The impact of industry characteristics on vertical integration strategies.

ABSTRACT

The purpose of the paper is to examine the impact of industry characteristics on vertical integration strategies. The empirical test was based on a framework that included two vertical integration strategies (levels of vertical integration strategies and trends of vertical integration) and three industry characteristics (competition, uncertainty, and industry's average company size). The statistical analysis of a sample of 316 manufacturing companies resulted in the findings that the levels of vertical integration were influenced by all industry characteristics considered while the trends of vertical integration were explained only by industry uncertainty.

1. INTRODUCTION

Vertical integration can be defined as the number of activities along the value chain that are performed within a single company. Vertical integration is an attractive strategy if its benefits (such as low transaction costs and increased market power) outweigh its costs (managing several value chain activities within in single organization). However, the extent to which a company can realize these benefits is heavily dependent upon industry factors such as competition and uncertainty. Since these factors vary across industries, vertical integration strategies have been linked to industry-specific conditions (Chandler, 1962, 1990; Livesay & Porter, 1969; Lieberman, 1991; Tucker & Wilder, 1977).

As suggested by several researchers (e.g. Chandler, 1990; Quinn, Doorley, & Paquette, 1990; Stuckey & White, 1993), industry factors change overtime. If industry factors change, it can reasonably be implied that vertical integration strategies would also change (D'Aveni & Ravenscraft, 1994; Harrigan, 1986; Livesay & Porter, 1969; Stuckey & White, 1993). However, because the way industry factors change may be dependent upon each industry, it is likely that the resulting change in vertical integration will also be industry-specific.

The argument that vertical integration is industry-specific can be explained by institutional theory (e.g. DiMaggio & Powell, 1983; Scott, 1987). According to this theory, a successful strategy will be considered legitimate in the "field" of the companies that have adopted that strategy. For example, if a few firms from an industry successfully change their strategy by performing vertical disintegration, vertical disintegration may subsequently become a legitimate strategy that would spread among other firms in the same field or industry. The purpose of this paper is to empirically test the argument that vertical integration strategy is industry-specific.

2. HYPOTHESES

The framework that will be used to derive the hypotheses of the study includes two vertical integration strategies and three industry characteristics. Vertical integration can be discussed in terms of the levels of vertical integration as well as the trends of vertical integration. While the levels reflect the extent to which a company is vertically integrated, the trends indicate how the extent of a company's vertical integration levels change over time.

The working proposition of this research is that vertical integration strategies can be predicted by industry characteristics. To understand how industries can affect vertical integration strategies, it is necessary to identify their relevant characteristics. Several factors can be used to categorize industries. This study focuses on three industry characteristics: competition, uncertainty, and size. Both the traditional economic literature (e.g. Auerbach, 1988) and the economic history research (Chandler, 1962, 1990) have suggested that the intensity of competition in an industry has a major impact on vertical integration strategies that firms adopt. This suggestion was supported by a number of empirical studies conducted in the strategy research (Balakrishnan & Wernerfelt, 1986; Harrigan, 1986). Given the links that exist between industry competition and vertical integration, this study will formulate the following hypotheses:

Hypothesis 1. A firm's levels of vertical integration will be predicted by industry competition.

Hypothesis 2. A firm's trends of vertical integration will be predicted by industry competition.

The second industry factor is uncertainty. Researchers agree that uncertainty is a significant determinant of vertical integration (Balakrishnan & Wernerfelt, 1986; Harrigan, 1986; Miles & Snow, 1986). For example, when an industry's uncertainty is high, being in several vertically integrated businesses would adversely affect flexibility, as no company can excel in several activities at the same time. So in highly uncertain industries, companies will concentrate on their core activities where they can achieve high flexibility and rapid adaptation (Miles & snow, 1986), and consolidate technologies and skills into core competencies (Prahalad & Hamel, 1990). Thus it can be hypothesized that:

Hypothesis 3. A firm's levels of vertical integration will be predicted by industry uncertainty

Hypothesis 4. A firm's trends of vertical integration will be predicted industry uncertainty

Size can also affect vertical integration strategies. According to the size literature (e.g. Blau, 1970; Blau, Falbe, McKinley, & Trace, 1976), as size increases, the span of control of each supervisor widens, leading to economies in managerial supervision. Thus companies tend to seek cost advantages by growing in size. Since growth strategies include increasing the levels of vertical integration, it can be argued that as compared to industries with small companies, industries with large companies would be characterized by higher levels of vertical integration. This relationship leads to the following hypotheses:

Hypothesis 5. A firm's levels of vertical integration will be predicted by industry's average company size

Hypothesis 6. A firm's trends of vertical integration will be predicted by industry's average company size

3. VARIABLES

The dependent variables are levels and trends of vertical integration. In studies of vertical integration, measurement is the most controversial issue. This paper measured vertical integration by the ratio of the number of vertically related businesses over the total number of businesses. This definition was operationalized using an instrument proposed by Davis and Duhaime (for detail of this instrument, see Davis & Duhaime, 1992). Based on this instrument, an index of vertical integration was estimated as the ratio of the number of integrated segments over the total number of segments. For example, if a firm has 5 segments, and two of those segments are integrated, the ratio of vertical integration would be 0.40 (2 over 5).

Using this formula, an index of vertical integration was computed for each of the 18 years that are were covered by this study (1980-1997). From this index, two statistics were calculated. First, a company's level of vertical integration was obtained by averaging 18 annual indices of vertical integration. Second, a company's trend of vertical integration was determined by the trend of annual indices over the 18-year period. A time series analysis was performed on the yearly indices of vertical integration to identify each company's trend of vertical integration.

The independent variables are industry characteristics competition, uncertainty and size. Competition was measured as a 20-firm concentration ratio. The scores assigned to industry competition were 1 (low competition), 2 (average competition), and 3 (high competition). Following Harrigan (1985), uncertainty was measured as the growth rate of sales in an industry. The scores for uncertainty were 1 (low uncertainty), 2 (average uncertainty), and 3 (high uncertainty). Finally, size was measured as the average number of employees for companies in an industry. The scores were 1 (small size), 2 (average size), and 3 (large size).

4. DATA COLLECTION AND ANALYSIS

The empirical analysis was based on a sample of 316 companies selected in the manufacturing sector. These companies were randomly selected among parent companies that were in existence from 1980 to 1997. The source of data was Compustat. Compustat provides the information on each company's SIC code and business segments. Two statistical techniques were used to conduct the empirical tests, ANOVA and the Pearson Chi-Square.

The levels of vertical integration ranged from a minimum of 0 (no business segment was vertically integrated) to a maximum of 1 (all business segments were vertically integrated). The mean of all companies' average levels of vertical integration was 0.4429. 155 companies were above this mean, and 161 companies were below. The analysis of the trend of vertical integration indices over 18 years resulted in two categories of companies, each with two distinct groups (table 1).

Interestingly, the sample was evenly divided between the companies that did not change their levels of vertical integration and the companies that changed their levels. The category of 158 companies that did not change their vertical integration levels included 70 companies that kept the same level during the 1980-1997 period, and 88 companies whose levels varied slightly, but the change was not statistically significant. In the category of 158 companies that significantly changed their vertical integration levels, 74 companies increased their vertical integration levels, but 84 companies vertically disintegrated by decreasing their vertical integration levels.

The analysis of mean differences presented in table 2 shows that overall the four groups had different levels of vertical integration (F = 9.463 at p = .000). When individual means are compared from one another (using the Bonferroni test), the mean for the group of companies that had the same levels of vertical integration over the period (0.2904) was significantly different from the means of the other groups (difference was significant at respectively p = .000, p = .000, and p = .029). The means of the other three groups were not statistically different among them.

The companies that maintained the same levels of vertical integration over the entire period had the lowest levels of vertical integration, and many of these companies were not at all integrated (their average level was 0). The companies that increased their levels of vertical integration also had the highest levels, closely followed by companies that slightly changed (change not significant) their levels of vertical integration.

5. HYPOTHESIS TESTING

The paper developed six hypotheses. Hypotheses 1, 3 and 5 stated that the levels of vertical integration would be predicted by respectively industry competition, industry uncertainty, and industry's average company size. To test these hypotheses, ANOVA was performed because the dependent variable (vertical integration levels) was continuous while the independent variables (industry characteristics) were categorical. As can be seen in table 3, hypotheses 1, 3 and 5 were supported. This suggests that companies' levels of vertical integration were dependent on how competitive and uncertain industries were, and on the size of companies in their respective industries. Therefore, the levels of vertical integration were dependent on industry factors.

Hypotheses 2, 4 and 6 suggested that vertical integration trends would be predicted by industry competition, uncertainty, and average company size. To test these hypotheses, Chi-Square was used because the dependent variable (vertical integration trends) and the independent variables (industry characteristics) were all categorical. Table 3 shows that uncertainty was the only industry characteristic that determined the trends of vertical integration (hypothesis 4). Competition (hypothesis 2) and size (hypothesis 6) did not affect vertical integration trends. So, the trends of vertical integration were partially dependent on industry factors.

6. CONCLUSION AND IMPLICATIONS

The purpose of the paper was to empirically examine whether vertical integration strategies were industry-specific. Vertical integration levels were explained by all three industry characteristics considered. However, vertical integration trends were explained by only one characteristic. The conclusion that can be drawn from these findings is that vertical integration is mostly industry-specific. The results suggest that if a company's industry is known, it will be possible to speculate whether the company's vertical integration levels would be higher or lower.

Since however most industry characteristics do not explain vertical integration trends, it can be implied that some vertical integration decisions are perhaps made based on managerial choice, beyond simply responding to the requirements from a company's industry. Therefore, as argued by Hrebiniak and Joyce (1985), it is safe to say that theories founded on environmental determinism (e.g. institutional theory, industrial organization) as well as those based on strategic choice (e.g. resource-based approach) are all relevant to how organizational decisions are made.

This study can help practicing managers in their vertical integration decisions. Specifically, when trying to determine the appropriate levels of their company's vertical integration, managers should attempt to align the levels to the average levels in their industry. However, managers should also refer more to their organization's characteristics when it comes to the decision of how to change their vertical integration levels, as these decisions are less affected by industry factors.

TABLE 1. DISTRIBUTION OF COMPANIES
BY VERTICAL INTEGRATION TREND *

Trends of vertical integration                # Companies   Frequencies
                                                            (%)

            Same levels                       70            22
No change   No significant change in levels   88            28
            Sub-total                         158           50

            Increase in levels                74            23
Change      Decrease in levels                84            27
            Sub-total                         158           50

TOTAL COMPANIES                               316           100

* The significance level considered in
the time series was p = .05

TABLE 2. ANALYSIS OF MEAN DIFFERENCES *

Group of companies                    Mean

Same levels of vertical integration   0.2904
No significant change in levels       0.5153
Increase in levels                    0.5185
Decrease in levels                    0.4275

F (9.463) was significant at p = .000

* ANOVA was used for mean comparison

TABLE 3. HYPOTHESIS TESTING RESULTS

HYPOTHESES     F-value   Chi-Square   Significance level

Hypothesis 1   5.777     --           P = .003
Hypothesis 2   --        5.229        P = .515
Hypothesis 3   8.741     --           P = .000
Hypothesis 4   --        16.848       P = .010
Hypothesis 5   3.042     --           P = .049
Hypothesis 6   --        3.466        P = .749

7. REFERENCES

Auerbach, P., Competition. The Economics of Industrial Change, Basil Blackwell Ltd., Oxford, UK, 1988.

Balakrishnan, Srinivasan and Wernerfelt, Birger, "Technical Change, Competition and Vertical Integration", Strategic Management Journal, Vol. 7, 1986, 347-359.

Blau, P.M., "A Formal Theory of Differentiation in Organizations", American Sociological Review, Vol. 35, 1970, 201-218.

Blau, P.M., Falbe, C.M., McKinley, W., and Tracy, P.K., "Technology and Organization in Manufacturing", Administrative Science Quarterly, Vol. 21, 1976, 20-40.

Chandler, Alfred D., Strategy and Structure: Chapters in the History of the American Industrial Enterprise, The Belknap Press, Cambridge, MA, 1962.

Chandler, Alfred D., Scale and Scope, The Belknap Press, Cambridge, MA, 1990.

D'Aveni, Richard A. and Ravenscraft, David J., "Economies of Integration versus Bureaucracy Costs: Does Vertical Integration Improve Performance?", Academy of Management Journal, Vol. 37 (5), 1994, 1167-1206.

Davis, Rachel and Duhaime, Irene M., "Diversification, Vertical Integration, and Industry Analysis: New Perspectives and Measurement", Strategic Management Journal, Vol. 13, 1992, 511-524.

DiMaggio, P.J. and Powell, W.W., "The Iron Cage Revisited: Institutional Isomorphism and Collective Rationality in Organizational Fields", American Sociological Review, Vol. 48, 1983, 147-160.

Harrigan, Kathryn R., "Vertical Integration and Corporate Strategy", Academy of Management Journal, Vol. 28, 1985, 397-425.

Harrigan, Kathryn R., "Quick Change Strategies for Vertical Integration", Planning Review, Vol. 14, 1986, 32-37.

Hrebiniak, Lawrence G. and Joyce, William F., "Organizational Adaptation: Strategic Choice and Environmental Determinism", Administrative Science Quarterly, Vol. 30, 1985, 336-349.

Lieberman, Marvin B., "Determinants of Vertical Integration: An Empirical Test", The Journal of Industrial Economics, Vol. 39 (5), 1991, 451-466.

Livesay, Harold C. and Porter, Patrick G., "Vertical Integration in American Manufacturing, 1899-1948", The Journal of Economic History, Vol. 29, 1969, 494-500.

Miles, Raymond E. and Snow, Charles C., "Organizations: New Concepts for New Forms", California Management Review, Vol. 28 (3), 1986, 62-73.

Prahalad, C.K. and Hamel, Gary, "The Core Competence of the Corporation", Harvard Business Review, May-June 1990, 79-91.

Quinn, James B., Doorley, Thomas L., and Paquette, Penny C., "Technology in Services: Rethinking Strategic Focus", Sloan Management Review, Winter 1990, 79-87.

Scott, W.R., Organizations: Rational, Natural and Open Systems, Prentice-Hall, Englewood Cliffs, NJ, 1987.

Stuckey, John and White, David, "When and When not to Vertically Integrate", Sloan Management Review, Vol. 34, 1993, 71-83.

Tucker, Irvin B. and Wilder, Ronald P., "Trends in Vertical Integration in the US Manufacturing Sector", The Journal of Industrial Economics, Vol. 26, 1977, 81-94.

Richard T. Mpoyi, Middle Tennessee State University