Bryan Caplan  

Is Average Over?: Two Equivocal Graphs

Great Moments in Crowd Sourcin... Noah Smith on Redistribution...
Here are the unredacted graphs from "Tell Me What 'Average Is Over' Looks Like."

Graph #1 is from 1978:


Graph #2 is from 2008:


What's equivocal?  Although the 2008 graph definitely has more inequality, the 2008 labor market consistently offers more continuous rewards for achievement.  In 1978, "some college" was rarely more lucrative than "high school only," and "post-BA degree" was rarely better than "college degree."  The two lines virtually overlap until the 80th percentiles.  In 2008, in contrast, the marginal payoff for jumping up a single educational category is visible throughout the entire distribution.

Think about it this way: In 1978, there were effectively only three educational levels.  By 2008, there were clearly five.  If "average is over" means moves toward bimodality - as Tyler has often claimed in personal conversation - then 1978 fits his story better than the world of today.

Comments and Sharing

COMMENTS (13 to date)
Peter H writes:

It also matters what the slope of the lines is. The HS dropout/grad lines have substantially positive slopes for their entire distance in the 1978 case, but are much flatter in 2008.

For example, a 20th percentile college grad in 1978 was as high income as about a 60th percentile dropout. In 2008, the 20th percentile HS grad makes as much as a 90th percentile dropout (I think - the graphs are pretty small)

What I'm getting at is that intra-group variability dropped a ton, such that being in one of the bottom two groups (HS dropout and HS grad only) is a near guarantee of a low income in a way it wasn't in 1978.

TallDave writes:

Celebrate income diversity.

Brian writes:

"Celebrate income diversity."

Exactly. I'm not sure why progressives and liberals, who seem to bow at the altar of diversity, suddenly get all shaken up about diversity of wealth and income. If diversity is such a good thing, why make an exception when it comes to wealth?

The reality is that growing variation in wealth and income is good for the same reason that diversity is good in others areas--it means we are expanding the range of what's possible and thereby increasing liberty. What's not to like?

joe writes:

There's something missing from the analysis.

If you're finished college and have a median income in that bracket, you would be probably in the, say 80 percentile of a high school graduate if you didn't went to college.

That should somehow factor into the analysis , not sure how.

Cmprostreet writes:

Can someone explain the Y axis to me? It looks like its saying that in 1978 a high school dropout in the 30th percentile of earnings earned $500,000 that year (in 2009 dollars). I know that's wrong, but I can't figure out what I'm misinterpreting. Is the axis in cents and not dollars? Is it earnings over a longer timeframe or larger group than an individual? Do I have a reading comprehension fail? Thanks.

Dustin writes:

The links in the earlier related post displayed that the Y axis is lifetime earnings.

R Richard Schweitzer writes:

Overlay those graphs with those from the mortality and morbidity tables of the same periods.

Then adjust.

Consider just the change in American (born) Negro mortality rates (longevity)in those 30 years that would have comprised most of the "lifetime" over which the earnings accrued.

Consider what groups made up each of those "lines" such as "drop-outs," and HS only and their relative mortalities, etc.

It's not so simple.

This is what happens when we "select" our own facts.

Cmprostreet writes:


Thanks! I didn't click through the earlier links to avoid revealing the years then, but I should have checked there before posting.

Guest2 writes:

Law graduates clearly show bi-modalism. I am not sure what, if anything, your graphs show.

The difficulties that growing numbers of law school graduates are having finding jobs is by now a well known problem (Wall Street Journal (9/24/2007, A1 and A11), but particularly interesting is “a survey of about 650 Chicago lawyers published in the 2005 book 'Urban Lawyers' [which] found that between 1975 and 1995 the inflation-adjusted average income of the top 25% of earners, generally big-law firm lawyers, grew by 25% -- while income from the other 75% actually dropped.” (WSJ 9/24/07 A11)

True to type, this local stratification of the ranks of Chicago lawyers is an especially interesting example of occupational bifurcation, that is, of “something happening at the top,” where conditions of extreme competition and hierarchicalization predominate, resulting in large divergences over seemingly small underlying differences. This last point is key.

Law school salaries really demonstrate tri-modal bifurcations: $0 (i.e., unemployed or part-time work, not necessarily in the legal sector); $60k; and $120k. Over the years 1996, 2000, 2006 (Fig. 1., Fig. 2., Fig. 3.), this demonstrates a noticeable divergent stratification process.

"The result was, for the first time, the two peaks of a bimodal distribution, with nearly 14% of salaries reported at the $125,000 level and 48% of the new lawyer salaries falling in the $30,000 to $50,000 range. [But] By 2006, ... the obvious and dramatic stratification of the market for new lawyers was conspicuous, with 44% of new lawyers earning between $40,000 and $60,000 and 17% of new lawyers earning the $135,000 or $145,000 starting salary common to big law practice."
The successive graphs clearly show the emergence of increasing bimodalism over time.

But what is noteworthy is that each of the bifurcations (employed/unemployed vrs $60k vrs $120K) that are emerging among law school graduates are indicative of underlying instability, as depicted here:

The article on law school enrollments supports the belief that we are currently in a "buying stampede" as shown here; the further we are from the back-wall (and the closer to the front-edge), the greater risk that small changes will result in massive shifts in herd behavior. In terms of this model, any increase in "fear" could have this effect, and supports the plausibility of a cusp-catastrophic bifurcation, or series of bifurcations, occurring in higher education.

nl7 writes:

Guest2 - re lawyer bimodialism, that doesn't disprove this graph so much as add nuance to the specific case of law school attendance (especially to a law school outside the top echelon, with diminished reach into NLJ250). But even the lawyers at the lower mode making $40k or $50k are often going to have lifetime earnings north of $1M and even $2M. Twenty years at a first year salary of $50k is already $1M, and even these lawyers will receive raises or promotions to junior partner. Meanwhile, others will switch professions or drop out of the workforce (maybe forever, maybe for children) so the impact of bimodalism is spread out.

The same is true for the top mode around 160 (and lesser modes around 125 and 145), since some markets and firms don't offer the same lockstep as say Cravath and since many associates will drop to different firms later in their careers, and some will switch out of law or take a year or two off for children or government work.

So while a bimodal distribution itself would suggest a couple long flat lines (where a hefty percentile of people are all making the same salary), this effect is muted within the law because its inputs are reflected over lifetime earnings.

And of course, many other post-BA degree jobs are going to have conflicting salary distributions and will cancel out a lot of the effects of the law degree. Something like 8% of the over-25 population has a post-BA degree, so law degrees are a small portion of that total.

Guest2 writes:

nl7, Thanks for the reply!

"So while a bimodal distribution itself would suggest a couple long flat lines ... this effect is muted within the law because its inputs are reflected over lifetime earnings."

Are you assuming that law degree holders can get jobs that pay? Link to your data?

The reason that I ask is because I've worked with recent law degree holders, and they aren't making that much. Otherwise, they would be unemployed.

In fact, what ends up happening, is they go back to school for another degree (LLM or something) from an even more prestigious school, and double their student loan debt burden. And, of course, this adds to credential inflation, making it rougher on everyone else, especially minorities that earned degrees in the 1970s, 1980s.

Guest2 writes:

More on law schools from Dec 17, 2013 WSJ, which reported that the American Bar Association said that the number of first-year law students fell 11% this year across the 202 U.S. law schools that the group accredits.

The total — 39,675 full and part-time first year students — was one shy of 1977's total, when the nation had far fewer ABA-approved schools.

The decline highlights the crunch in the legal profession in the wake of the recession.

Lucrative work in fields like mergers and acquisitions dried up as the economy stalled and law firms cut back on hiring.

Guest2 writes:
Comments for this entry have been closed
Return to top