In The Long Run…
Will Wilkinson posts the following interesting chart highlighting the remarkable long-term consistency of the growth trend in US per capita income:
Even something as cataclysmic at the Great Depression doesn’t fundamentally alter the equation. On the other hand, it did take fully fifteen years for incomes to return to where they were before things fell off the cliff and that’d be an awful long time to wait for “the long run.” In a gloomy mood, I recall the lesson of The Black Swan, namely that our habit of regarding the Great Depression as a worst case scenario is really nothing more than a convention. It was very bad, and substantially worse than any other modern downturns. Ergo, we use it as a baseline of how bad things could get in the economy. But for all we know, things could get much worse.
One assumes that probably they won’t. That, indeed, they probably won’t get anywhere near as bad. But it could have been the case that over the past two weeks I more-and-more got the sense that policymakers had a handle on the problems and were working to address them. Instead, day-after-day the reverse seems to be the case.
45 Responses to “In The Long Run…”
Register
RSSLogin
Topic Cloud
ArchivesBlog RollAbout Matt YglesiasContact Me
Donate
October 7th, 2008 at 8:52 am
RE Matthew’s comment “Ergo, we use it [Great Depression] as a baseline of how bad things could get in the economy. But for all we know, things could get much worse.
One assumes that probably they won’t. That, indeed, they probably won’t get anywhere near as bad.”
—————–
That’s what the Anasazis said.
October 7th, 2008 at 8:52 am
I suspect that graph would look a bit different, meaning less consistent, if you drew a similar line for inflation growth alongside.
October 7th, 2008 at 8:55 am
It would be really interesting to see median income plotted in exactly the same way.
October 7th, 2008 at 8:59 am
The graph appears to start at about $1300 in 1820 and ends at $30K in 2008.
According to this inflation calculator, inflation alone would account for a rise to $20K. Essentially, real incomes have only improved by 50% of the last 188 years.
October 7th, 2008 at 8:59 am
I’d be more sanguine about this, if it wasn’t Wilkerson arguing (much like Bryan Caplan) that democracy sucks (Bring back the Randarchy!).
Anyways, I expect that any country with a stable government that isn’t disrupted by war, and in fact the world overall, has a similar curve. We can then note that if you look at total market capitalization/GDP, the markets have been over-priced for 20 years. It has taken a great deal of effort to keep the stock prices high, but that effort has been pointless and stupid (unless you’re really well off), and then thing was bound to come down sooner or later. Well, it’s later.
But it could have been the case that over the past two weeks I more-and-more got the sense that policymakers had a handle on the problems and were working to address them. Instead, day-after-day the reverse seems to be the case.
Well, they had a handle on the pretense that everything was ok, but no, they haven’t got a handle on it. We are unfortunately reaching the point where it will be impossible to get a handle on it until the big unwind has run its course.
Sorry, dude. The upside is we have plenty of resources, lots of food, lots of educated people. We kinda sold a bunch of factories, unfortunately. So we can repair most of this. (Really, back when the markets were really unregulated, business depressions were common and nasty, and we survived.) It won’t be pleasant, I expect. (Probably more unpleasant for me than for you.)
The real issue is what happens in politics: many people are going to have the impulse to go fascist. This kind of ugly situation often results in ugly government. That’s the really dangerous part.
max
['So let us pray, or light candles, or feed the hungry or at least say mazel tov. And keep your powder dry.']
October 7th, 2008 at 8:59 am
@xtophr - “2005 dollars” means inflation’s factored in.
October 7th, 2008 at 9:02 am
Two thoughts: (1) Looks like the greatest growth came during FDR’s presidency, so maybe in part a tribute to New Deal economics (though having the economy fall off the cliff in the preceding years, or a war like WWII, is not something we should want to see again), and (2) the per-capita growth may have continued in recent years, but that growth has not been evenly distributed; what the graph does not show is that it has been eaten up entirely by the very top income earners, and so the perception of growth may not be shared by most of us.
October 7th, 2008 at 9:03 am
But for all we know, things could get much worse.
Highly, highly unlikely. We have stronger institutions now and a better understanding of how the economy works. (Studying what went wrong during the Depression is one of Bernanke’s specialties.) Also, as the economy diversifies and globalization continues, there are more stabilizers. For example, droughts aren’t nearly as a big a problem as they used to be.
That’s why recessions have generally gotten less severe over time.
October 7th, 2008 at 9:05 am
Our local theater troup, which has the delightfully Commie name of “People’s Light and Theatre”, is currently running a production of Aeschylus’s “The Persians”. Ellen McLaughlin’s adaption, actually.
The local area has been Republican in the past but that is changing. Nevertheless, it was interesting to watch the audience to see who squirmed and stirred restlessly at Aeschylus’s shafts from 2480 years ago.
Of how the stupid hubris of Empire can bring down disaster — the death not only of our loved sons but of our nation. Of how an incompetent son (Xerxes) can lead his people into disaster due to his need to surpass a virtuous father (Darius). Of how a people’s obediance can turn to raging fury against their leaders when the consequences hit home.
Putting on a production of “The Persians” during a disasterous , unnecessary war shows simply foresight. But having a major financial collapse hit during the season — that’s priceless.
October 7th, 2008 at 9:06 am
On the other hand, it did take fully fifteen years for incomes to return to where they were before things fell off the cliff and that’d be an awful long time to wait for “the long run.”
And the fifteen years (and the graph in general) is a good reminder for our schmibertarians who claim that we oughtn’t have Social Security or similar programs. Things work themselves out on the graph–eventually–but in the meantime we’re talking about real people.
October 7th, 2008 at 9:16 am
Two things - this is in real 1990 dollars, for those of you thinking in terms of inflation.
However, this graph is still idiotic/unreadable. I minored in math (ok, 10 years ago), and it makes my head hurt to figure out how I’m supposed to read it. Setting aside my inability to figure out what they’re up to, though, it clearly doesn’t show what Matt thinks it does. Note, for instance, that it shows per capita GDP not quite doubling between 1950 and now, and shows it roughly doubling between 1900 and 1930 — which would indicate that per capita GDP was growing at double the rate then as it is now.
Lesson (especially to Matt!): You can’t read a graph on some kind of perverse, logarithmic scale (actually, this one is more broken than that - like I said, it makes my head hurt) as if it was on a standard, linear scale. The actual numbers don’t show anything like what you think they do.
October 7th, 2008 at 9:20 am
Lesson (especially to Matt!): You can’t read a graph on some kind of perverse, logarithmic scale (actually, this one is more broken than that - like I said, it makes my head hurt) as if it was on a standard, linear scale.
The whole point of a log scale is that steady growth shows up as a straight line.
October 7th, 2008 at 9:23 am
The graph above also raises the question: WHO has been getting the increasing pile of money?
Answer: Not you guys
http://en.wikipedia.org/wiki/Image:United_States_Income_Distribution_1967-2003.svg
October 7th, 2008 at 9:25 am
Big disagreement with Harvey. A logarithmic scale is the only way to show data like this - having the vertical scale represent constant fractional changes is the most informative way to view data that varies substantially from its mean.
For example, looking at the Dow Jones Industrial Average on a linear scale from the Depression to today, you can hardly make out major financial crises of the 20th century because of the range of values sampled by the index. When you’re looking at a number that doubles every 20-80 years like this one over such a huge span of time, a logarithmic scale is the only choice.
October 7th, 2008 at 9:29 am
The whole point of a log scale is that steady growth shows up as a straight line.
Yeah, I know. Let me reword more sensibly, after another look at this graph.
Using a logarithmic scale, obviously, is fair enough, although the line markers (first for thousands, then for tens of thousands) are confusing as hell, and make it harder to read. My point is that if you actually try to figure out the numbers represented on this graph which crowds way too much data into way too little space, you’ll see that the actual slope isn’t as steady as Matt is making it out to be - an easy illusion to create with a logarithmic graph. Looking at it visually it’s easy to come away with the impression that growth has not slowed, but if you look at the numbers (as I pointed out) it sure seems to have slowed greatly.
One more time: a well-executed logarithmic graph is fine, but this one is badly executed, and both Matt and its creator are reading it improperly. This needed a big, clear graph with lots of intervals marked — if it had this, it would make the opposite point of what this Cato guy is trying to make, and that Matt is swallowing uncritically.
Anyway, I think so. The graph is so small and crowded that it’s hard to be sure.
October 7th, 2008 at 9:32 am
I’m not a math major, but something about that graph looks off. The physical space occupied by $1000 is much larger between 1k and 2k than it is between 9k and 10k. And then the gap increases dramatically from 10k to 11k. It seems to me that the graph gives you a false sense of how stable the growth rate has been. In reality, growth has been much slower after WW2 than in the period from 1890 to 1945, even though the slope of the curve seems to stay roughly the same.
October 7th, 2008 at 9:43 am
The variation in gap size is because it’s a log-scaled plot. You’re not reading it quite correctly; the first interval above 10K is from 10 to 20, not 10 to 11.
-dms
October 7th, 2008 at 9:51 am
Zach, I too would like to see the median (half below and half above) income graph. I assume the graph shown is the mean (total income divided by population). The weakness of the latter is frequently highlighted by the example of the average income of two people in a bar, Bill Gates and most anybody else. I suspect the median graph would reflect much more variation among periods of time, rather than essentially a straight line as depicted in the mean graph.
October 7th, 2008 at 9:52 am
People, this graph uses a log scale. Look at the y axis - the space from 1000 to 10,000 is equal to the space from 10,000 to 100,000. And notice how the horizontal lines get closer and closer together? It’s a perfectly reasonable representation of the rate of growth.
So, Harvey Lobster, try reading it again. The increase from 1950 to about 2000, where the graph ends, is not less than double. It’s three-fold - from about 10,000 to about 30,000 in constant dollars. The increase from 1900 to 1930 is less than double - from about 3,000 to about 5,000.
If you were to carry the line out to 2010, it would just about reach 40,000. What the graph shows is that, on average, per capita income roughly doubles about every 30 years, with the rate of growth slightly higher after 1950 than before 1903. Of course, we may be about to fall off a cliff, like they did in 1929-1932.
(Remember, this is per capita income in constant 1990 dollars - so all income is allocated over the entire population, including children, and the use of 1990 dollars wrings out inflation).
If you drew this graph on ordinary graph paper, the shape would be a steeply rising curve.
I understand that this is a blog for English majors, but jeez louise, didn’t anyone here pass Algebra II?
October 7th, 2008 at 9:55 am
Two points:
a) It’s not clear to me that you can even calculate GDP accurately prior to 1950 for comparison purposes. What’s the value of a buggy whip? Or a Morgan horse?
b)Much of the rise in GDP is the accumulating capital that is passed on from parents to descendents over many generations
c) But we have also SPENT much of our inheritance — we have CONSUMED our non-renewable resources. The oil wells of Texas. The iron deposits of north central USA. The coal deposits of Appalachia.
October 7th, 2008 at 10:01 am
Ask the Anasazis what happens when the resources run out.
I would suggest that you ask our Republican political leaders but they would probably lie. If they decide to tell the truth, it would go something like this:
a)You pick up your bags of gold and move elsewhere. Find another nest to shit in.
b) And if you don’t have no gold, you are truly fucked.
c) And don’t blame us. We are waving the American flag.
Even as we renounce our American citizenship. Well, actually we renounced our citizenship in our hearts long ago — we just doing it publicly now.
October 7th, 2008 at 10:02 am
One also might wish to recall that the people living in the actual early 1930’s didn’t have the benefit of a graph from the future telling them it would all work out.
No one ever actually knows until after the fact.
October 7th, 2008 at 10:20 am
If you drew this graph on ordinary graph paper, the shape would be a steeply rising curve.
I’ll try for the third time - hopefully more competently - to make my point, while admitting that I misread the graph the first time.
My point is that if we did, in fact, put it on ordinary graph paper
1) It would, indeed, be a steeply rising curve
2) We would be able to accurately see that the slope of that curve has not, in fact, been consistent.
Let’s look where, on this graph, per capita gdp doubles. It hits 2,000 some time in the mid to late 1850s (again, the problem with badly marked intervals). It hits 4,000 in the late 1890s - roughly 40 years to double. It hits 8,000 roughly in the early 1940s - a little over 40 years to double. The scale changes, and we can’t easily see where it hits 16k, but we can see that it hits 10k in about 1950 and 20k in roughly 1980 - much faster per capita GDP growth.
Then we have the chart ending at an unmarked year at an unmarked per capita GDP. It looks to me like the curve is noticeably flattened - and that’s my point. By not marking enough intervals, the graph obscures a period of accelerated gdp growth, and very possibly obscures one of flattened gdp growth.
My discussion hasn’t been particularly good, and I made errors the first time around, but my point remains: this chart suppresses important differences by using a logarithmic scale without proper size & markings.
October 7th, 2008 at 11:00 am
I think what the Lobster is saying would be more easily illustrated if you tried drawing an actual straight line on the graph. By contrast, it would show that the line isn’t really all that straight.
You could take the log of the GDP numbers and do a simple first order linear regression, then you could see which eras exceeded or fell short of average long-term growth. You might also see that there might be a non-linear trend if you did a higher order regression.
October 7th, 2008 at 11:04 am
2) We would be able to accurately see that the slope of that curve has not, in fact, been consistent.
Seriously you’re digging a deeper hole here. The point of a constant slope on a log plot is that the fractional growth is constant… the rate of return if you’re looking at the growth of value of an investment over time, for example. Matt’s observation of “the remarkable long-term consistency of the growth trend in US per capita income” is exactly right if you realize that he’s talking about the rate of growth in fractional terms (%/yr and not $/yr).
October 7th, 2008 at 11:12 am
Ok I’m not going to get into elementary discussions of graphing on log scales (thanks to Zach et al for trying to bring the others up to speed on how it works..). I’d like to second the call for median income to be plotted the same way and maybe some higher order moment like the variance to see how inequality has changed over time.
October 7th, 2008 at 11:17 am
gdp as a function of population says nothing about the distribution of wealth.Clearly for many(if not most) people the 1920’s were no better than the 1900’s. In addition, we are all assuming that we are witnessing the decline in american power and empire. Clearly the past 100 years were a time of American dominance. It is hard to imagine that the next 100 years, with increased globalization will show the same linear progression. I wonder what the english graph would look like for the 100 years before world war 2 (english empire)and the next 60 years after.
October 7th, 2008 at 11:48 am
Look, this is one graph, not the Statistical Abstract of the United States. The point is that GDP per person, adjusted for inflation, roughly doubles every thirty years, but that there’s been one significant period of aberration - the Great Depression followed by the Second World War - that lasted long enough to cause considerable dislocation to every-day life. What’s controversial about that?
October 7th, 2008 at 11:49 am
I never got the sense that policymakers were getting a handle on the problem, or even that they were concerned about the seriousness of the problem. For one thing, every time they talked about people’s “confidence” in the economy I felt they were just saying it was basically a collective delusion. It’s not. The US does have a problem with people not being confident enough, but it has far more fundamental problems which many businesses are now seeing which is why they lack confidence in the economy, and until those underlying problems are addressed, dealing with the “lack of confidence” will do nothing.
For another, the proposals for dealing with the issue look remarkably like more giveaways to campaign contributors. If the policymakers were really concerned about the economy they would not be worried about getting reelected so much as they would be worried about being killed by an angry mob in the streets.
Another indicator is that policymakers don’t seem to have consulted any actual economists, only various former business moguls that now have cushy sinecures given them by Bush, or previous administrations.
Basically what I think has happened is that the normal system whereby policymakers come to understand that a problem is real has broken down. Rampant cronyism in all of the various federal offices has replaced the few people who actually knew what they were doing, and produced real information to policymakers, with conservative hacks who have no idea what is going on and only memorize lines about the free market uber alles. The president, who caused this problem, has no idea these people actually had a role because, frankly, when everything was going well, they didn’t. Congress, on the other hand, doesn’t really understand that ALL the people with any understanding of the situation have been replaced.
October 7th, 2008 at 11:51 am
I cut the figure and pasted it into a graphics program. Overlaying straight lines on the curve (not really the best way to fit it, but without the numbers it will do for a rough estimate)
Pre civil war growth looks like its about 1.2%/yr
Civ war-depression growth ~ 1.8%/yr
Post WWII growth ~ 2.2%/yr.
It looks like constant growth, but it isn’t.
October 7th, 2008 at 12:59 pm
As an engineer I am mildly amused by the confusion generated by this graph. Suffice it to say that this graph demonstrates exactly what MY says it does, and it is the correct representation to do so. Outside of some tremendous variation around the Great Depression and post WWII recovery, the growth per year has been remarkably robust. The amazing thing about the disruption is that it didn’t seem to generate any kind of offset in the long term growth pattern. It’s as if there was some kind of underlying, fundamental trend (maybe productivity growth) driving this data. Severe economic misallocation was able to substantially disrupt the trend, but only temporarily. I don’t have any explanation for the overshoot, but if you drew a linear regression of this data, you would find a strong tendency in the data to return to the regression mean.
October 7th, 2008 at 2:21 pm
[sigh]
OK. Get a piece of paper with a straight edge and hold it up to your screen. Look at how it fits the straight line on the graph. You’ll see that Njorl is right. The graph is slightly concave, and if you wanted to you could divide it into three sections with radically different growth rates. Something that doesn’t matter much over a short distance, but makes a tremendous difference when you extrapolate it for long times.
However, the part of the curve before 1890 is very smooth for a reason — they weren’t collecting good measurements and it’s estimated from data that isn’t very good. Maybe it should be discounted somewhat.
And all through, to convert to $1990 dollars they have to make lots of assumptions about inflation. Maybe what we’re seeing with this almost-straight line is precisely those assumptions. I mean, how good is our estimate of how many loaves of bread got baked in the USA in 1890? Not very. How good is our estimate of the cost of a loaf of bread in 1890? We have lots of records showing prices in different places. How do you compare a loaf of 1890 bread with a loaf of modern pre-slice enriched white Wonder Bread? It wasn’t as white. It didn’t have its vitamins removed and then 8 of them added back — they hadn’t discovered vitamins. It didn’t come sliced. There’s basicly no comparison. But we assume they’re the same thing and then what? Should we assume that a loaf of rosemary foccacio that costs $6.70 compared to the Wonder Bread $2.20 is worth 3 times as much? To estimate inflation you have to make some sort of assumption like that.
And so we might wind up with a straight line estimate for per capita GDP because that’s how the inflation rate got estimated.
On the other hand, I’ve seen this claim before. There was a communist economist, I think he was from hungary, who claimed that each society has a natural rate of increase for per capita GDP, and whenever something happens that makes it deviate from that then it changes the other way until it gets back to the GDP it would otherwise have had. He showed that happening to the USSR after WWII. The war hurt them very badly but then they grew fast and then when they caught up they went back to the old line. He claimed the rate was significantly lower for czarist russia versus under the communists, but he specifically refused to speculate whether the increase was worth the suffering and deaths of the transition.
His guess was that the increase comes because of technological innovation, and that techie innovation comes from prototyping — you make new improvements on top of the last set of improvements, and you can’t make them until the last set has worked. The prototyping process continues no matter what, but the actual implementation in the wider economy varies.
I don’t remember his name or how to find any links about him.
October 7th, 2008 at 2:55 pm
Two points:
a) It’s not clear to me that you can even calculate GDP accurately prior to 1950 for comparison purposes. What’s the value of a buggy whip? Or a Morgan horse?
Brad Delong has spent his most of his post-doc career doing just this. (Basically, trying to measure the per capita GDP back to literally the stone age). It makes for interesting reading, at least by the standards of academic economic papers.
October 7th, 2008 at 5:29 pm
I more-and-more got the sense that policymakers had a handle on the problems and were working to address them. Instead, day-after-day the reverse seems to be the case.
Policy makers don’t have a clue. Not a single clue. What has been called the free market became the most manipulated distorted market in the history of man. I’m talking about the credit market. The credit market is the father and mother of capitalism. It’s been wrecked.
To fix it the Fed and Treasury are attempting to manipulate the market some more. A lot more. A lot lot lot more. Uncle Sam is now the lender of last resort for the entire economy but the problem is he has to borrow the money to do it. That takes money away from the real market. He is lending to damaged entities taking bad paper as collateral. The borrowers are then putting the money into Treasuries. No new ‘liquidty’ results. Just crowding out of other borrowers. It’s a circle jerk.
No matter how bad you think it is, it’s worse. Everything you know is wrong. Everything that is about the economic status quo. Most Americans are going to be significantly poorer for a long long time. Uncle Sam will not be able to help as soon enough nobody will want to lend to him anymore and interst rates will spike on the debt and debt service will soar as a percentage of the economy.
Eventually the Fed might have to just print money. Which would be signing its own death warrant but it will be eaasier to do than face lynch mobs for Ben and the boys.
October 7th, 2008 at 7:02 pm
How good is the 19th century data? The graph shows no trace of the Depression of 1837, or the 1871 panic though both were major upheavals at the time.
Re: Most Americans are going to be significantly poorer for a long long time.
Prediction: we’ll have all this (mostly) behind us by the time of the next presidential election.
March 11th, 2009 at 7:45 am
It is the coolest site,keep so!
March 13th, 2009 at 4:11 am
Great site. Good info
March 17th, 2009 at 3:15 am
If you have to do it, you might as well do it right
tramadol
March 22nd, 2009 at 7:09 am
tramadol
Very interesting site. Hope it will always be alive!
March 22nd, 2009 at 11:10 am
buy viagra online
Great site. Good info
April 2nd, 2009 at 8:43 am
If you have to do it, you might as well do it right
buy cheap viagra
April 14th, 2009 at 10:42 am
Great site, Good info
viagra