Isn’t (1) more or less just regression to the mean?
Travis Downssays:
(1) is true only for a very strange definition of “did better”. That is, if you define “did better” in the specific relative way: “the probability they did better than their parents”. I don’t think that’s how anyone defines it in common usage! If a lawyer making $500/hr has a son who ends up making $450/hr, whereas an Amazon warehouse worker making $10 has a daughter who ends making $15, no one would reasonable say the warehouse working is “doing better” than the lawyer!
It is almost a mathematical certainty that the riches of the rich will have children who are less prolific earners, and highly likely that people at the very bottom of the income scale will have children who end up making somewhat more than their parents – simply by regression to the mean. You could show the same “effect” showing that children of the shortest parents “do better” in height than children of the tallest parents, and so on for almost any tangible factor that you can assign to people that tends not to run away to infinity or zero.
I think the author actually has a good point that following particular people gives a different view on the issue than looking at quintile snapshots over time, but it shouldn’t be presented as “the poor are doing better than the rich”, but rather there is still non-zero income mobility and hence regression to the mean is still a thing.
One can easily make the point that the top earners are taking a huge portion of the income pie: and pointing out that the top earner quintile isn’t entirely static over time doesn’t invalidate that.
One can easily make the point that the top earners are taking a huge portion of the income pie: and pointing out that the top earner quintile isn’t entirely static over time doesn’t invalidate that.
In such matters, it is useful to quantify:
It turns out that 12 percent of the population will find themselves in the top 1 percent of the income distribution for at least one year. What’s more, 39 percent of Americans will spend a year in the top 5 percent of the income distribution, 56 percent will find themselves in the top 10 percent, and a whopping 73 percent will spend a year in the top 20 percent of the income distribution.
Maybe you knew about these numbers and they are what you expect. I don’t think that they are what I expected. I submit to you that many people would be surprised by these numbers.
Travis Downssays:
Those numbers aren’t at all what I’d expect if we were talking about “recurring” employment or business income.
However, my first inclination would be to wonder if it is using a more broad definition of income, such as what the IRS uses, which includes capital gains such as the sale of stock, real estate or small business.
It woudn’t surprise me to find out that 12% of the population dispose of property, such as a house, that puts them in the top 1% (a bar of only a few hundred thousand) at least once in their life, and similarly for the similar statistics (and for the lower thresholds you are in the range of one-time payouts for things like severance payments, pension commutation, etc).
The fact that this income is recognized in a single year despite usually accumulating over many many years is more a quirk of the income tax system than any deep insight on income mobility.
However, if the study excluded capital dispositions and other one-off events of a similar nature, and the figures were as above, then indeed I would be very surprised!
Travis Downssays:
… but the point is taken: talking about the distribution between the 1% and the 99% or whatever other division for a single year is not very illuminating if every year there is a large amount of flux between the groups so that the actual income per person, averaged over some years, is much more equitable.
Rather than second order statistics such as “12% of people are in the 1% at least once in their lives”, we could just calculate what we want directly: take the lifetime income of the population at a moment in time and give the statistics on that, so the 1% are those with the highest 1% lifetime income. Of course, this is kind of hard because of the population demographics: how do you snapshot lifetime income of a population ranging in age from 1 to 110 or whatever? That difficulty is probably one reason you don’t see that, but it seems solvable.
Why would we care about differences in lifetime income? The larger point is why should we care about differences in wealth or income. This is rarely spelt out, but I think it needs to be spelt out if you are going to be able to argue in favor of a specific metric.
That is, before you can argue that a metric is the right one, I think you have to define the problem.
That’s not to say that things like the gini coefficient are uninteresting… but who is to say that it is the “right” measure?
Travis Downssays:
Well I feel like this larger point is outside of the scope of either article, and falls in the obvious and/or implicit category.
To be explicit then: I think the basic idea is to understand if the fruits of labor and capital and whatever else are being distributed “fairly” or “evenly enough” or whatever term you’d like to use. Now this is obviously a touchy, political subject, fraught with complications, and hard to come up with any specific answer – but one can simply accept that many or most people don’t want the distribution to be too uneven. These people would think that a statistic like “50% of the wealth of a country accumulates to the top 1% of it’s population” or that “a CEO makes in 1 second what an employee makes in a month” are problematic (these are totally made up).
Without taking a particular stance on what the correct number is, one can see that some ways of measuring this make a lot more sense than others. If 50% of the wealth accumulated to the top 1%, but the top 1% rotated around every minute so that over 100 minutes every got their kick at the golden goose, even the most communist among us would probably be satisfied! There is some sense “massive inequality” – but only on very short timescales, and when you average over any timescale that matters, it’s all dead even.
That’s my impression of the tack this article and other similar articles are taking: they are arguing that at any moment in time there is a large amount of inequality – but that this is deceptive because there is a large amount of flux in the income between measurements, so when averaged over a lifetime or something reasonable, everything is more even. This argument has some merit – but presenting metrics relating to the chance someone has of being in the top N% over one year of their life seems like it obfuscates what you’d like to know: which I think is more something like “lifetime income” or perhaps income averaged over 10 or 20 years, or some more complex metric, like total lifetime income integrated against some utility function or whatever.
Essentially I’m saying the most interesting metric for “fairness” advocates is probably somewhere in the middle between one year population snapshots and metrics like “% of earning more than parents” or “chance of having a single high earning year”.
many or most people don’t want the distribution to be too uneven
Maybe, but is that a good thing?
I call this a culture of envy and I have been advocating that we live differently. I submit to you that envy makes people unhappy.
Travis Downssays:
It’s hard to defend envy per se, since that word already carries a lot of negative connotations, but I think it is perfectly rational in some cases to measure yourself against others, when the gains of others serve as a good signal for what your fair share is.
Eyeing your neighbors BMW and being envious of it isn’t particularly rational since there is nothing inherently unfair about your neighbor buying a BMW and likely you have very little in common economically with your neighbor – you just happen to live next to each other.
On the other hand, if you found out that all your colleagues, who perform identical work to you (probably hard to measure in your profession, so maybe imagine another one), were paid 10 times the amount? Something like “envy” is entirely rational here: not because you begrudge your colleague the money, but because you know your work was apparently “worth” 10 times what you were being paid and apparently you were duped out of it, due to poor negotiation skills, information asymmetry, whatever.
One can construct other scenarios where something like “envy” is somewhat rational, at least if you take a simple form of fairness as a desirable – when the wealth of others serves as a signal for the size of the pie that is being split up, and lets you calculate that you aren’t getting your share.
The question then is are statistics about income percentiles that we were discussing more like the “neighbors BMW” case or the “colleague paid 10x for identical work” case? That depends partly or entirely on your views on redistribution and other economic factors (of course, you might be fine with the 10x scenario – but I think that would make you at outlier and with a big hill to climb to convince most people otherwise).
In a market economy, if others are paid $X for their services, and you suddenly learn that you were charging $X/10, then you should quickly start charging more.
BTW I’d be super happy to learn than my colleagues earn 10X as much as I do. That’d make my salary negotiations quite easy.
But that’s not what country-wide inequality is. It is not the case that unemployed carpenters can charge as much as Harvard-educated surgeons, if only they could get around to it.
You have used the language of the pie on several occasion. This assumes that we are playing a finite game. That is, if you earn more, then others have to earn less.
But it does not work like that. If you pay Harvard-educated surgeons half as much, you won’t see the income of unemployed carpenters rise.
It is not a finite-sum game.
Oh. It can be. That’s what rents are. Rents are bad.
If the US is more unequal because of rents… I’d love to see the research on that.
Travis Downssays:
BTW I’d be super happy to learn than my colleagues earn 10X as much as I do. That’d make my salary negotiations quite easy.
Sure, yes – if that is possible in your scenario. In some cases the opportunity may have passed, negotiation may not be possible, whatever.
But that’s not what country-wide inequality is. It is not the case that unemployed carpenters can charge as much as Harvard-educated surgeons, if only they could get around to it.
Correct, which is what I was getting at at the end of my post. It is clear that different people of different abilities, education and a long list of other questions will generally receive different incomes, but the question is how different? The surgeon will probably make more than the carpenter, but how much more? 50% more? Three times more? A million times more?
You have used the language of the pie on several occasion. This assumes that we are playing a finite game. That is, if you earn more, then others have to earn less.
But it does not work like that. If you pay Harvard-educated surgeons half as much, you won’t see the income of unemployed carpenters rise.
I don’t understand “finite game” in this context, but it kind of does work like that, on a “local” and “instantaneous” basis, doesn’t it? Of course, there is no particular relationship between the surgeon and the carpenter, so I wouldn’t expect any income decrease for the surgeon to go directly to the carpenter, but it goes to someone. Perhaps it goes to other doctors, non-surgeons or not educated at Harvard. Perhaps it lines the pockets of the hospital owners, or health insurance companies. Perhaps it is spread around, bit by bit, among all the people who pay less for surgeries (ha!).
Where these decisions are made, it is often very much a zero sum game, at least in an instantaneous sense. Consider a corporation trying to allocate its costs among a huge variety of different employees, ranging from menial jobs, to various professional and executive-type workers, up through the C-suite and, ultimately, the owners. That’s a fixed pie to slice up – anyone getting more means someone getting less. Of course, this is true only instantaneously and locally: if you decide to pay everyone the same today (perhaps with some fixed owner vs labor split), regardless of position, then tomorrow you might find yourself without a lot of the high skilled workers you need to earn the revenue you decided to split up so evenly.
The same kind of pattern repeats itself throughout the economy on larger scales: one could certainly decide to even everything out via taxation and redistribution – again, it’s a fixed pie “instantaneously” – but go too far and you just make almost everyone poorer.
So I guess a relevant question is what is the current distribution the most efficient, considering the dynamic nature of the system? If there is a more efficient distribution, is it achievable in practice? Some might argue that it is worth making the pie smaller, if you can change the distribution such that most people are richer (in an absolute sense).
Some might argue that it is worth making the pie smaller (…)
In most of our economies, the pie is growing at a rate faster than the population, so that the pie per capita is growing. In fact, it is essentially growing exponentially (a percentage point or more per year).
So I submit to you that it is a bad model to consider the pie as something fixed to be distributed.
It is not even close to a zero-sum game when you consider the economy as a whole. It is exponentially growing pie and we have good reasons to think that policies and cultural frames of mind have a direct impact on the size of the growth. We know this because some countries with equally smart people and with great natural ressouces have far, far lower economic growth over time than we do in North America. We cannot exactly control economic growth but we know it is the result of human behavior (new wealth does not come from trees or the ground).
So I reject the notion of a fixed pie that we can shrink or redistribute.
A much more reasonable question to ask is “how much year-on-year growth are you willing to sacrifice to get a more equal distribution”. My answer to that question, I will argue, is none.
Some might have a mental model of Jeff Bezos became wealthy at the expense of others. However, it is much more likely that Jeff Bezos contributed to making the pie X% larger, and he got a fraction of this X% as a compensation. On the whole, the society did not become poorer because Jeff Bezos got richer, the opposite is true. If you take out Jeff Bezos and his wealth from the pie, then the rest of the pie got larger.
We don’t know very well how to increase economic growth, but we sure know one way to reduce it: making life harder for people like Jeff Bezos.
It is simply a fact that the most productive people contribute disproportionally to economic growth. I like carpenters and cashiers will enough, but they typically contribute little to per-capita growth. Entrepreneurs contribute a lot more. (I use the general definition of an entrepreneur.)
Now, people are often quite willing, in principle, to argue that we should reduce growth. Who cares about growth?
What to combat climate change? Let us drop huge taxes on carbon… so what if it lowers the economic growth!
But there are a few things to consider.
Growth is an exponential factor. This changes all of the math. It seems that people are not too excited by the 4.1% GDP growth in the US last quarter. Who cares, right?
But the difference in the size of the pie between a 1% growth and a 4% growth over time is enormous. If you live in the second society, over time, you will be better off even if the distribution is a lot more uneven (as long as the distribution does not become less even with time).
Between WW2 and the fall of the Berlin wall, East Germany saw an economic growth of about 4% (for the entire set of decades) whereas West Germany had 4% per year for decades. The net result is that in 1990, even the people at the top of the East German society were poorer than almost anyone in West Germany. Had the wall not fallen, East Germans would probably still be driving cars designed in the 1950s, they would probably still frequently live in appartements without a private bathroom.
As long as wealth distribution remains more or less the same, there is almost no amount of real per capita growth that you should be willing to sacrifice. And even under an increasing dispersion of the wealth, if the growth of the pie per capita is sufficient, you are still better off. The power of an exponential is just tremendous.
Some necessary and useful expenses are already growing at a rate faster than overall economic growth. Health certainly is, and so is education. It should be immediately clear, from first principles, that no service can continuously grow at a rate faster than the whole economy without growing as a fraction of the whole economy. There is serious talk out there about how it is fine for a mature economy to spend 80% of its wealth on health. I don’t disagree per se, but even if you agree with the 80% figure, you have to realize that once you reach 80%… growth at a rate faster than the whole economy is no longer mathematically feasible. So it must stop somewhere. There are basically two kinds of solutions to achieve a sustainable outcome. Firstly, you reduce growth in health-related expenses. And so we are clear, cutting off health benefits for the top 1% won’t do it. If you are going to bring down the growth of health-related expenses, it is going to affect most people. Secondly, you keep economic growth as high as possible.
The larger point I am making is that if you try to lower economic growth, it will force you into making hard choices that will affect everyone far sooner than you’d like.
I am sure that, like me, you believe that everyone should have access to the best healthcare there, in the sense that if a debilating disease can be cured, or a condition controlled, you want people to receive the therapy, irrespective of the cost. That’s an ideal to aim for. I am sure you agree that the best way to achieve this result is to aim for an economic growth that is at least as high as the rise in the cost of medical care.
Zuckerberg wants to wipe out all diseases. I agree with his sentiment. I think we should all agree with it. But it is much easier to achieve this goal in a society that has more than enough to pay for medical bills.
Conclusion.
Maybe the ideal system would be one where wealth is evenly distributed and growth is fantastic. We don’t know how to build something like that. But given a choice between more growth and more equality, we should almost always go with more growth.
It is a lot easier to wipe out poverty with a 4% annual per capita real growth than with a 4% growth over four decades. There is just no comparison.
Exponentials win out in the end.
========
As an aside, I often have arguments on this blog about the fact that people living longer lives is a bad thing because it leads to overpopulation.
It is same kind of problem. People think nothing of increasing the number of children per woman by 50%. There are countries where there are policies with such goals. But people are very concerned about the thought that we could increase lifespan by 50%.
It is bad math! The number of children per women is an exponential factor, the lifespan is a linear factor. Double the lifespan, you double the people… double the number of children per women, you jack up the exponential growth.
That’s why countries like Japan and Germany, despite setting records in longevity, have falling populations… whereas the countries where the longevity is greatest often have lurking overpopulation problems.
===========
I want to come back on the notion that envy can be just rational. I want to fight back against this notion. It is tempting to rationalize envy, but I think we must resist.
I don’t think it is. I think it is a evolved instinct that bring us back to a time when most of us were starving. If your neighbor had a lot more meat than you did, it might genuinely mean that you (or your kids) might die soon.
If Jack gets a lot of sex and you get none, this means that he will have offspring whereas you won’t.
So there are good evolutionary reasons for us to be upset when others have more.
Animals are envious all the time.
Travis Downssays:
I understand that future growth is related to distribution choices you make today and acknowledged that explicitly. I never said the pie is fixed (that was your idea based on my use of the word pie) – only that for practical purposes it is instantaneously fixed at any given moment. If you grow 100 bushels of corn this year, you have exactly 100 bushels of corn – regardless of the fact that due to growth you might have 1000 bushels 10 years from now. You still need to decide now to distribute those 100 bushels (and yes, your decisions might affect how much corn you have in the future). I think we agree on this relationship and I agree with everything you say until, but not including My answer to that question, I will argue, is none.
Exponentials win out in the end, but we have finite lifespan so a reasonable (and self-interested) person may prefer more equality now and less growth, since the difference in growth is at some point too slow to make up for the more skewed distribution over their lifetime. Most people (I think!) are not playing the really long game where they care about getting to some technological singularity 1,000 years from now: they mostly care about the world they live in, and perhaps their children, etc.
Of course if you can choose between 1% and 4%, the difference is staggering, even over a typical lifetime. That was the difference between a well-functioning, in some senses world-leading economy and a totally disfunctional system. The types of choices we usually debate in modern western economies on other the hand are probably a small fraction of a percent: even that still “trumps everything” over time, but a long time…
The actual numbers do matter here: just like an exponential algorithm can beat a polynomial one for many values of N, maximum growth might take “too long” to benefit the individual.
The growth numbers we are talking about for modern economies are in the low single-digit percents. Over a 60 year “adult” lifetime (i.e., live to 80 but for 20 years you don’t care about this stuff or can’t vote, etc), 2% growth is only a factor of 3.3 and 3% growth is only a factor of 5.9. So you are not exactly in the realm that the exponential overwhelms everything else even by the end of a lifetime, at least at the poor end of the scale. Unsurprisingly, these growth factors have been dropping over time as our ingenuity hasn’t been able to keep the exponential trend steady – so 2% or 3% seems like a reasonable future estimate.
Imagine the two points: the instantaneously optimal[1] distribution without considering future growth, and the highest growth distribution. I submit to you that the ideal is somewhere between those points, not at either end, and the ideal will also vary by individual: based on things like your expected remaining lifespan, how much you care about the world you leave to the future generations, how much you believe growth as usually measured actually flows through to human well-being (or whatever factor you are optimizing for), etc.
Furthermore, the distribution matters even if you want to optimize for the really long run – if growth is maximized and exponential, but all the growth accrues only to a small portion of the population, then we become “exponentially wealthy” only on average. That is, as you point out, your conclusions only hold if a large part of the population doesn’t get a smaller slice of the pie. Who is to say that the maximum growth rate strategy would hold their slice of the pie constant? I think there is already some evidence it won’t.
I’m not sure about Bezos specifically since his wealth is based on the price of Amazon stock, which is somehow created “out of thin air” and includes predictions of future growth, but it is obvious to me that on the retail side Amazon’s revenues come at the expense of others being poorer (fewer revenues). That is, if Amazon sells 200 billion of goods a year, some (most) of that comes out of the pockets of other sellers, who are poorer for it. Perhaps not all of it, because the increased convenience and efficiency may increase the size of the pie (total sales), but that effect is small in an instantaneous sense: if you take the Amazon sales out, the remaining pie is probably not larger (but growth might be higher).
There’s no problem with that! That’s how the market works, and this kind of thing should be encouraged. People are buying more from Amazon, and less from Walmart or Sears or whatever because it offers a better experience. You don’t need to pretend that all of Amazon’s growth came out of thin air, rather than at the expense of others, for your argument to work: you just point out that Amazon succeeding increases the total size of the pie, and future growth and here we agree.
One of the big problems I see in various economic decisions is the idea that “everyone wins” with any particular decision. Like “free trade” is a win for everyone. Of course not! Free trade is a good idea and I’m a strong component, but it is obvious to me that there specific people, regions and industries within some countries that are big losers. The small, disperse benefit of free trade will never make up a 3 or 4-fold reduction in earning power. Rationally, those people should be against free trade, since it’s strictly worse for them, over their lifetime – unless you insist they should show a level of altruism far beyond normal.
So I think it’s fine to say that Bezos’ wealth came at the expense of some others: it certainly did to some extend (and those people were often super-rich themselves) in the direct sense that Amazon disrupted many existing businesses. However, the kind of society where Bezos can get super rich and disrupt all those industries has high growth and people may be better off on average.
Finally, there seems to be an assumption (maybe I’m wrong) that inequality and high-growth go hand-in-hand: that as you move towards the high-growth end of the scale, equality necessarily decreases since the highest growth strategy is allocate most resources to a limited number of the most skilled entrepreneurs. I think this is true only up until a point: things start to fall apart in various ways, politically and practically at some point. So even if one subscribes to your notion of optimizing for maximum growth at any cost, it worth investigating which side of the maximum growth peak we are on: perhaps some places are already on the far side of the peak and it is possible to increase growth and increase equality at the same time.
[1] I’m just sidestepping the issue of what the “best” instantaneous distribution is, but imagine its something on the Pareto frontier that achieves the best average utility over the population with some utility function, and some reasonable estimate of the instantaneous impacts of the selected distribution. Basically what you get with “1 person, 1 vote” and perfectly rational voters and then just hand-waving away all the game-theoretical problems with voting (or perhaps invoking a benign dictator).
Finally, there seems to be an assumption (maybe I’m wrong) that inequality and high-growth go hand-in-hand: that as you move towards the high-growth end of the scale, equality necessarily decreases since the highest growth strategy is allocate most resources to a limited number of the most skilled entrepreneurs. I think this is true only up until a point: things start to fall apart in various ways, politically and practically at some point.
I don’t think that’s what I wrote. At least, that’s not what I meant.
The question I was asking was whether you should be willing to sacrifice economic growth to reduce inequality. I think you should not.
Travis Downssays:
I don’t think that’s what I wrote. At least, that’s not what I meant.
The question I was asking was whether you should be willing to
sacrifice economic growth to reduce inequality. I think you should
not.
Right, then we can just prune that branch of the discussion then.
It does raise the question though: if reducing inequality increases growth, why bother discussing the hypothetical tradeoff between growth and equality on the other side of the peak? After all, both sides now want to move in the same direction, just for different reasons.
It’s still interesting, sure – but if doesn’t apply to a particular country/region/whatever it’s mostly academic.
Scott Hesssays:
WRT #2, maybe you decided to become vegetarian because you were searching for something which wasn’t otherwise present in your life?
Mesays:
Searching for a purpose in life (which likely is related to depression) may make you more likely to become a vegetarian, I guess. Correlation isn’t causation – you don’t know the direction, or whether there is a common cause explaining both factors.
Isn’t (1) more or less just regression to the mean?
(1) is true only for a very strange definition of “did better”. That is, if you define “did better” in the specific relative way: “the probability they did better than their parents”. I don’t think that’s how anyone defines it in common usage! If a lawyer making $500/hr has a son who ends up making $450/hr, whereas an Amazon warehouse worker making $10 has a daughter who ends making $15, no one would reasonable say the warehouse working is “doing better” than the lawyer!
It is almost a mathematical certainty that the riches of the rich will have children who are less prolific earners, and highly likely that people at the very bottom of the income scale will have children who end up making somewhat more than their parents – simply by regression to the mean. You could show the same “effect” showing that children of the shortest parents “do better” in height than children of the tallest parents, and so on for almost any tangible factor that you can assign to people that tends not to run away to infinity or zero.
I think the author actually has a good point that following particular people gives a different view on the issue than looking at quintile snapshots over time, but it shouldn’t be presented as “the poor are doing better than the rich”, but rather there is still non-zero income mobility and hence regression to the mean is still a thing.
One can easily make the point that the top earners are taking a huge portion of the income pie: and pointing out that the top earner quintile isn’t entirely static over time doesn’t invalidate that.
One can easily make the point that the top earners are taking a huge portion of the income pie: and pointing out that the top earner quintile isn’t entirely static over time doesn’t invalidate that.
In such matters, it is useful to quantify:
Source: New York Times
Maybe you knew about these numbers and they are what you expect. I don’t think that they are what I expected. I submit to you that many people would be surprised by these numbers.
Those numbers aren’t at all what I’d expect if we were talking about “recurring” employment or business income.
However, my first inclination would be to wonder if it is using a more broad definition of income, such as what the IRS uses, which includes capital gains such as the sale of stock, real estate or small business.
It woudn’t surprise me to find out that 12% of the population dispose of property, such as a house, that puts them in the top 1% (a bar of only a few hundred thousand) at least once in their life, and similarly for the similar statistics (and for the lower thresholds you are in the range of one-time payouts for things like severance payments, pension commutation, etc).
The fact that this income is recognized in a single year despite usually accumulating over many many years is more a quirk of the income tax system than any deep insight on income mobility.
However, if the study excluded capital dispositions and other one-off events of a similar nature, and the figures were as above, then indeed I would be very surprised!
… but the point is taken: talking about the distribution between the 1% and the 99% or whatever other division for a single year is not very illuminating if every year there is a large amount of flux between the groups so that the actual income per person, averaged over some years, is much more equitable.
Rather than second order statistics such as “12% of people are in the 1% at least once in their lives”, we could just calculate what we want directly: take the lifetime income of the population at a moment in time and give the statistics on that, so the 1% are those with the highest 1% lifetime income. Of course, this is kind of hard because of the population demographics: how do you snapshot lifetime income of a population ranging in age from 1 to 110 or whatever? That difficulty is probably one reason you don’t see that, but it seems solvable.
Why would we care about differences in lifetime income? The larger point is why should we care about differences in wealth or income. This is rarely spelt out, but I think it needs to be spelt out if you are going to be able to argue in favor of a specific metric.
That is, before you can argue that a metric is the right one, I think you have to define the problem.
That’s not to say that things like the gini coefficient are uninteresting… but who is to say that it is the “right” measure?
Well I feel like this larger point is outside of the scope of either article, and falls in the obvious and/or implicit category.
To be explicit then: I think the basic idea is to understand if the fruits of labor and capital and whatever else are being distributed “fairly” or “evenly enough” or whatever term you’d like to use. Now this is obviously a touchy, political subject, fraught with complications, and hard to come up with any specific answer – but one can simply accept that many or most people don’t want the distribution to be too uneven. These people would think that a statistic like “50% of the wealth of a country accumulates to the top 1% of it’s population” or that “a CEO makes in 1 second what an employee makes in a month” are problematic (these are totally made up).
Without taking a particular stance on what the correct number is, one can see that some ways of measuring this make a lot more sense than others. If 50% of the wealth accumulated to the top 1%, but the top 1% rotated around every minute so that over 100 minutes every got their kick at the golden goose, even the most communist among us would probably be satisfied! There is some sense “massive inequality” – but only on very short timescales, and when you average over any timescale that matters, it’s all dead even.
That’s my impression of the tack this article and other similar articles are taking: they are arguing that at any moment in time there is a large amount of inequality – but that this is deceptive because there is a large amount of flux in the income between measurements, so when averaged over a lifetime or something reasonable, everything is more even. This argument has some merit – but presenting metrics relating to the chance someone has of being in the top N% over one year of their life seems like it obfuscates what you’d like to know: which I think is more something like “lifetime income” or perhaps income averaged over 10 or 20 years, or some more complex metric, like total lifetime income integrated against some utility function or whatever.
Essentially I’m saying the most interesting metric for “fairness” advocates is probably somewhere in the middle between one year population snapshots and metrics like “% of earning more than parents” or “chance of having a single high earning year”.
many or most people don’t want the distribution to be too uneven
Maybe, but is that a good thing?
I call this a culture of envy and I have been advocating that we live differently. I submit to you that envy makes people unhappy.
It’s hard to defend envy per se, since that word already carries a lot of negative connotations, but I think it is perfectly rational in some cases to measure yourself against others, when the gains of others serve as a good signal for what your fair share is.
Eyeing your neighbors BMW and being envious of it isn’t particularly rational since there is nothing inherently unfair about your neighbor buying a BMW and likely you have very little in common economically with your neighbor – you just happen to live next to each other.
On the other hand, if you found out that all your colleagues, who perform identical work to you (probably hard to measure in your profession, so maybe imagine another one), were paid 10 times the amount? Something like “envy” is entirely rational here: not because you begrudge your colleague the money, but because you know your work was apparently “worth” 10 times what you were being paid and apparently you were duped out of it, due to poor negotiation skills, information asymmetry, whatever.
One can construct other scenarios where something like “envy” is somewhat rational, at least if you take a simple form of fairness as a desirable – when the wealth of others serves as a signal for the size of the pie that is being split up, and lets you calculate that you aren’t getting your share.
The question then is are statistics about income percentiles that we were discussing more like the “neighbors BMW” case or the “colleague paid 10x for identical work” case? That depends partly or entirely on your views on redistribution and other economic factors (of course, you might be fine with the 10x scenario – but I think that would make you at outlier and with a big hill to climb to convince most people otherwise).
In a market economy, if others are paid $X for their services, and you suddenly learn that you were charging $X/10, then you should quickly start charging more.
BTW I’d be super happy to learn than my colleagues earn 10X as much as I do. That’d make my salary negotiations quite easy.
But that’s not what country-wide inequality is. It is not the case that unemployed carpenters can charge as much as Harvard-educated surgeons, if only they could get around to it.
You have used the language of the pie on several occasion. This assumes that we are playing a finite game. That is, if you earn more, then others have to earn less.
But it does not work like that. If you pay Harvard-educated surgeons half as much, you won’t see the income of unemployed carpenters rise.
It is not a finite-sum game.
Oh. It can be. That’s what rents are. Rents are bad.
If the US is more unequal because of rents… I’d love to see the research on that.
Sure, yes – if that is possible in your scenario. In some cases the opportunity may have passed, negotiation may not be possible, whatever.
Correct, which is what I was getting at at the end of my post. It is clear that different people of different abilities, education and a long list of other questions will generally receive different incomes, but the question is how different? The surgeon will probably make more than the carpenter, but how much more? 50% more? Three times more? A million times more?
I don’t understand “finite game” in this context, but it kind of does work like that, on a “local” and “instantaneous” basis, doesn’t it? Of course, there is no particular relationship between the surgeon and the carpenter, so I wouldn’t expect any income decrease for the surgeon to go directly to the carpenter, but it goes to someone. Perhaps it goes to other doctors, non-surgeons or not educated at Harvard. Perhaps it lines the pockets of the hospital owners, or health insurance companies. Perhaps it is spread around, bit by bit, among all the people who pay less for surgeries (ha!).
Where these decisions are made, it is often very much a zero sum game, at least in an instantaneous sense. Consider a corporation trying to allocate its costs among a huge variety of different employees, ranging from menial jobs, to various professional and executive-type workers, up through the C-suite and, ultimately, the owners. That’s a fixed pie to slice up – anyone getting more means someone getting less. Of course, this is true only instantaneously and locally: if you decide to pay everyone the same today (perhaps with some fixed owner vs labor split), regardless of position, then tomorrow you might find yourself without a lot of the high skilled workers you need to earn the revenue you decided to split up so evenly.
The same kind of pattern repeats itself throughout the economy on larger scales: one could certainly decide to even everything out via taxation and redistribution – again, it’s a fixed pie “instantaneously” – but go too far and you just make almost everyone poorer.
So I guess a relevant question is what is the current distribution the most efficient, considering the dynamic nature of the system? If there is a more efficient distribution, is it achievable in practice? Some might argue that it is worth making the pie smaller, if you can change the distribution such that most people are richer (in an absolute sense).
In most of our economies, the pie is growing at a rate faster than the population, so that the pie per capita is growing. In fact, it is essentially growing exponentially (a percentage point or more per year).
https://goo.gl/images/5CCxaZ
So I submit to you that it is a bad model to consider the pie as something fixed to be distributed.
It is not even close to a zero-sum game when you consider the economy as a whole. It is exponentially growing pie and we have good reasons to think that policies and cultural frames of mind have a direct impact on the size of the growth. We know this because some countries with equally smart people and with great natural ressouces have far, far lower economic growth over time than we do in North America. We cannot exactly control economic growth but we know it is the result of human behavior (new wealth does not come from trees or the ground).
So I reject the notion of a fixed pie that we can shrink or redistribute.
A much more reasonable question to ask is “how much year-on-year growth are you willing to sacrifice to get a more equal distribution”. My answer to that question, I will argue, is none.
Some might have a mental model of Jeff Bezos became wealthy at the expense of others. However, it is much more likely that Jeff Bezos contributed to making the pie X% larger, and he got a fraction of this X% as a compensation. On the whole, the society did not become poorer because Jeff Bezos got richer, the opposite is true. If you take out Jeff Bezos and his wealth from the pie, then the rest of the pie got larger.
We don’t know very well how to increase economic growth, but we sure know one way to reduce it: making life harder for people like Jeff Bezos.
It is simply a fact that the most productive people contribute disproportionally to economic growth. I like carpenters and cashiers will enough, but they typically contribute little to per-capita growth. Entrepreneurs contribute a lot more. (I use the general definition of an entrepreneur.)
Now, people are often quite willing, in principle, to argue that we should reduce growth. Who cares about growth?
What to combat climate change? Let us drop huge taxes on carbon… so what if it lowers the economic growth!
But there are a few things to consider.
But the difference in the size of the pie between a 1% growth and a 4% growth over time is enormous. If you live in the second society, over time, you will be better off even if the distribution is a lot more uneven (as long as the distribution does not become less even with time).
Between WW2 and the fall of the Berlin wall, East Germany saw an economic growth of about 4% (for the entire set of decades) whereas West Germany had 4% per year for decades. The net result is that in 1990, even the people at the top of the East German society were poorer than almost anyone in West Germany. Had the wall not fallen, East Germans would probably still be driving cars designed in the 1950s, they would probably still frequently live in appartements without a private bathroom.
As long as wealth distribution remains more or less the same, there is almost no amount of real per capita growth that you should be willing to sacrifice. And even under an increasing dispersion of the wealth, if the growth of the pie per capita is sufficient, you are still better off. The power of an exponential is just tremendous.
The larger point I am making is that if you try to lower economic growth, it will force you into making hard choices that will affect everyone far sooner than you’d like.
I am sure that, like me, you believe that everyone should have access to the best healthcare there, in the sense that if a debilating disease can be cured, or a condition controlled, you want people to receive the therapy, irrespective of the cost. That’s an ideal to aim for. I am sure you agree that the best way to achieve this result is to aim for an economic growth that is at least as high as the rise in the cost of medical care.
Zuckerberg wants to wipe out all diseases. I agree with his sentiment. I think we should all agree with it. But it is much easier to achieve this goal in a society that has more than enough to pay for medical bills.
Conclusion.
Maybe the ideal system would be one where wealth is evenly distributed and growth is fantastic. We don’t know how to build something like that. But given a choice between more growth and more equality, we should almost always go with more growth.
It is a lot easier to wipe out poverty with a 4% annual per capita real growth than with a 4% growth over four decades. There is just no comparison.
Exponentials win out in the end.
========
As an aside, I often have arguments on this blog about the fact that people living longer lives is a bad thing because it leads to overpopulation.
It is same kind of problem. People think nothing of increasing the number of children per woman by 50%. There are countries where there are policies with such goals. But people are very concerned about the thought that we could increase lifespan by 50%.
It is bad math! The number of children per women is an exponential factor, the lifespan is a linear factor. Double the lifespan, you double the people… double the number of children per women, you jack up the exponential growth.
That’s why countries like Japan and Germany, despite setting records in longevity, have falling populations… whereas the countries where the longevity is greatest often have lurking overpopulation problems.
===========
I want to come back on the notion that envy can be just rational. I want to fight back against this notion. It is tempting to rationalize envy, but I think we must resist.
I don’t think it is. I think it is a evolved instinct that bring us back to a time when most of us were starving. If your neighbor had a lot more meat than you did, it might genuinely mean that you (or your kids) might die soon.
If Jack gets a lot of sex and you get none, this means that he will have offspring whereas you won’t.
So there are good evolutionary reasons for us to be upset when others have more.
Animals are envious all the time.
I understand that future growth is related to distribution choices you make today and acknowledged that explicitly. I never said the pie is fixed (that was your idea based on my use of the word pie) – only that for practical purposes it is instantaneously fixed at any given moment. If you grow 100 bushels of corn this year, you have exactly 100 bushels of corn – regardless of the fact that due to growth you might have 1000 bushels 10 years from now. You still need to decide now to distribute those 100 bushels (and yes, your decisions might affect how much corn you have in the future). I think we agree on this relationship and I agree with everything you say until, but not including My answer to that question, I will argue, is none.
Exponentials win out in the end, but we have finite lifespan so a reasonable (and self-interested) person may prefer more equality now and less growth, since the difference in growth is at some point too slow to make up for the more skewed distribution over their lifetime. Most people (I think!) are not playing the really long game where they care about getting to some technological singularity 1,000 years from now: they mostly care about the world they live in, and perhaps their children, etc.
Of course if you can choose between 1% and 4%, the difference is staggering, even over a typical lifetime. That was the difference between a well-functioning, in some senses world-leading economy and a totally disfunctional system. The types of choices we usually debate in modern western economies on other the hand are probably a small fraction of a percent: even that still “trumps everything” over time, but a long time…
The actual numbers do matter here: just like an exponential algorithm can beat a polynomial one for many values of N, maximum growth might take “too long” to benefit the individual.
The growth numbers we are talking about for modern economies are in the low single-digit percents. Over a 60 year “adult” lifetime (i.e., live to 80 but for 20 years you don’t care about this stuff or can’t vote, etc), 2% growth is only a factor of 3.3 and 3% growth is only a factor of 5.9. So you are not exactly in the realm that the exponential overwhelms everything else even by the end of a lifetime, at least at the poor end of the scale. Unsurprisingly, these growth factors have been dropping over time as our ingenuity hasn’t been able to keep the exponential trend steady – so 2% or 3% seems like a reasonable future estimate.
Imagine the two points: the instantaneously optimal[1] distribution without considering future growth, and the highest growth distribution. I submit to you that the ideal is somewhere between those points, not at either end, and the ideal will also vary by individual: based on things like your expected remaining lifespan, how much you care about the world you leave to the future generations, how much you believe growth as usually measured actually flows through to human well-being (or whatever factor you are optimizing for), etc.
Furthermore, the distribution matters even if you want to optimize for the really long run – if growth is maximized and exponential, but all the growth accrues only to a small portion of the population, then we become “exponentially wealthy” only on average. That is, as you point out, your conclusions only hold if a large part of the population doesn’t get a smaller slice of the pie. Who is to say that the maximum growth rate strategy would hold their slice of the pie constant? I think there is already some evidence it won’t.
I’m not sure about Bezos specifically since his wealth is based on the price of Amazon stock, which is somehow created “out of thin air” and includes predictions of future growth, but it is obvious to me that on the retail side Amazon’s revenues come at the expense of others being poorer (fewer revenues). That is, if Amazon sells 200 billion of goods a year, some (most) of that comes out of the pockets of other sellers, who are poorer for it. Perhaps not all of it, because the increased convenience and efficiency may increase the size of the pie (total sales), but that effect is small in an instantaneous sense: if you take the Amazon sales out, the remaining pie is probably not larger (but growth might be higher).
There’s no problem with that! That’s how the market works, and this kind of thing should be encouraged. People are buying more from Amazon, and less from Walmart or Sears or whatever because it offers a better experience. You don’t need to pretend that all of Amazon’s growth came out of thin air, rather than at the expense of others, for your argument to work: you just point out that Amazon succeeding increases the total size of the pie, and future growth and here we agree.
One of the big problems I see in various economic decisions is the idea that “everyone wins” with any particular decision. Like “free trade” is a win for everyone. Of course not! Free trade is a good idea and I’m a strong component, but it is obvious to me that there specific people, regions and industries within some countries that are big losers. The small, disperse benefit of free trade will never make up a 3 or 4-fold reduction in earning power. Rationally, those people should be against free trade, since it’s strictly worse for them, over their lifetime – unless you insist they should show a level of altruism far beyond normal.
So I think it’s fine to say that Bezos’ wealth came at the expense of some others: it certainly did to some extend (and those people were often super-rich themselves) in the direct sense that Amazon disrupted many existing businesses. However, the kind of society where Bezos can get super rich and disrupt all those industries has high growth and people may be better off on average.
Finally, there seems to be an assumption (maybe I’m wrong) that inequality and high-growth go hand-in-hand: that as you move towards the high-growth end of the scale, equality necessarily decreases since the highest growth strategy is allocate most resources to a limited number of the most skilled entrepreneurs. I think this is true only up until a point: things start to fall apart in various ways, politically and practically at some point. So even if one subscribes to your notion of optimizing for maximum growth at any cost, it worth investigating which side of the maximum growth peak we are on: perhaps some places are already on the far side of the peak and it is possible to increase growth and increase equality at the same time.
[1] I’m just sidestepping the issue of what the “best” instantaneous distribution is, but imagine its something on the Pareto frontier that achieves the best average utility over the population with some utility function, and some reasonable estimate of the instantaneous impacts of the selected distribution. Basically what you get with “1 person, 1 vote” and perfectly rational voters and then just hand-waving away all the game-theoretical problems with voting (or perhaps invoking a benign dictator).
Finally, there seems to be an assumption (maybe I’m wrong) that inequality and high-growth go hand-in-hand: that as you move towards the high-growth end of the scale, equality necessarily decreases since the highest growth strategy is allocate most resources to a limited number of the most skilled entrepreneurs. I think this is true only up until a point: things start to fall apart in various ways, politically and practically at some point.
I don’t think that’s what I wrote. At least, that’s not what I meant.
The question I was asking was whether you should be willing to sacrifice economic growth to reduce inequality. I think you should not.
Right, then we can just prune that branch of the discussion then.
It does raise the question though: if reducing inequality increases growth, why bother discussing the hypothetical tradeoff between growth and equality on the other side of the peak? After all, both sides now want to move in the same direction, just for different reasons.
It’s still interesting, sure – but if doesn’t apply to a particular country/region/whatever it’s mostly academic.
WRT #2, maybe you decided to become vegetarian because you were searching for something which wasn’t otherwise present in your life?
Searching for a purpose in life (which likely is related to depression) may make you more likely to become a vegetarian, I guess. Correlation isn’t causation – you don’t know the direction, or whether there is a common cause explaining both factors.