Economics, perhaps more than any other area of social inquiry, aspires to the status of a science. Like astronomers, we economists acknowledge that our inability to conduct controlled experiments puts us at a disadvantage. But instead of minimizing this disability, all too often we have accentuated it by failing to implement those straightforward measures that are taken as routine in the physical sciences. In particular – and as was forcefully pointed out by Dewald, Thursby, and Anderson in the American Economic Review (AER) for September 1986 – we have not even taken the standard precautions needed to ensure that our work can be replicated independently by others.

What is required to ensure replicability? The editors of the AER, in the issue referred to above, announced that it was their policy to “publish papers only where the data used in the analysis are clearly and precisely documented, are readily available to any researcher for purposes of replication, and where details of the computations sufficient to permit replication are provided” (p. v). In the case of applied general equilibrium (AGE) work, this can amount to a tall order.

To take just one well-known example – the Australian ORANI model – a listing of the model's data base and parameter file runs to 646 pages (Kenderes and Strzelecki 1991). The equations of the model itself, their interpretation and illustrative simulations are documented in a 372-page monograph (Dixon, Parmenter, Sutton, and Vincent 1982), and in a journal literature too voluminous to cite here.