Real wage stagnation, year 41
Remember the Economic Report of the President? Of course you do. It’s where we get our annual data on real wages (and apparently some other stuff, too). The 2014 edition was released on March 10. As you may recall, I made my first post on the declining real wage trend through 2011 and was literally the first person to notice 2012’s further slight decline.
The good news is that, in what is now Table B-15 rather than B-47, real wages advanced somewhat in 2013, from $294.31 per week (in 1982-84 dollars) to $295.51, an increase of 0.4%. Woo hoo!
The bad news, of course, is that this is still 13.5% off the peak real weekly wage of $341.73, achieved in 1972. One swallow doesn’t make a spring, and all that.
Interestingly, last week Paul Krugman felt compelled to argue that real wages aren’t going up all that fast, but so what if they did? He said that basically, this was something primarily only visible in the real hourly (my emphasis) wages of production and non-supervisory workers, which happens to be one of the components of Table B-15. However, he was reporting based on the Bureau of Labor Statistics’ monthly reporting of this stat.
As he puts it, the folks he is criticizing are saying that “a dangerous acceleration in the pace of wage increases is already underway. Time to raise interest rates!” His response to them is fine as far as it goes, but he misses that even on the terrain of the supposedly most rapidly increasing measure, there is no there there.
| Feb.2013 | Dec.2013 | Jan.2014(p) | Feb.2014(p) | |
|---|---|---|---|---|
|
Real average hourly earnings(2) |
$8.73 | $8.81 | $8.83 | $8.86 |
|
Real average weekly earnings(2) |
$294.96 | $295.22 | $295.69 | $295.08 |
|
Consumer Price Index for Urban Wage Earners and Clerical Workers |
229.180 | 230.919 | 231.233 | 231.344 |
|
Average hourly earnings |
$20.00 | $20.35 | $20.41 | $20.50 |
|
Average weekly hours |
33.8 | 33.5 | 33.5 | 33.3 |
|
Average weekly earnings |
$676.00 | $681.73 | $683.74 | $682.65 |
|
OVER-THE-MONTH PERCENT CHANGE |
||||
|
Real average hourly earnings(2) |
-0.3 | -0.1 | 0.2 | 0.3 |
|
Real average weekly earnings(2) |
0.2 | -0.6 | 0.2 | -0.2 |
|
Consumer Price Index for Urban Wage Earners and Clerical Workers |
0.6 | 0.3 | 0.1 | 0.0 |
|
Average hourly earnings |
0.3 | 0.2 | 0.3 | 0.4 |
|
Average weekly hours |
0.6 | -0.6 | 0.0 | -0.6 |
|
Average weekly earnings |
0.8 | -0.3 | 0.3 | -0.2 |
|
OVER-THE-YEAR PERCENT CHANGE |
||||
|
Real average hourly earnings(2) |
0.1 | 0.8 | 0.8 | 1.5 |
|
Real average weekly earnings(2) |
0.4 | 0.2 | 0.4 | 0.0 |
|
Consumer Price Index for Urban Wage Earners and Clerical Workers |
1.9 | 1.5 | 1.6 | 0.9 |
|
Average hourly earnings |
2.0 | 2.3 | 2.3 | 2.5 |
|
Average weekly hours |
0.3 | -0.6 | -0.3 | -1.5 |
|
Average weekly earnings |
2.3 | 1.7 | 2.0 | 1.0 |
Source: Bureau of Labor Statistics (link above), footnotes omitted
First of all, we should remember that a couple months’ trend is not really equal to a 41-year trend. That kind of error we’ll leave to Reinhart and Rogoff.
Second, we can see the “scary” number: nominal hourly earnings increased from $20.00 in February 2013 to $20.50 in February 2014, or 2.5%.
Third, we know that we actually want to adjust that for inflation; hence we find that real hourly earnings went from $8.73 in February 2013 (1982-84 dollars) to $8.86 in February 2014, or just 1.5%,
Fourth, what Krugman appears to miss is that average weekly hours have fallen substantially since 1972. In fact, as the BLS table shows, they fell by half an hour, from 33.8 to 33.3 hours, from February 2013 to February 2014, or 1.5%.
You can see where this is going now: Real weekly earnings went up by 0. More precisely, 0.04%. And you’ll note that the February 2014 figure is lower than the full-year 2013 level. So there is actually no increase to explain in the production/non-supervisory workers series.
But Krugman does hit the nail on the head: “What’s so bad about rising wages?” And the answer, of course, is “Absolutely nothing.” (“Say it again!”)
Cross-posted at Middle Class Political Economist.
Ken:
I was just commenting on the stagnation of household income for families and the more likely decrease of it for 90% of the households. Good timing on the post.
What is the real hourly wage and real weekly earnings deflating both series by the PCEPI rather than the CPI-W back to 1972?
There is a divergence in the 2 deflators starting in 1980.
The CBO reports (in Table 1 that real hourly wages at the 50th percentile have risen particularly for women in the 1979-2009 period.
Good graph, Marmico. But I think the new FRED messed it up. I see that for “Units” you chose “Index (Scale value to 100 for chosen period)” but FRED omits the chosen period date, and apparently defaults to 2009. It makes your graph look like the two lines converge at 2009 rather than diverge from 1980. Good grief!
I picked a “U.S. Recession peak date” of 1980-01-01 for both series, and the divergence you point out showed up.
I added GDPDEF to the graph, indexed to the same date, and it closely followed the lower of your two lines. But I’m not linking to the revised graph, because FRED has become untrustworthy.
Have you seen Andrew Kliman’s “Clarifying ‘Secular Stagnation’ and the Great Recession” at New Left Project —
http://www.newleftproject.org/index.php/site/article_comments/clarifying_secular_stagnation_and_the_great_recession
Kliman: “A remarkable recent study of U.S. household income by Armour, Burkhauser, and Larrimore looked at after-tax income, including transfer payments such as Social Security benefits and noncash income such as medical-insurance benefits, and adjusted for changes in household size. The authors found that the inflation-adjusted income of the middle 20% of households rose by 34% between 1979 and 2007, while incomes of the lowest and next-to-lowest 20% groups rose by 32% and 31%, respectively. And during the 1989–2007 sub-period, income inequality actually declined: these groups’ income growth outpaced that of higher-income households.”
Kliman pesters my mind.
FRED has become untrustworthy. It may be a temporary bug. Typically I link to the FRED page, as it is (was) interactive, as opposed to the image.
I’ve read the work of Armour et. al.. It is similar to the CBO’s Trends in the Distribution of Household Income Between 1979 and 2007.
“The authors found that the inflation-adjusted income of the middle 20% of households rose by 34% between 1979 and 2007, while incomes of the lowest and next-to-lowest 20% groups rose by 32% and 31%, respectively. And during the 1989–2007 sub-period, income inequality actually declined: these groups’ income growth outpaced that of higher-income households.”
Somebody is dreaming or in a different economic dimension. Every other national study including Angry Bear’s Spencer England’s Labors Share http://angrybearblog.strategydemo.com/2009/10/labors-share.html have shown decreasing real wages for Labor. If you look closer at the study, I believe you would see what you have claimed in terms of “Social Security benefits and noncash income such as medical-insurance benefits;” but, they ignored other benefits received by Labor such as these employer benefits “pensions, taxes other than income and payroll taxes (such as corporate, excise, and capital gains taxes), or the wide variety of in-kind transfer benefits (e.g., the Earned Income Tax Credit, school lunches, SNAP, child care, housing assistance, and energy assistance?”
It fries my mind, Run. I would expect data like that from the right, but not from a New Left Project…
Arthurian:
What are we talking about here? You are being a tad too cryptic as I did not look at the politics of any of them and I was interested in the numeric of the studies being cited. Are you discussing Burkhauser, Larrimore, and Simon or Kliman? Or have I slipped into something in which I am unaware?
Arthurian, if you adjust for household size without adjusting for number of incomes per household, you are sure to bias the figures upward. Well, that’s true if you do or don’t adjust for household size. In any event, focusing on household income is a fundamental mistake. You have to look at individual income. The fall in individual income is what has necessitated the increase in number of incomes per household, which is mostly increased female labor force participation.
Ken:
There is more to this than what is being told. I did find a brief study refuting the Burkhauser, Larrimore, and Simon study cited which some of our posters are using to claim that Labor did see magnificent increases in wages. It was then I began to understand what is happening here. Rather than cite such costs as the YOY increase in the cost of healthcare as a negative to Labor’s share, the study has used it (in the form of healthcare insurance) as increases to Labor’s wages in the form of ESI, Medicare and Medicaid which reflects the cost of healthcare. The offset would have been the decreases in 401k/pension contributions by companies, decreases in TANF, SNAP, school lunches, EITC, etc. The original study left those offsets out and included everything relating to healthcare (with its extreme growth) which skews Labor’s share upwards. The original study by Burkhauser, Larrimore, and Simon is flawed as it cherry-picked the numbers (healthcare) to use and left others out.
Run, I read Kliman, and quoted Kliman in response to Marmico. You quoted part of that excerpt and seemed to refer to it as what I have claimed. To be clear, I’m not claiming anything and, in particular, I am NOT among some of your posters who may be using some study “to claim that Labor did see magnificent increases in wages”. Just to be clear.
Your comment to Ken just above is the kind of thing that I am looking for:
“Rather than cite such costs as the YOY increase in the cost of healthcare as a negative to Labor’s share, the study has used it (in the form of healthcare insurance) as increases to Labor’s wages…”
Yeah, that agree with what I was thinking, and if you are saying there’s something not right about it, I certainly agree with that, too.
Ken, What does it mean to “adjust for household size”? If the ‘average household’ today is half or twice as big as it was N years ago, then I suppose an argument could be made that half or twice as much household income would be the right number. Of course, if people today have to work three jobs to live as well as their parents did with one job, something’s still not right. But all this is very fuzzy for me, so far.
RE your statement that “In any event, focusing on household income is a fundamental mistake. You have to look at individual income.”
That reminds me of something Kliman said:
“Figure 1 depicts trends in household borrowing (there are no data specifically for ‘workers’).”
I don’t know, maybe Kliman is wrong. I would just like all of this to be a little less fuzzy in my head.
What is the real hourly wage and real weekly earnings deflating both series by the PCEPI rather than the CPI-W back to 1972?
Okay, I’ll answer my own question.
Real hourly wage +16%
Real weekly earnings +6%.
There is no wage stagnation using PCEPI* as the wage or earnings deflator.
*It rose from 24.07 in 1972 to 107.2 in 2013 or a factor of 4.45
marmico:
Not ignoring you; but, I would suggest that even then the numbers are heavily skewed due to the top 10% of households receiving greater amounts of income and being able to spend.
marmico: We know from the Social Security debate that chained calculations give a lower inflation figure than the standard approach. So you need to justify why your deflator is better than the one chosen by the experts at BLS.
Even if we accepted your figures, I am not sure how you cannot be straining to call 6% in 41 years anything other than “stagnant.”
Arthurian: Yes, you have got the right idea about adjusting for household size. You divide household income by average household size to get what’s effectively a household per capita figure. Again, using household income at all is a mistake unless you adjust for the average number of incomes per household. If you start with individual data instead, you don’t need either of those calculations.
The PCEPI is the preferred deflator in the measurement of the Personal Income and Outlays component of the National Income and Product Accounts, by the Congressional Budget Office and by the gods that pull the monetary levers at the Federal Reserve.
Focus on the hourly wage. For all we know the weekly earnings may be a function of an increasing ratio of part time to full time workers going back to 1972. A mean hourly real wage increase of 16% in 41 years is not stagnant.
Yes, you have got the right idea about adjusting for household size. You divide household income by average household size to get what’s effectively a household per capita figure
No, you adjust household income by the square root of total household occupants. See CBO Appendix A.
Okay, Thomas is now trapped in his “justification” narrative. It actually was a reverse onus. If he was bright,
1. He would have said that the male hourly wage is stagnant. Fair enough and I would agree. My response is SO WHAT. If nonsupervisory males want to participate in a rising standard of living, then they must become more skilled. A 25 year old male in 2014 has not improved his skill set (as measured by years of education) relative to a 25 year old male in 1972. Nonsupervisory women are doing just fine since 1972 as they have improved their skill set.
2. He would have said that nominal personal consumption in the NIPA must be 20% higher than recorded, the percentage difference between the CPI-W and the PCEPI since 1972. That’s over a trillion dollars or about a 15% increase in economic activity as recorded by the Bureau of Economic Analysis. .Does not compute.
marmmico:
Just doing a little research here besides having to work once in a while :). Is it true, the PCEPI does not include a housing deflator in its calculation which is the largest component for CPI? Housing weights as a part of the CPI rose from 20% in 1997 to 25% in 2010 before falling back to 24% in 2013. That is pretty sizable part of the CPI which appears to be substantially less or ignored by the PCEPI.
CPI also uses a fixed basket of goods where the PCEPI varies its basket of goods believing that if 90% lean hamburger become too expensive, the consumer will substitute 80% lean hamburger. This would be a lowering of the standard of living which it appears the PCEPI does not recognize. Granted it may be more accurate month to month; but, it would not make it a better measure of a standard of living.
As far as the comments to Ken, now you are being an ass and I do not like asses. Ken is a pretty nice guy. What you have said is true about males as they are caught up in the lack of education as opposed to females; but then and maybe rightfully so, the emphasis has been on educating females in those studies typically reserved for males. Maybe society will get back to offering an education.
In any case, it does not change what has taken place with income dispersal. The largest percentage of income gain has bee heavily skewed to the upper 10% of households. Within that 10%, it is heavily skewed to the 1% and the 1/2 of 1 percenters.
Marmico: Thanks for the info on how to adjust by size of household. It appears to me that using the square root definitely pushes upward reported household income relative to a per capita adjustment. As average size per household falls, as it has, the decline in the divisor is less if you are using the square root; hence, you report household income as higher. This is on top of the fact that the calculation reported does not account for the increase in incomes per household. So households are still the wrong unit, IMO.
The PCEPI is a chained index, and it is subject to the point Run made, that substitution itself (the key to chained indices including chained CPI) is a sign of a deteriorating standard of living.
Hourly is not the right measure. If there is more involuntary part-time work, people are being paid less. And even if it were right, 16% in 41 years is still stagnant.
I am off to a conference so I won’t be checking back in until Sunday or Monday.
Are you presenting at the conference, Ken?
The argument is not about mean income dispersion. No one argues that the basic measure of dispersion, the gini index, is rising for both market income (Piketty, Saez), and after tax, after transfer income (CBO). The latter is the redistribution (or mitigation) mechanism which has become less effective since 1972.
As Lambert says the return on capital (the “stock”) is rising faster than the return on income (the “flows”).
Is it true, the PCEPI does not include a housing deflator in its calculation which is the largest component for CPI?
No. See an FRBSF Economic Newsletter here for discussion of “shelter”.
I will repeat. Insignificant annual differences compound to significant multi-decadal differences.
Asses like Run75441 need to justify why nominal NIPA GDP is understated by trillions.
Hourly is not the right measure. If there is more involuntary part-time work, people are being paid less.
Part timers get paid less per hour to do the same nonsupervisory (noncognitive) work that full timers do. That’s a non-issue.
I agree that at this stage of the business cycle the part timers for economic reasons is elevated relative to prior cycles. That’s a non-issue.
Now your job is to reconcile that the ratio of part time to full time rose from 18.8% in 1972 to 23.4% in 2014 thereby reducing weekly earnings. You can only account for ~1.5% of that 4.6% (call it one third) difference due to economic reasons. Plus, there is a 1994 data “definitional” break in the time series.
That’s why you focus on the hourly wage.
Kenneth: “if you adjust for household size without adjusting for number of incomes per household, you are sure to bias the figures upward.”
Richard Serlin expands on that a bit in his “Important points that are rarely made on income inequality statistics”
http://richardhserlin.blogspot.com/2014/02/important-points-that-are-rarely-made_2.html
Kliman resurfaces at Truthdig:
“It is reasonable to want statistics that reflect the particularly rapid inflation that consumers have experienced. Nonetheless, to offer another analogy, the EPI’s procedure produces a lopsided growth of productivity relative to compensation in much the same way that you would make a car lopsided by deflating the tires a little on one side and a lot on the other. The numbers cannot be fairly compared.”
http://www.truthdig.com/report/print/are_corporations_really_stealing_workers_wages_20140409