How much are wages really going up?

Friday, 25 November 2022

Pay data seems to be a classic case of being able to prove anything with statistics.

ANALYSIS: The Government and the Opposition had very different takes on what was happening to people’s pay when Stats NZ published its latest labour market figures earlier this month.

Finance Minister Grant Robertson said wages were growing “faster than inflation”, but National Party leader Christopher Luxon said the wage data showed New Zealanders “were continuing to go backwards”.

Neither were wrong, exactly.

Robertson appeared to be referring to data from Stats NZ’s Quarterly Employment Survey (QES), which estimated average ordinary-time hourly earnings were 7.4% higher in mid-August than they were at the same time last year.

But Luxon was referencing Stats NZ’s adjusted Labour Cost Index (LCI) which showed wage costs rising at the much more sedate pace of 3.7%.

Kāpura has rolled out living wage adjustments to all employees. Staff at one of their venues, Huxley's, talk to Stuff about the decision to pay the living wage.

* Meet the bar owner who's signed up for a Fair Pay Agreement for hospo workers

* Countdown signs off 19% wage increase over two years for supermarket workers

Every year Stats NZ also estimates wage changes from its Household Labour Force Survey. That last measured the rise in median hourly earnings from wages and salaries during the year to June at 6.8%.

One of the reasons economists are keeping a close eye on pay at the moment is to check whether we may be entering a “wage-price spiral” when wage rises first play catch-up and then start to fuel inflation.

But what is not clear is whether any of the statistics currently collected by Stats NZ are suitable for shining a light on that.

Waikato University associate professor Michael Cameron says he finds Stats NZ quite trustworthy, but “you need to know ‘how the sausage is made’.”

What’s the issue with the QES numbers?

Nothing much as far as they go.

Every quarter Stats NZ analyses the payroll data of just over 4000 employers, including almost all the major public and private sector employers, and a representative sample of smaller businesses.

The weightings of the data supplied by small businesses are then bumped up to compensate for the fact that not all small businesses are included in the survey.

A small chink in the QES data is that it excludes any pay information for about 100,000 workers, mostly those employed in farming and aquaculture, mainly because Stats NZ assumes pay on farms can be hard to measure.

Former Transport Minister Michael Wood announced a $61 million Government boost for bus driver pay to $30 an hour for urban drivers and $28 for regional drivers.(Video November 2022)

Intentionally, it doesn’t include any information on earnings for the self-employed.

The QES survey is based on payroll data for a week in the middle of each quarter.

So, for example, when Stats NZ reported on November 2 that data for the quarter ended September 30 showed average ordinary-time hourly earnings rose 7.4%, that information was actually from further back in mid-August.

Assuming pay is not likely to be rising much faster or slower in agriculture than in any other industry, then the QES data should provide a comprehensive, if slightly dated, snapshot of changes in pay.

But that may not be the same as wage inflation.

After all, if pay has gone up 5% because we have become more highly-skilled and hence 5% more productive, then the cost to employers of getting a set amount of work done will not have increased.

And the LCI data?

This is where Stats NZ’s ambitious “adjusted LCI” measure should probably come in.

Workplace Relations Minister Michael Wood and National Party MP Scott Simpson clash over the Fair Pay Bill.

It attempts to disregard pay rises that reflect an improvement in the quality of work being performed, so it can provide a reliable measure of the changing cost to employers of getting work done to a set standard.

To compile its adjusted LCI index, Stats NZ tracks the pay of a total about 6500 individuals or positions in the private and public sector.

They are designed to be reflective of the different jobs people do across the economy.

The data is gathered from about 2000 employers, also around the middle of each quarter.

Employers are asked to explain why they gave those 6500 people or positions pay rises, and the proportion of any rise that was due to someone gaining a new qualification or skill, for example, would be screened out of the statistics.

If any of the individuals whose pay is being tracked changes jobs or is promoted to a significantly different role, then they will need to drop out of the index altogether and be replaced by another employee doing their original role.

That’s because Stats NZ can’t assume the quality of the work they are doing won’t also have changed.

The snag?

There are at least three reasons to suspect the LCI might significantly underestimate wage inflation.

Stats NZ acknowledges it will underestimate wage growth if employers are mainly offering pay rises to new hires rather than existing staff tracked by the LCI.

There is strong evidence, at least internationally, that employers have recently been doing exactly that.

The United States’ Pew Research Centre reported in July that the typical US worker who changed jobs in the year to the end of March saw their pay rise by 9.7% more than the rate of inflation, while the typical worker who stayed in their job saw their real wage decline by 1.7%.

First Union says it's a promising sign for Fair Pay Agreements in the supermarket sector.

Another issue is the tendency for employers to offer more fancy job titles to staff in partial lieu of pay rises, or so-called “job-title inflation”.

Stats NZ agrees that if all the sales reps in a business end up as sales directors or “senior vice presidents of sales” that would probably be reflected in the survey as a change in the quality of work being produced, even if those employees were basically the same people doing the same job.

The third reason to be cautious is a basic “sniff test”.

To marry up Stat NZ’s LCI and QES data, the quality of work people were doing in the economy would need to have risen by about 3.5% in the past year.

That far exceeds any recent measure of labour productivity increases, which instead have rarely fallen outside of the band of 0.5% to 2% over the last 25 years.

ANZ chief economist Sharon Zollner points out there is evidence productivity has been falling, not rising, internationally and says she is confused by the LCI data.

“In the US, they estimate there has been a negative productivity shock, so productivity-adjusted wage growth is higher, not lower than wage growth.

“Stats NZ is implicitly suggesting that people have become more productive, and we all know we have become less productive. I’m not saying they are wrong. I just don’t understand it.”

Is there another way to measure wage inflation?

Maybe, though nothing with any official hallmark.

ANZ economist and former Stats NZ statistician Kyle Uerata cautions that doing a separate productivity-adjusted wage measure is an interesting exercise, but “fraught with conceptual discrepancies at all levels”.

That said, one crude approach would be to take the growth figure for average total hourly earnings from the QES data and adjust that for growth in productivity.

Finding the best measure of productivity growth opens up another can of worms, in part because hours worked won’t be the same as paid hours, and because of the contribution to GDP by the self-employed.

But the QES data shows New Zealanders captured by that survey were paid for just under 71 million hours in the week in mid-August that the QES covered.

That was up 2.9% on a year prior, while GDP for the June quarter was up only 0.4% on the same period the previous year.

By that measure there was a drop of just over 2.4% in productivity over the approximate period and productivity-adjusted wage inflation would have been running at a super hot 10%.

It may be a very rough and ready calculation, but it would be in keeping with Zollner’s assumption that labour costs are rising faster than wages, and not slower as the adjusted LCI might suggest.