The modelling of regional and local economies, especially through using the
input-output (IO) approach, is a standard practice for many economic consu
ltants and yet quite often this is based on little if any primary data on (
interregional) exports and imports. This paper argues that the relative siz
e of the trade sector in final demand, and its importance in the literature
on regional growth, is such that current models are at best misspecified a
nd are at worst likely to provide biased and possibly misleading results, e
specially with regard to planning and policy at the local level. On a more
practical level, this paper shows that with certain minimum information, es
pecially with regard to exports and imports, a hybrid local IO table can be
constructed that provides reasonable accuracy ill terms of estimates of IO
multipliers vis-ci-vis a complete survey table (which is almost prohibitiv
ely expensive to construct). This is in contrast to the results obtained fr
om the more usual LQ approach which, primarily because it underestimates tr
ade, are seriously biased.