Subal C. Kumbhakar

One of the many things I learned from him is not to complain about negative referee reports. Papers get rejected but in his view that does not mean the research is meaningless or bad or that you should stop.
Bill Cooper was a man who wore many hats and who had many talents. For brevity’s sake, I will focus on the work for which he is most known, that on efficiency using Data Envelopement Analysis (DEA) which he invented with Charnes and Rhodes in 1978. Also I will only comment on our joint works. Working on the stochastic frontier approach (SFA), developed by Aigner, Lovell and Schmidt (1977), I joined the economics department of the University of Texas at Austin (UT) in 1986. Bill was a professor in the business school at UT. In those days SFA was viewed as the archrival of DEA. Although the Business School was located in the same building, I avoided his office, since I was a student of Dennis Aigner. However, it seemed odd not to read some of the DEA paper, especially when the two main architects had offices in the same building. So sometime at the end of 1986 I went to the Center for Cybernetic Studies for some of Bill and Abe’s working papers. The secretary informed me that I could not get any of those papers without their approval and suggested that I talk to Professor Charnes, whose office was at the Center for Cybernetic Studies. Based on what I had heard about Abe, I did not have the courage to go to his office and ask for his permission.

Two years later (in 1988) I met both of them in Chapel Hill at a conference that Knox Lovell organized with Arie Lewin. This was my first ‘real’ conference and I met many of the top people working on DEA and SFA. I remember talking to Bill at the airport because we were waiting for the same flight. The SFA group (Knox Lovell, Peter Schmidt, Robin Sickles and some juniors like Paul Bauer and myself) was also attending the conference. Although there were some disagreements on issues of functional form, distributional assumptions and presence of noise, I was relieved to see that there was no real fight between the DEA and the SFA group. At the end of the conference Knox said: “It was fun …., I did not lose my shirt.” This gave me the impression that he had also been apprehensive regarding the possibility of acrimonious exchanges. The disagreements are still debated today, although the gap has narrowed down in recent years.

Although I encountered Bill again at many conferences (mostly on efficiency and productivity) after 1988, my real interaction with him was through some of his PhD students (Bardhan and Kwinn) and a visitor (Xuelin Yu, an engineer from Tianjin, China) who came to UT to work with Bill on DEA.
My first academic work with Bill started with the Chinese aggregate time series data that Xuelin had. We examined performance of Textiles, Chemicals and Metallurgical Industries after the reform. The paper “DEA and stochastic frontier analyses of the 1978 Chinese economic reforms” (Cooper, Kumbhakar, Thrall and Yu) was published in Socio-Economic Planning Sciences 29, 1995, pp. 85-112. While working on the paper our conversations were often orthogonal because I was not a proponent of DEA and Bill was not an econometrician. However, he was never an extremist arguing that SFA had no value; his interest was to come up with a method that could bridge the gap between DEA and SFA. We did this in a paper using data on Texas secondary schools. We used DEA first to find the schools that were fully efficient. This information was then used, in the form of a dummy variable, in the OLS regression to test whether the relationships for efficient and inefficient schools were the same. The paper “New Uses of DEA and Statistical Regressions for Efficiency Evaluation and Estimation With An Illustrative Application to Public Secondary Schools in Texas” (Arnold, Bardhan, Cooper and Kumbhakar) was published in Annals of Operations Research 66, 1996, pp. 255-277.

We continued working on this idea of joint utilization of the DEA and SFA methods. To make sure that our results were not driven by the Texas school data, we decided to analyze it further via simulation. The results were reported in “A Simulation Study of Joint Uses of Data Envelopment Analysis and Statistical Regressions for Production Function Estimation and Efficiency Evaluation” (Bardhan, Cooper and Kumbhakar), Journal of Productivity Analysis 9, 1998, pp. 249-278. In the first draft Bill added another coauthor who was probably not even aware of his co-authorship and knew nothing about the content of the paper. At the end, his name was dropped. I have never seen any one more generous than Bill, perhaps to a fault, so far as co-authorship is concerned.

This issue of generous co-authorship came up again in the last two papers that I worked with him. Both these papers deal with joint advertising vs. service specific advertising in military recruitment. Bill was very fond of this topic. The first of the two papers is: “Alternative Statistical Regression Studies of the Effects of Joint and Service Specific Advertising on Military Recruitment” (Brockett, Cooper, Kumbhakar, Kwinn and McCarthy), Journal of the Operational Research Society (2004). The story behind the first paper goes like this. Bill had a PhD student (Michael Kwinn, a Major at that time in the U.S. military) with whom he and Brockett wrote a DEA paper. Bill asked me to assign Michael (who took my econometrics class) to work on a paper using the military recruitment data that he was using for his dissertation. The idea was to use the same method we used in the Texas school paper. Michael asked a Captain in the army to run the SFA models. Bill happily added the Captain as a coauthor.

When military recruitment fell short of targets during the war with Iraq, Bill saw the opportunity of addressing the same issue (as above) using the military recruitment data again. So we wrote another paper with some additional insights on advertising. A marketing expert was added as a coauthor (in addition to the five in the other paper) to add values on the advertisement side. The paper, as far as I recall, was first submitted to Management Science. But MS was taking too long and Bill could not wait. So he wrote to the editor that the issue was too important to wait and withdrew the paper. Finally he submitted it to Socio-Economic Planning Sciences on the condition that the editor would do a quick review. The editor kept his promise. In the final version of the paper another coauthor was added. As a result, we ended up publishing the paper “Estimating Elasticities with Frontier and Other Regressions for Use in Evaluating Different Advertising Strategies for U.S. Army Recruiting” Socio-Economic Planning Sciences 42, 2008, 1-17, with seven coauthors (later on a reviewer joked about the number of authors in private conversation).

After moving to Binghamton we lost touch other than through the occasional e-mails. Because of his deteriorating health he stopped attending conferences (so I missed seeing him) but never stopped writing papers. Whenever I e-mailed him asking about his health, he talked about it as little as possible, focusing on his research and not on personal issues. He decided to postpone his knee surgery because that would take his time off from research. His photographic memory helped him to work on several projects at the same time (parallel processing) and switch discussions on age old topics seamlessly. In casual conversations in his office, he would talk about his research with Charnes in the 1950s, before I was born, in such detail as though it was yesterday. He kept his papers (all versions) in several file cabinets. If you asked for a copy, he would know exactly where to look for it, take out the whole file and put it back after the paper was copied. Unlike others, Bill used to carry a bag from the hotel to the conference site with 50 to 100 copies of the paper he would present in a conference. I offered help to carry the bag a couple of times while walking with him but he would not let anyone else carry his bag. He would always wear a tie and a jacket no matter whether he was flying or in his office working, except perhaps on the weekends – although he would still be working in his office even then.

Bill never used a computer, at least before 2001 (when I left UT, Austin). Once he told me that his secretary wanted to teach him how to check his e-mails. He refused. In every morning the secretary would printout of all his e-mails, Bill would then handwrite his response on the printed version. Finally, the secretary would type and e-mail them. It worked well in the sense that no e-mail was unanswered – except perhaps some were late because secretarial help was not available during the weekends or holidays (though Bill himself would work 24/7). He handwrote all his papers, no cutting and pasting, filling the whole space in the first line of the pad and then gradually moving to the right in a way that the last line would have only a couple of words, forming a kind of inverted triangle. He would fill up several writing pads in this way for every paper. Once something had been written, he would resist deleting anything.

One of the many things I learned from him is not to complain about negative referee reports. Papers get rejected (he talked about his DEA paper rejected in the Quarterly Journal of Economics): but in his view that does not mean the research is meaningless or bad or that you should stop. Although he and Abe had some rough encounters with Evans and Heckman on the Bell merger issue, he never took things personally. To him the essence of this debate was to emphasize on cross-validation of results with alternative theories. He showed us the thick file containing the correspondences with editor of MS in which some of the exchanges/debates were published.

Bill was a mentor to me in many respects, and I will miss him dearly.

Subal C Kumbhakar, USA
August 2012