4  New Zealand’s productivity performance

4.1  Aggregate productivity

Figure 1 presents partial and multifactor productivity estimates for the market sector of the New Zealand economy. The productivity series used in Figure 1 are also reported in Appendix 2. The output index used in forming each of the productivity series is a chained Fisher output index that has been constructed using industry volume GDP and implicit prices as elemental series. The input index used in forming the multifactor productivity series is a chained Fisher input index that has been constructed using industry hours worked, capital stock, and labour and capital cost data as elemental series. The corresponding chained Fisher input index for the labour productivity series was constructed using industry hours worked and labour cost data. The chained Fisher input index for the capital productivity series was formed using industry capital stock and capital cost data.

The productivity series shown in Figure 1 represent accumulated growth from 1988. The percentage difference in productivity between two years is found by taking the ratio of the index value at the last year to the index value at the first year and then subtracting 1.[6]

Figure 1 – Multifactor and partial productivity estimates

Changes in input utilisation arising from business cycle fluctuations are reflected in productivity estimates. Although during slack periods some labour is usually shed, workers that are retained or do not have their hours reduced are often underutilised. Underutilisation of the capital stock tends to be relatively greater, because the capital stock cannot be shed as easily as labour. For this reason, annual productivity growth is usually strongly positively correlated with annual output growth (see Figure 2).

Figure 2 – Annual output and productivity growth

One way of accounting for changes in input utilisation arising from business cycle fluctuations is to calculate average productivity growth between two consecutive cyclical peaks or troughs in the level of activity. Providing input utilisation rates are the same at consecutive cyclical peaks or troughs, this is a valid method to account for changes in input utilisation and gives a measure of trend productivity growth over the classical cycle. Alternatively, trend productivity growth can be measured over the growth cycle. This is done by measuring average growth rates between three consecutive points at which the economy is deemed to be on trend. On-trend points can be identified in one of two ways. First, a variety of survey based measures of input utilisation can be used to judge when the economy is on-trend. Second, statistical filters can be used to measure the economy’s trend level of output and on-trend points identified where the trend level of output and actual output are equal.

The period covered by these productivity series is relatively short and consequently there are few business cycles over which to compare trend productivity growth. The annual output series for the market sector suggests that peaks in the level of economic activity occur in 1989 and 1997 and troughs in the level of economic activity occur in 1990 and 1998. McLellan (2001) also argued that cyclical peaks and troughs in the level of economic activity in New Zealand occurred in these years based on the Haugh (2001) aggregate production GDP series. In addition, Downing, Janssen, McLellan and Szeto (2002), who used the Quarterly Survey of Business Opinion (QSBO) to identify growth cycles for the New Zealand economy, suggested a growth cycle between 1993 and 1999.

Table 1 panel I reports trend productivity growth over the classical cycles identified by McLellan (2001) and the growth cycles identified by Downing et al (2002). The geometric average growth rate has been used to calculate trend productivity growth.

Between 1989 and 1997 trend growth in multifactor productivity was 0.82% per annum. Trend capital productivity growth was almost identical to trend multifactor productivity growth, and labour productivity growth was somewhat weaker at 0.73% per annum. Trend multifactor and capital productivity growth measured from the trough in 1990 to the trough in 1998 were slightly higher than trend growth measured from the peak in 1989 to the peak in 1997. Trend labour productivity growth was essentially the same whether measured from peak to peak or from trough to trough.

The trend growth estimates for multifactor productivity, labour productivity and capital productivity measured over the growth cycle identified by Downing et al (2002) are 1.19% per annum, 0.91% per annum, and 1.29% per annum, respectively. These trend productivity growth estimates are higher than those obtained for the classical cycle, whether measured from consecutive cyclical peaks or consecutive cyclical troughs.

Figure 1 indicates there may have been a change in New Zealand’s multifactor and capital productivity growth around 1993, with these series showing an upward trend after 1993 in contrast to the period before 1993. This is also evident from Table 1, panel II, which shows an increase in average multifactor and capital productivity growth in the period 1993 to 2002, compared to the period 1988 to 1993. However, it is difficult to conclude that there has been a structural improvement in New Zealand’s multifactor productivity growth given the short time period covered by the data. Formal time-series tests for structural breaks in New Zealand’s productivity will require productivity data that cover a longer period.[7]

Nonetheless, some recent research using longer time series has suggested the New Zealand economy experienced a structural break in the early 1990s. For example, Razzak (2002) argued that trend labour productivity growth during the 1990s was different than in the previous two decades. Similarly, Buckle, Haugh and Thompson (2002) found evidence of a significant change in New Zealand’s GDP growth characteristics dating back to 1993.