysis of the decoupling of officially announced SIC codes and actual business areas warn against over-confidence in formal classification schemes. Existing empirical studies in organizational ecology, strategic management, and IO economics rely heavily on SIC codes in the classification of organizational populations [42,78,79]. SIC codes are also frequently used as a basis for formulating industrial policies [31,78]. However, this study shows that formal SIC codes are merely a representation, not a precise reflection, of the organizational reality.

Another policy implication of this study is that government regulation may have unintended consequences in the behavior of firms. Governments may announce new regulations in an industry when there is a need to prevent, change or redirect certain firm behaviors in the industry [38,39]. However, the findings from this study imply that firms under such institutional pressures may choose to escape the industry rather than conforming to the pressures, and explore business areas outside their stated business area. Therefore, policy makers should take into account how new industrial regulations may have the unintended consequences of firms fleeing the affected industry.

Despite the above-mentioned potential contributions, this study has limitations that should be addressed by future studies. First, the statistical models used in this study may have endogeneity issues, since there are many variables that are interrelated with each other in complex ways, and there also may be omitted variables affecting variables and possibly the error term. To mitigate this issue, the models employ fixed-effects and various control variables. Using fixed-effects solves the problem of unobserved heterogeneity between different firms. I also added year fixed-effects to deal with endogeneity in time-variant variables. In addition, the models control for several variables at three different levels—affiliate, business group, and industry—to alleviate problems arising from omitted variables. Li’s [80] study on various methods of dealing with the endogeneity problem suggests that the combination of firm fixed effects, year fixed effects and meaningful control variables mitigates the endogeneity issue. Yet, I acknowledge that there may be remaining endogeneity issues in this paper that may be better resolved by future studies employing more sophisticated econometric methods.

Second, findings from this study are limited in their generalizability. The empirical setting of this research was large business groups in Korea, known as chaebols. Although Korean chaebols are one the best-known examples of business groups [10], they still may not be the most representative type of this special organization form. Indeed, chaebols are noted as the most patrimonial example of this particular organizational form compared to different types of business groups, such as Keiretsus in Japan and the Guanxiqiye networks in Taiwan [10,29]. Therefore, readers should exercise caution in generalizing the findings from this study to business groups from countries other than South Korea. Future studies should comparatively analyze the dynamics of business groups from diverse and different cultural backgrounds.