Jon’s post on Big-time college sports draws on work by Robert Frank, who treats high performance in college sports as a positional good.
By an interesting coincidence, Frank gave a seminar here in Brisbane on Friday and stayed for a very interesting chat afterwards. He argued that the growth in inequality in the US has been positively harmful to the middle class, even though their income has been roughly stationary since the 1970s.
One argument is that expenditure on positional goods by the top quintile has negative externalities such as the need to buy larger cars to protect yourself against SUVs and the need to buy more expensive clothes to appear decently-dressed in various contexts, such as job interviews. . Another has to do with informational cascades creating higher aspirations for consumption. The one-line message is “relative income matters”. Frank sees this as a big factor in declining savings rates, increasing household indebtedness and (I would infer) growth in the current account deficit.
Although I agree with a lot of this, I think another version of relative income is more significant in the case of the US. Incomes rose rapidly and across the board in the postwar period, and this followed a lot of equalisation during WWII and the New Deal. By contrast, since 1970, they’ve been flat for middle income households and have actually declined for the bottom 20 per cent of households[1]. This means that when people compare their current consumption to their own past peak levels or to their parents’ they may well find that it has declined.
This is even more likely to be true at a disaggregated level. Lots of new goods like computers have come on the market since 1970. While more variety is welfare-improving, if budgets are fixed, expenditure on new items must be financed by a reduction in consumption of old items. If people aspire to avoid such reductions, they must increase consumption expenditure.
As Frank observed in discussion, there is nothing in this that would have seemed surprising to economists in the 1950s and 1960s when Duesenberry’s work on previous-peak consumption models was influential. But this model has been almost entirely displaced by Friedman’s permanent income model and its successor the life-cycle model, even though the empirical performance of these models is not as good as Duesenberry’s. Clearly there are strong ideological/methodological preferences at work here.
fn1. Gregg Easterbrook ran hard with the idea that the bottom 20 per cent is made up largely of recent immigrants, so that native-born households have had rising incomes. But as I recall this idea didn’t stand up to scrutiny.
{ 24 comments }
cure 09.19.04 at 6:29 pm
What was the objection to Easterbrook’s argument? I’d be willing to accept flaws in his argument, but not without some data.
As far as household data are concerned, the relatively stationary position of household income has as much to do with the declining size of households since 1970 (because of young adults moving out earlier, divorce, etc.) as anything, right? It would be quite surprising if household size dropped from 3.5 to 2.5 and household income did not stagnate(these numbers are inexact though the point remains).
Further, though I too find SUVs rather ridiculous, I find it more ridiculous to claim that their presence on the roads causes the poor to buy more expensive cars. If one’s car purchase is based on protecting oneself from SUVs, why not protect oneself from moving vans and semis? Or minivans, which despite their size seem to create little of the anti-SUV animosity? And though us economists say things like “informational cascades creating higher aspirations for consumption,” do we not mean that people simply are spending beyond their means?
I propose a quick test: Find the correlation between savings rate and Gini coefficients in nations worldwide, or simply among the different US states. I don’t think you’ll find that strong of a correlation.
Steve Carr 09.19.04 at 6:45 pm
John, with all due respect, the idea that an average middle-class householder could reasonably feel that their current consumption has declined relative to their parents is utterly implausible. In the first place, you can look at any survey of what the average middle-class house had in 1970 and what it has today, and there is simply no comparison. We have much more stuff today, and a much wider variety of stuff.
Just (or even more) important, products today are of far higher quality. Compare the food that was available at my local supermarket when I was a kid in the 1970s, versus the food that you can find at any Stop n Shop (let alone a Whole Foods) today, or a Dodge Dart to a Honda Civic, or televisions, or kitchen appliances, and in every case a middle-class householder today is vastly better off than his or her parents were.
Then there are the macroindicators: life expectancy is up by seven or eight years, a significantly higher percentage of people have graduated from college and an even higher percentage have attended some college, crime has now been stagnant or falling for almost a decade, the Internet is ubiquitous (giving average Americans access to infinitely more information than their parents could have dreamed of), homes are bigger, etc.
This doesn’t mean that things are easy for middle-class people — although I think the main source of difficulty is for parents, because of the rising cost of education. But the notion that we live less well is simply wrong.
The one caveat here would be that we work longer hours — if not individually, then collectively with the arrival of women in the workforce. But it’s far from clear that this is entirely (or even largely) a bad thing, unless you can confidently declare that we were better off having a male-dominated workforce. I also think the equation of work=pain is perhaps a bit facile.
Steve Carr 09.19.04 at 6:54 pm
Also, while Easterbrook completely misstated the point about immigration — which draws on the work of Gary Burtless — his conclusion is actually relevant to your claim here and has not been discredited. Easterbrook said that if you factored out immigration, there was no rise in inequality since the 1970s. This is false. But in this post, you’re not talking about inequality. You’re talking about whether middle-class people of today are worse off relative to middle-class people of thirty years ago. And here the impact of immigration is absolutely relevant in looking at the statistics. If you factor out immigration (and factor in declining size of households), the stagnation in American median income disappears.
There’s no doubt that middle-class living standards did not rise as fast between the mid-1970s and 2004 as they did between 1945 and 1970. But I think the numbers suggest that, controlling for immigration, native-born middle-class Americans live much better than their parents did.
bob mcmanus 09.19.04 at 7:52 pm
But,but this would say bad thing about Ronald Reagan! Impossible! He saved us all, and helped us all.
Must be wrong.
I would have only anecdotal evidence at this point, and a question derived from watching the life of my grandparents, dead these ten years, as to the intangible value of a secure job for a lifetime, a trust in a secure retirement and no fear of loss of medical care.
John Quiggin 09.19.04 at 9:56 pm
“I propose a quick test: Find the correlation between savings rate and Gini coefficients in nations worldwide, or simply among the different US states. I don’t think you’ll find that strong of a correlation.”
Frank mentioned the same test and claimed that the correlation was strong – more inequality means less savings.
Actually, I think the argument ought to apply only to median household savings, expressed as a proportion of income. With high inequality you could have a lot of saving at the top.
Walt Pohl 09.19.04 at 10:36 pm
Evidence in favor of your comment about Duesenberry: I just recently sat in on a graduate macro class, and the professor covered the permanent income and life-cycle models, but not Duesenberry’s model. (What is the model, anyway?)
John Quiggin 09.20.04 at 12:30 am
There’s a summary of Duesenberry and other models here
John Quiggin 09.20.04 at 1:00 am
Steve, price indexes are adjusted for quality, so these benefits are factored in. (As I note, benefits from variety are a different and more complex story). And you need to have a consistent time period – the decrease in crime over the last decade only takes it back to rates prevailing in the 1970s.
You get much the same story if you look at wages as at household incomes. Declining household size has been offset by increased hours of work per person.
Steve Carr 09.20.04 at 3:37 am
John, the CPI’s adjustments for quality have historically been very weak. Only eleven percent of the items in the sample are adjusted at all for quality in a given year (www.bus.umich.edu/KresgeLibrary/ Collections/Workingpapers/wdi/wp681.pdf), which given the pace of change in most products seems implausible. More importantly, until very recently, for instance, the index has only used hedonic adjustments for PCs, televisions, and apparel, adding VCRs/DVDs, stereos, and some appliances in 1999/2000. The Boskin Commission’s estimate — which suggested that the CPI overstated inflation by 0.6% a year because of unearned quality — may be high (although I think most neutral observers see it as reasonable), but I think there’s little doubt that, if you account for quality, middle-class incomes and lifestyles have improved significantly since 1970.
As for the consistent time period: life expectancy at birth has risen by seven years since 1970, and life expectancy at age 65 has risen by three years (20%), which means that a middle-class person can expect his children and his parents to live a lot longer than someone in 1970 could (and that he himself will live longer, too): http://www.cdc.gov/nchs/data/hus/tables/2003/03hus027.pdf. As for education, in 1970, 11 percent of the population had finished four years or more of college. Today, it’s 25 percent. 7.4 million people were enrolled in college in 1970, today — despite the decline in the size of the college-age population — it’s around 13 million. Healthier and better-educated certainly should count for something.
I also don’t understand your point about the increase in variety and fixed budgets. All budgets are fixed somehow. And it’s true that if I buy X, then it may mean I can no longer buy Y. But if someone today can choose to buy X or Y, while in 1970 he had to buy Y, that is unequivocally welfare-enhancing.
John Quiggin 09.20.04 at 4:00 am
‘I also don’t understand your point about the increase in variety and fixed budgets. All budgets are fixed somehow. And it’s true that if I buy X, then it may mean I can no longer buy Y. But if someone today can choose to buy X or Y, while in 1970 he had to buy Y, that is unequivocally welfare-enhancing.”
I agree. My point is about demand implications not about welfare. If, as a result of the availability of Y, a person must choose either to consume less of X or to save less, a peak-consumption/habituation model implies that they may decide to save less.
John Quiggin 09.20.04 at 4:04 am
An important issue is the stock-flow distinction. Because incomes rose steadily from 1940 to 1970, someone who has worked from 1970 to now has a higher lifetime income than someone who worked from 1940 to 1970, even if their income is still at the 1970 level.
Steve Carr 09.20.04 at 4:10 am
Oh, okay. On reflection, the post confused me a bit because I wasn’t sure what the phrase “more significant in the case of the US” was referring to: the macroreality of declining savings and higher current-account deficit, or the question of whether middle-class welfare has actually declined since 1970.
Ted 09.20.04 at 4:12 am
“[…] what the average middle-class house had in 1970 and what it has today, and there is simply no comparison. We have much more stuff today, and a much wider variety of stuff.”
But the difference is technology, not economics. There have been some amazing improvements in materials, and electronics, and manufacturing processes that have made better quality manufactured goods less expensive. To some extent that’s the result of investment of capital, but to a greater extent it’s the result of intellectual efforts of engineers rather than of political economists.
I think it would be interesting to speculate about how much better off we’d be now compared to 1970 if technology were held constant — if we could build more or fewer of the same sorts of factories, but not new kinds of factories.
Steve Carr 09.20.04 at 4:41 am
Ted, you can’t divorce technology from economics. There’s a reason the Soviet Union demonstrated essentially zero technological innovation over the course of its history when it came to goods that actually improved the lives of the average Soviet citizen.
Carlos 09.20.04 at 4:24 pm
Ted, why can’t we divorce technology from economics? We know that economic incentives improve the rate of technological change but surely other factors matter. After all implying that technical progress is completely dependent on the economic infrastructure is a bit too marxist, isn’t it? The economy can provide more incentive to make better consumer products, but unless scientific research provides the means to actually make them, nothing will happen.
Carlos 09.20.04 at 4:34 pm
Sorry, I meant to say Steve.
msw 09.20.04 at 11:38 pm
“What was the objection to Easterbrook’s argument?”
My own objection is that there’s no sharp, out of trend dip in median income in 1986 as a result of the amnesty for illegal immigrants.
“As far as household data are concerned, the relatively stationary position of household income has as much to do with the declining size of households since 1970 (because of young adults moving out earlier, divorce, etc.) as anything, right? It would be quite surprising if household size dropped from 3.5 to 2.5 and household income did not stagnate(these numbers are inexact though the point remains).”
Go with the median individual wage, then. It peaked in 1973. And why would the decreasing size of the household affect its income? It’s the number of wage earners that counts, not the size of the household. I’m not sure where Prof Quiggin is getting the idea that household income has been flat since the 1970s – according to the census bureau it’s up since 1970. (http://www.census.gov/prod/2002pubs/p60-218.pdf) Since the median wage is slightly down, this presumably represents the entry of women into the workplace.
I’m suspicious of hedonic pricing changes. It seems to me that there are plenty of examples of technology replacing a high quality product with a lower quality, but cheaper one – think Pergo floors and high-fructose corn syrup. And medical technology has advanced, but it seems to me that the diagnosis of a qualified medical professional is at least as significant a part of your medical dollar as the pill you get. When was the last time an American got to talk to a MD for more than 10 minutes? Are there anhedonic pricing adjustments for talking to a Nurse Practitioner (if you’re lucky) instead?
Cars are functionally identical to the cars of the 1970s. The shinier finish and superior cup-holders seems like a great example of a positional good to me.
Jason McCullough 09.20.04 at 11:41 pm
“You’re talking about whether middle-class people of today are worse off relative to middle-class people of thirty years ago. And here the impact of immigration is absolutely relevant in looking at the statistics. If you factor out immigration (and factor in declining size of households), the stagnation in American median income disappears.”
What’s the theory here? That immigrants count at one end of the time period, but not at the other?
Jason McCullough 09.20.04 at 11:43 pm
“You’re talking about whether middle-class people of today are worse off relative to middle-class people of thirty years ago. And here the impact of immigration is absolutely relevant in looking at the statistics. If you factor out immigration (and factor in declining size of households), the stagnation in American median income disappears.”
What’s the theory here? Wnhy should immigrants count at one end of the time period, but not at the other?
Also don’t see working hours included in the income numbers…..
msw 09.21.04 at 1:10 am
“What’s the theory here? Wnhy should immigrants count at one end of the time period, but not at the other?”
Because the numbers surged in the 1970s. Scott Martens has some great graphs over at pedantry (http://pedantry.fistfulofeuros.net/archives/000472.html)
But of course, there are countervailing demographic trends. The seventies is when the baby boomers first entered the job market as low-skill entry level workers. They’re now at the peak of their earning power. I’d be interesting to check the relative size of the boomer generation vs immigration.
msw 09.21.04 at 1:21 am
“Then there are the macroindicators: […] a significantly higher percentage of people have graduated from college and an even higher percentage have attended some college”
I would think that this is a great example of Frank’s theory. In my mind, there are three reasons to go to college:
1. Because it’s inherently worthwhile to the student. Life of the Mind, a four year vacation, easy access to sex and drugs, etc
2. Because it’s an unwelcome but increasingly necessary credentialing process.
3. Because you’ll feel like a loser if you don’t.
Given that the costs of college have increased significantly, and the quality of a college education presumably hasn’t (I’m pretty sure that it’s a less reliable source of sex and drugs than it was 30 years ago), we would expect students motivated by #1 to decrease significatly.
#2 is a pure loss to the individual. If you now *need* a college degree to earn a middle class income, you’re a worse off than the guy in the bell bottoms who could earn his way into the middle class with only a (free) high school degree.
#3 is what Frank is talking about. An expense that didn’t exist (or not to the same degree) 30 years ago that is now necessary to “keep up with the Jones”
John Quiggin 09.21.04 at 4:01 am
msw, I was quoting Frank in saying that incomes were roughly stationary. Table A-1 in the Census report you cite gives an increase in median household income from 36000 in 1973 (the pre-recession peak) to 42 000 in 2002. so I guess it would be better to say “median incomes have increased only modestly”. Clearly this increase is almost entirely due to greater labour force participation of women.
cure 09.21.04 at 4:53 pm
msw, it’s simple not true that median income peaked in 1973. Let’s look at the Full-Time Year-Round worker table (http://www.census.gov/hhes/income/histinc/p36.html). Among men, income has been quite stagnant at the median since 1973 (39 v. 40k in 2001, inflation adjusted). Among women, incomes are up from 22k to 30k, inflation adjusted. The biggest stagnation has been among white men; men of other races have shown substantian wage growth.
The true stagnation has been among men with high school education or less. This is partially because of a loss of low-skill labor, partially because college degrees act as a selection effect (justified or not) in a way they didn’t in the 1970s.
msw 09.21.04 at 6:10 pm
I don’t see what the point is of dividing up the population into different groups and then pointing out that this group makes more than that one. You wouldn’t look at the 1950s and say, sure Italian americans were making more, but look at the sad fate of Boston Brahmins (that may be a social vistory, but it’s not an example of increasing aggregate standards of living). The data you reference confirms my point – the median income for an american male was more or less the same in 1973 as it was in 2001 (extending the series through 2004 would be interesting – *mean* incomes have decreased in the last couple years, so I doubt median incomes are up).
That the average black men is richer and the average white men is poorer now than 30 years ago might be a net increase in justice. Even better would be a world where both races are richer. There’s no reason this has to be a zero-sum game. The big injustice here is that something close to 0% if the 50% increase in per capita GDP in the last 30 years ended up in the pocket of the average american.
I’d argue that the number for women is a social rather than economic story. But if you want to claim that the increasing income of women marks an increase in the living standard of the average american worker, then they should be included in the sample with men. Would that help your point? Yes, their incomes have increased, but they’re still significantly less than the male number and they now make up 41% of the work force vs 30% in 1973. Would ignoring the sex of the worker increase or decrease the median wage?
(If you just take the average of the two wages, weighted by workforce participation, you get .59 * 40136 + .41 * 30420 = 36152.43 vs .7 * 39578 + .3 * 22391 = 34421.90. Or about a 5% increase. Or about 45% shy of what we would expect if a rising tide really did raise all boats)
Comments on this entry are closed.