Median-itis and The Great Stagnation

Here is my NYT column from today, on themes relevant to The Great Stagnation.  I won't rehash this entire discussion, but I would like to focus on this one column excerpt:

From 1947 to 1973 – a period of just 26 years – inflation-adjusted median income in the United States more than doubled. But in the 31 years from 1973 to 2004, it rose only 22 percent. And, over the last decade, it actually declined.

I am noticing that some reviews or commentaries (and here) are citing per capita income growth as a response to my argument.  It is true that per capita income grows at a slower rate post-1973, but my argument is about the slowing down of median income growth and that is a much stronger shift.  The productivity data also tell a glum story.   

CPI bias can change those numbers in absolute terms (see comments from Russ Roberts), but it also changes the pre-1973 median income growth numbers and arguably more so.  The gap remains and TGS refers to the living standard for the average person or household in the United States, not the total amount of innovation, which remains quite high.  They're just not innovations with the same trickle-down or broad-based effects as in an earlier era.

Kindle eBooks are themselves a good example.  It's a real improvement for a lot of us — especially travelers – but even the median reader, much less the median American, doesn't have a Kindle or buy eBooks.  As I argued in The Age of the Infovore, the big gains of late have gone to the extreme information-processors.  

I've seen in the MR comments (and elsewhere) a lot of anecdotal comparison of recent gains vs. earlier gains in technology.  Don't we now have this, don't we now have that, and so on.  Of course.  Median incomes have risen somewhat.  But, when it comes to the average household, the published numbers for median income are adding up and trying to measure those gains and it turns out their recent rate of growth really has declined.  Most serious researchers who work in this area use and accept these numbers as the best available (though they do not in general advocate my causal interpretation; see for instance Mark Thoma or Jacob Hacker).  

If the numbers for median income growth are low we ought to take that seriously, as does Scott Sumner.  We are not cheerleaders per se (BC: "I'm baffled why Tyler would focus on slight declines in American growth when the world just had the best decade ever."  Is it then wrong to focus on any other problems at all?  I also was one of the first people to make the "best decade ever" argument, which I still accept.)  Medians also matter for the political climate, even though the median earner is not exactly the median voter.  Adam Smith's welfare economics was basically that of the median, a point which David Levy has made repeatedly.

I'm also being called a "pessimist" a lot.  Yet in my view our current technological plateau won't last forever.  That's probably more optimistic than the Hacker-Pierson approach, which requires a Progressive revolution in economic policy (unlikely), although it is not more optimistic than denying the relevance of the numbers.

I'll soon blog some remarks on changing household size as another attempt to avoid confronting the facts about slow median income growth.


Surely it is possible for median income to stay flat, while median the median life improves (or is unchanged). There are surpluses that don't enter into median calculations.

Of course surpluses have accrued to extreme information processors first - it's in the nature of the respective beasts.

However, if there is one thing that we've learnt about technology driven surpluses it is that they rapidly make it down to everybody, thanks to that nifty low cost to scale. The IRC of the information processors don't have to change much to become the Facebook of the masses.

Even a technological plateau that has us frozen at this technological movement leaves plenty of space for the technological frontier to move (albeit at slower rates).

Well said, Tyler.

However, let's try framing this question in a different way: the shape of the class structure has been changing away from equality.

In 1945-1973, there were a lot of people in the middle, and the top wasn't all that far away from the median. Today, the shape of the class structure is more stretched out. The top is almost out of sight and there isn't a huge bulge around the median.

"inflation-adjusted median income": I take it that that refers to median individual income, not household? There must be some lower age limit, I assume? 18? 22?

Median household income may be stagnant but the income of a median household need not be. It is vital to remember that these are snapshot figures and do not track households over time. Experts who use these figures may understand that distinction, but the public, and most non-expert commentators, do not.

When Tyler says things like "The gap remains and TGS refers to the living standard for the average person or household in the United States..." he is blurring that line. It's an irresponsible use of words.

You don't mention whether you are speaking of median household income or median person income. If you're talking about the median household, is the change due to change in household composition, e.g. rise of single adult families, whether single person or single mom/dad.

Good issue, but I know just enough to worry about the numbers. How do we account for divorce? If a couple divorces and produces two poorer households, how is that counted? How do we account for changes in the cost of living? CPIs are notoriously poor for this purpose -- think about the cost of housing as a start. If this is one year's income, has the permanence of income fluctuations changed? Etc. I think the first thing we need to do here is take a close look at the evidence, which in my experience is trickier than it sounds.

Tyler, you say focus on the numbers. Well then, let's focus on the numbers. The REAL numbers that have meaning. Let's add in non-cash compensation. Let's account for immigration and household composition. Then we can have a discussion. Without that being included, we have nothing.

Laughable to say non-cash compensation doesn't count because it isn't worth what it's buying. Isn't that like saying, let's take all the income spent on entertainment out of the equation, because that isn't worth what is spent on it. Preferences are preferences.

I am reading the Great Stagnation now. It is a persuasive argument. But how does it jive or not with the argument in Myths of the Rich and Poor that we are working for less time to buy what we need even as median incomes have not risen much?

I have been perplexed by this since I read the book.

"From 1947 to 1973 — a period of just 26 years — inflation-adjusted median income in the United States more than doubled. But in the 31 years from 1973 to 2004, it rose only 22 percent. And, over the last decade, it actually declined."

For the vast majority of 1947 to 1973 there were two very important things making this happen. 1, the gold standard, with almost no inflation, any gains made by employees was theirs to keep. 2, increase in the workforce as a percentage of people. When 70% of the population makes no money at the start and at the end of the time period it becomes only 40%, then the shift of median moves further and further away from 0 baseline.

From 1973 until 2004 the entire time has been a period of inflation at the behest of the federal government. Thus, every increase in wages won by the employee was degraded by the devaluation of the currency and the high cost of all the environmental regulations imposed on the people who pay the wages.

As for the last decade, why exactly are we overlapping 2002 through 2004 with 2001 through 2011? Why not stick to the 2004 through 2011 time frame, or shorten the earlier period to cover until 2001? The reason for the decline has several reasons to it. Workforce participation levels have been declining making the number of 0's larger. The government has been aggressively devaluing the currency with inflation, perhaps to lower the debt burden. An ever increasing burden of regulation costs which competes with the money available to pay employees. The extreme globalization of the economy where we are competing more and more with third world nation wages and regulations which prevents companies from being capable of paying employees more or to be priced out of the market.

One of the real returns to technological change is more leisure time, the value of which is not measured directly in the GDP figures. That leisure time can take the form of shorter work week, longer vacations, later entrance into the work place, earlier retirement, and even consuming more "leisure" on the job, i.e. treating your work as fun.

I'm not sure how to do it, or if it's been done, but if the annual value of leisure time consumed is added to a time series on GCP, how would the rates of growth change, if at all?

To paraphrase a famous quote: "The greatest benefit from technological change is the easy life".

I have not read the book yet. But does Tyler consider the possibility that the period 1947-1973 is the historical aberration, rather than 1973-present? Looks of weird things going on in this period, e.g., little immigration:

Like andy weintraub upthread leisure time also struck me as a factor. See, Measuring Trends in Leisure: The Allocation of Time over Five Decades by Mark Aguiar and Erik Hurst

From the abstract: In this paper, we use five decades of time‐use surveys to document trends in the allocation of time. We document that a dramatic increase in leisure time lies behind the relatively stable number of market hours worked (per working‐age adult) between 1965 and 2003.... This increase in leisure corresponds to roughly an additional 5 to 10 weeks of vacation per year, assuming a 40‐hour work week. We also find that leisure increased during the last 40 years for a number of sub‐samples of the population, with less‐educated adults experiencing the largest increases....

Increases in leisure are just a symptom of market satiation and the fact there are not enough new and valuable creations to justify working as long. They support Tyler's position.

Getting copies of The Great Stagnation in other formats.

I have a request. I have several people who do not own Kindles. Is their some way to get a PDF copy of the document? I am happy to buy the copies but I need something that can be printed.

Other publishers do this (, though not many. Please let me know.

Eric Grossman

How much did you pay for it Thomas?

Several ideas:
1) In a talk by Kevin Murphy on wage inequality, he discussed how real wage growth has been much higher at higher wage levels. The premiums for an undergraduate and graduate degree have risen. Murphy's interpretation was:
a) Demand for high skills has increased.
b) In response, more people were going to college to attain high skills.
c) BUT the number of college graduates has NOT followed the number of college attendees. People are getting to college and then failing. Hypothesis: the marginal student may have an inferior K-12 education.

2) If inequality is driven by productivity growth among highly educated (i.e. higher returns to education), then is the growth in inequality a bad thing?

3) I recall a paper several years ago that argued the price of consumption goods purchased by low income individuals has risen at a slower pace than the price of consumptions goods purchased by high income individuals. The poor have experienced lower inflation than the rich. If this is true, the CPI can bias things in RELATIVE terms.

And the usual:
4) How much has the growth in the cost of health care taken out of income growth?

5) If this is based on household data, to what extent has average household size been declining? (you allude to this...)


I've become frustrated with claims of a decade without economic growth. Are we to believe that the internet, smart phones, and autotuned YouTube videos aren't economic growth?

Lord says: "Increases in leisure are just a symptom of market satiation and the fact there are not enough new and valuable creations to justify working as long. They support Tyler's position."

There may not be enough new and valuable creations at an attractive price. So, since leisure is relatively cheap, we "buy" more of it. But since there is a trade off between leisure on the one hand, and goods and services on the other, shouldn't increases in leisure production still be valued and included in some measure of production, i.e. GDP? And if they were, would the recent GDP growth rate be higher?

Here's the flaw I can't get past, though. You can make the same argument for going from the 1950's to the 1970's. Our living standard may have doubled, but we had pretty much the same things in 1973 that we had in 1953. Cars, refrigerators, dishwashing machines, etc. were also around 20 years earlier, yet living standards doubled. In fact, I have a hard time thinking of disruptive innovations during that period.

Cowen (and Scott Sumner) would probably argue that this period was the "tail end" of the first wave of industrialization. In other words, television, cars, refrigerators, dishwashing machines, etc were discovered years before, but it wasn't until this period that they were fully implemented in all their lifeway-changing ways.

Compare that with the Information Revolution of the past three decades*. Much of the tech underpinning it was discovered earlier, and most of it is implemented now. Yet aside from a brief period from 1994-2000, we didn't get the massive rises in median income that we had in the implementation period in the 1950s.

*I say "three decades" because the Information Revolution is not just the rise of the Internet. There were a whole ton of computer innovations that really came into their own in the 1980s, ranging from transistors to miniaturized computers and so forth.

One critique is over shifting purchasing power. For example Louis the 14th could not buy an iPhone with all his resources and I can, therefore I am richer than Louis the 14th because I have greater purchasing power. Of course that ignores the reverse, that Louis the 14th could buy Versailles and an army of peasants and I cannot. Or that Thomas Jefferson could buy and sell people and I cannot. Does that mean Thomas Jefferson and Louis the 14th had greater purchasing power than any American today? So you should be careful when making comparisons between decades and looking at purchasing power. For the peanut gallery, consumption ≠ income. Assets ≠ income. Wages = income.

I would also challenge people like andy weintraub to produce evidence of this growing leisure time during the great stagnation. All the evidence I've read says the opposite; people are working more now than ever.

The median household has gone from picking the fruit to being the fruit being picked.

In the 1960s, a house with a yard in the San Fernando Valley with a satisfactory public school nearby was affordable for the average white married couple with three kids _with just the husband working_.

The downsides: 1. Smog was becoming much worse. 2. Crime was about the same in the SFV as today, but on the way up instead of the way down.

The big difference I see for people in the middle of society is that respectable middle class family formation is much more stressful today. The tradeoffs between housing costs, commute time, schools, childcare and other investments in children are much more stressful. Having an iPod is nice, but it doesn't do much for making family formation easier.

Obviously, the postwar San Fernando Valley where I grew up was a Golden Age of affordable family formation. Still, it's worthwhile to focus on the biggest things in life, such as being able to afford to get married, have children, and educate them well, rather than get distracted by the current wide availability of Twitter.

Can't the stagnation of median income be explained away by a two- or three-decade error in valuing American labor relative to the rest of the world? (I'm trying not to use the word "bubble"...)

If that were the case, couldn't you explain media wage stagnation simply as a reversion to the mean? That would also indicate that things would go back to normal once our deflation/stagnation met the ROW's inflation. Of course, by that time a lot of goods and services may be so cheap that we'll start to see demand saturation, which ought to cause everybody to stagnate again.

Is median income controlled for age?

If you have an aging population (less than 58 during this period) median income would increase--you earn more at 55 than you did at 45. Did you control for age?

A coupl'a people upthread twigged it: The median family or household then is NOT the median family or household now. A time-series of median incomes is meaningless: " is a tale told by an idiot, full of sound and fury, signifying nothing."

All I have to say, is what I have been saying.....and have been wondering about, or worried about; in 1979 I made over $8 an hour at a summer job at a local plant and over $12 an hour with overtime.

Granted, times were different....inflation, etc..and this was a Union plant, since moved South.

But, 30 years later....a Counselor with a Master's degree is paid from $12 plus an hour.....millions of workers hope to find a $12 an hour job, considered a good wage in today's work environment.....30 years later.

Here is my two cents worth as a non-US resident (although I lived in and visit the US regularly);

When I compare (not in a scientific way, just my impression the US lifestyle with other developed countries such as Europe, Australasia etc) it doesn't seem as far ahead as it used to say 30 years ago even though on a pure PP basis the gap remains the same. The US average lifestyle used to seem an impossible "film star" level to those of us in the rest of the developed world, but not any more. So Tyler's point that if median income had grown at previous rates it would now be 50% higher ($90k versus $60K) resonates with me. But my view is that the US consumer seems less far ahead only because manufactured goods now are so much cheaper now, due to partly innovation, partly scale, partly offshoring of the manpower intensive bits, probably mostly experience and know-how. Basically the extra income the US consumer gets versus other developed nations has diminishing returns in terms of quality and quantity, especially when compared with the past. So in my view this supports those who claim that PPI is understating growth in incomes, the medium income now buys so much more stuff which hasn't been reflected accurately in PPI hedonic adjustment.

Even if I accept that median incomes (whether family or individual) have stagnated over the past 30 years or so, I don't think this is because of reduced rate of innovation, like other respondents I have a hard time believing you can quantify this or it being true. I believe the reason is that the economy is much more service oriented now. The share of manufacturing in US GDP has fallen from 28% in 1953 to 11% now and services have correspondingly risen to replace it. As a result the median individual is now much more likely to be working in services than in manufacturing. Service industries are more fragmented and less susceptible to rent extraction by workers than capital intensive manufacturing industries (compare trying to organize the workers in a fast food restaurant with those in a steel mill) meaning service industries can pay their workers closer to their marginal value. This is why in my view the median worker has seen their incomes stagnate, or not grow so quickly, in other words it is due to the job growth over the past thirty years being mostly in service industries.

For example, my higher education was ridiculously cheap by current standards. Full tuition at Rice U. in 1976-77 was $2,300. I don't even remember what the fees were at UCLA for my MBA in 1980-82 they were so minimal.

Is it possible to buy your new e-book in hard or soft cover to give to someone who does not have a computer but who would like to read it?


John C.Gardner

What about healthcare benefits? Since 1973, there has been a DRAMATIC shift towards healthcare benefits instead of wages. Could this all be just a mistake to include healthcare benefits?

Chris T, I am really looking forward to the future in 10 years. People on welfare now have PS3s and big screens, in a few years, a lot of us will have little TVs in the tub. Or maybe I just envy Diddy too much lol.

I'd love to hear what Paul Romer has to say about all of this.

Comments for this post are closed