Improving Information for Social Policy Decisions -- The Uses of Microsimulation Modeling: Volume II, Technical Papers (1991)
Chapter: Short-Term Issues
simplified tax and transfer program simulations in addition to the jobs component. The model was heavily used and extensively modified, but there was never time for documentation or to develop a generalized software framework. Operating efficiency was also a problem; in fact, the model sometimes brought down the whole computer. KGB was usable only by its creators and their research assistants, and its use ceased when its creators left HHS/ASPE.
The ad hoc approach, with the advantage of fast response and the disadvantages of idiosyncratic behavior, limited lifetime, and lack of ability to be useful in environments other than the original technical group, has characterized most microanalytic simulation systems that have been built. If the adage that “programs are what you write for your use, while software is what you write that is to be used by others” is correct, few if any microsimulation systems exist today—they are really programs masquerading as software.87 This is not to say that all previous systems are similar to KGB in their development or to deny the role of additional experience in developing software over time. Hindsight is often a powerful teacher, yesterday’s generalized system often is viewed as rigid and limited because of heightened understanding of what is possible now.
The pattern of incremental funding that has characterized the research and evaluation responsible for microanalytic simulation activity has been a major contributor to the current situation; over a significant period of time such a pattern implicitly denies the existence of a medium-run perspective. Another contributing factor has been the persistence of a “not invented here” mentality that channels scarce investment funds into multiple directions at the expense of possible greater benefit from more central development. But centralized development is not necessarily superior to decentralized development; rather, the “not invented here” attitude blocks awareness or consideration of the question. We believe that it is very important to separate short-run and medium-run issues to ensure that short-run issues do not determine medium-run outcomes, a year at a time. Obviously, for consistency medium-run strategy and goals will affect how short-run issues are addressed.
Short-Term Issues
Our principal short-term question is as follows: Given the apparent success of implementing SPSD/M in a microcomputer-based environment, what are the benefits and costs of moving TRIM2 to a similar environment or reimplementing it in that environment?
Desktop environments are appealing for microanalytic simulation as well as for many other kinds of computer uses. Compared to alternate scales of
