Abstract:
Software project escalation has been shown to be a widespread phenomenon. With few exceptions, prior research has portrayed escalation as an irrational decision-making process whereby additional resources are plowed into a failing project. In this article, we examine the possibility that in some cases managers escalate their commitment not because they are acting irrationally, but rather as a rational response to real options that may be embedded in a project. A project embeds real options when managers have the opportunity but not the obligation to adjust the future direction of the project in response to external or internal events. Examples include deferring the project, switching the project to serve a different purpose, changing the scale of the project, implementing it in incremental stages, abandoning the project, or using the project as a platform for future growth opportunities. Although real options can represent a substantial portion of a project’s value, they rarely enter into a project’s formal justification process in the traditional quantitative discounted cash-flow-based project valuation techniques. Using experimental data collected from managers in 123 firms, we demonstrate that managers recognize and value the presence of real options. We also assess the relative importance that managers ascribe to each type of real option, showing that growth options are more highly valued than operational options. Finally, we demonstrate that the influence of the options on project continuation decisions is largely mediated by the perceived value that they add. Implications for both theory and practice are discussed.
Introduction:
An escalation situation exists when there is “decision making in the face of negative feedback about prior resource allocations, uncertainty surrounding the likelihood of goal attainment, and choice about whether to continue” (Brockner, 1992, p. 122). The bulk of prior work on escalation has sought to understand apparently irrational instances of escalation where actors persist in courses of action that they could (or should) have known were destined to fail (Staw, Sandelands, & Dutton, 1981; Brockner, 1992). The main goal has been to isolate nonrational factors (i.e., those leaving the project payoff structure unaffected) that reinforce escalation tendencies. For example, managers are more likely to escalate when they have a greater need for self-justification, when sunk costs are higher, and when they hold asymmetric information about project status (Keil, 1995).
Despite the primary focus on nonrational factors, escalation researchers have acknowledged the possibility of other rational factors that promote escalation in the presence of negative feedback. An early literature review by Staw (1981) included two rational factors—probability of future outcomes and value of future outcomes—as determinants of continuation tendencies. Bowen (1987) argued that instances of escalation often involve equivocal information about project status and future prospects and that in such cases the common understanding of escalation as an irrational process of throwing good money after bad would not necessarily hold. In fact, there can be numerous subtleties not readily apparent to external observers that could promote project continuation in spite of negative feedback about project status. In the context of information technology (IT) investment projects, Keil and Flatto (1999) suggested that one of these subtleties is the presence of real options in a given project. In particular, they argued that, in instances that appear to be unwarranted (i.e., irrational), escalation may actually be warranted (i.e., rational) escalation were the value of real options taken into account. In this research, we examine whether real options actually do increase the propensity to engage in what we call warranted continuation in escalation situation, that is, a normatively rational continuation of a troubled IT project that has continued uncertainty about goal attainment.
Opportunities to embed real options are pervasive in IT projects (Benaroch, 2002). A project embeds real options when managers have the opportunity (but not the obligation) to adjust the future direction of the project in response to external or internal events. These adjustments can take the form of deferring the project, switching the project to serve a different purpose, changing the scale of the project, implementing it in stages, abandoning the project, or using the project as a platform for future growth opportunities. Yet, real options, which can represent a substantial portion of a project’s value (Taudes, Feurstein, & Mild, 2000), rarely enter into a project’s formal justification process in practice. Instead, quantitative project valuations are typically based on traditional discounted cash flow techniques such as net present value (NPV) that ignore option value (Busby & Pitts, 1997). Naturally, formal post-mortems are likely to focus on how actual experience comported with expectations on this same NPV basis.
Even though real options are rarely considered explicitly, some studies beyond the IT context indicate that operational managers often implicitly recognize the value of real options (Busby & Pitts, 1997; Kogut & Kulatilaka, 2004). There- fore, there may be instances in which projects that appear to external reviewers employing an NPV logic to have undergone an irrational escalation actually underwent a warranted continuation owing to implicit recognition by operational managers of real options external to the original justification decision. However, this would only be a pervasive phenomenon if it were true that operational IT managers were likely to place a significant value on real options and were more prone to continue troubled projects when they do.
In this research, we investigate the effects of real options on IT project continuation in escalation situations. Our approach is to use a conjoint study to examine how the presence of one or more different option types affect managers’ perceptions of project value and their expressed likelihood of continuing investment in troubled projects. In particular, we study escalation scenarios where there is negative feedback and considerable uncertainty about goal attainment and where managers should be indifferent toward continuation from a traditional NPV perspective (i.e., the NPV is set at zero). Making the base-case indifference toward continuing allows us to better isolate the effects of having one or more real options present.
The key question for us is not whether IT managers might place some positive value on real options. Prior research outside IT (Busby & Pitts, 1997) suggests that they often do, and we see no obvious reason why either the IT project context or the escalation context per se would prevent managers from recognizing option value. Rather, we see the key research issues as being: (i) how strongly option value translates into an increased propensity to continue a troubled project and (ii) what differences exist in how different types of real options (i.e., to switch use, change scale, stage investments, abandon, or strategically grow a project) are valued in escalation situations. The first issue gives an indication of how pervasive the phenomenon of warranted continuation due to recognition of options might be in practice. The second issue can provide some initial insights into possible biases in how options are valued.
Our results show that the presence of real options does lead to a tendency toward continuing projects (i.e., warranted continuation), and this tendency increases with the number of real options that are present. Second, our results show that in escalation situations managers place a much higher value on options that spawn new investment opportunities (strategic options), as compared with options that allow them to reconfigure such elements as the timing, scale, and scope of an investment (operating options). Some options, such as the option to abandon, are given very little comparative value.
Negative feedback about project progress and uncertainty about goal attainment are very common occurrences on IT projects (Wallace, Keil, & Rai, 2004). Decisions about whether to continue or to terminate these troubled IT projects are among the most difficult that IT managers face. To come to a full understanding of these escalation situations and to advise managers appropriately, researchers must endeavor to sort out the rational forces from the nonrational. This study complements the large body of previous work examining nonrational factors that promote unwarranted escalation by considering a pervasive but subtle rational factor—the presence of real options—that could lead to warranted continuation.
The remainder of the article proceeds as follows. In the next few sections, we review the relevant literature on escalation, describe real options analysis (ROA), and use real options theory to develop our hypotheses. Then we describe the methodology and data collection and present our analyses and results. Finally, we discuss the theoretical and normative implications of these results, identify directions for future research, and conclude with a summary of the study’s key contributions.
Summary of Real Option Types:
An Overview of the Conjoint Experiment Approach:
Conjoint analysis is a multi-attribute judgment analysis technique based on Anderson’s (1981) Information Integration Theory that involves a posteriori decomposition of the respondent’s decision process, here project continuation (Louviere, 1988). There are three central elements in a conjoint research design: attributes, conjoint profiles, and part-worth and overall utilities. An attribute refers to a decision criterion (presence of various options) that respondents might use to evaluate the dependent variable (willingness to recommit). The overall value assigned to the dependent variable is referred to as its overall utility. Willingness to recommit can therefore be viewed as a multi-attribute decision-making problem in which managers integrate their knowledge of various attributes to arrive at an overall assesment about their willingness to recommit. The contribution of each attribute toward the formation of the overall utility of a project is called its part-worth utility. Different combinations of attribute levels are called treatments or conjoint profiles.
The conjoint technique requires respondents to make a series of judgments about a dependent variable based on a set of attributes from which the underlying structure of their cognitive system can be statistically inferred. A series of conjoint project profiles with different combinations of attribute levels were presented to each respondent, and the respondent provided an assessment of the dependent variables for each project profile. The five embedded real options were the attributes, and perceived option value and willingness to recommit were the dependent variables.
Based on the options included and excluded in each project profile, each respondent provided an assessment for the two dependent variables. (Details of the precise steps in the data collection process are described later in the data collection section.) The underlying structure of the respondents’ cognitive models regarding the influence of various options on the respondents willingness to recommit can be statistically inferred from these judgments by analyzing the responses at the individual and aggregate levels using multiple regression (Louviere, 1988).
The analyses are roughly comparable to repeated measures experimental de- sign and explicitly account for the nonindependence among the assessments of multiple project profiles by a single respondent. To identify attributes that are statistically significant at the aggregate level, the regression coefficient for each attribute is averaged across individuals and the sign of this coefficient indicates the direction of the relationship between the attribute and the dependent variable (Vancouver & Morrison, 1995). A Z-statistic aggregates the T statistics derived from the individual-level analysis for each attribute to assess the statistical significance of the attribute in predicting the dependent variable. Such analysis at both the individual respondent level and aggregate level increases the predictive ability of the model (Moore, 1980). These significance tests are supplemented with Hays’ (1973) ω2 , which measures the variance explained by each attribute and is used to assess the relative importance of each attribute (Louviere, 1988). Thus, beta values indicate the directionality and ω2 indicates the relative importance of each option.
Real Options and Escalation Rationality:
A key finding of this study is that the presence of options increases managers’ willingness to continue a troubled project. The central mediation role of perceived option value further suggests that managers recognize and implicitly value the embedded real options, and are, therefore, acting rationally in their decision-making process of whether to continue it. A large and statistically significant path coefficient of .757 (p < .001) between perceived option value and managerial willingness to recommit as well as the large variance explained by it (76%) suggests strong support for our idea that embedded real options will engender warranted continuation in troubled projects.
Continuation of a project with zero NPV should be largely explained by the value of the real options that it carries. The relationship between the total value of options (computed as a weighted sum of the options in each project, with weights derived from the regression model) and managers’ willingness to continue a troubled project is illustrated in Figure 3. The figure illustrates that managers are closer to the termination end of the scale and remain below the neutral point (i.e., 5) until the option bundle exceeds the standardized .5 level. The .5 level for perceived option value is equivalent to the value the average respondent placed on configurations with an average bundle of options present (two to three). Thus, managers appear to err on the side of being too conservative and risk averse when few real options are embedded. Our conjoint design included a conjoint profile in which there were no embedded options and the NPV was set to zero (profile 2 in Table 2). Other things being equal, subjects should be expected to be neutral toward continuing a zero NPV project with no embedded options. However, the mean willingness to continue score of 2.08 for this conjoint profile was well below the midpoint of 5, indicating neutrality. In terms of the frequency distribution, 84.3% of the respondents leaned against continuation of the project when no options were embedded. This suggests that there was considerable uncertainty in managers’ minds when they assessed the project’s future.
Relative Importance of Various Real Options:
A second important insight from the results is the relative importance that managers ascribe to each type of real option (beta weights in Table 3). Consistent with Hypothesis 2, managers ascribed more weight to strategic than to operational options. An embedded growth option had the highest ascribed importance in managers’ perceptions of a project’s option value. This result is consistent with both of the rationales presented earlier and with survey results reported by Busby and Pitts (1997). On one hand, it could be that managers ascribe more value to growth options because such options capture the complete value of one or more additional assets, rather than modifying the value of a single asset as with operational options. Or it could be that the framing explanation holds sway, that is, managers prefer options that decrease the probability of any loss over those that decrease the severity of potential losses should a loss occur.
Among the remaining options, managers ascribed the most value to the option to switch use. This option was much more highly valued than changing scale, staging investments, or abandonment. One possible explanation may be that some managers viewed the option to switch use as affecting payoffs similarly to a growth option, rather than as a tool to contain downside losses as with the other operating options. While the switch use option is traditionally conceived as an option that is exercised only when the original use has proved infeasible, software assets have two unique properties compared to other assets such as furnaces, manufacturing plants, or industrial machines in which real options analyses were originally conceptualized. The malleability of software suggests that it can be more easily modified than physical assets and, therefore, IT applications should be more easily repurposed compared to other kinds of assets. Moreover, due to the low cost of reproduction, a software-based system need not be discontinued from its original application before it can be repurposed. For example, a system originally developed for internal use can be sold to another firm even while the developing firm continues to use it. That is, due to their high malleability and low cost of duplication, software assets can be used simultaneously for their original purpose and for different purposes. These potential new purposes may be seen as a call option on separate but related assets. Viewed in this light, a switch use option can be interpreted as providing value in a way that is similar to growth options, which increases the chances of at least breaking even rather than decreasing the extent of the most severe losses.
Managers perceived the abandonment option as less valuable than the other options, which is consistent with prior survey results (Busby & Pitts, 1997). Because the abandonment option is likely to be exercised only on the least successful projects, the low value placed on this option is consistent with the framing argument presented earlier, where managers are less appreciative of options that only serve to curtail severe losses. Another possible explanation for the low value placed on abandonment is the difficulty of exercising this option in practice. Exercising this option is particularly disruptive and may cause morale and credibility problems among team members and other stakeholders, because they may have become personally invested in seeing the project completed. In contrast, exercising the growth and switch use options may even boost morale and evoke a sense of accomplishment among project stakeholders. Using similar logic, neither reducing the scale of a project nor staging investments is likely to be viewed as negatively as exercising the option to abandon. Changing scale implies that the project may still be implemented, though on a smaller scale than originally anticipated. The option to stage implies that it has been acknowledged all along that continued funding is contingent upon achieving interim milestones. In both cases, the degree of perceived failure may be less than that associated with the abandonment option.
To develop this line of reasoning further, it is possible that the signaling effects associated with exercising different options will lead managers to view some as being more valuable than others. Managers are likely to ascribe more weight to options that increase the opportunity to create a positive impression regarding project status. Abandonment is not likely to be exercised unless the manager believes that the project cannot be brought back on track, and even in these instances, the manager may be inclined to avoid abandonment in order to save face. In contrast, the option to switch use carries a positive connotation because it signals that the investment can be profitably salvaged by channeling the project in a new direction. This may explain why the option to switch use has the highest perceived value after the growth option. Exercising the option to change scale in response to a troubled project situation would signal that, although the project is troubled, the scope of the problem is still containable. Just higher than the abandonment option is the option to stage investments. Exercising this option signals that some, but not all, of the project can be salvaged and that managers had been taking a prudent approach to funding the project to that point. Overall, the pattern of results obtained suggests that managers’ ascribe more value to options that allow for positive signaling upon exercise.