CAPITAL BUDGETING PRACTICE
UNDER
ENVIRONMENTAL UNCERTAINTY
2
PREFACE
PREFACE
3
SUMMARY
SUMMARY
4
TABLE OF CONTENTS
TABLE OF CONTENTS
PREFACE ...2
SUMMARY ...3
TABLE OF CONTENTS ...4
1.
INTRODUCTION ...6
2.
LITERATURE REVIEW ...8
2.1 Review of studies toward capital budgeting practice ... 8
2.2 Concluding thoughts on previous research... 9
3.
THEORETICAL FRAMEWORK ...9
3.1 The Investment Principle and Capital Budgeting ... 10
3.2 Environmental Uncertainty ... 10
3.3 Uncertainty and Managerial Flexibility ... 11
3.4 Flexibility and Capital Budgeting ... 11
3.4.1 DCF Based Methods ... 11
3.4.2 Methods Dealing with Uncertainty ... 12
3.5 Conclusion ... 13
3.6 Capital Budgeting Methods and Their Costs ... 14
3.6.1 Intrinsic cost factors ... 14
3.6.2 Relative costs factors... 16
3.7 Hypotheses ... 17
4.
RESEARCH DESIGN ... 18
4.1.1 Measuring Environmental Dynamism & Complexity ... 18
4.2 Measuring Usage of Methods ... 19
4.3 Control variables ... 20
5
TABLE OF CONTENTS
4.4 Survey Design, Delivery and Response ... 21
5.
ANALYSIS AND RESULTS ... 23
5.1 Sample Description ... 23
5.2 Univariate Analysis ... 24
5.2.1 Capital Budgeting Methods versus Firm Characterstics ... 24
5.2.2 Capital Budgeting Methods versus Uncertainty Variables ... 26
5.3 Multivariate analysis ... 28
6.
CONCLUSION & DISCUSSION ... 31
APENDIX ... 33
A.1 Search field Kompass.com ... 33
A.2 Survey ... 34
B.1 RESULTS PROBIT MODEL 1 ... 36
B.2 RESULTS PROBIT MODEL 1 & 2 ... 37
REFERENCES... 40
6
INTRODUCTION
1. INTRODUCTION
Why do firms commit good money to risky investments? The answer is simple, to make a future profit. But how do managers decide on which projects to pursue and which not? Theory describes a fair amount of methods that allow managers to aggregate large amounts of information into meaningful projections of expected future profitability of an investment opportunity. Based on this information the managers makes a decision whether to invest or not. This process is called capital budgeting and the managers choice of a method is the primary unit of research of this thesis. These methods can be categorized in three different categories. The naïve methods, such as the payback period (PB), DCF based methods, such as Net Present Value (NPV), Internal Rate of Return (IRR) and Discounted Cash Flows (DCF), and the extended methods, such as Decision Tree Analysis (DTA), Real Options Valuation (ROV) and Simulation techniques (SIM). Because of the diversity in capital budgeting methods, ranging in sophistication and adherence to economic theory, it is an interesting question to ask what moves manager toward a certain capital budgeting method for basing investment decisions on. The literature always advocated the use of DCF based methods over the use of naïve methods, because they incorporate the main principles of time and risk and the use of cash flows. However, empirical results of previous studies show that managers do not always use sophisticated methods in their investment decisions; Naïve methods are still very popular.
This study seeks to explore a possible explanation for this behavior by arguing that a decision for a certain method is a consideration of cost and benefit. In context of capital budgeting, benefits would mean improved decision making based on calculations made with a certain method. This thesis builds a case that one reason for moving towards a certain method might be the way these methods deal with managerial flexibility under environmental uncertainty. Assuming that future managerial flexibility under uncertainty has a significant influence on the actual profitability of a project, this flexibility has a certain monetary value. Therefore one might expect that managers facing investment opportunities under higher uncertainty are inclined to choose those methods that are able to put a value on flexibility to base their investment decisions on. This paper argues that extended methods are better able to handle uncertainty as compared to naïve or discounted cash flow methods. Secondly, this thesis argues that there is a difference in cost of usage among the different methods which should be considered. Costs of performing and interpreting these analyses are not uniform among the different methods. This paper argues that extended methods are more costly in usage than DCF based methods and DCF based methods are more costly than naïve methods.
The consideration of cost and benefit of using a certain method is then consolidated in an expectation of behavior towards the decision for a certain method under environmental uncertainty.
Under low environmental uncertainty a manager is expected to be inclined to:
• choose for a DCF based method over PB because of its theoretical superiority.
• Choose a DCF based method over extended methods because both types of methods yield the same results and extended methods are more expensive in use.
Under high environmental uncertainty a manager is expected to be inclined to:
7
INTRODUCTION
• choose an extended method over a DCF based method because extended methods yield better results under high uncertainty.
The main goals of this paper are to deliver a theoretical basis for above hypotheses and secondly, test these hypotheses empirically;
8
LITERATURE REVIEW
2. LITERATURE REVIEW
The main reason for me to write a paper on this subject was dissatisfaction with the (lack of) explanations scholars who previously studied capital budgeting practices gave for phenomena they discovered. For instance, there was an apparent shortfall in the degree of adoption of theoretical sound capital budgeting methods. The main rhetoric in standard textbooks is to use the theoretical superior discounted cash flow method and avoid using methods like payback period which are crude and not founded in economic theory. However, Naïve methods are still very popular among managers. Secondly, it also appeared that the degree of adoption of DCF based methods was negatively correlated with firm performance. In this chapter the results of previous empirical studies on this subject are discussed.
2.1
REVIEW OF STUDIES TOWARD CAPITAL BUDGETING PRACTICE
Capital budgeting is the process of planning and managing expenditures for projects. It is a process that helps managers determine in which projects should be invested and in which not. This definition is based on the underlying assumption that firms should always maximize shareholder wealth, thus decisions should always support this. In a world of limited funds for investment in projects, capital budgeting is even more important. Within the realm of capital budgeting, many methods exist which offer the manager various degrees of informational quality to base their decision on. Existing literature makes a clear distinction between naïve methods and sophisticated methods. Methods like payback period (PB) are often called naïve, where methods based on discounted cash flows (DCF) are called sophisticated. Klammer (1973) defines capital budgeting sophistication as using discounted cash flow analysis and using a formal method of risk assessment. Haka et al (1985) use a similar definition by saying sophistication is based on whether naïve or theoretical superior methods are used. More recently a third group of methods came to life. The methods are extensions on DCF methods in that they implement optionality in their calculation methods and are thus more flexible than DCF based methods. I call this group the extended methods. Many scholars have started their studies with the assumption that capital budgeting sophistication increases profitability through improved decision making however, empirical results show that the managers’ choice of capital budgeting method does not appear to fall in line with this assumption.
For instance, the capital budgeting sophistication and firm performance school of research could not prove the relation that sophistication had a positive impact on firm performance. Klammer (1975) found that firms using the payback period method actually performed better than firms using discounting techniques. Pike (1984) found a negative relationship that was significant at the 5% level. Haka et al (1985) did not find any significant long-term improvements in market performance of firms adopting sophisticated capital budgeting techniques. A more recent survey performed by Farragher et al (2001) found negative relationship at the 10% significance level. The survey conducted by Kim (1982) is the only survey that found a significant positive result.
9
THEORETICAL FRAMEWORK
attributed this to macro environmental improvements like cheaper access to IT resources and better management education. Pike (1996) conducted a longitudinal study based on four replicated surveys, executed at approximately five-year intervals between 1975 and 1992. Many of his findings were consistent with the findings of Sangster (1993). He found that usage of DCF based methods with each survey increased. Unlike Sangster (1993) he still found a relationship of sophistication and firm size. Arnold and Hatzopoulos (2000) studied the degree of adoption of sophisticated methods and benchmarked it against results from studies of pike (1982, 1988 and 1996) and McIntyre (1985). The authors concluded that the degree of adaptation of DCF based methods increased. Although there is a trend of increasing adoption of more sophisticated methods, naïve methods like the Payback Period still are used widely among managers. This is underlined by the study of Graham & Harvey (2001). They found high popularity of PB among their US sample. Brounen et al (2004) replicated their study and found similar results for their European sample. Both found around 60% of the CFO always or almost always used the PB method. The fact that naïve methods like PB are still very popular is interesting at the very least.
Schall and Sundem (1980) conducted a research on the relationship between environmental uncertainty and capital budgeting methods employed by managers. They found that the use of sophisticated capital budgeting techniques tend to decline as environmental uncertainty increased. They argue that sophisticated methods may be more difficult to apply hence be more expensive in more uncertain environments. This could be due to the greater costs of accurately estimating the values of the variables used in more sophisticated approaches in highly uncertain environments. The cost of obtaining the desired data may be great and the measurement error large in highly uncertain environments, making sophisticated adjustments not worth the effort (Schall & Sundem, 1980).
2.2
CONCLUDING THOUGHTS ON PREVIOUS RESEARCH
By looking at the fact that the PB method is still as popular as it is, one might argue that managers do not share the assumption of the practical superiorness of sophisticated methods like NPV in project appraisal. Secondly, empirical proof that sophistication, at the very best, does not improve firm performance indicates that just using DCF based methods for project appraisal might not be the right way to improve decision making. A partial explanation might be given by Schall & Sundem (1980), that DCF based methods are no proper methods to base capital budgeting decisions on under environmental uncertainty. In the next chapter I try to put forward a more compelling explanation why DCF based methods are not suitable to valuate investment opportunities under uncertainty.
3. THEORETICAL FRAMEWORK
10
THEORETICAL FRAMEWORK
3.1
THE INVESTMENT PRINCIPLE AND CAPITAL BUDGETING
It is generally accepted that DCF based methods are superior to naïve methods because it allows for adjustments for time value and risk. When future pay off and their risk can be determined, the Net Present value of the expected future cash flows can be calculated. This could be easily applicable in a few cases in the real world, i.e. for zero coupon treasury bonds. However in the business environment, future pay offs and risk of a project more frequently than not is difficult to estimate. The lack of knowledge about risks and future pay offs is called uncertainty. Since DCF based method rely on reliable knowledge of the amounts and riskiness of future cash flows for reliable results, imperfect knowledge on these variables makes the results of a DCF analysis imperfect.
Because DCF based methods lack tools to adjust for uncertainty it is considered to be unfit to base resource allocation decision on under environmental uncertainty. This is underlined by for instance Lenos Trigeorgis in his book “ REAL OPTIONS: managerial flexibility and strategy in resource allocation” (2001), where he disserts that DCF based methods are not able to valuate managerial flexibility under environmental uncertainty and that therefore results based on these analysis methods are not reflecting true value. Pindyck & Majd (1987) show how using a simple NPV rule can lead to gross errors when valuing a investment opportunity under uncertainty.
Uncertainty and its relationship to capital budgeting is further explained in the next paragraphs.
3.2
ENVIRONMENTAL UNCERTAINTY
Uncertainty is different from risk; both are concepts that emerge from randomness. Risk is variability in which events have measurable probabilities. Uncertainty arises from imperfect knowledge about the way the world behaves or will behave in the future. It is the inability to assign probabilities to the outcomes of events.
In organizational theory, environmental uncertainty is divided into two dimensions: complexity and dynamism. Environmental complexity describes the number of units with which interaction is required and the extent to which an organization must have a great deal of sophisticated knowledge about products, customers and so on (Aldrich, 1979). Dynamism is the extent to which change occurs in the environment. Increasing dynamism means increasing difficulty of predicting future events. Complexity uses the number of elements as an explanatory variable while dynamism uses the rate of change of the elements as an explanatory variable.
11
THEORETICAL FRAMEWORK
3.3
UNCERTAINTY AND MANAGERIAL FLEXIBILITY
It is generally accepted among scholars that flexibility under uncertainty has value. Flexibility relates to the managers’ capacity to adjust course of action from prior defined plans and to change and/or exploit opportunities resulting from environmental changes and can be considered a company-specific skill or a resource. Flexibility is assumed to be one of the most important requirements for firms to survive and prosper in turbulent and unpredictable environments (Dreyer & Grønhaug, 2004). According to Lau (1996), flexibility has become the most important factor in achieving competitive advantage. In other words:
Under uncertainty, flexibility has value
3.4
FLEXIBILITY AND CAPITAL BUDGETING
3 .4.1 DCF BASED METHODS
An increasing number of academics and practicing managers are now convinced that the standard approaches like DCF to corporate resource allocation have failed to take into account flexibility in future actions. Dean (1951), Hayes & Abernathy (1980) and Hayes & Garvin (1982) recognized that standard capital budgeting methods employed by managers too often undervalue investment opportunities, leading to underinvestment and loss of competitive position. They argue that these methods have failed because it does not properly allow for modeling of the managers ability to adapt and revise decisions in response to unexpected changes in the environment.
Managerial operating flexibility and strategic adaptability are crucial to capitalizing successfully on favorable future investment opportunities and to limiting losses from adverse market developments or competitive moves (Trigeorgis, 2000). The DCF based analysis typically assumes that the project is committed to a predetermined path, thereby treating uncertainty as a value destroyer. It does not take into account that managers initial expectations might change due to the fact that uncertainty gradually resolves over time. The real market place is characterized by future uncertainty, causing initial expectations of its state and its actual state to be dissimilar. Flexibility to change a course of action or to limit losses when future states of the world gradually become apparent is valuable. The DCF-analysis works over an expected scenario, assuming a fixed sequence of actions within the project. A DCF-analysis is therefore not able to value managerial flexibility in the presence of uncertainty. Hence, the DCF-analysis is not an accurate valuation model to value investment decisions under uncertainty in the presence of managerial flexibility.
DCF based methods do not have the proper tools to measure the value of managerial flexibility under uncertainty
12
THEORETICAL FRAMEWORK
3 .4.2 METHODS DEALING WITH UNCERTAINTY
Extended methods like Decision Tree Analysis, Simulation techniques and Real Options Valuation are methods that try to solve the uncertainty problem. They are discussed in the following paragraphs. This discussion is by no means meant to be an extensive review of the methodologies; only the concepts are explained.
DECISION TREE ANALYSIS
Decision Tree Analysis (DTA) attempts to account for uncertainty and the possibility of later decisions by management. DTA helps management to structure the decision problem by mapping out all feasible alternative managerial actions, contingent to certain possible states of the world in a hierarchical manner. As such it is particularly useful for analyzing complex sequential decisions when uncertainty is resolved at distinct points in time (Trigeorgis, 2000). DTA addresses the problem of the deterministic or passive view of a DCF valuation by allowing taking different contingent scenarios into account. A traditional DCF analysis is basically the valuation of only one, or most likely, path through the branches of a decision tree. The recognition and analysis of possible other paths through uncertain states of the future costs considerable more effort and is thus more costly. DTA incorporates the ability of management to take decisions when future states of the world become clear in the case it is not yet committed to the project. DTA explicitly takes into account that outcomes can differ from initial expectations. In its simplest form, a DTA can be seen as a representation of a sequence of sub-projects in which decisions can be modeled that are contingent on some future state of the world. In case of favorable market conditions the manager might decide to go to the next stage of the project. In case of unfavorable market conditions he might decide to abandon the project and recover a certain amount of salvageable value. This flexibility is explicitly mentioned in DTA in terms of salvage value whereas DCF doesn’t.
SIMULATION TECHNIQUES
Another approach towards dealing with uncertainty in the business environment is simulation techniques. Traditional simulation techniques use repeated random sampling from probability distributions for each of the variables that has an impact on cash flows of a project. In this way, the uncertainty of a future cash flow forms its own risk factor. Simulation techniques can handle complex decisions with a large number of interacting variables in the presence of uncertainty. It attempts to imitate a real-world decision by using a mathematical model to capture the important functional characteristics of the project as it evolves through time and encounters random events (Trigeorgis, 2000). The output of such an analysis would be a set of probability distributions of all project cash flows for a given management strategy. The expected cash flow should then be discounted by their corresponding risk adjusted discount rate which is derived from the variability of the distribution of the cash flow. A single value NPV can be derived for clear cut decision making.
REAL OPTIONS VALUATION
13
THEORETICAL FRAMEWORK
means that a course of action that is apparently unprofitable on a simple NPV basis may be worth pursuing because it allows extra freedom of action. In this example, the acceptance of the first negative NPV project gives the option to expand in other markets in the future. Using ROV, the user can properly analyze these elements by thinking of investment opportunities as a collection of options on real assets by using option pricing techniques that are used in financial options. This is similar to the American call option, which gives the owner the right and not the obligation to acquire an asset at a specific date for a specific price. The owner of the option will exercise the option only when it is in his best interest to do so. There exists a close analogy between investment opportunities or real options and financial call options (Trigeorgis, 2000). When real options exist, an investment opportunity can be seen as a call option on a portfolio of the project value including the other real call options.
3.5
CONCLUSION
Concluding can be said that, at the conceptual level, extended methods are best able to deal with uncertainty in the business environment. Naïve methods are the least appropriate in most cases because they lack the ability to take into account risk and time value. DCF based methods ability diminishes with increasing uncertainty because of its inability to put a value on flexibility.
Figure 1
Uncertainty versus Usability per method
Based on this insight only, one could conclude that extended methods should yield the best results in all cases and should therefore be always used. However, empirical results on the adoption of extended capital budgeting do not support this statement. For instance, Graham & Harvey (2001) found that only 27% of the American managers always or almost always use Real Options Valuation and 14% use simulation techniques. When taking usage costs into consideration another picture emerges.
A more in depth discussion about costs is presented in the following paragraphs.
14
THEORETICAL FRAMEWORK
3.6
CAPITAL BUDGETING METHODS AND THEIR COSTS
The choice of a method will depend on the consideration of cost and benefit of that specific method, relative to the cost and benefit of alternative methods. It would be a quite reasonable assumption that for instance a decision maker selects a naïve method over a complex method, if the costs of using the more complex alternatives exceed its benefits. This chapter first goes into the intrinsic cost differences of the capital budgeting methods under investigation.
3 .6.1 INTRINSIC COST FACTORS
Intrinsic cost factors are costs that are inherent to properly using the various methods. The height of the costs depends very much on the method itself. For example, compared to a DCF analysis, PB methods do not require forecasting cash flows after the cut-off period. Neither does it require a discount factor in its calculation, which is considerably difficult to estimate. This makes the PB method cheaper in use. On the other hand, usages of extended methods are more expensive than DCF methods because more complex models are employed. Performing such an analysis and interpreting the results require a high degree of knowledge of financial and statistical theory.
The costs of the methods under investigation are compared by looking at the differences in cash flow
forecast requirements, discount rate requirements and model complexity.
The payback period is a simple method for making decisions concerning projects acceptance. The payback period is defined as the time it takes to earn back the initial investment. It simply subtracts the future cash flows from the initial investment and it looks at the time it takes when the investment is paid back. Since the payback period ignores cash flows after the cut-off or the time the initial investment is repaid, PB does not require estimating cash flows after this period. These limitations are also the reason why the payback period is so popular: It is simple, cheap in use and easy to communicate to people not familiar with financial theory.
DCF based methods views a project as a set of cash in- and outflows on various points in time that would be incurred and received if the project was undertaken. The NPV is the value of all future cash flows in the present today, discounted by an appropriate discount rate. A rigorous estimation of the discount rate of a certain project requires an estimation of the weighted average cost of capital (WACC) plus some sort of risk factor for the project in which is to be invested. Finding the proper discount rate for investment opportunities can be particularly easy for some types of investments, for instance investment in US treasuries, which require the risk free rate as a discount rate. On the other hand it could be very difficult in case of for instance investment in the biotechnology sector. DCF based methods are considered more expensive than PB because of stiffer cash flow forecast requirements, and the need for the estimation of a discount rate. In addition, the model complexity is higher.
15
THEORETICAL FRAMEWORK
or continuous) and decision nodes where the manager has the possibility to adjust the course of the project. In terms of cash flow forecasting this would mean that for all contingencies a forecast should be made. A traditional DCF analysis is basically the valuation of only one or most likely, path through the branches of a decision tree. The recognition and analysis of possible other paths through uncertain states of the future costs considerable more effort and is thus more costly. The decision-tree approach will strike some businessmen as complex. Certainly it is more complicated than rule-of-thumb approaches (Magee, 1964).
Table 1
Method Cash flow Discount rate Model complexity Intrinsic costof methods
PB Cash flow forecast until cut-off
None required.
Low: Add up cash in and outflows. The time it
takes until investments are repaid is PB.
IRR Cash flow forecast until end of project
None required for calculation
Moderate: Discount with arbitrary rate; repeat
process until IRR is found for NPV = 0 on trial and error basis.
NPV/DCF Cash flow forecast until end of project
Discount rate required
Moderate: Discount all cash flows with discount
rate.
DTA Multiple Cash flow forecasts contingent on certain future market conditions until end of project
Discount rate required
High: Model possible future market conditions
and decision nodes. Calculate their monetary impact. Select those cash flows resulting from most rational decision on certain market conditions. Discount selected cash flows.
SIM Expected cash flows and their risk are determined based on underlying variables and their probability distributions
Discount rate required
High: Set up model that represents the project.
Determine variables that have impact on cash flows. Determine probability distributions of variables. Randomly pick values from distribution and determine distribution of cash flows. Discount cash flows with adjusted discount rates.
ROV Multiple Cash flow forecasts contingent on certain future market conditions until end of project
Discount rate required
High: discount expected cash flows. In addition
value flexibility using some type of option pricing model.
With simulation techniques, cash flows and their variability or riskiness are estimated using underlying economical variables and their probability distributions. From these distributions, values are drawn and then entered in an econometrical model. The model then calculates the impact of the values of the underlying variables on the cash flows of the project. This is repeated a sufficient number of times so that a probability distribution of the cash flow is generated. The mean value of this distribution is then regarded as the expected cash flow. The variance in the distribution represents the riskiness. Simulations can handle complex decision problems under uncertainty with a large number of variables which can interact with one and another across time. However it is difficult to capture all the interdependencies. The algorithms which estimate the impact of the underlying variables on the expected cash flows require high knowledge of mathematical and econometrical theory which might incline managers to outsource model building to expensive external experts. The data requirement for this method is very high; for all underlying economic variables a considerable amount of data should be available for the probability distributions. In addition, for each project, a specific simulation model should be constructed since no project is alike.
16
THEORETICAL FRAMEWORK
sophisticated knowledge of option valuation, increasing its cost of usage. A number of professional managers have been concerned that, although the analogy relating managerial flexibility to options has intuitive appeal, the actual application to capital budgeting must be too complex for practical application (certainly more complex than DCF techniques) (Trigeorgis, 2000).
Table 1 is a summary of the intrinsic costs of using the various methods. Basically the PB method is the cheapest in use, DCF based methods are more expensive and the extended methods are most expensive.
3 .6.2 RELATIVE COSTS FACTORS
17
THEORETICAL FRAMEWORK
3.7
HYPOTHESES
The discussion of abilities and costs of the various methods boils down in three hypotheses:
HYPOTHESIS 1)
PROBABILITY OF USE OF DCF-BASED METHODS (NPV, IRR, DCF) INCREASES WITH A DECREASE IN ENVIRONMENTAL UNCERTAINTY
Under low uncertainty, expected cash flows and discount rates can be precisely determined. Because there is low uncertainty a DCF based analysis should yield the same quality of results as an extended method. In this case, a manager is expected to be inclined to choose a DCF based method over an extended method because extended methods are more expensive in use. In addition, under low uncertainty, a manager is expected to be inclined to choose a DCF based method over PB because of its theoretical superiority.
HYPOTHESIS 2)
PROBABILITY OF USE OF NAÏVE METHODS (PB) INCREASES WITH AN INCREASE IN ENVIRONMENTAL UNCERTAINTY
When uncertainty is high, a manager is expected to be inclined to abandon DCF based methods. Where increasing uncertainty reduces the managers’ ability to estimate expected cash flows and their appropriate discount rate(s), in combination with the fact that a traditional DCF analysis does not allow for valuation of flexibility under that same uncertainty, its usefulness under uncertainty is questionable. It might make sense that a manager does not want to waste valuable resources to make a relatively expensive DCF analysis. Short cutting the process of information seeking and dissemination by applying a rule of thumb method like PB can considerably reduces the effort and costs of analysis.
HYPOTHESIS 3)
PROBABILITY OF USE OF EXTENDED METHODS (DTA, ROV, SIM) INCREASE WITH AN INCREASE IN ENVIRONMENTAL UNCERTAINTY
18
RESEARCH DESIGN
4. RESEARCH DESIGN
This chapter goes into the design of the empirical investigation of the behavior of the manager towards capital budgeting methods under environmental uncertainty. Its goal is to deliver a set of data which enables us to test whether or not the financial manager behaves in the hypothesized manner as defined in chapter 3.7. In the next chapter measuring environmental uncertainty will be discussed. After that, measuring capital budgeting usage will be discussed. Then control variables will be discussed and after that the survey design and delivery will be discussed.
4 .1.1 MEASURING ENVIRONMENTAL DYNAMISM & COMPLEXITY
Like Schall & Sundem (1980), this study tries to find a relation between capital budgeting practice and environmental uncertainty. They defined five variables which represent a firm’s environment, size, level of debt, firm beta and industry beta. This study takes another approach to measuring environmental uncertainty; uncertainty is divided in the two separate constructs of both environmental dynamism and complexity. Both constructs are measured by means of perceived measures instead of archival measures. This definition will be explained in the next paragraph. Thereafter the measurement of environmental uncertainty will be treated.
A rather large body of scientific research has addressed the problem of operationalizing and measuring the concepts in a variety of contexts. Two main types of measurements are distinguished; archival measures and perceived measures. Both archival and perceived measures of environmental dynamism and environmental complexity have been utilized frequently in the strategy literature (Rasheed & Prescott, 1992). The archival measure of environmental dynamism has been consistently operationalized as the volatility of either industry sales or operating earnings. The archival measure of environmental complexity has been frequently operationalized as the sum of squared market shares in an industry (Harrington & Kendall, 2005). For perceived or self-reported measure of environmental uncertainty, dynamism, or complexity, most studies have considered dynamism only or combined dynamism and complexity into one uncertainty or turbulence construct (Harrington & Kendall, 2005). In their study, Harrington & Kendall (2005) found that knowledgeable managers’ perceptions match archival measures of dynamism and complexity. As a result this study measures the uncertainty concept using perceived complexity and dynamism variables, replicating the measures of Harrington & Kendall (2005).
Dynamism and complexity of a firms environment are assessed by asking survey questions about the firm’s industry. The questions used in this part of the survey are replicated from previous studies of Brews & Hunt (1999) and Harrington & Kendall (2005). The survey contained 11 questions concerning perceived environmental uncertainty of the firm. Each question, either assessing perceived dynamism or complexity, uses a 10-point Likert-type scale. The 10-point scale was used to ensure sufficient psychological and statistical spacing as well as to force respondents away from an automatic “middle of the road” or average response. The specific questions are shown in the appendix and discussed below.
Dynamism Scale. This dimension refers to the volatility or unexpected change in a firms environment.
19
RESEARCH DESIGN
questions concerning dynamism in a firms primary industry were included in the survey to measure perceived environmental dynamism. These seven items were intact measures used in previous research by Brews and Hunt (1999) and Harrington & Kendall (2005).
The following items are used:
1. Volatility in sales, on an annual basis 2. Volatility in earnings, on an annual basis 3. Rate of change in government regulations 4. Rate of change in technology
5. Rate of product obsolescence
6. Degree of pressure to develop new products/ applications 7. Degree of difficulty in forecasting industry trends/ changes
Complexity Scale. This dimension implies a consideration of the number, diversity, and distribution of
elements in a firms environment. The most recent archival measures of environmental complexity have two main dimensions: the number of firms in an industry and their relative inequalities in market share (Harrington & Kendall, 2005). For the self-report survey, four items were included to measure complexity; these four items were taken from previous surveys of Hart & Banbury (1994) and Harrington & Kendall (2005).
The following items are used:
8. Degree of technological complexity
9. Degree of complexity in the general business environment 10. Degree that your actions affect your competitors
11. Number of firms in your industry, compared to other industries
4.2
MEASURING USAGE OF METHODS
This paragraph focuses on measuring the usage of capital budgeting techniques for evaluation of projects. This study uses the same methodology of Graham & Harvey (2001), Brounen et al (2004) and Hermes et al (2006) by using the 5 point Likert scale, asking about the frequency of usage, ranging from never (0) to always (4). By using the same methodology, it is possible to compare our results with theirs.
In the section of naïve methods PB is selected; Graham & Harvey (2001) found that 57% almost always used this method. Brounen et al (2004) found around 55% and Hermes et al (2006) found that 84% of Chinese CFO’s almost always uses the PB method. ARR was also included in the survey, however due to very low response on this method it was excluded from the analysis.
20
RESEARCH DESIGN
In the section of extended methods, Decision Tree Analysis, Simulation and Real Option Valuation is inquired about. Graham & Harvey (2001) found that 27% almost always used the ROV method and 14% almost always use simulation techniques. Brounen et al (2004) found that around 40% almost always uses ROV 18% almost always uses simulation techniques. Hermes et al (2006) did not discern added these questions in their research. DTA was added to the survey as well because it is a distinct methodology, and important to the hypotheses.
4.3
CONTROL VARIABLES
This thesis is interested in the relationship between choice of a capital budgeting method and environmental uncertainty. To measure if uncertainty affects the choice of capital budgeting methods we must also measure other variables that might affect this choice and include this in the analysis. Although these variables are not relevant to this particular study, they have been found to significantly affect the decision on a particular method in numerous previous studies. By including these variables we can separate their effect from the uncertainty variables.
4 .3.1 COMPANY SIZE
Company size has found to be related to the choice of capital budgeting methods. Graham & Harvey (2001), Brounen et al (2004) and Arnold & Hatzopoulos (2000), among others found significant evidence of this. The effect that was mainly found was that size was positively related to the usage of DCF based methods. Smaller firms were more likely to use PB methods (Graham & Harvey, 2001). This size effect can be attributed to the fact that large companies are able to employ a full time staff for capital budgeting tasks. Secondly, large firms make large capital expenditures so the total cost of performing analysis is smeared out over the entire project. Relative costs of capital budgeting are thus lower for large companies through economies of scale. Firm size has been measured by total assets by many previous studies. It is considered a superior measure for this kind of research because it is more directly related to capital budgeting, compared to total revenues.
OWNERSHIP
Graham & Harvey (2001) found that publicly owned companies used DCF based methods significantly more compared to privately owned companies. This may be explained by the fact that financial markets and especially stock markets have developed over time. The shareholder maximization concept has gained in importance which has created pressure on the firms’ managers to use DCF based methods over naïve methods.
PERCENTAGE OF FOREIGN SALES
21
RESEARCH DESIGN
MANAGERS CHARACTERISITCS
Graham & Harvey (2001) found that CFO’s with an MBA degree were more likely to employ sophisticated methods compared to non-MBA CFO’s. Brounen et al (2004) found a similar relationship. Concerning the age and tenure of the CFO, Graham & Harvey (2001) found that older managers and managers with longer tenures are more inclined to use the PB method. Better trained CFO’s enables him to better understand and use more sophisticated methods.
4.4
SURVEY DESIGN, DELIVERY AND RESPONSE
The survey consists of three parts and is presented in appendix A2. The first part asks about firm characteristics such as size in assets, industry, country, ownership, foreign sales and business breadth, as well as CFO characteristics such as tenure, education and age. The second part of the survey asks the respondent to rate their primary industry. 7 questions were included that measure perceived environmental dynamism and 4 questions were included that measure perceived environmental complexity. All the questions use a 10 point Likert-type scale to ensure sufficient psychological and statistical spacing as well as to force the respondent from an automatic middle of the road response. The third part goes into the firms capital budgeting practice. This part asks about the capital budgeting method, the hurdle rate, and the estimation method of cost of equity and debt. All the questions use a 5 point Likert scale, asking about the frequency of usage, ranging from never (0) to always (4). This methodology was also used by Graham & Harvey (2001), Brounen et al (2004) and Hermes et al (2006), making it possible to compare the results with theirs.
In order to obtain data, the survey was distributed among 2156 CFO’s of firms in various countries in western Europe. The addresses of the firms and the names of their respective CFO’s were obtained from the compass database. The search function of Kompass allows the user to make specific search queries. A screen dump of the search screen is showed in appendix A1. I selected the following search options: producing company, headquarters, more than 250 employees and for availability of the name of the CFO and the company address. This yielded 2156 Belgian, Danish, French, German, Dutch and Swiss firms.
Table 2
Country Response Total sent % response Response per country Belgium 45 363 12% Denmark 41 247 17% France 27 475 6% Germany 13 271 5% Netherlands 49 624 8% Switzerland 16 175 9% Blank 1 Total 192 2156 9%22
RESEARCH DESIGN
responses were obtained through the website and 159 by mail, yielding a total of 192 responses. The 33 respondents who visited the website took, on average, 8 minutes and 4 seconds to complete the survey. The response per country is outlined in Table 2.
The response rate of 9% is comparable with recent studies in this field. Graham and Harvey (2001), Brounen et al (2004) and Hermes et al (2006) respectively obtained 9%, 5% and 16% response rates.
23
ANALYSIS AND RESULTS
5. ANALYSIS AND RESULTS
This chapter describes the survey results. It investigates whether the choice of capital budgeting methods are affected by environmental uncertainty. First it discusses the sample description, and then a univariate analysis will be presented. Finally a multivariate analysis is presented in order to control for other factors that might affect the choice of capital budgeting methods.
5.1
SAMPLE DESCRIPTION
Table 3
Industry Frequency (#) Percent (%)Firm Characteristics Basic Materials 32 16,7% Capital Goods 35 18,2% Consumer Goods 31 16,1% Energy 9 4,7% Financial 3 1,6% Healthcare 11 5,7% Services 24 12,5% Technology 28 14,6% Transportation 8 4,2% Utilities 8 4,2% Retail 1 ,5%
Ownership Frequency (#) Percent (%)
Public 51 26,6%
Private 140 72,9%
Size in Assets Frequency (#) Percent (%)
(in Euro) <10m 3 1,6% 10-24m 19 9,9% 25-99m 70 36,5% 100-499m 66 34,4% 500-999m 6 3,1% 1-5bn 17 8,9% >5bn 5 2,6%
Foreign Sales Frequency (#) Percent (%)
(% of total sales) 0 29 15,1
1-24 33 17,2
25-49 23 12,0
>49 103 53,6
Business Breadth Frequency (#) Percent (%)
Single line 69 35,9%
Multi line 107 55,7%
24
ANALYSIS AND RESULTS
In tables 3 and 4 the sample description are presented. It describes the frequencies and percentages of the respondents firm and CFO characteristics, giving a global view on the population sample.
Table 4
Age Frequency (#) Percent (%)CFO Characteristics (in years) <40 38 19,8% 40-49 74 38,5% 50-59 67 34,9% >59 10 5,2%
Tenure Frequency (#) Percent (%)
(In years) <4 70 36,5%
4-9 62 32,3%
>9 49 25,5%
Education Frequency (#) Percent (%)
Bachelor 27 14,1% Other Master 20 10,4% Msc 47 24,5% MBA 69 35,9% >Master 24 12,5%
5.2
UNIVARIATE ANALYSIS
In this paragraph the results of the univariate analysis are presented. The statistical method employed here is the independent students T test. With this method the mean frequencies of use of a capital budgeting method are compared among the two groups per firm and CFO characteristic. It tests whether the means are significantly different from each other.
5 .2.1 CAPITAL BUDGETING METHODS VERSUS FIRM CHARACTERSTICS
25
ANALYSIS AND RESULTS
Table 5
PB IRR NPV DCF DTA ROV SIMCapital budgeting methods used % 3 & 4 Scores 69% 55% 63% 59% 7% 10% 2% Mean score 2,85 2,29 2,51 2,51 0,67 0,61 0,27 Total assets <100m (0) 2,85 1,99 2,04 2,15 0,72 0,68 0,15 >100m (1) 2,91 2,61 2,92 2,88 0,62 0,58 0,40 0,29 2,88*** 4,19*** 3,69*** 0,65 0,75 2,30** Ownership Private (0) 2,96 2,21 2,39 2,47 0,68 0,64 0,29 Public (1) 2,59 2,57 2,82 2,65 0,65 0,51 0,20 1,74* 1,48 1,82* 0,77 0,19 0,67 0,8 Foreign Sales <50% (0) 2,66 2,13 2,42 2,54 0,79 0,69 0,34 >50% (1) 3,03 2,45 2,54 2,52 0,56 0,51 0,17 1,90* 1,44 0,55 0,08 1,55 1,08 1,65* Business breadth Single (0) 2,93 2,26 2,52 2,41 0,65 0,52 0,17 Multi (1) 2,81 2,30 2,52 2,57 0,70 0,69 0,34 0,60 0,18 0,03 0,78 0,30 0,95 1,44 Age of CFO <50 year (0) 2,94 2,33 2,47 2,48 0,73 0,67 0,29 >50 year (1) 2,75 2,26 2,55 2,64 0,60 0,56 0,25 0,93 0,32 0,33 0,75 0,90 0,65 0,44 Tenure <4 years (0) 2,86 2,09 2,50 2,43 0,63 0,63 0,30 >4years (1) 2,92 2,43 2,55 2,62 0,68 0,61 0,24 0,30 1,50 0,22 0,91 0,37 0,09 0,53 Education <MBA (0) 2,84 2,35 2,68 2,50 0,59 0,55 0,31 >MBA&PhD (1) 2,90 2,29 2,35 2,55 0,73 0,66 0,23 0,32 0,28 1,51 0,24 1,00 0,61 0,76
Most respondents use the Payback Period (PB) always or almost always (3 or 4 score on the scale). 69% of the respondents choose this method. Brounen et al (2004) found similar popularity of PB among their European sample. Graham & Harvey (2001) found that 57% of the American managers always or almost always used the PB method, the third most popular method behind NPV (75%) and IRR (76%). Hermes et al (2006) found 79% of the Dutch respondents and 84% of the Chinese respondents always or almost always use the PB method, coming in second behind NPV (89% in the Dutch sample) and IRR (89% in the Chinese sample). This sample shows popularity of the DCF based methods that is similar with the sample of Brounen et al (2004). With regard to the extended methods (DTA, ROV, SIM), this sample shows significant less popularity among managers compared to the other studies. For instance, Graham & Harvey (2001) found 27% of the American managers always or almost always use Real Options Valuation and 14% for simulation techniques, against 10% found in this study.
The fact that PB is used most often by the managers in this sample is interesting because its shortcomings are discussed in textbooks for many decades.
26
ANALYSIS AND RESULTS
1% level) more likely to use a DCF based method. The same size effect was recorded by Graham & Harvey (2001), Brounen et al (2004) and Hermes (2006). However, this study found no effect of firm size on the use of PB, where previous studies found a negative size effect. Looking at ownership of a firm, private firms are significantly (10% level) more likely to use the PB method. Public firms are significantly (10% level) more likely to use the NPV method. Similar results were found by Graham & Harvey (2001). When taking foreign sales into account, the use of PB is significantly higher (10% level) in firms with high foreign sales. This result contradicts the theory that high foreign sales results in more FX risk and thus needs an adjusted discount rate. However, foreign sales might have been interpreted by the respondents as sales within Euro carrying countries as well. FX risk is not relevant in that case. The opposite is true for the extended methods, where low foreign sales companies use these methods more (10% significant for SIM). CFO characteristics seem to have little influence on the choice of a capital budgeting method in this sample. No significantly different means were found.
5 .2.2 CAPITAL BUDGETING METHODS VERSUS UNCERTAINTY VARIABLES
Table 6
(1) (2) (3) (4) (5) (6) (7) Capital budgeting & Environmental Dynamism Mean Score 4,65 5,41 4,42 4,90 3,94 5,85 5,44 PB 0, 1 & 2 scores (0) 4,30 5,32 4,43 4,69 4,00 5,88 5,23 3 & 4 Scores (1) 4,81 5,45 4,42 4,99 3,92 5,84 5,54 1,61 0,48 0,05 0,96 0,26 0,12 1,09 IRR 0, 1 & 2 scores (0) 4,59 5,38 4,44 5,07 4,22 5,93 5,38 3 & 4 Scores (1) 4,70 5,43 4,41 4,76 3,71 5,79 5,49 0,35 0,19 0,12 1,07 1,77* 0,44 0,41 NPV 0, 1 & 2 scores (0) 4,63 5,36 4,39 5,24 4,43 6,01 5,50 3 & 4 Scores (1) 4,67 5,44 4,44 4,70 3,66 5,76 5,41 0,13 0,29 0,17 1,84* 2,43** 0,81 0,34 DCF 0, 1 & 2 scores (0) 4,67 5,59 4,45 4,96 4,14 5,60 5,49 3 & 4 Scores (1) 4,64 5,29 4,40 4,86 3,80 6,03 5,74 0,09 1,11 0,15 0,35 1,16 1,36 0,28 DTA 0, 1 & 2 scores (0) 4,69 5,43 4,47 4,95 3,89 5,89 5,46 3 & 4 Scores (1) 4,21 5,14 3,86 4,64 4,64 5,43 5,29 0,83 0,57 1,05 1,22 1,37 0,78 0,34 ROV 0, 1 & 2 scores (0) 4,65 5,40 4,43 4,90 3,91 5,91 5,45 3 & 4 Scores (1) 4,68 5,53 4,37 4,95 4,28 5,37 5,37 0,07 0,29 0,12 0,11 0,75 1,05 0,19 SIM 0, 1 & 2 scores (0) 4,67 5,38 4,39 4,90 3,90 5,83 5,48 3 & 4 Scores (1) 3,75 7,00 6,00 5,00 5,75 6,75 3,50 1,83* 1,76* 1,53 0,10 1,85 0,86 2,20**27
ANALYSIS AND RESULTS
This procedure was also employed in previous studies by Graham & Harvey (2001), Brounen et al (2004) and Hermes (2006).
The environmental dynamism variables are coded as:
(1) Volatility in sales; (2) Volatility in earnings;
(3) Rate of change in government regulations; (4) Rate of change in technology;
(5) Rate of product obsolescence;
(6) Degree of pressure to develop new products/ applications and (7) Degree of difficulty in forecasting industry trends/ changes. The environmental complexity variables are coded as:
(8) Degree of technological complexity;
(9) Degree of complexity in the general business environment; (10) Degree that your actions affect your competitors;
(11) Number of firms in your industry, compared to other industries.
28
ANALYSIS AND RESULTS
The first thing that stands out is that the univariate analysis does not return strong results. Only a few variables are statistical significant with the usage of methods. In table 8 the results are summarized in a simple sign analysis. Positive means that the respondents who reported they almost always or always use a specific method reported higher than average on the uncertainty variables. Negative means the opposite; respondents who always or almost always use a certain method report lower than average on the uncertainty variables.
Table 8
PB IRR NPV DCF DTA ROV SIMSimple Sign Analysis Methods versus Uncertainty variables (univariate) Significant (Positive) 1 1 1 1 1 0 2 Insignificant (Positive) 6 5 4 3 2 5 6 Significant (Negative) 0 1 3 1 0 0 2 Insignificant (Negative) 4 4 5 5 7 7 1
The most interesting results come from users of PB and NPV. Firstly, the users of PB appear to show a weak tendency to report higher than average on the uncertainty variables. Only one variable is significantly positive though; “degree of technological complexity” (8). Six variables show a positive, but insignificant relation. Four variables show a negative insignificant relation.
NPV users seem to show a tendency to report lower values on some of the uncertainty variables, meaning that NPV is used less in uncertain environments. NPV users report significantly lower numbers for variables “rate of change in technology” (4), “rate of product obsolescence” (5) and “number of competitors” (11). Variable “Degree that your actions affect your competitors (11)” shows a significant positive relation though.
There is no clear tendency apparent from users of IRR, DCF, DTA and ROV.
Based on these results it becomes difficult to conclude anything about the hypotheses. There is no clear indication that managers prefer either naïve methods or extended methods over the DCF based methods in case of high environmental uncertainty, as predicted by the hypotheses.
However, the differences from mean analysis does not allow for adjustment by the control variables which might have an impact on choice of capital budgeting. This is done by the multivariate analysis which includes control variables. This is discussed in the next chapter.
5.3
MULTIVARIATE ANALYSIS
In the previous paragraph the univariate analysis showed very weak statistical results. This paragraph presents a more advanced analysis in the form of a Probit analysis. The main benefit of a multivariate analysis is that it allows for control of factors that influence capital budgeting choice other than the environmental uncertainty variables, which are central in this thesis. With this analysis we can control for the impact of the other explanatory variables on the choice of capital budgeting methods.
29
ANALYSIS AND RESULTS
uncertainty factors. The second model adds control variables to see if the model improves and if the uncertainty relation still holds when other determinants are added.
Identical to the univariate analysis, the dependent variables are converted into dichotomous variables. Where the manager indicated that he always or almost always uses a certain capital budgeting method (3 or 4 score) the variable is converted into 1, and otherwise into 0. The environmental uncertainty variables are kept in their original form, in a 1 to 10 scale. Concerning the control variables, I chose foreign sales, ownership and size in assets variables because these had the greatest influence on the choice of capital budgeting method according to the univariate analysis (see table 5). Again, the conversion of the control variables into dichotomous variables is the same as in the univariate analysis. Because of their rather large size, the table for capital budgeting methods is placed in appendix B.
In table 9 the results of the second model are summarized in a simple sign analysis. In this table, the positive or negative beta’s (or slope of the regression line) are counted, where positive betas means that the respondents who reported they almost always or always use a specific method reported higher than average on the uncertainty variables. Negative means the opposite; respondents who always or almost always use a certain method report lower than average on the uncertainty variables.
Table 9
PB IRR NPV DCF DTA ROV SIMSimple Sign Analysis Methods versus Uncertainty variables (Multivariate) Significant (positive) 3 0 0 1 1 0 3 Insignificant (positive) 7 7 3 1 3 3 4 Significant (negative) 0 0 3 1 1 0 1 Insignificant (negative) 1 3 5 4 5 7 2
The results appear to be improved, compared to the univariate analysis. Using the multivariate model which controls for control factors, the use of the PB method shows a stronger tendency to increase under higher environmental uncertainty. When looking at the improved model, which includes the control variables, the users of PB report significant higher (at the 10% level) values on uncertainty factors “volatility in sales” (1), “rate of change in technology” (4) and “degree of technological complexity” (8). For all the other factors except factor “degree of complexity in the business environment” (9), the relation is positive but insignificant.
30
ANALYSIS AND RESULTS
The use of the simulation techniques shows a tendency to increase under higher environmental uncertainty. Because of the relatively low number of users of SIM, controlling for all factors was not possible. Only size in assets was included as a control factor. Users of SIM report significant higher (at the 10% level) values on uncertainty factors 2 (volatility in earnings), 5 (rate of product obsolescence) and 9 (degree of complexity in the business environment). For factor 7 (degree of difficulty in forecasting industry trends) the relation is significantly negative. For all the other factors except factor 1 (volatility in sales) and 8 (degree of technological complexity), the relation is positive but insignificant.
31
CONCLUSION & DISCUSSION
6. CONCLUSION & DISCUSSION
This paper argues that the choice of capital budgeting methods may be related to the level of environmental uncertainty a firm is facing. The argument boils down into three hypotheses:
(1) PROBABILITY OF USE OF DCF-BASED METHODS (NPV, IRR, DCF) INCREASES WITH A DECREASE IN ENVIRONMENTAL UNCERTAINTY
(2) PROBABILITY OF USE OF NAÏVE METHODS (PB) INCREASES WITH AN INCREASE IN ENVIRONMENTAL UNCERTAINTY
(3) PROBABILITY OF USE OF EXTENDED METHODS (DTA, ROV, SIM) INCREASE WITH AN INCREASE IN ENVIRONMENTAL UNCERTAINTY
This paper has investigated these hypotheses by gathering information on the use of capital budgeting techniques employed and perceived environmental uncertainty variables among managers. This information was obtained via a survey sent to 2156 CFO’s active in corporations in 6 developed European countries. 192 usable responses were retrieved, yielding a 9% response rate.
With this information, we carried out statistical analyses, comprising of standard differences of means tests and multivariate regression analyses to see whether environmental uncertainty has an impact on the choice of a capital budgeting method. We focused on whether perceived environmental uncertainty variables had a significant impact on employed capital budgeting methods.
In summary, the univariate analysis returned weak results. The significance of the correlation between usage of methods and the uncertainty variables was disappointing. The results give no clear indication that uncertainty plays a role in the manager’s decision on using a specific method. Based on these results it becomes difficult to prove or disprove any of the hypotheses.
When performing the multivariate probit analysis, which includes the control variables “ownership”, “percentage in foreign sales” and “size”, the results improved slightly. Firstly, usage of PB seems to positively correlate significantly with 3 uncertainty variables. 7 variables correlated positively, although insignificant. Only one variable showed insignificant negative correlation. These results are weakly supportive of the second hypothesis. Secondly, the probit analysis revealed that NPV usage negatively correlates significantly with 3 uncertainty variables. 5 variables negatively correlate insignificantly. 3 variables show positive but insignificant correlation. These results are weakly supportive of the first hypothesis. Thirdly, SIM usage positively correlates significantly with 3 uncertainty variables. 4 variables positively correlate insignificantly. 1 variable showed negative significant correlation and 2 variables showed insignificant negative correlation. These results are weakly supportive of the third hypothesis. It needs to be noted here that there were a low number of respondents who used simulation techniques, hindering proper statistical analysis. It was not possible to include all three control variables here. The results for the IRR, DCF, DTA and ROV methods were inconclusive. Based on these weak indications, it becomes difficult to make strong conclusions about the hypotheses.
32
CONCLUSION & DISCUSSION
A second proposed way of improvement is the method of data collection. Data collection by means of survey has been problematic and has received criticism by the likes of i.e. Rappaport (1979) and Schall & Sundem (1980). Performing a large scale field study should address the problems of gathering data using surveys.
33
APENDIX
APENDIX
34
APENDIX
35
36
APENDIX
B.1 RESULTS PROBIT MODEL 1
PB IRR NPV DCF DTA ROV SIM
Est. Z Est. Z Est. Z Est. Z Est. Z Est. Z Est. Z
37
APENDIX
B.2 RESULTS PROBIT MODEL 1 & 2
(1) (2) (1) (2) (1) (2) (1) (2) (1) (2) (1) (2) (1) (2) Intercept 0,14 0,16 0,06 -0,41 0,29 0,02 0,26 -0,14 -1,2*** -1,15*** 1,31*** -1,01*** -1,56*** -1,83*** 0,6 0,54 0,25 -1,48 1,26 0,09 1,12 -0,49 -3,67 -2,9 -4,24 -2,77 -2,96 -2,92 1) 0,08 0,09* 0,02 0,04 0,01 0,01 0 0 -0,06 -0,04 0 -0,02 -0,11 -0,11 1,61 1,71 0,36 0,86 0,14 0,12 -0,09 0 -0,82 -0,53 0,07 -0,26 -0,92 -0,85 Foreign 0,11 0,25 -0,08 0,11 -0,54* -0,21 0,52 1,28 -0,41 0,58 -1,78 -0,77 owner -0,47** 0,12 0,18 0,06 0,07 -0,31 -2,07 0,6 0,79 0,78 0,22 -0,91 size 0,11 0,44** 0,48** 0,62*** 0,14 -0,13 0,42 0,51 2,22 2,43 3,12 0,44 -0,48 0,91 McFadden 0,01 0,04 0 0,04 0 0,03 0 0,05 0,01 0,04 0 0,02 0,02 0,05 Intercept 0,35 0,41 0,08 -0,39 0,24 -0,03 0,54* 0,11 -1,24*** -1,08** -1,39*** -1,01** -3,58*** -4,11*** 1,2 1,19 0,28 -1,15 0,83 -0,08 1,88 0,34 -3,1 -2,28 -3,66 -2,31 -3,43 -3,42 2) 0,03 0,02 0,01 0,03 0,01 0,01 -0,06 -0,05 -0,04 -0,04 0,02 -0,01 0,25 0,28* 0,49 0,44 0,19 0,56 0,3 0,26 -1,11 -0,87 -0,55 -0,59 0,28 -0,19 1,64 1,75 Foreign 0,15 0,26 0,08 0,12 -0,55* -0,21 0,75 1,37 -0,4 0,3 -1,84 -0,8 owner -0,48** 0,12 0,18 0,05 0,07 -0,3 -2,12 0,52 0,8 0,23 0,23 -0,9 size 0,07 0,43** 0,48** 0,61*** 0,14 -0,13 0,57 0,36 2,18 2,44 3,07 0,44 -0,48 1,14 McFadden 0 0,03 0 0,03 0 0,03 0 0,05 0 0,04 0 0,02 0,09 0,13 Intercept 0,5** 0,45 0,15 -0,29 0,29 0,1 0,27 -0,11 -1,15*** -0,8** -1,26*** -1,01*** -2,76*** -3,02*** 2,26 1,59 0,73 -1,07 1,33 0,38 1,24 -0,43 -3,69 -2,1 -4,31 -2,82 -4,68 -4,5 3) 0 0,02 -0,01 0,01 0,01 -0,01 -0,01 0 -0,07 -0,13* -0,01 -0,02 0,14 0,15 -0,05 0,44 -0,12 0,32 0,17 -0,25 -0,15 -0,09 -1,05 -1,65 -0,12 -0,25 1,44 1,46 Foreign 0,16 0,28 -0,08 0,11 -0,62** -0,22 0,81 1,42 -0,42 0,57 -1,99 -0,82 owner -0,49** 0,1 0,18 0,07 0,15 -0,3 -2,17 0,46 0,8 0,29 0,44 -0,89 size 0,07 0,42** 0,48** 0,62*** 0,17 -0,13 0,42 0,33 2,15 2,43 3,14 0,53 -0,47 0,9 McFadden 0 0,03 0 0,03 0,01 0,03 0 0,05 0,01 0,07 0 0,02 0,06 0,08 Intercept 0,27 0,13 0,38 -0,01 0,76*** 0,58** 0,33 0,04 -1,03*** -0,09** -1,32*** -1,07*** -2,09*** -2,29*** 1,06 0,45 1,56 -0,04 2,99 2 1,32 0,12 -2,88 -2,15 -3,96 -2,83 -3,72 -3,71 4) 0,05 0,09* -0,05 -0,05 -0,09* -0,11** -0,02 -0,04 -0,09 -0,1 0,01 0 0,01 0 0,97 1,69 -1,07 -0,93 -1,83 -2,28 -0,35 -0,76 -1,23 -1,3 0,11 -0,05 0,1 -0,01 Foreign 0,16 0,27 -0,08 0,11 -0,57* -0,21 0,77 1,39 -0,41 0,6 -1,87 -0,8 owner -0,54** 0,13 0,24 0,08 0,13 -0,3 -2,35 0,59 1,04 0,37 0,4 -0,89 size 0,07 0,42** 0,49** 0,62*** 0,19 -0,13 0,44 0,32 2,15 2,49 3,15 0,61 -0,47 0,95 McFadden 0 0,04 0 0,04 0,01 0,05 0 0,05 0,02 0,06 0 0,02 0 0,03 SIM PROBIT Analysis Model 1 & 2
