Input-output coefficients and multipliers based upon them are increasi
ngly being used to evaluate the impact of changes in agricultural poli
cies, yet there are a number of specific difficulties with regard to t
heir use in this context. This paper articulates various criticisms wh
ich can be made, and tests their validity by comparing multiplier fore
casts with out-turns from a series of eight input-output tables descri
bing the Welsh agricultural sector. The predictive performance deterio
rates rapidly as the time between base matrix and forecast year increa
ses; more attention needs to be paid to the forecasting of final deman
d, spill-over effects into other sectors, and the functional relations
hip between labour input and gross output.