Department of Management Science & Engineering

Number 7/Fall 2000

Investment Science Newsletter

By Professor David G. Luenberger
luen@stanford.edu

Investment Science Home

Back to Newsletter

Forum

CONTENTS

FORUM

THE WEBSITE

NEW PROJECTS

SHORT COURSE NEWS

NEW DISSERTATION

PRICE CYCLES IN COMPETITIVE CAPITAL-INTENSIVE INDUSTRIE

This is the seventh newsletter describing events in the Investment Science program at Stanford. It has been a busy period recently, with active development of the theory and participation in interesting and important projects in industry. Each of these newsletters contains a short technical article as well as general news about investment science. In this issue, we are extremely fortunate to have Marius Holtan as a guest author of a technical article on cycles. Dr. Marius Holtan, of Onward, Inc., received his Ph.D. from Stanford University in 1997, and he is an expert in both the theory and application of investment science. One of his projects has been the application of investment science to industries characterized by heavy capital expenditures and price cycles. He has shown that conventional methods for valuing and managing capital assets in these industries are grossly inferior to methods based on modern techniques of investment science. Indeed, over the long run, a company that employs the newer strategies will likely grow substantially faster than the industry as a whole, benefiting from the cycles rather than being frustrated by them. I believe you will find his article fascinating and informative.

The Website

Remember that the Investment Science web site is available. It is located at http://www.investmentscience.com. You will find copies of past issues of this Newsletter at the site, as well as additional information about investment science projects and software tools. Your comments on the site are welcome.


New Projects

In addition to continuing work, we have carried out some new interesting projects this year, including several student projects done for local companies done in conjunction with the Investment Practice course taught at Stanford by Robert Luenberger. In addition, two students worked on a comprehensive project for HP Labs, and we have carried out a very interesting new area of investment science at Cadence Design Systems, Inc. The span of application areas is increasing beyond project valuation and management to design of pricing systems, study of business models, development of R&D strategies, and enhancement of intellectual property values. Sound financial theory coupled with strong modeling and computation is a formula that has enormous potential.


Short Course News

Short Course News Investment Science for Industry was given at Stanford in April. We had a very good group in attendance and very much enjoyed the interaction. The course content continues to evolve, partly because of advances we make at the university and partly because of the feedback and ideas that we get from participants of the course. The course will next be given this September 14-15. Several past participants have referred colleagues to the course, and this has become a good mechanism for spreading the ideas of investment science. I hope you will consider presenting the idea of the course to your colleagues. You may refer them to the course though the web site http://www.stanford.edu/~luen or to the investmentscience.com web site mentioned above.


New Dissertation

Mark Erickson obtained his Ph. D. in Management Science and Engineering, at Stanford University this past June. His Dissertation is titled, "Real and Financial Options: Theory and Lattice Techniques." This work makes a substantial contribution to both the theory and practical application of investment science. He presented a comprehensive theory for valuing options that are defined in terms of observable quantities that do not have prices. For example, one might have an option based on the weather or on the size of the market for a certain product. He also presented an outstanding analysis of various lattice methods for representing uncertain processes and showed how to construct lattices that are both computationally efficient and technically accurate. His work is an important piece within the investment science framework.


Price Cycles in Competitive Capital-Intensive Industries
By H. Marius Holtan, Ph.D.

Overview
Severe earnings fluctuations, or cycles, coming about as a result of large variations in the product's price, is a defining characteristic of many industries. Prominent examples of such industries are petrochemical, semiconductor, paper, and construction. In fact, most if not all commodity industries experience serious earnings fluctuations.

Capacity planning can be difficult in a cyclical environment. In particular, while cycles may at first glance appear to be periodic they also exhibit strictly random behavior. This randomness, coupled with large fluctuations, long lead times from initiating capacity construction to final use, and large capital expenditures makes capacity planning especially hard.

Capacity planning becomes easier, however, if we understand the fundamental causes of cyclicality. With this knowledge we can devise a capacity planning strategy that takes advantage of good opportunities and minimizes mistakes during especially difficult times. While uncertainty will always influence the end result, a structured approach based on an understanding of the fundamentals driving the industry should significantly improve the odds.

Some often cited reasons for cyclicality range from unavoidable cyclical fundamentals in the industry to the presence of crazy eddies, i.e. irrational behavior on either the supply or the demand side, to herd behavior, to industry competition.

While we do not preclude the presence of any of these reasons as a cause for industry cycles, recent investment science research has shown that competition stands as a significant contributor to severe industry fluctuations. Moreover, companies that are aware of the effects of competition and equipped with a capacity planning strategy that incorporates competitive effects may expect to dominate their industry in the long term.

The next section will discuss the basis of the research. The discussion is kept at a qualitative level. We then turn our attention to the competitive advantage gained by using the competitive knowledge-based strategy versus a deterministic net present value strategy.

Competition and Industry Cycles
The main motivation of companies is to improve profitability. In a typical commodity industry this means cutting costs/improving production efficiencies or increasing production. Influencing the price is generally not an option for companies in a commodity-type industry because in such industries the price is set by the overall market demand and supply.

Production volume can be changed in either of two ways: buying other companies' capacity or constructing more capacity. To understand the effect of competition on cyclicality we focus on capacity construction, as buying other's capacity does not alter industry supply.

We now describe a typical scenario of supply decisions over time according to our industry model.

Capacity construction occurs at some point after demand has been rising and current and future expected industry earnings are high. At this point the return on investment of building a new production facility is sufficiently high to cover the capital market's requirement of expected return relative to the level of accompanying risk. As long as demand rises more capacity is built. At some point though, because demand is fundamentally uncertain, demand will suddenly change direction and start decreasing. This will cause companies to stop initiating new construction.

The result of the above description is that the industry will experience first a boom and then a bust. While demand is rising but before capacity under construction becomes available, companies experience a seller's market. Limited supply and ample demand generates high prices and correspondingly high earnings. When new capacity becomes available the price level flattens or perhaps decreases slightly, but as long as demand grows profits remain high.

The boom will come to an abrupt end, nonetheless, when demand shifts direction. Not only does demand decline, supply will continue to increase for some time as construction started prior to the downturn provides new capacity. Both effects combine to put severe downward pressure on the price. The result is that the industry will experience a significant decrease in earnings and losses will occur for the less efficient suppliers.

It is important to point out that we do not assume any irrational behavior on the part of the suppliers. Every supplier has the same objective, which is to maximize the market value of their company. The outlook at the point in time when a company is making its investment decision is positive with an expected return that satisfies the capital market's requirement of expected return relative to the level of accompanying risk. It is just that investing in capacity is risky and at some point the risk will negatively affect the industry.

With regard to phenomena such as irrational behavior and herd mentality occurring during a boom interval, in a well functioning capital market it is not likely that they will persist on a large scale because such markets are quick to punish mistakes through means such as bankruptcy, restructuring, or takeover.

In the above scenario we defined the time when a company would invest in new capacity as the time when a new investment would satisfy the capital market's requirement of return versus accompanying risk. For this statement to be useful, except for the odd CEO with an intuitive grasp of market value, it is necessary to have a valuation model that incorporates the competitive element. In fact, not even the CEO with an intuitive grasp of market value can usually provide an exact value. In contrast, a valuation model will quantify an opportunity and therefore numerically verify the value of a good opportunity and the savings for avoiding a bad opportunity.

The basic idea of the valuation model that emerges from our competitive industry research is as follows: Suppose there were a futures market for the industry's product. Suppose further we sold all future output of a new plant in the futures market. Assuming that no other uncertainties impacts the plant's cash flows, the value of uncertainties impacts the plant's cash flows, the value of the plant's new cash flow would be the net present value of the discounted futures cash flow stream net of any production costs, using the riskfree rate as the discount rate. Thus, the futures price for delivery at some future point in time represents the value as seen from the point of view of the capital markets of a delivery of a unit of production at that point in time.

The above works nicely when a futures market is available, in which case all necessary information for valuation is publicly available. Our research describes a method for calculating synthetic futures prices when no futures market exists for a particular product.

There are two other characteristics that are worth mentioning. First is the fear of losing out by being preempted. That is the reason there always is some company that invests in new capacity at the very moment the value of the investment is sufficient to meet the capital market's requirement. Companies know that their choice is to either seize the moment or face the distinct possibility of being preempted by another company's investment the next day.

Second, competitive industries often operate in a price region where customers' sensitivity to price fluctuations is small. A small change in demand will result in a large change in price, causing both large upward and downward fluctuations in earnings for small demand variations.

Competitive Advantage of a Competitive Knowledge-Based Strategy
To compare the long term competitive advantage of following a competitive knowledge-based strategy versus following a strategy based on a deterministic net present value model we built a model that simulates industry conditions over a 40 year time horizon. The model works as follows: The industry is defined by a supply side, consisting of a number of companies, and a demand side. The total supply at a point in time is the sum of each company's capacity. The demand side changes randomly at the beginning of every month, the expected growth rate of demand being positive. After demand has changed we derive the price of a unit of production by equating supply over that month with demand during the month. The earnings of a company for a particular month is given by the product of its capacity and the price of a unit of production reduced by the cost of producing that unit.

Each company can attempt to increase its earnings by building more capacity. Building capacity, however, is costly and there is a construction lag. To make construction decisions each company relies on its own valuation model. A valuation model can be used to value a new unit of capacity or an existing unit of capacity. For the purposes of this study, there are two types of valuation models and we partition the companies into two groups according to which valuation model they use. One group uses a deterministic net present value model. The other group's valuation model is based on our research, which takes into account competition and demand uncertainty when valuing capacity.

Companies can also trade capacity. We assume a trade occurs each time that the estimated values of the two models differ by more than 10%. When this is the case the group with the lower estimated value will sell 5% of its capacity at the middle value point.

Each group's performance is measured by repeatedly simulating industry conditions. At the end of each run we collect for each group the overall level of cumulative earnings and the overall level of capacity. We assume that the group using the deterministic net present value model initially owns 90% of the total capacity.

Figures 1 and 2 show the results under two industry scenarios. In the first scenario the two groups are not allowed to trade capacity while in the second scenario the groups are allowed to trade. In both scenarios the groups can build new capacity.

The columns represent expected total return on capacity and earnings relative to initial market share. Thus, under scenario 1 the expected capacity growth of the group using the deterministic net present value model is 3.06 times relative to their initial capacity and expected cumulative earnings invested at the risk free rate is 9 times their initial market share. These results can be contrasted by the knowledge-based strategy group where the expected capacity growth is 23.64 times their initial capacity and expected cumulative earnings invested at the risk free rate is 8.17 times their initial market share.

Figure 1. Expected total returns on capacity and earnings when no trading of capacity is allowed between the two groups. DCF represents the supplier group using deterministic net present value model and MF represents the supplier group using the competitive knowledge based strategy.

Figure 2. Expected total returns on capacity and earnings when trading of capacity is allowed between the two groups. DCF represents the supplier group using deterministic net present value model and MF represents the supplier group using the competitive knowledge based strategy.

While the relative advantage of using the competitive knowledge-based strategy is significant when the two groups do not trade capacity, the performance is especially impressive when trading is allowed. In the latter case we expect the group following the competitive knowledge based strategy to increase its market share from 10% to about 64% while at the same time through timely trading capturing 76% of the earnings generated by the industry.

Conclusion
Cycles is a phenomena that has impacted certain industries for a long time. Our research shows that the existence of cycles can be viewed as a natural characteristic of a competitive economy. In this case, it is our view that behaving optimally within a cyclical environment, managing risk and aggressively pursuing new opportunities when they appear, will lead to superior results compared to trying to control or influence the cycles. The latter actions can potentially go against the natural behavior in a competitive industry and can thus reduce the beneficial aspects of economic competition.

The model emerging from the cycle research also provides a collection of synthetic futures prices. These futures prices can be used to calculate the value of an existing plant and also on a potential investment in new capacity. This valuation model can be used to generate a superior strategy for new capacity investments and also for identifying good opportunities for trading capacity with other industry participants.

The long term effect of adhering to the competitive based investment strategy was shown to be significant, clearly outperforming the strategy based on an application of the discounted cash flow model.

The overall principles outlined here should apply generally and should therefore be useful for providing a basis for constructing an investment strategy for many cyclical industries. Nonetheless, a careful study of the industry with the purpose of extending and calibrating the model to a particular situation should be executed. The simulation results suggest that the return of such a study would justify the effort.