• Summary
• Downloads
• Explanatory Notes
• Related Information
• Past Releases

FEATURE ARTICLE: THE SIZE AND FREQUENCY OF WAGE CHANGES
By James Bishop, Research Economist, Reserve Bank of Australia (on secondment to the ABS)

INTRODUCTION

The decline in wage growth over recent years has been one of the most important developments in the Australian economy. This article explores some of the factors underpinning this decline by analysing the job-level micro data from the Wage Price Index (WPI). In particular, it decomposes aggregate wage growth, as measured by the WPI, into the frequency and size of wage changes.

This work is the result of collaboration between the ABS and the Reserve Bank of Australia (RBA). Some of these results have already been previewed in a speech by the Governor of the RBA on 18 October 2016 and in the RBA’s November 2016 Statement on Monetary Policy (footnote 1) . This article presents some additional analysis and incorporates the September quarter 2016 outcome for the WPI.

BACKGROUND

The level of wage growth is an important indicator of inflationary pressure in the economy and is a key driver of growth in household income. As such, the level of wage growth has important implications for the macroeconomy. Wage growth has declined markedly in recent years. The seasonally adjusted WPI rose by only 1.9% over the previous year, according to September quarter 2016 figures released today by the ABS, compared to growth rates of 3.8% in mid-2012.

One way to shed light on the decline in wage growth is by using the job-level micro data from the WPI. To compile the WPI, the ABS collect data on around 18,000 different jobs. These jobs are followed every quarter for a number of years and the ABS devote substantial resources to ensuring that wage changes for individual jobs are accurately reported and reflect market forces rather than changes in the quality and quantity of the work performed.

This article explores one way in which these job-level data can help us understand recent developments in wage growth. It begins with a simple decomposition. This decomposition is based on the notion that there are two ways in which firms can achieve a given reduction in wage growth. The first is by reducing the average size of the wage increases they pay. The second is by reducing the frequency of those wage rises. This article quantifies the relative importance of these two margins of adjustment to the recent decline in aggregate wage growth. It then shows which parts of the wage growth distribution have played the largest role in driving these shifts.

ANALYSIS

Frequency and size of wage changes

Using the WPI job-level data, it is possible to decompose aggregate wage growth into the frequency and size of wage changes (see Appendix 1 for details). The ‘frequency’ is the share of jobs that experience a wage change in a given quarter, while ‘size’ is the average magnitude of wage increases for those wages that do change.

The average frequency and the average size of wage changes have both fallen since 2012 (Graph 1). The frequency of wage adjustment is currently at its lowest level since at least 2000, with around 21% of all wages being adjusted each quarter, compared with 25% in 2012 (footnote 2). This implies that the average length of time between wage changes has risen from once every 4 quarters in 2012 to once every 4¾ quarters in 2016. This fall in the average frequency could reflect more wage freezes or longer delays in renegotiating wage contracts (footnote 3).The average size of wage changes (conditional on a wage change) has fallen from 3.6% in 2012 to 2.3% in 2016 and is now well below its 2000s average.

GRAPH 1 – FREQUENCY AND SIZE OF WAGE CHANGES, Quarterly*

To quantify the relative importance of these two developments, Graph 2 decomposes the change in quarterly wage growth between 2012 and 2016 into the contributions of frequency and size (see Appendix 1 for details). For comparison, it also shows the respective contributions to the peak-to-trough decline in wage growth during the Global Financial Crisis (GFC), which is the other major decline in wage growth in the WPI data.

GRAPH 2 – CONTRIBUTIONS TO THE CHANGE IN WAGE GROWTH, Quarterly, Change relative to previous peak

The decomposition shows that the declining size of wage rises has contributed more than two-thirds of the overall fall in wage growth since 2012 (Graph 2). The reduction in the frequency of wage adjustment has contributed the remainder. A similar pattern is also apparent across both the private and public sectors individually. In contrast, the adjustment in wage growth during the GFC reflected a 50–50 split between frequency and size. The larger role of the frequency margin during the GFC was in part due to the Australian Fair Pay Commission’s decision to freeze the Federal Minimum Wage and award wages in 2009. The decline in the average size of wage increases since 2012 has been larger than the adjustment during the GFC.

Fewer large wage rises

The main driver behind the recent fall in the average size of wage rises has been a substantial reduction in ‘large’ wage rises. The share of wage changes that were greater than 4% fell from around 29% in 2012 to 7% in 2016 (Graph 3). The share of jobs in this category has a disproportionately large effect on aggregate wage growth because the average wage rises in this category tend to be relatively large. To see this, Graph 4 shows the contributions of five wage growth categories to aggregate wage growth over time. These contributions capture two things: (i) the fraction of jobs receiving a wage increase in each range and (ii) the average size of those wage increases (see Appendix 2 for details).

GRAPH 3 – WAGE CHANGES OF DIFFERENT SIZES, Share of jobs that experience a wage change*

GRAPH 4 – CONTRIBUTION TO AGGREGATE WAGE GROWTH, Wage changes of different sizes*

In addition to the declining share of wage changes that are larger than 4% (Graph 3), the average size of those increases has fallen from 7½% in 2012 to 5¾% in 2016, which has weighed on aggregate wage growth. While the share of wage increases of between 3% and 4% has also fallen since 2012, this has had less of an effect on aggregate wage growth (footnote 4). The share of wage rises between 2% and 3% has more than doubled since 2012 and now accounts for almost half of all wage changes. The share of jobs receiving a rise of 0–2% has also grown, while the proportion of jobs experiencing wage falls has remained broadly stable.

The declining share of ‘large’ rises has been apparent across all industries, though the shift has been largest in mining and the industries exposed to mining, such as construction and professional, scientific and technical services (Graph 5). This has contributed to a substantial fall in the dispersion – or standard deviation – in wage growth across jobs in the economy (Graph 6). During the resources boom, there was a high dispersion in wage growth across jobs, with especially strong growth in jobs exposed to mining and weaker growth in the many other parts of the economy. The reduction in the dispersion of wage growth over recent years also reflects the presence of ‘downward nominal wage rigidity’ – namely, an unwillingness or inability on the part of firms to reduce nominal wages. As the average level of wage growth falls, the share of firms that desire to cut wages, but are unable to because of rigidities, rises. This can lead to a compression in the distribution of wage growth as the mean of the distribution falls (footnote 5).

GRAPH 5 – SHARE OF WAGE RISES LARGER THAN 4%, By Industry

GRAPH 6 – DISPERSION IN WAGE GROWTH ACROSS JOBS, Standard deviation of quarterly growth rate*

CONCLUSION

Wage growth has declined markedly in recent years. Analysis of the micro-level WPI data indicates that this reflects both a decline in the frequency of wage increases and in the average size of the increases. In particular, the share of jobs that experienced wage growth in excess of 4% has fallen sharply since 2012.

FOOTNOTES

1 Lowe’s (2016) summary of this analysis and the November 2016 Statement on Monetary Policy did not incorporate data from the September 2016 release of the WPI. <back
2 In this analysis, any wage changes of less than or equal to 0.1% (in absolute value) are classified as having a zero change, given that such small changes are more likely to reflect rounding than a true change in wages. For this reason, as well as several other factors, there will be slight differences between the level of wage growth implied by Graphs 1 and 4 and the headline WPI series contained in this publication. <back
3 Renegotiation lags matter because wages are often frozen during the negotiation period. It is difficult to disentangle the impact of renegotiation lags and other (more explicit) forms of wage freezes using the WPI data. The decline in average frequency could also reflect a longer-run shift towards contracts that make less frequent wage adjustments. <back
4 The 3–4% wage range includes wage increases of exactly 4%, but excludes wage increases of exactly 3%. A similar rule is used for the 2–3% category. The 0–2% category includes wage increases larger than 0.1% but smaller than or equal to 2%, while the ‘less than 0% category’ includes all wage movements less than -0.1%. <back
5 The correlation between the standard deviation and the mean of wage growth over the sample period is 0.78, suggesting that dispersion rises with the rate of wage inflation, as predicted by downward nominal wage rigidity effects. Downward nominal wage rigidity is also the most likely explanation for the decline in the frequency of wage adjustment over recent years. <back

APPENDIX 1: THE CONTRIBUTIONS OF FREQUENCY AND SIZE

Aggregate wage growth at a point in time can be expressed as the weighted-average wage growth of each job in the WPI survey,

where is the percentage wage change for job i in quarter t and is the WPI weight of job i in quarter t (). Using a simple identity, it is then possible to decompose wage inflation into two terms measuring frequency and size, respectively,

where is an indicator function that takes the value of one when its argument is true i.e. when job i experiences a wage change in quarter t. The ‘frequency’ term measures the fraction of jobs that experience a change in wages in quarter t, while the ‘size’ term is the average magnitude of wage increases for those wages that do change.

Using this identity, it is possible to decompose the change in the level of wage growth between 2012 and 2016 as,

where and are the contributions of frequency and size to wage growth in year t, respectively. To examine the contribution of frequency to the change in wage growth between 2012 and 2016, the size effect can be removed by holding the average size of wage increases constant at its average level in 2012 and 2016,

Similarly, the contribution of the size of wage increases can be estimated by holding frequency constant,

APPENDIX 2: CONTRIBUTION OF WAGE CHANGES OF DIFFERENT SIZES

Aggregate wage growth can be decomposed into terms due to wage increases of different magnitudes using the following equation,

where, for example, is the fraction of jobs that receive an increase of 1–2% in period t and is the average size of those increases.

REFERENCES

Reserve Bank of Australia, 2016, Statement on Monetary Policy, November

Lowe, P 2016, ‘Why Is Inflation So Low?’ Speech at Citi's 8th Annual Australian & New Zealand Investment Conference, Sydney

Wage Price Index, Australia, September 2016 (cat. no. 6345.0).