I am continuing my analysis of per capita GDP growth, but instead of looking at trends in total growth, I’m looking at where growth has ended up.
My results will likely surprise people on both sides of the partisan divide. Rising income inequality is likely a feature of growth in the American system, with catastrophic financial events being the main mediator of the effect. Likewise, the bottom half of the income distribution likely benefits modestly from financial collapse.
Growth and Income Concentration
Due to the properties of income distributions, increases in the median income are reflected proportionally at all points. This implies that GDP will increase with the median income. However, increases in inequality benefit those making more than the median income and are detrimental to those making less. However, the scale of the effect is such that decreases in equality will cause a rise in production.
It is possible, and indeed, fairly straightforward to determine the magnitude of each of these effects on the average income. This gives a snapshot of how much per capita GDP growth is from increasing the median against how much is from increasing the concentration of income at the top. I’ve named these effects the Median and Concentration effects respectively. This is a graph of the running (10-year) averages of those effects over time:
In general, growth in American since the end of WWII has been driven more by rising income inequality than by rising median wages.
While I generally caution against eyeballing time-series, the exceptions fall at fairly exceptional times in American economic history. The high average Median Effect in the 50s is well-established to fall during a period of post-war industrialization; it is one of the clearest times of real growth in history. The dip in the Concentration Effect was likely the stagflation event. Unexpected inflation is better for the bottom of the income distribution than the top.
The collapse of the Concentration Effect in the mid-2000s is because the average reaches 5 years forward; the great recession hurt top earners disproportionately. In fact, the Median Effect was positive during this time, implying that the bottom half of the distribution made out proportionally very well. The data does bare this out. (I’ll be the first to admit that those are modest gains in absolute terms.)
I’d caution against easy lessons from that graph; it’s easy to see spurious relationships in time-series graphs. Still, I think we can glean a few generalities from this:
- The norm is widening inequality. While I fully recognize that is does not imply ought, the meme that widening inequality is a new phenomenon has to do with more accessible data beginning in the late 60s, right before inflation took off. Longer datasets, like this one, imply the post-70s increases in inequality were a return to the norm.
- The Median Effect is taking the edge off the Concentration effect. For many people, widening inequality is tempered by a widening pie, though below some point, this won’t prove true.
- Large splits between the Median and Concentration effect disproportionately effect the extreme income earners.
- Equality is a mediocre economic goal. The two big dips in the Concentration Effect correspond to periods of serious financial distress. It follows that disproportionately making the rich worse off will help equality, but there is no reason it will “Robin Hood”. Just because equality is increasing does not mean much of the distribution is actually being helped.
- As a corollary, it may be better to focus on increasing the Median Effect rather than making reducing the Concentration Effect a goal. The mathematical disconnect between inequality and median income implies that marginal well-being for even the poorest can be lifted by policy that harms next to no one. The disproportionately deleterious effects of the Concentration Effect on the bottom of the distribution means that we cannot push the Concentration effect out of our minds, but we needn’t necessarily take all of it head on.
So, despite conventional wisdom, middle-class earners have not stagnated in earned GDP and are seeing rates in line with historical norms. However, there is a larger trend towards greater inequality, concentrating income at the top of the distribution and away from the bottom. Much of the growth of the past decades has gone to the richest.
In short, we are not generally becoming a nation of haves and have-nots, but we are certainly becoming a nation whose very poorest are seeing income fall.
Economic growth has been a theme on this blog as of late, and I’d like to continue to challenge a piece of economic orthodoxy. Namely, the idea that a healthy economy grows at the rate of 3-5% a year.
The facet I’d like to look at is that the empirical record suggests that a certain kind of linear growth as been much more the norm over the last 5 decades. Per capita GDP growth is roughly constant. Exponential growth of total GDP is a function not of economic reality, but rather population trends over the last few decades.
Per Capita GDP
Simply put, Per Capita GDP is linear. You can see this on a graph:
Assuming that each year the average person will make 650 dollars more than the year before explains 98.6% of the trend—a pretty good correlation. And while the correlation on an exponential model is only a hair less, it doesn’t have normally distributed errors and is therefore a weaker model.
So, we can fairly safely say the direction of GDP per capita has been positive and constant in recent memory. Assuming it will hold forever is a bit brash, but we’re going to assume that for the time being, per capita GDP will hold.
Consequences for Growth
Working from that assumption, we encounter a surprising result: growth is close to expected levels.
One of the apparently troubling trends is falling GDP growth, and in particular GDP per capita growth. Plotted against time there is a clear downward trend, though it is apparent that annual effects have a good deal of influence over any particular year’s growth:
A plot of errors suggests that a different model is in order, and I’m about to show just what that is. Since per capita GDP growth is expected to be 650 every year, we can calculate the expected GDP growth by dividing 650 by per Capita GDP that year. I’ve included the fit line (in blue) from the previous graph for comparison:
The 650/x form gives the curve there and for those interested in mathematical language, both hyperbolic and inversely related apply here. In lay terms, we expect growth to approach 0 at a slowing rate. In 2012, expected growth was 1.3%. In reality, it was 1.45%, which accounting for macroeconomic factors is about on par.
Where does the exponential component of GDP growth come from? From population growth. As that has slowed down, so has GDP growth. That, combined with the fact per capita GDP growth has been linear, it’s pushed growth rates lower.
We cannot cling to a 3-5% GDP growth rule of thumb. It’s not really justified by the historical record, but rather a happenstance of population and GDP level in the post-war era.
A better rule—though, one I caution will likely also break down at some point—is $650 dollars of linear growth a year. (I’m not quite sure what to say as far as a healthy range goes; it’s clear from the plot of changes that a fairly large range is possible, though mitigating that is a clear policy object.)
This constant per Capita GDP increase implies falling growth rates. Right now, we only expect GDP growth to be population growth plus per capita GDP growth, or 2.7%. That’s below the classically healthy 3% rate.
It’s time for a new standard.
Or, How the Hunger Games are Not Survival of the Fittest
I know the sequel came out yesterday, but can we take a moment to have a game theory moment about the final chapters of The Hunger Games?
Peeta, Katniss, and Cato have been chased to the center of the arena for a three-way duel to the death. Cato is the strongest of the three. Peeta and Katniss are protected from turning on each other by a rule-change that allows them to win if Cato dies first. The end is great, and if you’ve not read it, do it!
But we’re going to look at a more generic Hunger Games. Let’s assume the every-person-for-themself rule still stands. Further, let’s assume all three players—who we’ll still call Peeta, Katniss, and Cato—only care about winning. That is, none of this Peeta dying so Katniss can live drama. Let’s also say Peeta is the weakest, and Cato the strongest.
Who would we expect to win the game?
What Duel Would you Want?
The problem they each face is that they can only kill one of the others at a time. If they are successful, they then have to duel the other survivor. There is of course that possibility that two or three are killed in the carnage, but we’ll get to that in a moment. Who would each player like to duel if it came to that?
Our unsentimental Peeta would prefer to duel Katniss as his chances of winning are higher than if he took on Cato. Katniss would rather duel Peeta for the exact same reason. Cato would prefer to duel Peeta because he’s the weaker target.
Survival of the Weakest
This turns out to be enough to force an alliance between Peeta and Katniss. Any other arrangement leaves all three more open to the possibility of facing the more dangerous enemy in the duel round. The fact that the Duel phase might be skipped because Cato and Katniss (perhaps with Peeta’s help) fatally wound each other is no matter. Peeta only benefits from that. Katniss isn’t better off trying to head off a duel with Peeta by inviting one with Cato. And Cato isn’t going to take out Peeta and risk a duel with Katniss because Peeta’s odds are good.
This is a really fortuitous arrangement for Peeta. No one tries to kill him until after one of the others is dead. In fact, depending on exactly how you define the game and exactly how good everyone is relative to each other. Using my numbers, Peeta’s odds are about 5 times better for being guaranteed to survive the first round. Round figures, Peeta survives with 54% probability, Katniss 12%, and Cato 9%.
Interestingly—and troublingly for the Gamekeepers—the chances of no one surviving the game are right around 25%. Ouch.
What Can We Learn from This?
First of all, that Haymitch gets it. He’s the one who said that when the Gamekeepers made Katniss a 12, it was to make her a target. While a 24-way duel is significantly more complex, the intuition that being on top does not make for good odds holds. When the Gamemakers say Katniss is #1, they’re hoping she gets shot for shooting an arrow at them.
But further, we can say head-to-head-to-head competitions are not a good way to find out who is strongest in general. Think about three-way primaries. The lower 2 spend a lot of time humiliating the front-runner, which means that the top runner is more likely to get worn out of the race. The worst candidate, therefore, is the most likely to win.
Same is true of oligopolies. When fighting for market share, advertising campaigns may focus on taking from the lead product to the benefit of the worst product.
Indeed, Katniss’s internal monologue in The Hunger Games says as much. She knows that if she and Peeta make it to the end through their alliance, she has to kill him, rule-change notwithstanding. When Peeta is at his weakest, she is the one taking the risks that might knock her out of the game and get Peeta closer to a victory.
Let’s have a fanboy/girl moment for Suzanne Collins’ game theory!
There is a dawning realization in economic circles that the old standby economic advice, the economy is healthy between 3% and 5% growth, is dated. Paul Krugman caught up with that crowd today, testing the waters with a non committal editorial where he suggests that maybe that’s what up.
He touches on what I think one of the central oversights of his generation of economists is: his generation drove economic growth in the middle of the last Century.
A Different View of the Economy
Annual, per capita GDP growth is stubbornly positive in the United States. Indeed, since President Eisenhower left office, GDP per capita has risen every year except 2009. (2008 comes was a close call, squeaking at a positive .7% growth.) This is remarkable, considering. As an American, you can expect your income to grow year over year. Even in 2012, you were looking at just shy of 4% growth. Here’s the graph of this since the beginning of the Great Depression:
A lot of trends jump out to my eye, but let me caution restraint. You can almost never see underlying causation in a trend line. Still, let me cautiously put forward that the bulge between 1960 and 1985 is very likely the boomers coming of age and entering the workforce while having fewer kids than their parents. This meant there was more to go around while the number of people to spread it to decelerated. Let’s zoom into that. This is the post-war years, everything since 1947:
The standard models of neo-Keynesian models were incredibly influenced by the first half of this graph. This was the dataset they drew on. The importance of this to Fed models, fiscal policy, academic papers, banking structure, and in general the entire economic landscape is not easily overstated. And if you zoom back out to the first graph, it’s an aberration. Accelerating per capita GDP isn’t the norm.
In fact, the last two decades, smoothed growth has looked more like it did under Eisenhower than Nixon.
If we expect Per Capita GDP Growth to be where it is, what does that mean for the permanent slump theory?
For starters it’s nominally correct. To use a simple example, imagine the economy has GDP 2 and 2 people. That is, GDP 1 per capita. If the next year you have a birth you have a population of 3 and a population growth rate of 50%. To maintain GDP per capita, you have to increase GDP by the same percentage, 50%. The year after, if there is no birth, you need no growth to maintain the standard of living. If GDP increases by 50% again for whatever reason, you have GDP 4.5 and per capita GDP 1.5, or a 0.5 increase in the material standard of living for every person.
I would agree that we’re entering what looks like a permanent slump. But that is because our metric is broken. I will submit that one of the reasons the economy overheated was because 5% GDP growth when population growth is .7% is not the same as 5% GDP growth when population growth is 2%. The former is much, much hotter since it has fewer people to go to.
One thing that is fueling the pessimism about our current circumstance, I think, is that this is the deepest recession we’ve had for generations. Of course its taking time to catch up! There are some serious issues, some structural, some distributional that I’m leaving by the wayside. And covering that gap has been slower than previous recessions would have suggested. But we’re also having trouble because we’ve not had time to close the per-person production gap.
America must move past a broken measure of productivity. GDP growth is fundamentally fine for measuring how production changes year over year, but such a measure can mask how quickly these things are changing relative to the population. Krugman is correct we’re entering a slump. But with slumping birth rates, should we expect anything else?
Do we need 3% growth when population is not growing so fast?
Do you want to be the one to labor to make up for the births that never came?
To no one’s surprise, I yet again critically disagree with a post from Benjamin Studebaker. (We went to high school together and have been debating for as long as we know each other, okay?) In this case, I disagree with his analysis of what my alma mater and employer has said in response to the impending ban on same-sex marriage, domestic partner benefits (for straight couples too!!!), and substantially similar arrangements.
And that merits some full disclosure. I am an IU employee and alumnus, though I am in no way a representative of them in my opinions here. I am also openly gay. I have been involved in both discussions about HJR-6 and other LGBT issues with IU’s administration. I have friends and coworkers who would be seriously harmed if HJR-6 passed, and I agree with both Mr. Studebaker and my university that every single student would be harmed by Indiana’s loss of competitiveness in the labor market should domestic partnerships be banned.
That is to say, I have a horse in this race. But that cuts both ways. Those who would accuse me of bias, I would remind you that I also offer a first-hand testament of why Purdue’s silence is so frighteningly dangerous. There is almost never a neutral position for institutions to take. In the case of an illiberal threat, institutions like state universities should respond with a clear judgement: that harming other members of the community is out of sync with the values we wish to have.
The Non-Neutrality of Inaction
Before I get too far, I want to broadly affirm one of Mr. Studebaker’s points. He puts forward a heuristic that a formal institution should take no position that does not advance the interests of its categorical members. He uses the example of a truckers union; a trucking union should take no position that does not advance the interests of truckers. Towards the end of the article he points out this is at times at odds with the goals of formally liberal institutions when liberalism may disadvantage some or all members. In such a case, reconciliation between the heuristics is in order. This will be the issue of second part, but for now I want to address the issue of the “neutral” position.
I have chosen the formulation above—”take no position”—to both echo Mr. Studebaker’s original wording and differentiate from “advance a position”. Advancing a position (at least as I’m using it) differs from taking one in that it rules out the case of staying the course. This is critical because it rules out a strong-form Burkean play where you argue for the status quo during controversy simply because it is stable.
The case of HJR-6 is Indiana is instructive. HJR-6 would outlaw same-sex marriage (SSM) or anything substantially similar*, like IU’s domestic partner benefits. Similar benefits exist at Purdue, though they differ in the details. There is of course controversy at IU about whether or not those benefits should exist along almost identical lines to the controversy about HJR-6. Allowing HJR-6 to become law will undeniably cause substantive harm to IU’s recognized same-sex couples.
Those benefits throw into sharp relief several problems with the “advance no position” formulation. First, at some point the benefits passed. This argument says that under controversy, no change ought be made. But once made, the controversy remains and the same argument against going back is now in force. Under an “advance no position” rule, the status quo is privileged, which is at best an arbitrary way of settling these disputes and at worst it entrenches abuses of power.
Studebaker’s position that Universities ought to shy away from taking a position on what the state’s values ought to be is a formulation I could live by. But with many such general positions, I find it does not universalize well:
The idea that Hoosier values include or ought to include a principle of liberal toleration that protects LGBT people is the very idea presently being debated. Those who oppose this idea generally believe that Hoosier values are Christian moral values, and that these Christian moral values, while compatible with a principle of toleration in many other areas, are incompatible with tolerating or appearing in any way to endorse LGBT marriages. IU is an institution that these individuals fund, attend, and otherwise contribute to. For its president to make statements about the nature of “Hoosier values” that exclude their views is for it to openly oppose the interests of its own benefactors.
Mr. Studebaker would have IU implicitly take the stance of the status quo because interests at IU are on record as being for HJR-6. The trouble with this is that it therefore privileges the advocates of HJR-6 over the law’s victims. To put it none to plainly, not a single argument he makes there isn’t reversible. To implicitly endorse the General Assembly’s plan is to imply Hoosier values are those of homophobia and thus IU should oppose that on behalf of LGBTA objectors to the law.
Which doesn’t quite get us to making IU a proper agent to oppose the General Assembly, but it does clear the way for my next argument.
Self-defense by the Liberal State
The problem with supposing that IU should not pass a value-judgement on the illiberal—and let’s not make bones about this, it is illiberal to give some couples the option to marry and not others—but rather with the value judgement in the first place. The General Assembly should not have the option of taking benefits from my coworkers’ partners on the grounds of sexual orientation alone. In the idealized liberal state, this would be a settled matter—against the rules. Alas, we don’t live in such a world.
It does not follow that in such a case the best course of action is to do as we would in the idealized liberal state. Mr. Studebaker would say this opens a “Pandora’s Box”. Nay, the General Assembly opened it by putting forth an illiberal law with a credible chance of it passing. We are here to discuss what the proper course for IU is in light of that.
The argument follows as (and perhaps generalizes from) self-defense. In civil society, it is agreed that violence is not acceptable. However, it is also generally agreed that once a breach of that occurs, it is not improper to defend oneself. Anything else is an admission that life is not worth defending.
The same is true of liberal institutions. If their integrity as a liberal organization is threatened, it is necessary insofar as we agree liberalism is worth protecting that they protect themselves. The use of otherwise illiberal means becomes acceptable.
As shown above, there is no purely liberal path forward. IU’s endorsement of a position betrays members of the IU community regardless. Abstaining is to endorse the status quo. Given it is in an intractable choice because of the General Assembly’s illiberal action, we must decide what the best illiberal action for IU is.
IU ought to defend the liberal values of the state, even if other organs of the state won’t. Among this is the right to equal protection for all folks, even if that offends some of IU’s students, staff, and faculty’s sensibilities. IU’s statement that equality and tolerance are Hoosier values, even over the free objection of Hoosiers, is completely in line with IU’s mandate as a liberal institution in a free society.
There is a final piece that I feel should clinch this, though what has passed is sufficient. The harms here are not symmetric. Indiana’s universities are choosing between opposing substantive harm to students and staff against a fairly narrow condemnation of illiberal action. IU has chosen to oppose harm by affirming its commitment to the liberal program. Purdue has chosen to remain quiet when institutional inertia is presently pushing us towards those harms in the name of illiberal objections.
And so we find my argument up to this point. There is no “neutral” choice in a world already defined. Rise Against put this exceptionally well: “Neutrality means you don’t really care/Because the struggle goes on even when you’re not there.” IU has not started a campaign to persecute its homophobes in any way, but rather to protect it’s LGBT members from homophobia. That is the duty of organizations with a liberal commitment when faced with an illiberal threat.
To resolve the tension Mr. Studebaker has found on behalf of illiberal objectors, as Purdue has done, is a greater threat to liberalism than opposing erosion to the liberal framework of the state. IU, and not Purdue, has said the right thing.
*What does “substantially similar” mean? Well, it’s so vague that several avowedly anti-gay members of the General Assembly voted against HJR-6 on the grounds that such a loose wording had no business in Indiana’s highest law. If you’re a Hoosier, that is grounds to oppose this even if you support a law outlawing SSM and civil unions.
**Self-selection being what it is, this is none-to-surprising, but can I remind the audience that the Christian values Indiana’s legislature has gotten ahold of look nothing like the values of my Christian friends? I know this gets couched in the language of Christian values, but it’s worth at least a footnote that these are homophobic values, not categorically religious ones.
I finally sat down and read The Guardian’s piece “Why have young people in Japan stopped having sex?” by Abigail Hayworth. It has some problems, but compared to virtually all Western journalism about Japan, it is quite excellent. The problem posed by the heading is intrinsically fascinating. Perhaps most fascinating for me is that this is a clear consequence of the 1991 asset crash in Japan. More, it offers an important warning for many Western nations.
In the late 80s, asset prices in Japan soared and soared quickly. If you put 10,000 Yen in the stock market at the end of 1988, you would have had 18,520 just a little over 2 years later. Such growth begat investment which begat growth and so on. 9 months later you would have lost all that and more if you’d stayed in the market. (Of course, you’d have lost it because so many people left and perhaps you’d have left before the music stopped.) This is a fairly common phenomenon—indeed, the United States just went through a similar event in 2008.
Exactly how what happened next played out is to this day a source of huge academic disagreement. But, best as I can, here is my understanding. As the crash happened, people began buying less—if you’re losing money on stocks, you don’t buy goods! As the circular effects began to weigh on the economy, the Bank of Japan began selling more currency, making money less scarce. That dropped the price of money and disincentivized savings. Banks found it easier to get ahold of cash, so why pay savers more for it? In theory, savers were induced to become consumers. In practice this effect wasn’t nearly as great as the Central Bank hoped.
The controversy is often as to whether or not the problem was with the theory or the magnitude of the interest rate drop. Did the Bank of Japan not drop interest rates fast enough or was the Central Bank wrong to drop interest rates? For what it is worth, it is my position that the Bank of Japan should have dropped rates faster and the Diet should have increased government spending.
Whatever the case, the Bank of Japan effectively ran out of space by 1995, as interest rates approached the barrier of 0%. This, combined with a fiscal authority unwilling to engage in stimulus, meant the economy was adrift. There was nothing anyone willing could do to stop Japan from losing years to stagnant growth. By 1995, the falling tax base necessitated the government begin taking on significant debts. The stock market has never recovered, interest rates have never risen, and GDP growth remains low.
It has not been so much a Ushinawareta Juunen, a lost decade, but rather we are approaching 22 years of Japan adrift.
Sex on the Zero-Bound
The effect of this stagnation has not only been macroeconomic. In 1988, Japan’s per capita income actually surpassed the United States. By 1995, the gap was the largest it would ever be—proof that lowering the interest rates were good for something. By 1998, the gains were shrinking, and after stagnated. (2007 marked a turnaround, for those keeping score, with Japan growing again. The Ushinawareta Juunen could be dated 1995-2007 by this measure, later and shorter than often ascribed, but still enough to make a lasting impact on the citizenry of Japan.)
In simple terms, post-1995 Japan has been one where its people could not hope to make more money next year.
That is the backdrop against which the Guardian piece is set. Like in the United States, the Japanese experience has been that the younger generations with no lines on their resumes have borne the brunt of unemployment. More, dates are expensive. Weddings are expensive. Kids are expensive. Relationships and families, therefore, are something that the subjects of the Guardian piece cannot afford.
Let me be clear, boiling these kinds of generational changes down to macroeconomic trends is a tad simplistic. But nonetheless, I think it is important to keep this context in mind as we discuss the Japanese experience. If per capita GDP growth stalls, if we cannot be sure we can afford the expenses associated with long-term commitment, we will not make them. The Guardian piece, even with its over-reliance on a quirky sex-therapist, gives flesh to that generalization.
Further, and perhaps most saliently for many of my readers, Japan offers a glimpse into the future of the United States. We are a country with an aging populace, deep debts, zero-bounded interest rates, and a fiscal authority unwilling and increasingly unable to do what it will take to avoid drift. It is not hard to imagine that in 2030 Japan will be reading about how Gen Y and our successors have married at a lower rate and are eschewing sex.
A nation that is not fruitful will not multiply.
I was working on a question about government spending and investment when this graph popped out:
If you’ve been conscious as of late, you might have heard the ongoing trend was that government spending was on the rise, not at historic lows. I halted work on my question assuming something was wrong with my dataset. After all, Wikipedia has government spending around 40%.
I realized several minutes into doubting my dataset that I was wrong to do so. It’s true, the national account Government Spending (G) has been falling—but only because of a quirk of national accounting. The mechanical reason is that the estimates usually cited for government spending include transfer payments like welfare, social security, etc. The national accounts include those in Consumption. This is all very elementary and I felt silly for forgetting it. Still, a good reminder for everyone that the government is big, and that the larger estimates include a diverse range of expenditures.
It’s also a reminder of how the government has changed. The services that used to be the bread and butter of government spending have shrunk relative to GDP. Entitlement spending—which is almost always transfer spending—has increased. Needless to say, my feelings on this vary program to program.
Which measure you use depends mainly on the kind of question you’re asking. Want to know the economic burden (in terms of taxes and debt)? The larger 40% measure is your game. Want to know what kind of direct impact the government is having on the goods and services market? The national accounts are your go-to.
It’s a good reminder that even basic facts, like the size of the government, are subjective questions open to interpretation and discussion.