UT's Edirisinghe uses Business Analytics to Bring Fresh Eyes to Financial Analysis
Chanaka Edirisinghe, PhD, has a passion for understanding how things work. As a child the first thing he would do with a new toy was to disassemble it to see how it worked. As a teenager he started fixing analog wrist watches and radios. In college he majored in engineering, graduating with honors in mechanical engineering from the University of Sri Lanka in 1980 and graduating with an MS in industrial engineering and management from the Asian Institute of Technology in 1985.
Fortunately for the Business Analytics programs at UT, his quests for knowledge of how things work eventually led him to study analytics.
He completed a PhD in management science from the University of British Columbia in 1991. Then, in his early years at UT, he started building computers and servers and became one of the early faculty members in the College of Business Administration to host a dedicated web-server for the classes he taught.
This passion for understanding how things work has served him well in his studies of financial markets. And his background in Business Analytics has allowed him to create better ways to measure risk and to select stock portfolios.
The performance of his methods has made him a featured speaker at practitioner conferences around the world. And he has received many accolades from the academic community for creating a "fourth principle" for estimating investment risk.
Investors have traditionally used three tools to assess the risk of a particular stock: (1) the mismatch between book value and market value (2) the size of the company and the market risk, i.e. smaller companies are more volatile, and (3) the market risk, i.e. risk associated with the uncertainties of the market.
Conceptually, most would agree that another source of risk would be associated with how well or how poorly the company is being managed relative to its competitors. This is reflected in the "productivity" of the company, i.e. the efficiency of the company in converting dollars of input to dollars of output. For example, suppose two airlines have similar operating expenses and similar capital investments, but airline A has higher sales revenues than B. Then A would be considered more productive.
The comparisons among companies are not as straightforward as this simple example suggests. For example, how can you combine the 30 or more measures of performance from balance sheets and income statements to create a measure of productivity? One company may spend more on research and development and another may spend more on marketing. One may have higher sales margins and the other may have lower accounts receivables. What is the relative importance of these numbers in measuring productivity?
Edirisinghe's background in business analytics allowed him to bring a set of fresh eyes to this problem.
The quality of a given measure of productivity can be measured by its historical correlation with market performance. The selection of the inputs and outputs and the weights attached to each can be viewed as an optimization problem. Data analysis and optimization are the hammer and screwdriver of business analytics.
Edirisinghe is sometimes asked the obvious question: Does your method contradict the efficient market theory developed at the University of Chicago in the 1970's? The efficient market theory assumes that all information available about a company is already reflected in its share price. If this is true than there is no point in trying to optimize the selection of stocks, none are overpriced and none are underpriced.
"The efficient market theory holds in the long run," concedes Edirisinghe. "But in the short run there are inefficiencies. For example, I analyzed historical data on stock prices in the U.S., China, and Japan. I found that the U.S. and Japanese markets are more efficient than the Chinese markets. I attribute this to the better flow of information in the U.S. and Japan.
"Stock pickers such as Warren Buffet are betting on the inefficiency of the market. Their primary tools are the three measures of risk difference between market and book value, company size, and market risk. And certainly many investors would at least intuitively try to factor in their understanding of a company's productivity relative to its competition. But, unlike the other three measures, there has been no objective measure of a company's productivity until now."
Edirisinghe's ideas have caught the attention of both practitioners and academics around the world. In recent years, he has been the invited keynote speaker at financial risk analysis conferences in Zurich, London, Tokyo, Rome, and Toronto. His article, "Portfolio Selection under DEA-based Relative Financial Strength Indicators: Case of US Industries," published in the Journal of the Operational Research Society (2008), was awarded the Emerald Management Reviews Citation of Excellence. This prestigious award is given only to those management articles judged to be top-50 from the 15,000 articles reviewed annually by Emerald.
Edirisinghe's writing was featured with other leading scholars, including Nobel Prize winner Harry Markowitz, in the book Advances in Mathematical Programming: State of the Art in Stochastic Programming. This book was edited by George Dantzig, the father of mathematical optimization. Additionally, Edirisinghe has received faculty fellowships from Osaka University in Japan and the University of Canterbury in New Zealand to lecture and do research on their campuses.
One method that he used in his analysis is Data Envelopment Analysis, a technique designed to combine multiple performance measures into a single measure that best captures the information available. Since performance of different companies are affected differently by the conditions of the market, it is necessary to factor market uncertainty into portfolio selection; decision making under uncertainty (stochastic programming) is an area in which Edirisinghe is recognized as a leading scholar. The end result is a method that gives an additional dimension of risk measurement and a dynamic method of portfolio selection; i.e. a portfolio should vary as market conditions vary.
Christine Vossler (865-974-1762, firstname.lastname@example.org)