A source of bias in regional input-output models estimated from national coefficients: Fid-Bau Portal

A source of bias in regional input-output models estimated from national coefficients

Jones, Lonnie L. / Sporleder, Thomas L. / Mustafa, Gholam

Conclusions Most regional input-output models are constructed for purposes of providing a basis for making policy decisions relative to regional economic growth and development. Evidence suggests they are used increasingly for this purpose. A major implication of this paper is that information generated from regional input-output models developed by modifying a national model may lead to unjustified optimism by those who make such decisions. At a minimum researchers and decision-makers should be aware that this may be a problem in models estimated using the net export-import technique. The significant of this problem may be expected to vary among regions. That is, its significance will vary inversely with, among other things, the level of economic integration within the region. If examination of secondary data vis-a-visa priori knowledge of the regional economy leads the analyst to believe that this may be a significant problem, steps could be taken to help alleviate it. Even studies with extremely limited resources could include a survey of major sectors to obtain estimates of their actual exports and imports. Su [10] suggests that costs may be reduced by a survey conducted on theproportion of required inputs for each sector imported from other regions in the total inputs of that sector. Such an approach is amendable to a mail or telephone survey that could yield, at minimum costs, sufficient data on exports and imports. Resultant percentages could then be used in conjunction with the technical coefficients matrix in a manner similar to the traditional net export-import technique. The Delphi technique of multi-round surveys of industry experts also holds promise for obtaining such proportions.