in the existing input-output table into four. Another reason for splitting published input-
output sectors is so that different technologies can be applied to an industry in different
regions later in the database process. For example, Victoria’s electricity generating
industry uses brown coal while that in New South Wales uses black coal. Additional
sectors black coal, brown coal, black coal electricity generation and brown coal
electricity generation capture these known regional differences in technologies. The main
source for the sectoral split is unpublished Australian Bureau of Statistics (ABS)
commodity details data. Such data provide a split of sales for approximately 1,000
commodities to 107 industries, plus final users.
The preparation of the TERM database does not rely on gathering input-output tables for
each region in the database. Rather, by assuming that the same technologies apply to a
given industry in each region, it is sufficient to gather regional shares of national output
for each industry. The next step therefore is to obtain, for each industry and final
demander, an estimate of each statistical division’s share of national activity. To develop
a full input-output table for each region, we required estimates of industry shares (ie, each
region’s share of national activity for a given industry), industry investment shares,
household expenditure shares, international export and import shares, and government
consumption shares. Horridge et al. (2005) details the data sources.
These shares are used to split the national input-output table. Database preparation is
undertaken at a maximum level of disaggregation, thereby enabling the practitioner to
impose different technologies where necessary, as in the case of electricity generation.
Inter-regional trade matrices are computed on the basis of disaggregated regional
demands and supplies. The gravity assumption (i.e., the volume of trade follows an
inverse power of distance) is invoked for inter-regional trade using a distance matrix.
Note that where ever production (or, more rarely, consumption) of a particular
commodity is concentrated in one or a few regions, the gravity hypothesis is called upon
to do very little work. Because our sectoral classification was so detailed, this situation
occurred frequently. In addition, outside the capital cities, most Australian regions are
rural, importing services and manufactured goods from the capital cities, and exporting
primary products through a nearby port. For a given rural region, one big city is nearly
always much closer than any others, and the port of exit for primary products is also well
defined. For local commodities, we needed to ensure that local supplies were set equal to
local demands. Otherwise, we would have generated implausible trades, such as inter-
regional trade in childcare services.
The process of splitting the national database into more sectors, and then splitting the
new national table into many regions is highly mechanised. This means that the
methodology can be applied to new input-output tables as they appear, or to tables for
different countries.