Appendix 1 – Industry productivity database

To form aggregate and industry productivity series for the market sector of the New Zealand economy it is necessary to have data on values and volumes of output, labour and capital. This appendix discusses the data used to form an industry productivity database in more detail.

The level of industry disaggregation for which productivity series can be constructed is determined by the industry hours worked data. Compared with the production and income GDP accounts and the capital stock data, industry hours worked data on an ANZSIC basis are constructed at a more aggregate level. Therefore, even though data on production and income GDP (compensation of employees and operating surplus) and the capital stock are available for 31 industries (ie, the two digit industry level), hours worked data are only available for a more aggregated nine industries (ie, the one digit industry level). This made it necessary to form more aggregate industry output and capital series, which matched the corresponding industry hours worked level of industry disaggregation, using data on production and income GDP and the capital stock for the 31 industries. Appendix Table 1 provides a summary of how the various data were aggregated to form nine comparable industries for use in constructing aggregate and industry productivity series.

Output

Output data were primarily sourced from Statistics New Zealand’s System of National Accounts (1993) production based GDP series. Annual March year volume GDP at the two digit industry level, which are chained Laspeyres constant price series with a fixed weight Laspeyres tail from 1999 onwards, are available for the period 1988 to 2002. These data were then aggregated to form the nine industries in the industry productivity database. Aggregation of the industries was by simple summing. While this is theoretically unsuitable for chained data, advice from Statistics New Zealand, and experimentation, suggest that the overall result between this and chaining is very similar.

Annual March year nominal GDP data at the two digit industry level are available from Statistics New Zealand’s income GDP accounts for the period 1988 to 1999. These data were then aggregated to form the nine industries in the industry productivity database. Nominal GDP data for the period 2000 to 2002 were projected forward using nominal expenditure GDP data for the entire economy, following the method suggested by Diewert and Lawrence (1999, p. 283) (who also faced the problem of a shorter time series for industry nominal GDP compared to industry volume GDP). This method is outlined below.

First, an implicit output price for the entire economy for 1999 to 2002 was found by dividing nominal expenditure GDP by real expenditure GDP. Percentage changes in the implicit output price for the entire economy were then used to project forward the implicit output price for each of the nine industries. Finally, the projected industry implicit output price series were used to reinflate volume GDP to yield nominal GDP.

This method does not allow for relative changes in prices between industries. As a sensitivity test, industry level nominal GDP was projected forward using Producer Price Index output prices by industry. While some differences in the level of GDP occurred at the industry level, when used to weight up the industry outputs, the alternative projections had a negligible affect on the aggregate productivity results.