Autoreg: a computer program library for dynamic econometric models with autoregressive errors

Econometric Software

“…Bergstrom's graduate students over a number of years in the 1960s [194], as well as Ross Preston [197] during much the same time, quite independently all wrote, or were closely associated with the writing of, programs to implement or study specific econometric techniques -but apparently not originally for the conscious purpose of writing 'econometric software' for it's own sake. In contrast, others, among them Robert Hall, Michael McCracken, and Clifford Wymer in the mid-1960s [25,26,160,162,163,165,192,204,211], and David Hendry at the end of the 1960s [114,115], wrote programs the direct descendants of which still exist and which affected the development of other programs. Demonstrably, each of these economists, as well as others, consciously sought to develop more broadly defined and generally applicable software that has evolved over time, regardless of whether this destiny was at first foreseeable.…”

“…At the LSE particularly, work on specification search led ultimately to the development of GIVE and, later, PcGive, as software designed to foster the general to specific, or LSE, method of specification selection [110,112,113,115,146,181]. In this context, the development of software played an integral part in the intended improvement of econometric methodology, with this software specifically designed as an analysis tool to be used pedagogically.…”

“…Then there was so much a need to program, and comparatively so much less computer use than today, that practically any econometric software written prior to 1975 involves some aspect that can be declared a 'first,' or at minimum viewed as involving some type of pioneering effort. Although there were certainly instances of shared software use by (comparatively small groups of) economists during the period between 1960 and 1980 [33,35,89,115,131,160,196,222,223], it was usual nevertheless that an intending hands-on user generally needed to learn to program, at least to some degree, but often at a rather fundamental level. Until at least the mid-1970s, an applied economist wishing to employ a computer would commonly either need to start from scratch, or, at best, be provided a box of cards (or, sometimes, a paper tape), onto which were punched the source code statements for a program -even if by 1975, perhaps earlier, it was not unusual for the 'box of cards' to have been replaced by a machine-readable card image file.…”

Systems testing of wealth-aggregation restrictions in an integrated model of expenditure and asset behavior

“…Any other chosen instruments from outside thedefined system appear rather arbitrary. We did, however, experiment with extending the set of instruments to include some variables reflecting constraints on the personal sector's ability to borrow and estimating the asset equations on an equation-by-equation basis using the generalized instrumental variable estimator (GIVE) (Hendry and Srba, 1980) which, effectively, takes a weighted average of the possible instruments available, with the weights chosen so as to maximize the correlation between this composite variable and the endogenous variable concerned. The GIVE is obtained by regressing the endogenous variable on all possible instruments (see Harvey, 198 1, pp.…”

Domestic credit and the balance of payments in Ghana: A comment

“…3. Nearly all the equations in this section were estimated with the GIVE program [Hendry and Srba, 1980].…”