On the development of econometric modeling languages: MODLER and its first twenty-five years 1

Econometric Software

“…These pre-1969 versions of TSP also included the first vestiges of a symbolic language interface [203]. With a few exceptions -such as MODLER [209] and TROLL [63] -until the later 1970s the operation of other programs of all types was commonly controlled by numbers, letters, and labels located in fixed fields on punched cards or paper tapes. It took considerable time then simply to operate a program, given the need not only to program, but also to validate results on a case-by-case basis.…”

“…In September 1984, the first microcomputer-based economic forecasting service was introduced at the annual meeting of the National Association of Business Economists, combining a 250+ equation Wharton Quarterly Econometric Model of the United States with the MODLER software [209]. The solutions for a 12-quarter forecast horizon took less than 4 minutes.…”

“…Only certain individual developers have ventured into print to any significant degree [19,63,64,113,115,160,163,202,203,208,209,225,230,244]. Furthermore, although the user guides and reference manuals commonly provided with individual programs often provide some aspects of their history, these accounts tend to be presented selectively, ordinarily without technical details.…”

“…As a result, at the beginning of the 1980s, economists were ready to adopt the then emerging microcomputer and, by the middle of that decade, had begun to use it as the primary locus for analytical processing, including by 1985 the solution of econometric models of 600 and more equations [209]. However, it was only in late 1984 that it began to become common for econometric models to be mounted on more than a single computer [203].…”

Econometric software: The first fifty years in perspective

“…Whereas by late 1982 or early 1983 the microcomputer potentially provided the individual user a full 640K of RAM, drawing abreast of if not surpassing many mainframes in this respect, it is only much more recently, during the past ten years, that economic data have become widely available in machine-readable form and, in particular, easily downloaded from the Internet. It is only since the mid 1980s that it has been generally possible for someone working alone, with a personal computer and a data base, to construct and solve a large-scale econometric model within a short time period [281]. 4 Milton Friedman has described the process of computing regression parameters for a single equation as taking essentially an entire day in the late 1940s [111].…”