Dear Readers,
I came across an old article on Gantt Head dataing back to 2002. The article is an interview between the project and programme management web portal Gantt Head and Dr. Harry Markowitz titled
Gantt Head about Dr Harry Markowitz
Dr. Harry Markowitz, of the Harry Markowitz Company and a professor at the University of California at San Diego, has been credited as one of the creators of portfolio management theory. In the 1950s, he wrote that a portfolio of diverse investments is more likely than individual investments to reduce risks and produce a higher rate of return. His Modern Portfolio Theory (MPT) revolutionized the way investments were viewed. With new tools for estimating risks and returns, investment managers were able to create funds designed to suit investors’ individual risk thresholds while attaining desired returns. For his work, Dr. Markowitz was awarded the Nobel Prize in Economics in 1990.
Project Portfolio Management (PPM) arises largely from Dr. Markowitz’s work in portfolio theory. The treatment of projects as investments, managing groups of projects in portfolios and overseeing their execution and value to the organization as a group is at the core of PPM. In this light, Dr. Markowitz gives us some of his thoughts about the application of financial portfolio theory to corporate project management.
For your convenience the interview is detailed below.
gantthead: What are your thoughts regarding Project Portfolio Management (PPM)?
Markowitz: I’m not directly familiar with this management theory; my work has been primarily in the equities markets. The notion that projects can have expected returns, variances and covariances of return does appeal to me. However, when you apply MPT to a portfolio of securities, you mostly assume a budget constraint to indicate what transactions are allowed. Thinking about applying this to corporate projects, I have some concerns about doing this.
gantthead: What are those concerns?
Markowitz: With projects in corporations, there needs to be a matching on the production side that doesn’t happen in the securities market. Some projects may have human capital requirements that others may not. An example is an oil company that buys a department store. The oil company does this to be better diversified. But the management skills the oil company brings forward do not necessarily match the management needs of the department store. The time required for the oil company to learn how to run a department store may result in lost market share for the store.
Here is another example. Suppose I have a job shop that turns out products. I can’t necessarily take on work that does not fit the skills and capabilities of my shop. If my shop builds cabinets, I would not ask it to begin developing software simply because the company wants to diversify.
I would be cautious about applying MPT to corporate projects as though these are liquid assets. There are different constraints regarding projects, like management expertise, human skill sets, physical production capabilities and other factors that come into play. I’m not sure that the constraint side of the project portfolio problem has been properly modeled. Understanding how an organization’s experience in one product area may be applicable to other markets is not very clear.
gantthead: What about evaluating returns from projects?
Markowitz: As far as returns from these projects, part of the uncertainty is market uncertainty as with securities. But part of the uncertainty also has to do with skills available. In order to evaluate the potential of a project, the company must estimate whether it can design and produce a product of a quality and price sufficient to be sold in the face of present and future competition.
Also, projects in a portfolio may reinforce each other in a non-additive way. Microsoft is the 800-pound gorilla in software. They are known to use their dominance in the Windows operating system to reinforce their sales of other software products. Thus some projects may add to the business portfolio other than only by their contribution to the revenue stream.
gantthead: What about evaluating product risks?
Markowitz: With securities, what has become fairly popular are models of why things go up and down together. We see that sometimes certain industries do better than others, or large capitalization stocks do better than small capitalization stocks, or vice versa, and so on. The riskiness of the portfolio depends not only on the riskiness of its individual securities but also on the extent they tend to go up and down together: their correlation or covariance. Rather than estimating individual covariances, we build models of covariances. A lot of effort has gone into building such models.
Some kind of model of covariance may also be applicable to project portfolio selection. Project covariances might depend partly on market factors similar to those used in MPT. But they might also depend upon technology factors or management factors. For example, perhaps two projects depend on the same yet-to-be-proven technology. Or perhaps both depend upon prompt completion and both rely on the same person’s ability to estimate completion times, which may itself have yet to be tested.
gantthead: How do you think Portfolio Theory could apply to Corporate Management?
Markowitz: Perhaps portfolio theory–as is–is not applicable. An investment manager doesn’t have to run companies he purchases. Also, he doesn’t jeopardize the prospect of the companies whose shares he buys if he invests a small fraction of his portfolio in this one and a small fraction in that one. But if a company manager subdivides his resources too finely among many projects in order to diversify, he may give each project too little to succeed. The company (or Division) manager cannot pretend that he is selecting liquid assets subject to a budget constraint.
gantthead: How would you approach applying Portfolio Theory?
Markowitz: In the first instance at least, I wouldn’t think in terms of applying portfolio theory. I’d think in terms of doing a Systems Analysis or Operations Research analysis of the enterprise. In the past, I’ve applied various techniques–such as linear programming and simulation analysis–to various kinds of business decision-making problems. In the 1950s I applied quadratic programming to the portfolio selection problem. Now it is quite common to us a combination of quadratic programming and Monte Carlo analysis. I think this is a good idea.
When you approach a new area, you should start with a blank white-board and ask what’s essential in this problem. My first reaction to the problem of management of a multi-project department or enterprise is that it reminds me of a job shop in manufacturing. In a job shop, you apply resources—such as people and equipment—to perform tasks. When you succeed at one task, the job may require other tasks to be performed. There is always a chance of rejects; work may have to be scrapped. You can think of a division with projects involving design, manufacturing and marketing as a job shop. In engineering tasks, there is always a chance that you won’t succeed. And there is always a chance that someone else will do the design better or beat you to the marketplace. These are some of the project risks that must be considered in evaluating projects in a portfolio.
Before one could start building a Monte Carlo model of such an enterprise “job shop,” one would have to think about what kind of skill levels need to be distinguished, what kinds of details could be aggregated or ignored, what kinds of data bases are available or need to be developed. Just thinking about how to go about building such a simulation might provide worthwhile insights.
But let me emphasize that this is not my field. All I know is that in the typical investment situation one can finely subdivide one’s funds among many fairly liquid assets. The same cannot be said of portfolios of projects. But perhaps people who propose the analysis of portfolios of projects have figured ways to cope with this problem.
