Continual Corporate Entrepreneurial Search for Long-Term Growth

[For Management Science’s Focused Issue on Entrepreneurship]

Continual Corporate Entrepreneurial Search for Long-Term Growth *

Gaurab Bhardwaj Assistant Professor of Strategy and Management 229 Tomasso Hall Babson College Babson Park, MA 02457 781-239-5701, [email protected], 781-239-5272 Fax

John C. Camillus Donald R. Beall Professor of Strategic Management Katz Graduate School of Business 338A Mervis Hall University of Pittsburgh Pittsburgh, PA 15260 412-648-1599, [email protected]

David A. Hounshell David M. Roderick Professor of Technology and Social Change Departments of History, and Social & Decision Sciences Porter Hall 219G Carnegie-Mellon University Pittsburgh, PA 15213 412-268-3753, [email protected]

August 2005

* This paper is based upon fieldwork supported by the National Science Foundation under grant SBR-9872070, and by a grant from the Eleutherian Mills-Hagley Foundation. We thank Rajshree Agarwal, Nick Argyres, Glen Asner, Jay Barney, Jennifer Bethel, Bala Chakravarthy, Olivier Chatain, Raghu Garud, Dhruv Grewal, Sam Hariharan, Connie Helfat, Kate McKone-Sweet, Kent Miller, Heidi Neck, Srini Rangan, Miguel Rivera, Keith Rollag, and Shaker Zahra for helpful comments in early stages of this research. The paper benefited from comments by the focused issue conference participants, and especially two anonymous reviewers and focused issue editor Scott Shane.

Continual Corporate Entrepreneurial Search for Long-Term Growth Abstract We examine how established firms conduct continual entrepreneurial search for possibilities for long-term growth. Drawing on comprehensive internal documents of the DuPont Company over a twenty-year period, we develop a search process that is a departure from frequent depictions of search as local or random. Longitudinal field data show that corporate entrepreneurs follow a “moving, anchored search” for growth possibilities. Employing this framework as lens, we develop propositions. We find that corporate entrepreneurs are more likely to conduct search in new domains following events that cause them to expect future performance to change significantly and lastingly. This is in contrast to the literature that has typically modeled the initiation of search as a response to poor past performance. Because new domains are unexplored territories for corporate entrepreneurs, they utilize transitional levers that they perceive will facilitate the move from existing domains to new ones. These perceived transitional levers, however, typically prove inaccurate or incomplete. Content within domains is searched using anchors whose locations and numbers change. The combination of search process and content searched influences the particular growth possibilities discovered and created. Search and pursuit of growth possibilities is accompanied by the creation of new knowledge and new capabilities.

KEYWORDS Corporate Entrepreneurship Opportunities Search Long-Term Growth Search Process

i

1. Introduction When viewed over long spans of time, companies display puzzling arrays of new businesses and innovations in their continual entrepreneurial attempts at long-term growth. Over 140 years, Nokia grew and transformed itself from a forestry company to a leading mobile phone company. Between these two far-apart businesses lay others: rubber works, cable works, electric power, rubber boots, toilet paper, winter tires, and pagers (Steinbock 2001). Over 58 years, Wipro, a seller of vegetable oil and soap transformed itself into a leading IT software and services company while also entering consumer and baby products, hydraulic equipment, and medical systems (BusinessWeek 2003). Similarly varied patterns in continual corporate entrepreneurship can be found over long periods in the growth of Xerox, Corning, DuPont, and GKN (Chesbrough 2002, Dyer and Gross 2001, Hounshell and Smith 1988, Wall Street Journal 2004), and in other companies from different industries, nations, and periods (e.g., Arora et al. 1998, Chandler 1990). The patterns described are of new businesses pursued, but entrepreneurial decision making for long-term growth is even more varied and compelling when innovations and new businesses considered but eventually not pursued are included along with those that were (Table 1). Despite the prevalence and importance of such puzzling patterns in corporate entrepreneurship that is continual and spans long periods, the phenomenon has received little academic attention. These patterns result because the entrepreneurial pursuit of long-term growth involves distant returns (Bhardwaj 2000). Decisions have to be made in the face of high uncertainty and causal ambiguity. Circumstances keep changing and there is insufficient information, but decisions cannot be deferred until uncertainty is resolved. The chances of obtaining outcomes exactly as desired are remote. And outcomes and other relevant information emerge gradually over time. These distant probabilistic and temporal characteristics make continual corporate entrepreneurship for long-term growth complex and difficult. Possibilities for growth are not always obvious and given. Instead, they must be discovered and created – or searched (Baumol 1993, Simon 1976). Because the search involves ambiguity and uncertainty and spans long periods, it results in an array of growth possibilities that are hard to understand without knowing the

1

search process behind them. The possibilities subsequently pursued crucially impact firm growth and transformation (Holbrook et al. 2000, Noda and Bower 1996). Therefore, academics have defined entrepreneurship as the process of search and exploitation of opportunities, and have also concluded that the literature reveals little about how opportunities are actually discovered and created (Gaglio 1997, Shane and Venkataraman 2000, Venkataraman 1997). This lack of understanding is even greater regarding actual search processes for long-term growth in established firms. Hence, using field data from the DuPont Company that span twenty years, we develop process theory on this corporate entrepreneurship phenomenon. With this objective, our focus is on the search process, not on exploitation. The latter involves additional factors beyond the theory building scope of this paper. Further, we use the term ‘possibilities’ instead of the frequently used term ‘opportunities’ since threats also play a role in search. Moreover, of the possibilities created and discovered, some later emerge as opportunities but not others; and some may later result in new businesses while others are discarded.

1.1 Continual Corporate Entrepreneurial Search Research on corporate entrepreneurship has focused mainly on its antecedents and consequences (Covin and Slevin 1991, Zahra et al. 1999), and there have been few studies of corporate entrepreneurial search processes, despite their importance for outcomes (Gaglio 1997, Shane and Venkataraman 2000, Venkataraman 1997). Much more is known about search done by individual entrepreneurs, where their characteristics and cognitive and behavioral factors have been found to explain opportunity discovery and startup creation (Gaglio 1997, Shane 2000). But corporate entrepreneurship is different, especially when the aim is long-term growth and the time span is long. Search then tends to be a continual process of multiple, simultaneous efforts that engages many individuals. The literature would suggest that such search is likely to be enhanced by past entrepreneurial activities (Holcombe 2003), and influenced by past choices, strategic and structural contexts, sociopolitical and cognitive processes, existing knowledge and capabilities, and temporal economies of scope (Burgelman 1983, Chandler 1962, 1990; Helfat and

2

Eisenhardt 2004, Helfat and Raubitschek 2000, Holbrook et al. 2000, Noda and Bower 1996, Penrose 1995, Tripsas and Gavetti 2000). What we know less about is how this continual search for long-term growth is actually conducted over long periods and results in the puzzling patterns noted earlier. The widely used concepts in management of search as local or random have limited explanatory power for this phenomenon. In local search, the alternatives considered are marginally different from existing choice, and decision makers settle for a satisfactory rather than optimal solution (Simon 1976). If needed, aspirations are lowered until a local alternative becomes satisfactory (Cyert and March 1963). Local search is indeed be useful in explaining incremental entrepreneurial search over short periods, but it is unlikely to be helpful in looking at long periods that often see significant departures from the past, as illustrated earlier. Treating search as random has proved useful for purposes like modeling (Nelson and Winter 1982), but cannot explain actual decision making processes and the perspective of those making the decisions (Lovas and Ghoshal 2000). When applied to actual search data over long time spans (e.g., Table 1), local search cannot explain decision making well. And random search can account for possibilities considered only by ignoring the entrepreneurial process by which they were discovered. The few studies conducted on search over time clearly show that the processes are complex and go beyond local search. For instance, Ahuja and Lampert (2001) found from patents analysis that breakthrough inventions entailed going beyond local search to exploring novel, emerging, and pioneering technologies. And Garud and Van de Ven (1992) found that under different conditions of slack and ambiguity, firms show either trial-and-error or persistence in their search. There is also considerable evidence that different kinds of contexts have their own decision-making processes (Papadakis et al. 1998). Thus, we need to use field data to develop the continual corporate entrepreneurial search for long-term growth. Besides our focus on long-term growth, we depart from practices in the literature in two ways to better capture search realities in established firms. One, we look at all the multiple, simultaneous searches at a point in time, not just a select few internal corporate ventures. For completeness, this includes all growth possibilities discovered and created, not just those that later led to new businesses. Two, we look

3

at continual search over a long period. Besides the reasons noted, it is also because long-term growth in established firms is a target perpetually in the distant future. This is especially evident in research-driven companies. We are thus studying all the simultaneous searches pursued over a long time span. Due to our interest in how growth possibilities are actually discovered and created, our data, consistent with the literature, includes scientific research, innovation, new technologies, and new businesses deliberations (Ahuja and Lampert 2001, Baumol 1993, Burgelman 1983, Chandler 1991, Hounshell and Smith 1988, Shane 2000, Temin 1991, Tripsas and Gavetti 2000). In the next section we describe our research methods. Following that, we develop process theory and propositions. And in the final section we draw conclusions and suggest ideas for further research.

2. Archival Methods To develop process theory of continual corporate entrepreneurial search for growth possibilities, we adopted the case research method based on archival data because it is suited for studying processes that span long periods and under-researched complex phenomenon (Cuff 2002, Gaddis 2002, Eisenhardt 1989). The method requires using rich field data to develop theory, not test hypotheses based on a limited literature on the phenomenon; and using the research question to guide data collection from a theoretically relevant setting. More important than the number of cases is the richness of data used in developing theory (March et al. 1991, Tripsas and Gavetti 2000). Using an extensive archive of DuPont Company’s internal documents, and newspaper articles, we wrote a comprehensive case on corporate entrepreneurial search at the company from 1902 to 1921. We then generalized from a set of events in the case to develop process theory explanations and propositions (Garud and Van de Ven 1992). The modern DuPont Company traces its origins to February 26th, 1902 when an earlier family partnership was dissolved and a new entity incorporated (Chandler and Salsbury 1971). An executive committee, comprising more than a dozen senior managers, a few of whom were from the du Pont family, was formed to lead the 100-year old explosives company. In starting our study in 1902, there is little risk

4

of left-censoring data. In the next 20 years, entrepreneurial actions taken by the Executive Committee, managers, and scientists grew and transformed DuPont from a badly performing explosives company to the biggest and most profitable explosives company in the industry, and then to a giant multi-business enterprise, and set the foundations for continued growth (Chandler 1962, Chandler and Salsbury 1971, Hounshell and Smith 1988, McCormick and Folsom 2003). Ignoring the massive sales and profits from the First World War (WWI), this search helped raise sales from $26.1 million in 1904 to $84.6 million in 1925, and increase net income from $4.4 million in 1904 to $24.9 million in 1925 1 . In this period, DuPont faced all phases of the economic cycle and two external shocks – an antitrust lawsuit and WWI. DuPont permitted unrestricted access to its vast collection of internal documents, running into hundreds of linear feet of stacks, archived at the Hagley Museum and Library in Wilmington, Delaware. To understand continual corporate entrepreneurial search for long-term growth possibilities, we gathered comprehensive data from 1902-1921 on all scientific research done (whether the outcomes were useful or not), on all attempts at innovations (whether successful or not), on new technologies considered in relation to the previous two activities, and on all deliberations of new business possibilities (whether they later resulted in new business creation or not). These data dealt with the new business and innovation deliberations of the Executive Committee, long-range scientific research and innovation done at the Experimental Station, and studies on new technologies and markets done at the Development Department to support the work of the former two bodies. The documents included memos, summaries of Executive Committee meetings, letters, and reports from the Development Department and Experimental Station. We also used Wall Street Journal and New York Times articles on DuPont from 1900 – 1925 to triangulate internal data, and gather any relevant information not contained in company documents. We found these data spanning 20 years sufficient for developing process theory on entrepreneurial search.

1

Additional financial data and data on context, search process, growth possibilities generated, and subsequent decisions and outcomes can be found in an appendix available as an online supplement on Management Science’s website. A narrative case history is available from the first author. And detailed histories of DuPont during 19021921 can also be found in Chandler (1962), Chandler and Salsbury (1971), and Hounshell and Smith (1988).

5

The documents described conditions and considerations as they unfolded and the reasons behind them, and often contained information on future expectations. Because they were written for internal use, they were a valid source of information on actual decision making and on the views of decision makers at the time of the deliberations. In addition, multiple documents from the Executive Committee, Development Department, and Experimental Station contained information on the same deliberation, aiding triangulation, as did information from newspaper articles. To better capture actual decision making, in writing the case and developing theory, we took the perspective of decision makers based on facts and reasons spelled out in the documents. Although we studied a single company, we were not limited to a single instance of search, but search spanning twenty years which included thousands of deliberations regarding numerous growth possibilities. Data came from tens of thousands of pages of documents from different parts of the company. This extensive data included top-down and bottom-up, and induced and autonomous processes involving senior and middle managers, research supervisors, and scientists (Burgelman 1983). Given the importance of information to search (Shane 2000), in our theory building, we combined the process or “how” of search with the “what” or content specifics searched. This archival study began as part of a larger study on distant returns searches (Bhardwaj 2000). Four settings were studied: two path-breaking inventions of a drug and a herbicide, creation of a novel biotechnology startup, and the corporate entrepreneurship setting on which this study is based. Given the content differences across these settings, we initially expected to see different search processes emerge from the field data. The first two cases on invention led to the development of the “moving, anchored search” (MAS) process. Following recommended practice, we had been cycling back and forth across data collection, analysis, and writing (Eisenhardt 1989). We expected the remaining two cases to reveal different search processes but found, to our surprise, that the broad MAS process emerged also from data on individual and corporate entrepreneurship. Although the four settings differed in their content details, we concluded that the data revealed the same MAS process across all because at an abstract level all four settings involved distant returns – a series of choices and outcomes over long periods, and involving high

6

uncertainty and ambiguity. These temporal and probabilistic characteristics made the MAS process appropriate to all four distant returns settings. This finding is consistent with the literature. For instance, the Bower-Burgelman and evolutionary frameworks have been found appropriate to a variety of settings (Burgelman 1983, Lovas and Ghoshal 2000, Noda and Bower 1996). As in these studies, we found that the commonality of a broad framework did not preclude the discovery of specific theoretical differences across settings. Following the development of the MAS process and the discovery of its applicability across various distant returns settings, we gathered additional data to develop process theory explanations specific to continual corporate entrepreneurial search for long-term growth.

3. Findings Data showed that continual corporate entrepreneurial search for long-term growth possibilities can be characterized as a moving, anchored search process. The MAS process is located conceptually in between the polar extremes of local and global searches, which are mutually exclusive but not completely exhaustive concepts. MAS addresses how search is conducted and what is searched, thereby melding process and content. The universe of long-term growth possibilities can be vast with some possibilities known and many yet to be discovered and created. To deal with this high uncertainty and ambiguity, the process involves first selecting a broad domain where search will be conducted (Figure 1). Despite drawing a boundary for where search will be conducted, a domain still contains innumerable possibilities yet to be discovered or created. Decision makers thus choose a search anchor within the domain to guide further search for growth possibilities. Domain and anchor choices are not necessarily obvious or given. They are not chosen randomly; nor are they necessarily only marginally different from existing choices. Taking the domain as given, subsequent search is tethered to the chosen anchor and involves creating and discovering growth possibilities using the anchor as guide (Figure 1). Multiple anchors may also be chosen. Over time, search may shift to new anchors so that subsequent tethered search is far removed from earlier considerations. New search paths are thus created. Two kinds of search movements are

7

evident within a domain: tethered search movements around an anchor, and shifts of search to new anchors – hence, a moving, anchored search. Over time, there is also movement among domains as some are discarded and new ones entered (Figure 1).

3. 1 Entering New Search Domains Pursuing long-term growth often requires searching grounds untrodden for a company. The universe of specific possibilities can be vast, with some possibilities vaguely understood and others yet to be discovered. Owing to bounded rationality and incomplete information (Simon 1976), corporate entrepreneurs define a subset in the universe of growth possibilities – a search domain – within which to discover and create specific possibilities. A search domain is not the same as product or market but is a broader concept. As is evident today in life sciences and nanotechnology, search domains precede the creation of products and markets, which are a result of search within domains. Entering new search domains is different from and precedes technological and business diversification (Silverman 1999). The choice of which domain to conduct search in involves high uncertainty and ambiguity (Milliken 1987, Mosakowski 1997). It is not necessarily an obvious choice for corporate entrepreneurs. A domain draws a perimeter around subsequent considerations of possibilities, because much of the search is yet to be conducted. This subsequent search is explained in the next subsection. During the search, domains change infrequently, if at all. If, however, results or other information strongly suggest that search is being conducted in the wrong space or if the strategic context changes, a domain may be abandoned. Between 1902 and 1921, in their quest for long-term growth, DuPont managers established nine new search domains, after having operated solely in the explosives domain for 100 years. These new domains took DuPont into territories far removed from explosives (Table 2). 3.1.1 Deviations in Expectations. Data show that corporate entrepreneurs were more likely to take search into new domains following an event that caused them to expect future firm performance to deviate significantly and lastingly above or below their prior expectations. The literature has mainly

8

shown that search and risk taking rise with bad past performance and drop with good past performance (e.g., Cyert and March 1963), but a few have instead argued for the affects of expectations on risk taking (e.g., Bromiley 1991). Although temporally the 1908 move into the nitrocellulose domain did follow poor performance in 1907, discussions in documents show no causal link. Instead, they show that this domain was established in response to an antitrust lawsuit and other actions of the U.S. government (Table 2). DuPont again suffered poor financial performance in 1913 and early 1914 and from 1920-21, but these circumstances also did not cause entrepreneurial search to move into new domains. In all of these cases, DuPont responded to poor performance by trying to improve operational efficiencies and business strategy. Also contrary to the literature, the remaining search domains were entered in times of good performance, increasing search and risk taking. Documents show that the reasons for doing so were varied but were not due to good past performance, except partly in the case of “motor cars” (Table 2). Short-term decisions of operational efficiency and business strategy were indeed related to past performance as noted in the literature. But our complementary finding is that entry into new search domains, which changes the direction of entrepreneurial search, was due to expectations of significant and lasting deviations, deleterious or beneficial, in future performance. The difference in findings could be due to corporate entrepreneurs viewing recent past performance as transitory that required improving operational efficiency and business strategy, reflecting changes in search intensity as noted in the literature, but not entry into new search domains which reflect changes in the direction of search. The latter are highly uncertain moves that need considerable resources over many years. Such changes in search direction may, however, be essential when expectations form that future performance will deviate significantly and also lastingly. Events that cause such changes in expectations, described below, also signal limits to growth from current domains or novel growth possibilities in new ones. Large sample studies and non-field-based methods, which dominate the literature, may not pick up such causality. We found four kinds of events changed expectations and led corporate entrepreneurial search into new domains: discovery, anticipated shortages, external shocks, and anticipated surpluses. Whether

9

anticipatable or unpredictable and deleterious or beneficial, they resulted in corporate entrepreneurs concluding that future performance would change significantly and lastingly from their earlier expectations. We also found that events that did not cause such changes in expectations did not take search into new domains. Besides limits of current domains or the promise of new ones they signaled, these expectations-changing events converted uncertainty about the future into risk, influencing the specific choices of new domains. In 1802, Eleuthère Irénée du Pont, began search in the explosives domain when he discovered an opportunity while hunting. The quality of black powder available in the U.S. was poor and its price high. Due to his experience researching explosives in France under the pioneering chemist Antoine-Laurent Lavoisier, Eleuthère was confident he could provide a far superior product. He established the company. Search continued within the explosives domain for the next 100 years and DuPont grew, expanding into smokeless powder and dynamite. There were also other instances, in the years after 1921, of DuPont entering new domains on the basis of discoveries and inventions (Hounshell and Smith 1988). Anticipated shortages that could curtail growth in the thriving explosives business led to the creation of two new search domains. As search within the explosives domain continued with increased vigor after 1902, DuPont established two pioneering search domains, biomanufacturing in 1903 and nitrogenous compounds in 1904, to discover novel methods of producing two critical explosives ingredients whose supplies managers anticipated would be severely limited in the future, hurting the growing explosives businesses (Table 2). Glycerin, a natural by-product, was being used increasingly in innumerable industries, and nitrate of soda, the source of nitrogen for making black powder and nitric acid, could be purchased only from mines in Chile. New supply sources seemed unlikely. Inventing novel methods to make glycerin and nitric acid synthetically would obviate DuPont’s reliance on their fixed supply. Therefore, the Experimental Station began fundamental research in both pioneering domains. Success in both remained elusive, despite considerable investment, which DuPont could afford

10

because the explosives business was profitable and growing. There was little pressure to create new businesses. That changed in 1907, however, when DuPont faced the first of its two external shocks. In 1907, the U.S. government filed an antitrust lawsuit against DuPont and canceled large orders of smokeless powder while starting construction of its own plants. Congress also passed a bill preventing Navy’s explosives purchases from a monopoly. The company was about to lose a large portion of its revenues that were derived from these sales. An intense search hurriedly began for new businesses that would utilize the large impending idle capacity in smokeless powder plants. New businesses based on nitrocellulose, whose manufacture was an intermediate step in the manufacture of smokeless powder, would utilize at least some idle capacity. The nitrocellulose search domain was thus established in 1908. In 1910, the government again began buying DuPont explosives but search continued within this domain. A few years later, five new domains were established following another external shock and also due to anticipated surpluses at the company (Table 2). The start of WWI, in August 1914, deeply jostled search. Explosives demand skyrocketed. As production capacity multiplied many-fold, senior managers foresaw the massive excess capacity that would result after the war when sales would plummet. The most intensive search for new businesses began to prepare for this eventuality. New search domains were created by 1916: synthetic organic chemicals, inorganic chemicals, vegetable oils, varnishes and paints, and paper. The company sought new businesses with high profit potential that would utilize the anticipated massive idle capacity in explosives manufacturing, and would later also facilitate entry into other new businesses. There was one additional factor in entering synthetic organic chemicals, the first of the five search domains pursued in the war years. Supplies of two organic chemicals, diphenylamine and toluene, that DuPont needed for explosives manufacture and formerly purchased from Germany, were cut off with the advent of WWI. With the aid of its own research, DuPont was able to quickly start producing both. This constituted an unplanned move into synthetic organic chemicals, before the domain was formally established. Supplies from Germany of other organic chemicals, including synthetic dyes (artificial colors) and intermediates, were also cut off. Their prices skyrocketed, affecting numerous U.S.

11

industries. The severity of the shortages invited scrutiny from politicians and the government and a push began for domestic manufacture of dyes and intermediates. As the largest domestic producer and user of diphenylamine, also used also in making dyes, DuPont recognized the potential for a political and public backlash. It was using this organic chemical heavily to manufacture explosives for the war in Europe, which the U.S. had yet to enter, and profiting handsomely from it while domestic businesses clamored for dyes. The importance of aiding the “dye crisis” in the U.S., by beginning dye manufacturing was plain. The final domain of “motor cars” was entered at the end of 1917 as outlet for DuPont’s rapidly accumulating surpluses – war profits and a vastly increased labor force. Two Executive Committee members, Pierre S. du Pont and John J. Raskob, who was also the company treasurer, had been investing personally in the General Motors Corporation (GM) and were instrumental in DuPont’s buying shares of GM and Chevrolet, despite the Executive Committee recognizing that this domain was far removed from its current ones. But the duo argued that the young industry and GM promised large profits and high growth, and DuPont could sell its products to GM (Table 2). DuPont personnel, vastly increased during WWI, could also find employment in the industry instead of being laid off after the war. Share purchases afforded a quick entry into the new business, which did not require integrating with existing ones. By 1919, DuPont owned a third of GM stock, and Pierre became GM’s president from 1920 to 1923. In addition to these ten search domains that were entered, there were additional areas that saw brief considerations but were not established as new search domains: machinery and tools, waterpower, water distillation, railroads, magazine publishing, explosives plant construction, investments, banking, and cement. Most of these possibilities were quickly discarded for being far removed from existing businesses, although the new domains entered were also far removed from existing ones. But also, none of these transient deliberations had followed any event to cause expectations of significant and lasting change in future performance. Waterpower and magazine publishing searches could have lasted longer if additional circumstances or complementary searches had been more promising.

12

Before the formal setting of the synthetic organic chemicals domain, suggestions by DuPont scientists to begin the manufacture of dyes had been turned down. Similarly, a suggestion by the sales department to conduct research on the use of kerosene as motor fuel was dismissed. A few months later, following the establishment of the motor car domain, DuPont was aggressively pursuing motor fuels research. Formal entry into new search domains came after some deliberation because searching new content requires significant time and resources for research, innovations, and developing new skills, knowledge, and capabilities. A formal domain increases the likelihood that key decision makers would be engaged, and that search would receive requisite resources and support. A formal domain is also a signal about which search outcomes are more likely to later actually lead to new business creation. The lack of formal sanction of a search domain does not preclude its pursuit, as long as marginal cost of search is very low and can be done with resources already allocated (Mosakowski 1997). Search in novel domains often begins thus. But for it to continue over the long term it will eventually need resources that come with formally entering a search domain. This does not, however, suggest that domain ideas emerge only by design. But it does mean that for search in a domain to endure and yield new businesses and growth, the firm must, at some point, adopt the domain formally, whatever the original source of the idea. Expectations-changing events of discoveries, anticipated shortages, anticipated surpluses, and external shocks converted uncertainty about the environment’s nature and its effect on the company into risk (Knight 1921, Milliken 1987), affecting the timing and content of new domain choices that might have been very different in the absence of these events. Without worries of massive idle capacity, some domains pursued during the war years may arguably have never been pursued, whereas others may have been established much later and with less urgency. Expectations-changing events make urgent the need to explore farther away in content and begin the search more immediately in time. They do not, however, convert all forms of uncertainty into risk. Response uncertainty or causal ambiguity remains about what specific actions to take, what their outcomes might be, and how effective they could be (Milliken 1987,

13

Mosakowski 1997). Search within a new domain thus continues to convert what is unknown but knowable into the known (Mosakowski 1997), and response uncertainty into risk (Milliken 1987). As Shapira (1995) noted, there were no attempts at estimating probabilities. Interestingly, unlike the managers studied by Shapira (1995) who paid special attention to the potential for significant losses, those studied here were primarily playing an entrepreneurial role (Chandler, 1991, Temin 1991) and they paid as much attention to chances of significant gains as they did to chances of significant losses in entering new search domains and changing the direction of entrepreneurial search. This finding of paying equal attention to losses and gains is consistent with the argument that managers and entrepreneurs think and behave differently (Busenitz 1999, Shane 2000, Temin 1991). Thus, P1: Events that cause expectations of a significant and lasting change in future performance are more likely to result in corporate entrepreneurial search moving into new search domains. 3.1.2 Perceived Transitional Levers. New domains involve unfamiliar content and high response uncertainty or causal ambiguity about action-outcomes (Milliken 1987, Mosakowski 1997). And because moves into new domains are due to expectations of significant and lasting deviation in future performance, they must often be made without the lapse of much time. Documents show that such moves are made on the basis of things that are perceived by corporate entrepreneurs to facilitate the highly uncertain transition from existing domains to new ones by aiding in the conversion of response uncertainty into risk (Knight 1921, Milliken 1987). Accordingly, we term these facilitators perceived transitional levers. The specific perceived transitional levers in our study fall within the broader categories of markets or knowledge and capabilities. Perceived transitional levers play a role because causal ambiguity in new domains can only be reduced through search; but to be effective such search requires new knowledge and capabilities that is not possessed at the time search begins in a new domain. Those based on existing domains are rarely fully transferable but some of them are nevertheless seen as aids in getting search in the new domain started that, over time, will yield the required knowledge and capabilities, along with new growth

14

possibilities. For instance, the move into organic chemistry was made not because DuPont possessed the required knowledge and capabilities but because it did not, and those could only be developed through search. Interestingly, in the complete absence of capabilities relevant for a new domain, as was the case with biomanufacturing, nitrogenous compounds, and motor cars, the transitional lever was an in-house market. It did not ease search but assured a buyer for the results of the uncertain search, eliminating the chances of resources spent in vain, thus partly reducing the uncertainty of entering these new domains. Typically, perceptions of these transitional levers later prove inaccurate or incomplete. Besides bounded rationality and incomplete information about the unfamiliar content of new domains, cognitive biases also likely play a role because corporate entrepreneurs tend to perceive less risk in a situation, are optimists, tend to be overconfident, and are inclined to pursue opportunities even without possession of all needed resources and capabilities (Busenitz 1999, Shane 2000, Stevenson and Jarillo 1990). And some knowledge and capabilities can only be generated through search. Thus, incompleteness or inaccuracy of perceived transitional levers is highly likely, resulting in unexpected demands later as search spans long periods with immediate needs for resources and uncertain promises of future outcomes. The perceived transitional lever in both biomanufacturing and nitrogenous compounds domains was a ready market in-house for glycerin and nitric acid. The content of search in both domains was so far removed from anything DuPont had done thus far that it did not have needed research capabilities to leverage from the explosives domain (Table 2, 3). But a ready in-house market for the results of these searches facilitated new domain entries by justifying the spending of considerable resources over many years during which DuPont unsuccessfully performed fundamental research in these difficult areas. In entering nitrocellulose, the perceived transitional levers were the impending idle capacity in smokeless powder plants and nitrocellulose manufacturing capabilities; the latter a delicate art since nitrocellulose is flammable and can explode. Due to the urgency of finding new growth possibilities following the antitrust lawsuit, this domain dealt with nitrocellulose-based businesses because they would utilize at least some idle capacity and skills and thus facilitate the move. As within-domain search

15

proceeded, DuPont discovered that far less idle capacity could be utilized than it had hoped. Moreover, the capabilities in manufacturing, research, and distribution required for the search anchors pursued within this domain were far more and different than they had initially anticipated (Table 3). The perceived transitional levers proved highly inadequate. Moves into the five new domains in the war years were made on the basis of perceived transitional levers related to both markets and capabilities, which later proved inaccurate and inadequate. To ease search in these new domains, DuPont tried leveraging its infrastructure, manufacturing resources and capabilities, and sales force. Some of the domains were also seen as having links among themselves (Tables 2, 3). Initially, the Executive Committee did not want to pursue any new business that would be “to any large extent foreign to the company’s capacity” or require experimental manufacturing processes. They soon discovered, however, that there was little relationship between new businesses they were studying and manufacturing and distribution at DuPont. As search proceeded, it also became clear that synthetic organic chemicals, vegetable oils, and paints and varnishes would use almost no idle explosives plant capacity. At best, they would use only a portion of general infrastructure: grounds, buildings, railroads, roads, wharf, offices, laboratories, storehouses, storage tanks, boiler plant, power plant, hydraulic press equipment, water supply, pipelines, and acid plants. Large additional investments would still be required. The Executive Committee subsequently decided not to limit search to “allied” industries in pursuing search within these domains, but to seek as well those that would employ the “organization” to the fullest, including the large R&D and manufacturing labor force hired for war-related work that would otherwise have to be laid off, and that would use DuPont’s “knowledge and experience in chemical manufacture.” Explosives and dyes were also known to be linked by their use of some common raw materials, organic chemical intermediates, and production processes. For similar reasons, DuPont managers thought that the dye business would later facilitate entry into other businesses based on synthetic organic chemicals, such as ethical drugs. An intensive and expensive search began in the synthetic organic chemicals domain, despite the scarcity of raw materials and intermediates and DuPont’s

16

inexperience and lack of knowledge and capabilities in research, manufacturing, and distribution. It was soon evident that the links that DuPont thought existed among businesses across these domains were tenuous. The domains also proved much farther removed than anticipated from existing ones in terms of research, manufacturing, and distribution. Barring varnishes, paints, and dyes, other searches were abandoned. Those that continued, especially dyes, proved long and costly. But the company persisted not only because these were promising new businesses but also because they would help create new knowledge and capabilities that would later pave entry into other new businesses. The motor cars domain was entered on the basis of two transitional levers – creation of an inhouse market and employment for DuPont’s surplus labor force. DuPont’s products would find a customer in GM, and they did (Table 2). And DuPont’s large cadre of engineers hired for constructing explosives plants for the war would not have to be laid off. But, following the war, enough work could not be found for them in the motor car industry. DuPont, however, did manage to get research work from GM for its scientists. There was, however, so little linking this domain with others that DuPont ran GM independently of its other businesses. Moreover, because DuPont had entered motor cars with massive share purchases, the perceived transitional levers ultimately mattered little. In changing the direction of entrepreneurial search by taking it into new domains, the emphasis shifts from leveraging existing resources and capabilities to identifying and developing new, required ones (Chandler 1991, Helfat and Raubitschek 2000). Perceived transitional levers play a crucial role in this transition even though they later emerge inadequate and inappropriate. This finding contributes to the literature on dynamic capabilities by providing process insights on the transition from old to new capabilities (Helfat and Peteraf 2003). Thus, P2: Moves from existing search domains into new ones are made on the basis of transitional levers perceived to facilitate the highly uncertain moves, but often these perceived transitional levers later prove insufficient or inaccurate.

17

3.2 Search within Domains Once chosen, a search domain still contains innumerable specific growth possibilities that yet remain largely unknown. Hence, corporate entrepreneurs select one or more search anchors to determine where within a domain’s content to look for them (Figure 1). Process and content are thus fundamentally intertwined. An anchor’s choice is not always obvious owing to uncertainty and causal ambiguity in a new domain (Milliken 1987, Mosakowski 1997). We found that the various specific reasons for anchor choices (Table 3) may be categorized more broadly as exploratory or exploitative (March 1991). New anchors choices may reflect attempts at exploiting market growth potential, industry inefficiencies, or current knowledge and capabilities. Exploratory reasons for anchor choices are reflected in wanting to create new knowledge and capabilities, or searching content areas because they are very different. Once selected, subsequent deliberations are tethered, or local, to an anchor. An anchor acts as conceptual guide, focusing attention and bounding search to around itself. Over time, an anchor may be retained in the absence of promising findings because there is still local space left to explore (Garud and Van de Ven 1992). Or, it may shift to new anchors to explore significantly different areas in the domain’s content. Multiple anchors may be pursued simultaneously. Shifting search to new anchors and pursuing multiple anchors reflect dissatisfaction, not satisficing, and are explicitly meant to avoid local search (Simon 1976). Search within domains is about reducing causal ambiguity or trying to convert response uncertainty into risk (Knight 1921, Milliken 1987, Mosakowski 1997). Our data show that the content location of search deliberations and the number of anchors pursued affect growth possibilities discovered and created. 3.2.1 Search Location Matters. Specifically where in a domain’s content an anchor is located, or how it is conceptualized, influences the possibilities generated or precluded. For instance, as part of the search driven by the securing supplies anchor in the explosives domain (Table 3), DuPont was willing to explore new businesses that would provide itself supplies (e.g., it entered nitrate of soda mining in Chile) but made an explicit policy to not get into the business of selling supplies to others. When

18

approached to sell explosives-making machinery and construct explosives plants for others, it quickly dismissed these new business possibilities. DuPont made its machinery and constructed its plants, but doing them for others reflected a different search anchor that it was unwilling to establish. Likewise, within the nitrogenous compounds domain, it limited its search to the nitrogen fixation anchor for years. It could conceivably have pursued different possibilities through other anchors, but it did not. The possibilities generated also vary depending on whether a new business is treated as part of a tethered search, as a separate anchor, or as a different domain. Deliberations regarding rubber were initially tethered to artificial fibers in the nitrocellulose domain (Table 3), so the possibilities generated were limited to those that complemented artificial fibers. A few years later, rubber became an anchor in the synthetic organic chemicals domain, yielding different possibilities due to difference in this domain’s content compared to nitrocellulose’s. Years later, as a separate domain and with DuPont’s invention of neoprene, the first synthetic rubber, the possibilities generated were completely different and the rubber business became crucial to DuPont (Hounshell and Smith 1988). When the new business of paints was initially considered, it was in the nitrocellulose domain as a “novel” outlet for nitrocellulose (Table 3). The possibility discovered of boat-bottom paints was limited. Years later, when paints and varnishes became a separate domain (Table 3), one that was also linked with vegetable oils, different and far more attractive possibilities were discovered, resulting in DuPont entering these new businesses. In the latter case, the domain’s search content was different and nitrocellulose was no longing a required and limiting factor in generating possibilities, as it had been initially. Process of MAS and the content searched combined to influence the nature of growth possibilities created and discovered. Thus, P3: An anchor’s particular location in a domain’s content influences growth possibilities discovered and created by including those within the boundaries of its tethered search and excluding others. P4: Significantly different growth possibilities emerge when a new business or innovation is considered part of a tethered search, as a separate anchor, or in different domains.

19

3.2.2 Anchors’ Numbers Change. Minutes of the Executive Committee meetings show that to ensure search for long-term growth possibilities would continue unhindered, as a matter of policy, funds were not cut, even during lean times, for the Experimental Station and Development Department. Yet, the number of anchors pursued, their duration, and intensity of tethered search varied over time, affecting the possibilities discovered and created. Anchors in different parts of the domain’s content reflect breadth of search and their respective tethered searches indicate local depth (Katila and Ahuja 2002). Deliberations regarding search moves within domains showed tendencies towards a quicker increase to a higher number and then a quicker decline in the number of anchors when a domain’s content was rich. When sparse, the rise and decline tended to be gradual and the number of anchors not as high. Rich content allowed greater variety of ideas for exploring where search could be done in more depth to discover promising new information and growth possibilities. Creating new anchors did not require deep knowledge because they were just starting points and guides for search. With increasing familiarity and generation of new information, the number of anchors tended to rise rapidly. Outcomes from their respective tethered searches created new knowledge and capabilities that improved search productivity (Helfat and Raubitschek 2000, Mosakowski 1997) by showing anchors with promise and resulting in increased intensity of their tethered search. Anchors without promise were abandoned. Over time, the number of anchors thus declined while tethered search around surviving ones increased. The richer the domain content, the faster this pattern of breadth and depth could play out. Both the explosives and nitrocellulose domains saw this tendency, which was even more pronounced in the simultaneous searches across the five domains during the war years. In both biomanufacturing and nitrogenous compounds, domain content knowledge was so sparse that new anchors could not be created after the first one. The generation of useful knowledge was also slow, keeping domain content sparse. In biomanufacturing, despite trying, years of search did not yield knowledge that could have led to new anchors. But the domain was important so search continued for years around the unproductive fermentation anchor until changed circumstances diminished the

20

importance of the domain, and search tapered off (Table 3). In nitrogenous compounds, the content was also sparse but new knowledge did emerge gradually that aided the creation of another anchor. A third soon followed but was just as quickly abandoned. Search around the earlier two continued. Motor cars saw a similar gradual rise to a low peak and decline in anchors. DuPont could not search much of this domain’s content because GM was run separately and was doing its own search. The space left for DuPont to explore was too sparse for new anchors after the one for motor fuels was created. Search around the latter also declined when promising findings came from GM instead. The number of anchors pursued over time was thus an inverted-U function and tended to have steeper slopes and higher peak when domain content was rich. When domain content was sparse, the slopes were gradual and the peak lower. The decline in anchor numbers was accompanied by a rise in the intensity of tethered search around surviving ones when domain content was rich. If, despite searching, content remained sparse, the intensity of tethered search tended to gradually decline. Thus, search tended to yield rising variety of broadly defined growth possibilities initially, and as variety dropped, it led to better defined growth possibilities, unless domain content remained sparse. P5: Greater the available knowledge of a domain’s content, more the number of anchors pursued, with a rapid increase and decline in their numbers, and in the variety of growth possibilities generated. With decline in anchors’ numbers, tethered search tended to intensify around surviving ones, yielding fewer and better defined growth possibilities.

4. Conclusions Using DuPont’s decision making documents from 1902 to 1921, we developed process theory explanations of how continual corporate entrepreneurial search for long-term growth possibilities is conducted, a little studied but important phenomenon. We found that the process can be framed as a moving, anchored search. Employing this as lens, we developed specific theoretical explanations and propositions. In our study, we took the decision maker’s perspective, integrated process and content,

21

incorporated external and internal factors, and included all growth possibilities discovered and created and not just those that later led to new businesses. The process theory we developed describes and explains, but is not meant to predict success and failure of new businesses that result from search. Our explanations are early in the causal chain and many subsequent activities, choices, and circumstances will influence such eventual performance. We believe that the search process explanations and propositions developed in this field study can be generalized to other companies and periods, as long as the settings involve a series of highly uncertain choices and outcomes that span long periods. These two distant returns conditions, as argued earlier, make the use of moving, anchored search process as theoretical lens appropriate and the propositions generalizable. However, the content of search will vary across settings. Detailed accounts of corporate entrepreneurship do indicate support for this argument (e.g., Chesbrough 2002, Dyer and Gross 2001, Hounshell and Smith 1988, Steinbock 2001). This paper is a start to understanding continual corporate entrepreneurial search for long-term growth. We propose ideas for further research using field studies, computer simulations, and laboratory experiments. What other relationships exist between domains and anchors, and what more factors influence their selections? How is search affected by organizational mechanisms like structure, controls, incentives, hiring and exits, and team composition? How do different institutions and changes in them influence search? What impact do competitive dynamics and technological change have? In the period studied, DuPont did not reduce resource for search even during periods of poor performance. This may also be so in other research-driven firms but perhaps not in other types of firms. Knowing how search is affected by fluctuations in resources is thus important. Our setting also did not include the abandonment of any long-established search domains. Do firms find it easier to begin search in new domains than to abandon it in old ones? Does the latter require an external or internal shock? Building on the findings of this study, investigating questions such as these will help advance understanding of continual corporate entrepreneurial search for long-term growth, an important but little studied phenomenon.

22

Table 1: Some innovation and new business possibilities considered at DuPont, 1902-1921 Cartridges. Fuses. Pyrites. Timber. Ethyl Alcohol. Nitrate of Soda. Nitrate of Potash. Glycerin. Sulfur. Nitrogen Fixation. Sawmill. Cottonseed Hull Fiber. Blasting Caps. Loaded Ammunitions. Fusel Oil. Fulminate of Mercury. Muriate of Potash. Explosives Plant Construction. Artificial Silk. Electric Fuses. International Markets. Copper Shells. Artificial Leather. Celluloid. Boat Bottom Paint. Electric Air Drill. Cotton Purification. Quicksilver Mining. Fertilizers. Fireworks. Bags. Waterpower. Refining Refuse Salt. Cattle Feed. Copper Wire. Enameling Fuse Wire. Cartons. Tin Boxes. Cotton Hose Lining. Ammonia. Water Distillation. Rubber and Synthetic Rubber. Photographic Films. Paraffin from Waste Paper. Potassium. Railroads in Chile. Selling Explosives Machinery. Saltpeter. Magazine Publishing. Picric Acid. Cinematographic Films. Diphenylamine. Toluene. Time Fuses for Shrapnel. Dyes and Intermediates. Chlorine. Drugs. Vegetable Oils & Fats. Growing Castor Beans. Soap. Oilcloth. Waterproofed Material. Inorganic Chemicals. Rifles. Cotton Cloth. Varnish. Paints. Photographic Chemicals. Perfumes. Food Flavorings and Preservatives. Pulp. Paper. Pigments. Banking. Camphor. Motor Cars. Investments. Printing Ink. Cement. Saccharin. Motor Fuels. Tetraethyl Lead.

Figure 1: Moving, anchored search (MAS) process

Search domain Search anchor

1 Tethered search I: Select new search domain

II: Select search anchor and perform tethered search within domain

2

1

Shift in anchor

4

3 III: Shift search to new anchors and perform further tethered search within domain (similar effect as multiple, simultaneous search anchors)

IV: Over time, many search domains may be pursued

23

Table 2: Reasons for search domain choices at DuPont, 1802 and 1902–1921 Search Domain

Reasons for Domain Choice

Explosives (1802 – beyond 1921)

Profitable market opportunity discovered by founder in 1802 for better quality and cheaper black powder; founder’s research expertise in explosives.

Biomanufacturing (1903 - 1916)

Anticipated shortage of glycerin, an important explosives ingredient, available only as a by-product of tallow candle manufacture, laundry soap production, and slaughterhouse waste. Novel synthetic production method needed. Wine and beer distillation known to yield minute amounts of glycerin as by-product. Conducted fundamental research on bacteriological fermentation to raise yield.

Nitrogenous Compounds (1904 – beyond 1921)

Anticipated shortage of nitrate of soda used to make nitric acid for explosives manufacture; Chilean mines were sole global source; inventing an industrial nitrogen fixation method would yield from the atmosphere an inexhaustible supply of nitrogen that could be used to make nitric acid and other nitrogenous compounds.

Nitrocellulose (1908 – beyond 1921)

U.S. government filed antitrust lawsuit in 1907. In 1908, the government canceled large orders of smokeless powder from DuPont, started construction of its own plants, and Congress passed a bill preventing Navy’s explosives purchases from a monopoly. Utilize idle plant capacity by finding new uses for nitrocellulose, which was manufactured as an intermediate step in the production of smokeless powder.

Synthetic Organic Chemicals (1915 – beyond 1921)

Imports of organic chemicals from Germany ceased during WWI; severe shortages, skyhigh prices; some organic chemicals used in explosives manufacture also used in other industries; potential for public and political backlash. Find uses for anticipated massive idle plant capacity after war; utilization of “organization”; avert post-war layoffs; new businesses that would pave entry into others later.

Inorganic Chemicals (1916 – 1918)

Complement ongoing work on photographic and pharmaceutical organic chemicals, and paints and varnishes.

Vegetable Oils (1916 – 1918)

Utilize anticipated post-war excess capacity; many uses of vegetable oils in other industries - “nucleus” industry as gateway to others later; “non-organized” in manufacturing and commercial terms, but high growth and profit potential.

Varnishes and Paints (1916 – beyond 1921)

Utilize post-war excess capacity; skills and equipment required are simple in “general” paints and varnishes; processes, equipment, and raw materials similar to those of pyroxylin solutions already being produced; market complement to pyroxylin solutions.

Paper (1916 – 1919)

Utilize cotton purification plant used in explosives manufacture; war shortages of cotton rags in paper trade; DuPont could produce purified cotton fiber or cotton pulp that paper manufacturers could use as substitute for cotton rags.

Motor Cars (1917 – beyond 1921)

Need to invest large war profits. DuPont’s treasurer and Pierre S. du Pont, who had both been personally investing in GM, were instrumental in DuPont’s buying shares in GM and Chevrolet. Young auto industry profitable and promised strong growth. DuPont artificial leather, paints and varnishes, celluloid plastic, and rubber-coated fabrics could be sold to GM. DuPont’s large engineering department could construct auto plants for GM and others in the growing industry to avert post-war layoffs.

24

Table 3: Search anchor choices within search domains, 1902 – 1921 Anchors Within Domains Explosives (1902 – beyond 1921)

Reasons for Establishing Anchors

Consolidation (established 1902)

Consolidate and reorganize company’s interests spread across more than a hundred companies; consolidate U.S. explosives industry.

Securing supplies (established 1903)

Meet supply requirements not being met by market; ensure reliable supply of critical ingredients and components; obtain best possible prices through market or backward integration.

Related products (established 1903)

Complement explosives businesses.

Foreign markets (established 1905)

US industry consolidation achieved, so seek new avenue for growth.

Biomanufacturing (1903 – 1916) Bacteriological fermentation (established 1903)

Find synthetic means of making glycerin; fermentation known to yield minute amounts of glycerin in beer and wine distillation.

Nitrogenous Compounds (1904 – beyond 1921) Nitrogen fixation (established 1904)

Develop synthetic method to make nitric acid for explosives manufacture; an industrial nitrogen fixation method would yield inexhaustible supply of nitrogen from air that could then be used to make nitrogenous compounds.

Other methods (established 1908)

Try other methods to make nitrogenous compounds since nitrogen fixation had yielded little.

Fertilizers (established 1910)

DuPont had been selling small amounts of nitrate of soda, bought for explosives manufacturing, to local farmers for use as fertilizers. Began studies on entering fertilizers after many employee suggestions.

Nitrocellulose (1908 – beyond 1921) Artificial fibers (established 1908)

Use of idle plant capacity; nitrocellulose used in manufacture of artificial fibers that had well-established markets; artificial silk market seen to offer “phenomenal growth” prospects.

Celluloid (established 1909)

Use of idle plant capacity since nitrocellulose used in manufacturing celluloid, a plastic; well-established market. Photographic and cinematographic films base used celluloid coated with emulsion.

Coating materials (established 1909)

Boat bottom paint brought to company’s attention as “novel” outlet for nitrocellulose.

(Continued next page)

25

Table 3: Search anchor choices within search domains, 1902 – 1921 (continued) Anchors Within Domains Reasons for Establishing Anchors Synthetic Organic Chemicals (1915 – beyond 1921) Dyes & intermediates (established 1915)

Utilize some excess plant capacity after war; utilize DuPont’s “organization”; develop new scientific and manufacturing capabilities to pave entry into other organic chemistry based businesses. Links with other anchors that could lead to a “complete system” to utilize excess capacity. Increasing consumption, market shortages, fear of political and public backlash, high profitability, and possibility of discovering patentable chemicals.

Drugs (established 1915)

Some commonalities with dyes in raw materials, intermediates, and production processes. High profitability, and possibility of discovering patentable chemicals.

Rubber chemicals (established 1915)

Important but “undeveloped” field.

Food preservatives (established 1916)

Growth potential.

Perfumes and flavorings (established 1916)

Growth potential.

Photographic chemicals (established 1916)

Photography and movie businesses growing.

Inorganic Chemicals (1916 – 1918) Water soluble inorganic chemicals (established 1916)

Complementary to pharmaceutical and photographic chemicals then being studied. Some had production processes similar to that of some organic chemicals. Some use of excess capacity.

Pigments (established 1916)

Mineral colors could be used in varnishes and paints then being studied, and also in existing businesses of dyestuffs, pyralin, pyroxylin, and artificial leather. “Unorganized” and “underdeveloped” business.

Vegetable Oils (1916 – 1918) Oils and Fats (established 1916)

“Unorganized”, “uncompetitive”; products used in paints, varnishes, soaps, artificial leather, oilcloth, linoleum, and others then being studied.

Downstream businesses (established 1916)

Oils and fatty acids used in businesses such as soap and oilcloth. Also, obtain glycerin as by-product of soap making, and oilcloth related to artificial leather.

(Continued next page)

26

Table 3: Search anchor choices within search domains, 1902 – 1921 (continued) Anchors Within Domains Varnishes and Paints (1916 – beyond 1921)

Reasons for Establishing Anchors

“General” varnishes and paints (established 1916)

Varnishes would pave way for entry into paints since latter “less allied” to existing businesses.

“Mixed” varnishes and paints (established 1916)

Very different from “general” paints and varnishes businesses.

Paper (1916 – 1919) Purified cotton pulp and fiber (established 1916)

Excess capacity in cotton purification plant used in explosives manufacture could be modified to make pulp for paper industry, or to enter latter.

Motor Cars (1917 – beyond 1921) Plant construction (established 1918)

Alternate use of DuPont’s large engineering organization built to construct explosives plants during WWI; avoid layoffs.

Motor fuels (established 1919)

Post-war use of DuPont’s Development and Chemicals Departments to conduct research for GM for mutual gain; important problems related to motor fuels known.

27

REFERENCES Ahuja, G., C. M. Lampert. 2001. Entrepreneurship in the large corporation: A longitudinal study of how established firms create breakthrough inventions. Strategic Mgt. J. 22 521-543. Arora, A., R. Landau, N. Rosenberg. 1998. Chemicals and Long-Term Economic Growth. John Wiley & Sons, New York. Baumol, W. J. 1993. Entrepreneurship, Management, and the Structure of Payoffs. MIT Press, Cambridge, MA. Bhardwaj, G. 2000. Search for Distant Returns: A Decision-Making Process Model from Choices at DuPont for Invention, Entrepreneurship, and Growth. Unpublished Ph.D. Dissertation, U. of Pittsburgh. Bromiley, P. 1991. Testing a causal model of corporate risk taking and performance. Acad. of Mgt. J. 34 37-59. Burgelman, R. A. 1983. Corporate entrepreneurship and strategic management: Insights from a process study. Mgt. Sc. 29 1349-1364. Busenitz, L. W. 1999. Entrepreneurial risk and strategic decision making: It’s a matter of perspective. J. of Applied Behavioral Sc. 35 325-340. BusinessWeek. 2003. India’s tech king. Asia edition, October 13 issue. Chandler, A. D., Jr. 1962. Strategy and Structure: Chapters in the History of the American Industrial Enterprise. MIT Press, Cambridge, MA. Chandler, A. D., Jr. 1990. Scale and Scope: The Dynamics of Industrial Capitalism. Harvard U. Press, Cambridge, MA. Chandler, A. D., Jr. 1991. Creating competitive capability: Innovation and investment in the United States, Great Britain, and Germany from the 1870s to World War I. P. Higonnet, D. S. Landes, H. Rosovsky eds Favorites of Fortune: Technology, Growth, and Economic Development since the Industrial Revolution, 432-458. Harvard U. Press, Cambridge, MA.

28

Chandler, A. D., Jr., S. Salsbury. 1971. Pierre S. du Pont and the Making of the Modern Corporation. Harper & Row, New York. Chesbrough, H. 2002. Graceful exits and missed opportunities: Xerox’s management of its technology spin-off organizations. Bus. Hist. Rev. 76 803-837. Covin, J. G., D. P. Slevin. 1991. A conceptual model of entrepreneurship as firm behavior. Entrepreneurship Theory and Practice 16 7-25. Cuff, R. D. 2002. Notes for a panel on entrepreneurship in business history. Bus. Hist. Rev. 76 123-132. Cyert, R. M., J. G. March. 1963. A Behavioral Theory of the Firm. Prentice-Hall, Englewood Cliffs, NJ. Dyer, D., D. Gross. 2001. The Generations of Corning: The Life and Times of a Global Corporation. Oxford U. Press, New York. Eisenhardt, K. M. 1989. Building theories from case study research. Acad. of Mgt. Rev. 14 532-550. Gaddis, J. L. 2002. The Landscape of History: How Historians Map the Past. Oxford U. Press, New York. Gaglio, C. M. 1997. Opportunity identification: Review, critique and suggested research directions. J. Katz, R. Brockhaus eds Advances in Entrepreneurship, Firm Emergence and Growth, vol. 3, 139-202. JAI Press, Greenwich, CT. Garud, R., A. H. Van de Ven. 1992. An empirical evaluation of the internal corporate venturing process. Strategic Mgt. J. 13 93-109. Helfat, C. E., K. M. Eisenhardt. 2004. Inter-temporal economies of scope, organizational modularity, and the dynamics of diversification. Strategic Mgt. J. 25 1217-1232. Helfat, C., M. A. Peteraf. 2003. The dynamic resource-based view: Capability lifecycles. Strategic Mgt. J. 24 997-1010. Helfat, C. E., R. S. Raubitschek. 2000. Product sequencing: Co-evolution of knowledge, capabilities and products. Strategic Mgt. J. 21 961-979. Holbrook, D., W. M. Cohen, D. A. Hounshell, S. Klepper. 2000. The nature, sources, and consequences

29

of firm differences in the early history of the semiconductor industry. Strategic Mgt. J. 21 1017-1041. Holcombe, R. G. 2003. The origins of entrepreneurial opportunities. Rev. of Austrian Eco. 16 25-43. Hounshell, D. A., J. K. Smith, Jr. 1988. Science and Corporate Strategy: DuPont R&D, 1902-1980. Cambridge U. Press, New York. Katila, R., G. Ahuja. 2002. Something old, something new: A longitudinal study of search behavior and new product introduction. Acad. of Mgt. J. 45 1183-1194. Knight, F. H. 1921. Risk, Uncertainty, and Profit. Hart, Schaffner, & Marx, Boston, MA. Lovas, B., S. Ghoshal. 2000. Strategy as guided evolution. Strategic Mgt. J. 21 875-896. March, J. G. 1991. Exploration and exploitation in organizational learning. Org. Sc. 2 71-87. March, J. G., L. S. Sproull, M. Tamuz. 1991. Learning from samples of one or fewer. Org. Sc. 2 1-13. McCormick, B, B. W. Folsom Jr. 2003. A survey of business historians on America’s greatest entrepreneurs. Bus. Hist. Rev. 77 703-716. Milliken, F. J. 1987. Three types of perceived uncertainty about the environment: State, effect, and response uncertainty. Acad. of Mgt. Rev. 12 133-143. Mosakowski, E. 1997. Strategy making under causal ambiguity: Conceptual issues and empirical evidence. Org. Sc. 8 414-442. Nelson, R. R., S. G. Winter. 1982. Evolutionary Theory of Economic Change. Harvard U. Press, Cambridge, MA. Noda, T., J. L. Bower. 1996. Strategy making as iterated processes of resource allocation. Strategic Mgt. J. 44 (summer special issue) 159-192. Papadakis, V. M., S. Lioukas, D. Chambers. 1998. Strategic decision-making processes: The role of management and context. Strategic Mgt. J. 19 115-147. Penrose, E. 1995. The Theory of the Growth of the Firm, 3rd ed. Oxford U. Press, New York. Shane, S. 2000. Prior knowledge and the discovery of entrepreneurial opportunities. Org. Sc. 11 448-469

30

Shane, S., S. Venkataraman. 2000. The promise of entrepreneurship as a field of research. Acad. of Mgt. Rev. 25 217-226. Shapira, Z. 1995. Risk Taking: A Managerial Perspective. Russell Sage Foundation, New York. Silverman, B. S. 1999. Technological resources and the direction of corporate diversification: Towards an integration of the resource-based view and transaction cost economics. Mgt Sc 45 1109-1124. Simon, H. A. 1976. Administrative Behavior: A Study of Decision-Making Processes in Administrative Organization, 3rd ed. Free Press, New York. Steinbock, D. 2001. The Nokia Revolution: The Story of an Extraordinary Company that Transformed an Industry. AMACOM, New York. Stevenson, H. H., J. C. Jarillo. 1990. A paradigm of entrepreneurship: Entrepreneurial management. Strategic Mgt. J. 11 (summer sp issue) 17-27. Temin, P. 1991. Entrepreneurs and managers. P. Higonnet, D. S. Landes, H. Rosovsky eds Favorites of Fortune: Technology, Growth, and Economic Development since the Industrial Revolution, 339-355. Harvard U. Press, Cambridge, MA. Tripsas, M., G. Gavetti. 2000. Capabilities, cognition, and inertia: Evidence from digital imaging. Strategic Mgt. J. 21 1147-1161. Venkataraman, S. 1997. The distinctive domain of entrepreneurship research. J. Katz, R. Brockhaus eds Advances in Entrepreneurship, Firm Emergence and Growth, vol. 3, 119-138. JAI Press, Greenwich, CT. Wall Street Journal. 2004. Flexible formula of British survivor: From nuts to jets. March 16 issue. Zahra, S. A., D. F. Jennings, D. F. Kuratko. 1999. The antecedents and consequences of firm-level entrepreneurship: The state of the field. Entrepreneurship Theory and Practice 24 45-65.

31