Twenty-fifth Anniversary of Econometric Models on the PC

MODLER™ is the first comprehensive model building and use econometric software to have been implemented on the microcomputer. In September 1983 this software began to be used to create macro-econometric models for the (as then called) IBM PC. These multiple equation models were the first such models to be estimated on a this or any other type of microcomputer, for prior to September 1981 this software was available only on mainframes.

In September 1984, the Alphametrics Corporation and Wharton Econometric Forecasting Associates jointly established the first PC-based econometric forecasting service, based upon the MODLER™ software and the Wharton PCMark7 Quarterly Econometric Model of the United States Economy. This was both the first large-scale econometric model to be estimated and solved using the microcomputer, containing more than 250 equations, and the first such service that involved the common use of a specific econometric model on multiple computers of any type. Prior to that, with the exception of the 6-7 equation Klein Model I, it was extremely rare that any model was mounted and used on more than a single computer. After all, before 1984, it required considerable computing programming skills on the part of a user to install and operate such a model. Consequently, before then, very few economists had any hands-on experience of econometric models.

Within a month of the inauguration of this service, more than 100 different groups of users had signed up for the service and were creating specialized forecast tables, graphs, and charts displaying the results. By the middle of 1985, MODLER had begun to be used by virtually all the major econometric forecasting companies, including Chase Econometrics, Data Resources, Lawrence Meyer & Associates, and Townsend-Greenspan and Company, in the United States, but also the London Business School in the United Kingdom, and many firms in Europe and other parts of the world, including FERI in Germany, Ibermetrics in Spain, and Pitagora in Italy. Since 1984, more than 2000 distinct econometric models containing 50 or more equations have been created and used in the MODLER software, a history that is well attested in the selected references section of this website. Literally thousands of economists have employed the package, many in both its DOS and Windows versions. In fact, some still use the DOS version -- at least one copy that remains daily use dates as far back as 1987.

The conversion of MODLER from the mainframe package to the PC occurred rapidly, keeping pace as the PC evolved in its first two years from a machine with only 64KB of RAM and a 120KB diskette drive to one with 640KB of RAM and a 10 MB hard Drive. On the mainframe, MODLER had been used for models of 300 equations and more, but the PC at that time was both a new and more limited environment. On the PC, Klein Model I was estimated and solved in MODLER in early 1982, followed during the next two years by progressively larger models of the UK, US, and World economies, culminating in the inauguration of the above-mentioned first forecasting service in 1984. In 1985, as the PC ramped up in speed from 4.77 Mhz to 6-8 MHz, the 600+ equation Wharton Mark7 US quarterly econometric model was both estimated and solved using MODLER. Today, on the most modern machines, MODLER solves models of this size in two seconds or less, compared to the literal minutes it took in the mid 1980s. Actually, for comparison purposes, these same models can be (and are) solved today: indicating MODLER's legacy awareness, the identical model and solution files can still be used with the current Windows versions of MODLER. Among present day software packages, in terms of the ability to support the use of the same files after 20 years, MODLER is unquestionably rare. Among econometric and statistical software packages it is unique.