ABSTRACT
This chapter provides consistent and asymptotically efficient parameter estimates of a large demand system consisting of some 80 commodities which is consistent with weak decentralisation and also generalised strong separability. To achieve economy of computation, particularly for a system with a large number of commodities, it is desirable to exploit the block recursive structure of and in the estimation process. The conditions under which commodity aggregation is possible with the usual forms of aggregate that time series data provides are extremely restrictive. There is some justification in the argument that an aggregated demand system conceals large differences in the behaviour of the different items within the aggregate. Some particular microeconomic policies, public transport pricing or fuel policy, require estimates of price and income elasticities for detailed items of expenditure. If provided within a demand systems framework such estimates will be automatically corrected for failures of the usual ceteris paribus assumption of partial equilibrium analysis at least on the demand side.
