Price and quality both matter: differences between the unadjusted and adjusted labour cost index series

The unadjusted labour cost index (LCI) is an additional measure that is intended to complement the official LCI. Like the official LCI, the analytical unadjusted series measures changes in salary and wage rates for a fixed quantity of labour, but reflects quality change in addition to price change.

Quality and price changes are usually identified when respondents report changes to job descriptions and surveyed pay rates. If changes are caused by quality reasons such as performance of employees, changes in qualifications, responsibility, or the experience of new employees moving into the surveyed positions, these are excluded from the official index but are included in the unadjusted index.

Examples of how a specific position would be treated in the official LCI and in the unadjusted index are available in the technical notes of the quarterly LCI Hot Off the Press.

This article explores the distribution of changes in the pay rates that are used to compile the unadjusted LCI. (To print this article, please select landscape format.)

Distribution of annual movements

The September 2010 quarter release of the LCI showed that the unadjusted salary and ordinary time wage rates for all sectors combined increased 3.1 percent compared with the September 2009 quarter. This was up from a 2.9 percent increase in the year to the June 2010 quarter. In comparison, the official adjusted LCI increased 1.6 percent in the year to the September 2010 quarter, which was the same increase as in the year to the June 2010 quarter. The following table compares the distribution of annual movements for the unadjusted series with the official LCI for the June 2010 and September 2010 quarters:

Table 1

For the unadjusted series, 6 percent of the surveyed pay rates decreased in the year to the September 2010 quarter and 7 percent decreased in the year to the June 2010 quarter. However, in the official index, there were very few decreases in the sample after adjusting for quality change. The main reason for the decreases in both quarters was 'different person', which typically means that pay rates have fallen due to new employees moving into the surveyed positions. The pay rates of these new employees were lower than the pay rates of the incumbents, in many cases by more than 10 percent.  

The proportions of surveyed positions that did not show changes in pay rates were lower for the unadjusted series than the official series in both the year to the September 2010 quarter (36 percent versus 51 percent, respectively) and the year to the June 2010 quarter (37 percent versus 53 percent, respectively). These differences imply that there were pay rates reported in the sample that did change but were adjusted to remove the estimated effect of changes in the quality of labour input for the surveyed positions.  

The proportions in the sample that increased were higher in the unadjusted series than the official series, indicating that some pay rates had no change in the official index but showed increases in the unadjusted series. These shifts were also observed within the distribution of annual increases. In the year to the September 2010 quarter, 40 percent of the sample increased by no more than 5 percent for the unadjusted series, about the same proportion (41 percent) as for the official index. However, 18 percent of the sample increased by more than 5 percent in the unadjusted series compared with only 8 percent of the sample in the official index. These shifts imply that there were pay rates that increased but were adjusted in the official index to remove the estimated effect of quality changes in the surveyed positions. In the years to both the June and September 2010 quarters, 8 percent of unadjusted pay-rate increases were more than 10 percent. The main reason given for rises of 10 percent or more was that a different person was doing the job.

Reasons for increase

Respondents are asked to give one or more reasons for each pay-rate increase in the surveyed positions they report. The reasons for annual increases for the year to the September 2010 quarter are summarised in table 2. 

Table 2

Annual increases were commonly due to cost of living (35 percent) and collective employment agreements coming into effect (28 percent). This is consistent with the reasons for the official index over the same period, but the proportions were higher for the official index as it excluded increases for quality reasons (ie 'cost of living' and 'collective employment agreements' were 43 percent and 34 percent, respectively, in the official index).

For the unadjusted series, 'cost of living' and 'collective employment agreements' were the most common reasons for the smaller annual increases (up to 5 percent). For increases of more than 5 percent, reasons tended to shift to matching market rates and to quality-related reasons. Increases of more than 8 but not more than 10 percent were commonly due to 'match market rates and/or retain/attract staff' (32 percent) and 'performance of employee' (31 percent). Nearly half (47 percent) of those that increased by more than 10 percent were due to new employees moving into the surveyed positions.

For pay rates that decreased in the year to the September 2010 quarter, the main reason provided by respondents was that a different person was doing the job. Eighty-seven percent of those that decreased were due to new employees moving into the surveyed positions. Note, reasons can be due to combinations of price and quality change. These combinations can occur in the same quarter or in different quarters within the same year. This is why some annual decreases had price change reasons such as 'cost of living' and 'match market rates' in combination with 'different person'.

Median and mean increases

The median (ie middle) increase for all surveyed salary and ordinary time wage rates that rose in the year to the September 2010 quarter was 3.4 percent for the unadjusted series, compared with 2.9 percent for the official index over the same period.

The mean increase for all surveyed salary and ordinary time wage rates that rose in the year to the September 2010 quarter was 6.2 percent for the unadjusted series, compared with 3.6 percent for the official index.

The median and mean increases by sector for both series are as shown in table 3.

Table 3 

The median increases of the unadjusted series were higher than the median increases of the official index. There were more observations in the unadjusted series than the official index, and the annual increases were larger, as movements due to quality changes were included.

The mean increases of the unadjusted series also showed higher values than those of the official index, but gaps were wider than the gaps between the two series for the median increases. A high proportion of pay-rate increases in the unadjusted series were significantly higher than those included in the official index. Nearly one-third (31 percent) of all pay-rate increases in the unadjusted series were more than 5 percent in the year to the September 2010 quarter, with about half of these being increases of more than 10 percent.

Median and mean changes

As the unadjusted series had 6 percent of the sample that decreased in the year to the September 2010 quarter, table 4 shows a separate analysis that includes decreases.

Table 4

The median changes of the unadjusted series are now lower and close to the median increases of the official series. The number of observations that were included in the calculation was 32 percent higher than the official index (compared with 20 percent higher when decreases were excluded). Since the additional observations were decreases, this moved the medians down and closer to the median increases of the official index.

Similarly, the mean increases of the unadjusted series are lower than in table 3 because of the decreases being added to the calculation.

Different person

In the September 2010 quarter, 4 percent of the surveyed positions in the unadjusted series were affected by a different person doing the job.

A look at the surveyed positions that had 'different person' as the reason in each quarter from the September 2009 quarter to the September 2010 quarter provides a picture of their contribution to the unadjusted series.

Table 5

In the September 2010 quarter, 47 percent of the surveyed positions that changed employees were replaced by employees who had lower pay rates than incumbents. More than half of these (55 percent) had pay rates that were lower than incumbents' pay rates by more than 10 percent. Fifty-one percent were replaced by employees with higher pay rates, one-third of whom were paid over 20 percent more than incumbents.

Table 6 shows the median and mean movements of the pay rates of the new employees replacing incumbents for the September 2009 quarter to the September 2010 quarter.

Table 6

Unlike the official median and mean increases calculation, the above movements were calculated including pay rates that decreased and those that remained unchanged.

For all sectors combined, the median movement for four of the five quarters was no change. Both increases and small decreases were recorded for the mean movement.

Table 7 shows the median and mean increases for the September 2009 quarter to the September 2010 quarter, and table 8 shows the median and mean decreases, for surveyed positions where new employees replaced incumbents.

Table 7

Table 8

For the September 2010 quarter, the mean increase (21.7 percent) for all secters combined was higher than the mean decrease (13.9 percent). This had an upward influence on the gap between the unadjusted quarterly increase (1.1 percent) and the official quarterly increase (0.5 percent). In contrast, for the December 2009 quarter, the mean decrease (13.4 percent) was higher than the mean increase (12.9 percent). This had a downward influence on the gap between the unadjusted quarterly increase (0.5 percent) and the offical quarterly increase (0.4 percent). Note, the mean and median movements in this article are unweighted, whereas movements of the unadjusted and official series are weighted to reflect the relative importance of the surveyed positions within their occupations and industries.

In conclusion, the unadjusted series has significantly greater proportions of the sample showing both decreases and increases of more than 5 percent, most of which are quality adjusted in the official series. For nine in 10 of the unadjusted decreases and one in two of the unadjusted increases of more than 10 percent, respondents indicated that different employees had moved into surveyed positions.

On average, new employees who replace incumbents are on similar pay rates, although this fluctuates from quarter to quarter and has an impact on the size of the gap between the unadjusted and official series. How well this is measured partly depends on how well the sample represents entrances and exits of employees, and on whether the sample replacement practice is unbiased in this regard (eg in some cases, replacement employees are incumbent employees filling other positions rather than new employees filling the existing positions – this can happen when there is a delay filling vacancies in surveyed positions). In addition, the analytical unadjusted index tends to reflect the effect of turnover in, and the cessation of, existing positions, but not the price and/or quality effect associated with employees being hired to fill new positions. An unadjusted measure designed from scratch might make use of the average pay rate within each surveyed firm of all employees filling jobs in each surveyed occupation.

Back to Price Index News