I have been meaning to write this, and have resisted because it is (locally) political, and it may be a declaration of the end of the University of Vermont.
Our university, like many others, has started a budgeting procedure called IBB. This stands for Irrational Brainless Budgeting, although the administration claims it stands for Incentive Based Budgeting. In this post I want to explain why it cannot work, and why implementing it may have caused terminal damage to the university, and the 500 million years of data, and 100 years of theory I have to support these claims.
The idea behind IBB is that units (departments, colleges etc.) that produce more revenue should get a bigger budget. Thus, if a college can grow its revenue by increasing its enrollment it is granted a bigger budget. On the surface this makes sense. It takes money to make money, so if a unit is making more revenue it makes sense that they should get more resources. But this only makes sense on the surface. The problem is that income tends to be relatively fixed – the number of enrolled students is approximately fixed, as is tuition; grant and contract income is unlikely to increase dramatically. As a result, the best way to increase a unit’s revenue is to take revenue from other units. This is exactly what is happening at UVM. Right now you can take a “biometry” type class in any of four colleges (Arts and Sciences, Agriculture, Medicine, Natural Resources). The incentives favor this because units gain resources by increasing their revenue, regardless of the cost to other units or the university as a whole.
Animals have been performing this experiment for 500 million years. The most basal animals (sponges, jellyfish) have an IBB type reproduction. That is basically any cell can give rise to reproductive cells (the “revenue” of living things). As animals become more derived (“advanced”) there is a tendency to limit the cells that can become reproductive, until you get to animals such as vertebrates in which the reproductive cells are sequestered very early in development, in fact while development is still under maternal control.
The importance of this is that the only way a somatic cell (say a liver cell) can increase its fitness is to support the survival of the whole organism and reproduction by the gonads. That is, by only allowing reproductive cells to reproduce vertebrates have (nearly) eliminated competition within the organism, and provided the strongest incentive possible for cooperation among cells within the organism.
This is exactly the opposite of what IBB does. IBB encourages self-promotion of units regardless of any effects such self-promotion may have on the soma (i.e., the university). At this point I should mention that Sears used a similar budgeting procedure. Sears had other problems, but I am certain that their accounting methods contributed to their demise .
The theory on this is well developed, and dates back to at least the 1960s (Griffing 1977). Selection for populations of interacting genotypes. Proceedings of the International Congress on Quantitative Genetics, August 16-21, 1976). In recent years there have been major theoretical advances on the importance of group structure in crop improvement (e.g., Wade, Bijma, Ellen, Muir 2010 Group selection and social evolution in domesticated animals. Evolutionary Applications 3: 453-465). This is an active area of research, and one that has resulted in large increases in agricultural yields.
A conceptual example will illustrate the results nicely. If you plant 10 1 hectare plots with corn, you can either attempt to increase the yield per hectare by selecting the individual corn plants that produce the greatest yield (bushels of corn) or the hectare plots that produce the greatest yield. If you select the individual plants with the greatest yield, paradoxically the yield per hectare goes down. This really happens! (it happened in my thesis, which is how I discovered Griffing). It happens because the individual selection favors plants that aggressively “steal” resources from their neighbors, resulting in a plot of “mean” plants that fight for resources rather than cooperating to raise overall yield. On the other hand, selecting at the plot level promotes cooperation and establishes the best compromise between cooperation and aggressive self-interest that maximizes overall yield.
The implications for IBB should be clear. IBB promotes aggressive self-interest of the colleges and departments within a university. This will promote duplication of effort in different colleges, “poaching” of revenue resources, and ultimately increase the expenses to the university without increasing its revenue.
Finally, IBB is very much like a carcinogen. A carcinogen may not cause cancer, but it does promote it. More importantly, removing the carcinogen does not make the cancer go away. Similarly, IBB does not cause redundancies, but it does promote it, and the results are long lasting, if not permanent. In the biometry example at the beginning, IBB provided an incentive to develop redundant courses, which in turn provided incentives to hire faculty and devote resources to teach those courses. Removing IBB will not make those faculty and resources go away. Further, colleges and departments are notorious for not easily giving up resources. For this reason, I fear that the damage done by IBB may be permanent.