Targeting the collection of superior data for the estimation of the intermediate deliveries in regional input – output tables

It is well known that the intermediate deliveries in (regional) input – output tables are estimated more accurately if nonsurvey techniques are complemented by superior data, obtained via surveys, experts, or other reliable sources. Collecting superior data is costly and it is therefore crucial that the cells for which to collect superior data are well chosen. Several methods have been proposed in the literature. Some of these select a set of individual cells, others identify entire sectors. In the present paper we evaluate six of these methods by simulating the estimation of the regional intermediate deliveries matrices for each of twenty-seven Chinese provinces in 2002, on the basis of their matrices for 1997. Data from the true 2002 matrices are used to mimic superior data collection.