Why should this be the case? After all, employment is not necessarily the most direct measure of economic activity or of land use in an area. The value of output per person employed varies immensely between subsectors of the economy, and so too does the number of hectares of land needed to support one job. On both dimensions, it is salutary to compare, say, a petrochemical works with a multi-storey hotel. In view of such difficulties, the thought must occur fairly early in this chapter that the common use of employment data occurs faute de mieux - because there is usually nothing better on economic structures below the regional level of analysis; filling such a vacuum, government instituted an annual Census of Employment (CoE) from 1971; while this became less frequent after 1978, and sampling methods were introduced from 1984 onwards, its geographical versatility and access have been greatly improved as one may see below.