A nit-pick: take home pay is deflated by the CPI which includes health care. Because health care is a big part of the rising CPI in recent years, deflating take home pay with the CPI (which is what everyone does) distorts the purchasing power of take-home pay for people with generous health care plan.

Nit-pick nit-pick: Samuel also says that fringe benefits are high proportion of increasing compensation, so the more highly inflated health care benefits contribute disproportionately to the rise in compensation.

Suppose all of your health care costs are insured by your employer. Health care costs double but all other prices fall so that the overall CPI is unchanged.

All other prices have risen recently, particularly "volatile" food and energy prices.

Suppose that all of my health care costs are "insured by my employer", i.e. 100% of my health care premium contributes to my compensation as a fringe benefit. Health care prices double. All prices (including health care) rise four percent, but most prices rise much less. Non-health-care prices rise only one percent.

My health care premium is five percent of my total compensation and it doubles, so my nominal compensation rises five percent. I don't really consume more health care. I only pay more for it.

In this scenario, my "real compensation" rises two percent, but my buying power falls. All of the five percent rise is health care inflation. The balance of my income doesn't change, but it still loses loses buying power, so my compensation effectively falls.

Some economists spin figures one way while other economists spin them another way. We know less about the real world than we know about the way people spin figures. Russ is right about that, but he's inevitably a player in the spin game himself. So am I, of course.

Average household income above the 99th percentile rose over 200% in three decades while average income below the 90th percentile rose only 10% during the same period. In an earlier three decade period, income below the 90th percentile rose 80% while income above the 99th percentile rose only twenty percent. We've hardly discussed this change.

Falling household size accounts for a bit of the change, but average household size didn't fall 800% for households below the 90th percentile, and household size above the 99th percentile certainly didn't increase ten fold.

The household size explanation is off by an order of magnitude as an explanation of the observed change. So why are we satisfied with this explanation? We're only playing a political game with the figures. Right?

Fringe benefits are dominated by two roughly equal components. One is employer contributions for retirement benefits, including social security. The other is employer contributions for health care, including medicare and medicaid.

Retirement benefits grow with wages and there is no problem with deflating them with the overall CPI.

However, health care benefits grow with health cost and are the reasons fringe benefits grow faster than wages. Health cost generally rise two to three percentage points more than the CPI — occasionally, like last year the overall CPI is greater — deflating this share of fringe benefits with the CPIcauses estimates of real fringe benefits to overstate real fringe benefits growth.
I "guesstimate" the distortion is something on the orders of one percentage point annually.

"Read the whole article. He was partly responding to people who compare the latest data to 2000 which was the peak of the latest boom."

That isn't even true. The household series he uses peaked in 1999, if other people were trying to be devious they'd use that date. And if you are using macro lingo "the peak of the boom" was in March 2001.

It looks a lot more like he just looked at the data and said, "hmmm, there's a lot of wage growth between 1997 and 1998. I better make sure I include that one."