« White House On The Equity Premium | Main | Saudi Elections »


Nathan Newman (as he often is) is mad at me for failing to appreciate that the reason people think prices are growing faster than wages is that even though prices grow slower than mean wages, they have grown faster than median wages:

In fact, between the 1970s and about 1995, real wages fell for most workers as measured by median wages. There was a pickup in real wage gains in the late 1990s, but those gains have declined in the last few years and the massive declines in pension and health care benefits in the last few years are leaving many workers worse off in real terms.
Nathan then suggests readers "see EPI's State of Working America." So, to quote from EPI's State of Working America:
Between 1947 and 1973 productivity and real median family income both grew 104%, a golden age of growth for both variables. Over this era, there could be no doubt that the typical family fully benefited from productivity growth.

Yet starting in the mid-1970s, this lockstep relationship broke down. From 1973 to 2002, median family income grew at about one-third the rate of productivity (22% versus 65%).
In other words, median real wage growth slowed, but continued. Wages rise faster than prices. Other quotes, "From 1995 to 2000, median family income grew at an annual rate of 2.1% for whites, 2.9% for blacks, and 4.6% for Hispanics." Sadly, however, "Real median family income fell by over $1,300 (2.4%) from 2000 to 2002 (in 2003 dollars." In addition, "Between 1979 and 2000, for example, the real income of households in the lowest fifth (the bottom 20% of earners) grew 6.4%, while that of households in the top fifth (the top 20% of earners) grew 70%." In other words, median wages grow faster than prices.

February 12, 2005 | Permalink


TrackBack URL for this entry:

Listed below are links to weblogs that reference Newman!:

» Reply to Matt on Wage [non]-Growth from NathanNewman.org
Matthew Yglesias responded to my post on why most Americans don't think wages have increased much. I pointed out that... [Read More]

Tracked on Feb 15, 2005 3:57:56 PM

» Reply to Matt on Wage [non]-Growth from NathanNewman.org
Matthew Yglesias responded to my post on why most Americans don't think wages have increased much. I pointed out that... [Read More]

Tracked on Feb 15, 2005 4:04:49 PM


Uh, Matt. Median family income and median wages are not the same thing. Median family income could rise because the number of people per household at work rises, even while median wages fall. So the data you refer to are not necessarily relevant to the issue. Unfortunately, I don't have access right now to data to try to see what's what.

Posted by: Donald A. Coffin | Feb 12, 2005 9:23:14 PM

It's still deceptive to say that median wages grow slower than prices unless you account for immigration. Immigrants tend to be in the bottom fifth in income. It's entirely possible that the bottom 20% of wage earners in 2002 would have lower real wages than the bottom 20% of wage earners in 2000 even though the actual individuals who were in the bottom 20% in 2000 would have seen real wage growth. The set of individuals in the workforce in 2000 and the set in 2002 are not identical.

I'm not sure exactly what the effect of considering immigration would be, but I don't think it's fair to conclude from the data you have that real wages of individuals declined. I'm also somewhat doubtful that you can infer a meaningful trend from only three years.

Posted by: Xavier | Feb 12, 2005 9:27:20 PM

And, by the way, if mean wages are rising more rapidly than median wages, the distribution of WAGES (not income) will become more unequal. This does seem to be consistent with much of what we know about wage changes over the past 20 - 25 years.

Posted by: Donald A. Coffin | Feb 12, 2005 9:50:15 PM

I think the word you are looking for is "mode." It's been a few years since I had statistics, but I'm pretty sure mean and median are synonyms. But talking about median income is a fool's (or a republican's) game because it's like that joke where Bill Gates walks into a bar and one guy turns to his friend and says, "hooray, the average income of everyone in this bar is now $1 million/year!" His friend says, "don't be stupid, our income is still the same." And the first guy says, "now don't go starting with the class war!"

A bad joke, but anyway, median is where you sum up all the numbers and divide them by however many samples you have. It only takes a few outliers to give you a misleading picture. "Mode," the number almost no one uses, is the value that occurs most often. If you are trying to understand what is the typical case, mode is what you want.

Posted by: SadieB. | Feb 12, 2005 11:01:14 PM

Let's recap. NN says that median real wages fell between 1970 and 1995 which was followed by some real gains in the late 90s .

MY comes back with "... from 1973 to 2002...median family income grew" and "... between 1979 and 2000... real household income grew".

Besides skipping over the difference between wages and family income, you've used a different set of dates.

Posted by: Quiet Storm | Feb 12, 2005 11:03:11 PM

It's been a few years since I had statistics, but I'm pretty sure mean and median are synonyms.

No they aren't.
Arithmetic Mean:
The sum of a set of values divided by the number of values in said set.

The value amongst a set of values that would be in the middle were the values ranked in order of magnitude. Or, the value for which there are an equal number of other values larger and smaller than it in a set.

Posted by: WillieStyle | Feb 12, 2005 11:12:46 PM

SadieB: What you say is not correct. Mean is where you sum up all the numbers and divide them by however many samples you have. Median is the number such that 50% earn more than that and 50% earn less. Bill Gates paradox applies to mean but not median.

Posted by: Adrian | Feb 12, 2005 11:13:31 PM

Since the '70s the supply of labor has increased due to three factors - population growth, immigration and the entry of large numbers of women into the workforce. Immigration and population growth are nothing new - we've had large waves of immigrants before, and '73 was part of the way up the peak of the baby boom, so those two factors probably tracked demand for labor, but the women are new.

When the supply of a thing increases faster than the demand for that thing, the price of that thing (in this case, the wage for labor) decreases. Or, in this case, the increased supply of labor from women has more or less exactly offset the increased demand for labor due to productivity growth and thus wages have been stable. This has caused an increase in household income for two-earner households. The relatively slow growth in household income relative to productivity is driven by the increase in the number of single-parent households.

At the end of the '90s the increase in women's workforce participation began to stall - women and men are more or less participating equally, so there is no obvious pool of new women to enter the workforce, and so wages started back up. Unfortunately, the slowdown of 2000-2002 abbreviated that. As the economy recovers, wages should grow from here. Household income will increase with wages so long as adult household populations stablize.

Posted by: rvman | Feb 13, 2005 12:16:29 AM

I can understand why we're going to be in trouble in the near future (ten to twenty years).

Many of you, sadly, are not taking your thinking far enough. Yes, you understand theories and some data, but little more. Sorry...but it's my take.

The issue isn't wages compared to prices.

The issue is household income compared to household expenditures. Net result is household savings, which is not growing. Zip. And that's THE issue. Ask Greenspan or read his speeches and testimony. It's all there.

The wage discussion falls on its face once you factor in health insurance, and other major drivers of household expenses. Include declining support from some employers for 401(k)s, loss of DB pension plans, and elimination of shared costing of health benefit plans. All three of these issues are on either on the table for negotiation or have been eliminated by many employers. The trend is clear. And we're not going back to the "old" structures and relationships with employers. As well, such perks will be included in income determination by the IRS at some point; already under consideration if not approved recently. Result: more taxes, less household savings.

Moreover, Greenspan's recent remarks at the London conference were worth noting, though somewhat misleading. Yes, we're going to watch import prices rise at a faster rate eventually. But we will not experience an immediate gain in household savings, as he suggested. There will be a savings lag due to slow adjustment against such purchases. Household savings will likely decline for a while, and then households will attempt to reduce debt levels back to where they were when such price began to increase at a faster rate. Thereafter, we may see an increase in household savings if and probably only if other domestic prices do not rise along with some imports prices (a tag along effect).

The household expenditures model does not follow the CPI down the line, so there is a further deviation to consider. As I recall, the CPI formula was adjusted not long ago, and I don't believe that it is a true measurement of average household expenditure inflation (if I may be so bold as to use that word).

It would be foolish to assume that household savings will rise significantly within the next few years.

So what if wages go up moderately? Prices, if including import goods, are expected to rise on a comparable scale or perhaps greater. Household expenses will hammer most wage and income increases.

The issue is household savings. And that growth measurement is dead in the water.

Posted by: Movie Guy | Feb 13, 2005 1:49:33 AM

People are using these terms to switch from one concept to another without thinking about the differences -- the terms being tossed about are not the same.

When we talk about wages we generally use the BLS data for average hourly earnings-- what an individual hourly wage employee makes. Hourly employees are about 80% of total employment and that share has not changed significantly.

When you talk about income, the meaning changes significantly, especially if you talk about family income. One, you have to add salaried employees in to the calculation and those wages are higher. Second, you need to add "unearned" income from capital and income from family owned businesses. Third, you also need to include transfer payments. Fourth, if you are talking about family income you need to include the point that families now generally have two or more workers when that was uncommon 40 or 50 years ago.

The CPI is a very good sample of how average hourly wage earners spend their income. Moreover, the spending package is reestimated every decade so the weights in the CPI are redone every decade to adjust for these changes.

The BLS also does an survery of how other family groups spend -- for example the elderly -- so
information on this type of info is available.

Posted by: spencer | Feb 13, 2005 9:44:19 AM


Suppose that in a given year, the wages of every single person in the U.S. remain fixed, except for one person: the absolute top earner in the U.S. sees his income go from $1 billion to $100 billion. Prices go up 2% due to inflation.

At the end of the year, here's some truths:

Median wages are precisely identical to last year.
Average wages are way up.
99.99999% of the population is poorer than they were last year: they have the same amount of money to buy things that are 2% more expensive.
That one guy is a lot richer - he has 100x the money to buy things that are 2% more expensive.

Now, if you're a Republican, promoting policies for rich people, you report that the average American is making more money (which is true, in the same way that when Bill Gates walks into a bar, the average net worth of people in the bar goes way up).

If you're a non-Republican, promoting policies for average people, you report that 99.99999% of Americans are worse off than they were last year.

Both of those above statements about how Americans are doing are true. It may be that one of them is a lot more informative about how the majority of Americans are doing, and one of them is pretty deceptive about the situation, but both of them are true. If more people understood the difference between average and median (including Mr. Yglesias and some of the commenters in this thread...) perhaps this deception would be harder to pull off.

(Family income is an apples-to-oranges comparison. Two-earner families are on the rise, at least partly because families can't make ends meet without both adults working, because the wages of the one earner aren't keeping pace with price increases!)

Posted by: Anon | Feb 13, 2005 11:48:00 AM

A big issue is that old people have a different consumption pattern than others, most notably, much larger proportion of their expenses relates to medicine, be it drugs, extra insurance premiums, co-payments, supplies (say, wheelchairs), home care, etc. Medical inflation is MUCH faster than the other components of GNP or home budgets.

We have a deflation in electronic goods, and old people proportionally consume much less of them.

Actually, medical inflation is a terrible problem and a true crisis, we are already in crisis, and solutions offered by Republicans make the problem worse.

Posted by: piotr | Feb 13, 2005 1:05:42 PM

Let me try again.

The mean (or median) is where you take all your numbers, add them up, and divide by your sample size. No argument so far?

But the number I'm talking about, the one that can really tell you what is "typical" (ie the reason you are doing averages in the first place, right?) is the mode. It's the number that occurs most often. Here's an example: 1,1,1,5. The mean of these numbers is 8 / 4 = 2. But the mode is 1. The mean is twice the mode.

Whe you're talking about income, mode is what you want because if 75% of the population is making $10,000/year, and 25% is making $500,000/year, (not realistic, I'm just trying to make it an extension of the 1,1,1,5 example) you're going to get a median income of $135,000/year. That is not an accurate description of the majority of your sample. But it is the statistical average. Now you see why I'm calling the median a "Republican" number?

Posted by: SadieB. | Feb 13, 2005 5:02:43 PM

Sorry - I take back what I said about median. You're right, it is the value in the middle. I should have looked it up before posting. But I stand by what I said about the mode! If I weren't old enough to have had some good old-school lefty economics professors, I would never have heard of it either.

Posted by: SadieB. | Feb 13, 2005 5:07:41 PM

I know that wages and incomes are not the same measurement.

I recommended looking at household income and expenses because the standard reference to wages and prices leaves too much off of the table.

As an example, recall the most recent unemployment data. The news media reported the U3 total unemployed number of 5.2%, and, in general, failed to discuss the broader merits of U6 unemployment. The larger number, 9.3%, includes workers who are not receiving full health benefits (if any) or pension qualifications because their work schedules are being shuffled down to 38 hours or less per week. This is a common practice in some industries, on a week by week basis. So, just comparing the wages to prices doesn't offer much of a valid comparison to the costs that workers must absorb.

Source: Alternative measures of labor underutilization

While it's true that household income includes income other than wages, the bulk of such income for workers earning say $40K-50K or less is from wages as relates to monthly revenue streams. Sure, such individuals and families probably have money invested in 401(k)s and other investments, but the likelihood of any draw down is small on a monthly basis with exception of retirees. Just my opinion, of course.

The Bureau of Labor Statistics data, including the CPI, is good information. I question the CPI data in small part because it is distorted due to the manner in which it is weighted. First, any reported numbers that exclude food and fuel offer no reasonable measurement of household expenses. Second, the manner in which housing costs are calculated doesn't represent the full picture. Third, the latest change to one form of CPI measurement calls for substitution of similar goods; as an example, the price of steak (beef) rises, so chicken and/or pork are substituted. This is called the chained index (C-CPI). So, it pushes the consumer price index down, and assumes or implies that you are no longer buying steak. If you do, you don't fit in the model and your price increases for steak are not captured. Lastly, the core CPI is inadequate as a measurement of household expenditures.


Bloomberg, 27 Jan 05

"Doug Lee, president of Economics From Washington, an economic consulting firm based in Potomac, Maryland, has identified some "disturbing underlying patterns" in the consumer price index, which rose 3.3 percent in 2004, compared with a 1.9 percent rise in 2003. The core CPI, which excludes food and energy, rose 2.2 percent last year compared with a 1.1 percent increase in the preceding year."

"To what do we owe the tame 2 percent annualized increase in the core CPI in the fourth quarter? Why, our old friend, owners equivalent rent: the largest single component of the CPI -- 23.4 percent of the CPI and almost 29.8 percent of the core index -- and one that imputes a rental value to owner-occupied homes based on a survey of rental units."

"While the CPI's "top line isn't doing anything exciting, the truth is everything but housing is going up," Lee said.

Other implications of the Chained CPI - reduced government expenditures


If a person simply analyzes wages as compared to prices, any "cost" conclusions could be fairly inaccurate. This is why I recommended that the focus should be on household income, expenditures, and savings.

An opinion about future prices? Here's one...

"U.S. economic growth will slow this year while core inflation picks up, prompting more interest rate hikes by the Federal Reserve, top forecasters said in a survey released Thursday. It predicted the core CPI, a less volatile measure of inflation, will increase 2.3 percent this year and 2.4 percent next year, up sharply from 1.8 percent in 2004. While the consensus believes inflation, as measured by the Consumer Price Index, will ease slightly in 2005 to a 2.5 percent increase from last year's 2.7 percent rise, forecasters believe prices excluding food and energy will continue to creep higher."


Do I believe that average hourly wages will stay ahead of prices as we proceed into this cycle? No.

Posted by: Movie Guy | Feb 13, 2005 8:01:02 PM

The comments to this entry are closed.