Real wages during the recession

I've seen a few stories about recent trends in wages, and too many of them seem to be trying to make a big deal out of movements that - when in put in context - are really too small to say much of anything.

[Updated in an almost-certainly futile attempt to combat confirmation bias.]

Here are the SEPH data for the average hourly wage for those paid by the hour, and for average weekly earnings. They are are divided by the CPI, and rescaled so that they are expressed in constant 2010 dollars. All series are deseasonalised.

The most recent months are down, certainly. But given the scale of variation in these fairly noisy series, I don't see how anyone is ready to declare a trend just yet.

And if we're going to start seeing stories to the effect of "real wages are declining in weak labour market", I'm going to have to start asking pointed questions about why we weren't seeing "real wages are increasing during the recession" stories over the past couple of years.

Update: I was surprised - although I probably shouldn't have been - to see that immediate reactions on twitter largely consisted of "I reject the data in favour of my preferred narrative." So here are the LFS data. These are not seasonally adjusted, but the LFS does provide data for the median.

 

I don't expect the LFS numbers will cause anyone to revisit his preferred narrative any more than the SEPH numbers did. Confirmation bias is too powerful.

Comments

[Nick Rowe]

Aha! Thanks for posting this Stephen. My take-away from my vague recollections of the studies I've skimmed is that anyone who tries to argue that real wages are either strongly pro-cyclical or strongly counter-cyclical is arguing more than the data can support.

My maintained hypothesis is that real wages are roughly a-cyclical. Wages and prices are roughly equally sticky. But eyeballing your graph, I maybe get the sense that real wages are lagging pro-cyclical?? They fall in a recession, but lag the recession? So they actually fall most during the recovery, and hit their peak after the top of the boom?

[Stephen Gordon]

Remember this post?

My impression also was that real wages were acyclical, but I suppose you can put together anecdotal explanations for variations - in either direction - during a given episode.

[Nick Rowe]

God, I had forgotten that. Hmm. There must be *some* story (or stories) explaining real wages and the business cycle. But it's not a simple one.

[Frances Woolley]

Remember that during the recession the people who keep their jobs are people in health and education, whereas people who lose their jobs are people in construction and other more cyclical industries. If a whole lot of minimum wage workers in tourist or retail industries lose their jobs, whereas nurses, doctors and teachers keep their jobs, average wages will go up.

It usually takes a few years from recession to governments realizing that they have no money and have to freeze public sector salaries e.g. the big hit on public sector salaries were in say 83, 84 after the recession of 80, Rae Days were in mid 1990s after the recession of 1990,91, etc.

Try breaking it down by gender and seeing what happens. The cyclical pattern of male and female employment is quite different.

[Lorenzo from Oz]

What Frances Woolley said. Loss of employment in recessions is overwhelming in the private sector, who have lower average wages than the public sector. Even in the private sector, one suspects that loss of employment is not even across income groups.

[Robert McClelland]

Average wages. You must not of heard the one about how Bill Gates walking into a bar raises the average wage to a million dollars.

[Ryan]

This would probably be more fruitful if you separate earnings into quantiles and expand the measurement to include the transitionally unemployed.

I agree that looking at average wages is not very insightful, but that doesn't mean that there are no insights to be gleaned from tracking real wages over time.

Out of curiosity, how does the median wage compare?

[Adam P]

Stephen,

As I recall for the US data wages appeared to rise early in the recession and then flatenned out, though didn't fall. However, productivity growth surged causing unit labour costs to plummet (before also flatenning out).

This probably reflected the fact that the lowest producing, lowest paid labour was getting canned so those still employed were both better paid and more productive.

Nonetheless, in aggregate data if we measure labour in effeciency units or something like it then real wages did in fact fall.

[Livio Di Matteo]

Hi Steve:
According to the Stats Canada release last week, average weekly earnings are up over the last year in most provinces and for Canada as a whole - but they are down in Ontario and Nova Scotia. Ontario was apparently surpassed by Saskatchewan.

[Determinant]

As a purely informal survey, I regularly check the Public Service of Canada's jobs.gc.ca job board. The Public Service has a policy of posting all externally advertised jobs there, including executive level.

Public Service openings at the lower entry levels (PM-01, Policy Administrator, Level 1, is a broad and typical entry-level classification in the Core Public Service) have become more infrequent within the last six months. The government has announced an "austerity" drive so departments appear to have curtailed new hiring, especially for permanent positions.

This is a back-of-the-envelope record of Government of Canada hiring, but jobs.gc.ca gives you an unusually convenient single-point source to make such a survey.

[Nick Rowe]

Hmmm. The SEPH data show a big drop in hourly real wages in 2001, but the LFS data show nothing happening then. That's the only difference I see. But it's strange.

Sorry, but what does SEPH stand for, and any guesses as to why the difference in 2001?

[edeast]

Stephen I suppose you've seen that Statscan online data will be free next year.

[Stephen Gordon]

Yes, and I'm somewhat over the moon about that. Hopefully people will realise that they don't have to make up their own facts; they can go look up the real thing.

Nick: The SEPH is the establishment survey. And no idea what happened in 2001 or why the SEPH numbers show a decline, while the LFS numbers don't.

I may have to do another post comparing and contrasting the two surveys, asking why there's a difference and which is more reliable.

[K]

Stephen: "Hopefully people will realise that they don't have to make up their own facts; they can go look up the real thing."

...and *then* misrepresent it.

But seriously... Surprisingly public minded move from this government. I guess, after making it worthless last year, making it free was the next logical step. :-)

[Steve Roth]

Looking at median hourly because of the obvious problems of mean, and because weekly makes us think about leisure time: What Lorenzo said. If a larger number of people at the low end go unemployed, or if (relatively) more people at the high end get employed, the median goes up. Maybe what happened 2009-2010, with some reversal in 2011?

[rp1]

We're looking at a roll over here.

[Jon]

Real-wages appear countercyclical during demand shocks and procyclical during supply shocks.

[W. Peden]

Jon,

I'm not sure how reliable even that rule would be. British average real earnings rose in the Great Depression (by an average of 3.03% per year), the 1974-1975 recession (by an average of 1.69%), the 1980-1981 recession (by an average of 1.65%), the ERM recession (by 1.7%) and the most recent recession (by 0.2% in 2009).

The 1974-1975 recession was a supply-shock (complicated by crowding-out and accelerating inflation distortions) yet real wages rose.

Looking at the figures for each recession, it's notable that (a) the proportionate rise in the level of unemployment, once the fall in GDP is taken into account, bears some relation to the rise in real wages; (b) there is no consistent difference between supply-side and demand-side recessions; (c) given the long-term costs of unemployment, a flexible labour market becomes extremely important in a recession.

[Chris of Stumptown]

Do these series include part time and temporary workers?