News for hackers. Stuff that matters.
Re:Piketty’s Capital in a Nutshell by mrow84 on Monday April 28, 2014 @06:39AM (#7659120)
I haven't got the book, but the hypothesis it claims to offer evidence for is given in the introduction, which is available online at [0]:

"When the rate of return on capital exceeds the rate of growth of output and income, as it did in the nineteenth century and seems quite likely to do again in the twenty-first, capitalism automatically generates arbitrary and unsustainable inequalities that radically undermine the meritocratic values on which democratic societies are based."

Indeed, you yourself wrote an interesting blog post essentially regarding technicalities of that point [1].

You note that he has (as far as you have read) said little about the present, so he may not have substantiated the conditions necessary for his hypothesis in the present day, but it's a little unfair to suggest that a hypothesis doesn't exist.

[0] http://www.hup.harvard.edu/features/capital-in-the-twenty-fi...

[1] http://www.chrisstucchio.com/blog/2014/piketty_inequality_an...

Re:Piketty’s Capital in a Nutshell by yummyfajitas on Monday April 28, 2014 @06:54AM (#7659151)
If that is the claim of the book, he has not made any attempt whatsoever to support it as of the halfway point of the book. So far, all I've read from him is some assorted data suggesting r has historically been bigger than g except when it wasn't, and that inequality grows except when bad things happen.

If he wants to show that r > g has any effect on inequality, he needs to do a LOT more than he's done. Historic timeseries of the sort he presents are completely incapable of demonstrating this point - you need longitudinal microdata for that, together with a far more detailed theory than "r > g". If you simply have r > g (ignoring volatility, as he's done so far), and treat capital as homogeneous (as he's mostly done so far), then this will explicitly NOT happen:

http://www.chrisstucchio.com/blog/2014/piketty_inequality_an...

Maybe he does this in the rest of the book, but I'm far less hopeful as I read more. To paraphrase Eliezer Yudkowsky:

But the real cleverness is in how inequality was marketed. They left out the math.

http://lesswrong.com/lw/vv/logical_or_connectionist_ai/

Re:Piketty’s Capital in a Nutshell by saalweachter on Monday April 28, 2014 @08:15AM (#7659404)
> If he wants to show that r > g has any effect on inequality, he needs to do a LOT more than he's done.

I am not sure what you mean by this.

'r' is the rate of return of capital, 'g' is the rate of growth of the economy. If 'r' is greater than 'g', than money continuously invested with a rate of return of 'r' will grow faster than 'g', by definition.

Are you disputing the basic match? Are you disputing that 'r' is ever greater than 'g'? Are you disputing that a diversified portfolio will return at a rate of 'r' when the average rate of return of all capital investments is 'r'?

Re:Piketty’s Capital in a Nutshell by yummyfajitas on Monday April 28, 2014 @09:03AM (#7659582)
I mean exactly what I said - r > g does not have any intrinsic relationship with inequality. All it means is that (capital stock) / (yearly income) will grow.

There is no reason why a society with capital/income = 10 years has more or less inequality than one with capital/income=5 years.

Re:Piketty’s Capital in a Nutshell by jgalt212 on Monday April 28, 2014 @03:15PM (#7662212)
I definitely think there is. The bigger the difference between r and g with r > g the more wealth gets concentrated in the hands of a few as time increases (as long as there is some variability in income levels at t0). If everyone earns the same exact money, it makes no difference what the return on capital is--everyone's income and wealth will be exactly the same for all t (as long as savings rates are equal).

I will attempt to google an answer for you

Re:Piketty’s Capital in a Nutshell by riffraff on Monday April 28, 2014 @05:07PM (#7663071)
it depends on how the r is distributed doesn't it? Suppose there is Alice who owns 1M in bonds and doesn't work while Barbara and Charlie own a 100K house and are blue collars.

If the houses price goes up ten times while the bonds pay only 110% than total inequality goes down even if Alice still hasn't worked a day.

Re:Piketty’s Capital in a Nutshell by jgalt212 on Monday April 28, 2014 @05:58PM (#7663400)
yes, that's definitely true, but if there's only 1 r and all assets return r, I'd say the bigger r is relative to g, the higher the wealth concentration as t increases.

most of this math is above me (differential equations and finite difference methods), but the models are pretty easy to experiment with if you employ Monte Carlo methods.

In fact, that would be a nice side project: Piketty web app where the user can mess with various permutations of r and g and their distributions.

Re:Piketty’s Capital in a Nutshell by danielweber on Monday April 28, 2014 @09:47AM (#7659881)
'r' is the rate of return of capital, 'g' is the rate of growth of the economy. If 'r' is greater than 'g', than money continuously invested with a rate of return of 'r' will grow faster than 'g', by definition.

Using these definitions of r and g,

How can the return on capital exceed the growth of the economy in the long term?

If the return on capital is so good, governments should invest their pension funds in the stock market and then sit back and relax while all the free money pours in.

Re:Piketty’s Capital in a Nutshell by arethuza on Monday April 28, 2014 @07:12AM (#7659186)
I must say that, possibly because of my own background as a reformed AI researcher, that article on "Logical or Connectionist AI?" you refer to there is far more interesting than the original article!

[NB It was probably Drew McDermott's "A Critique of Pure Reason" that fatally damaged my interest in symbolic AI].

Re:Piketty’s Capital in a Nutshell by mrow84 on Monday April 28, 2014 @07:23AM (#7659227)
When you say that he gives "some assorted data suggesting r has historically been bigger than g except when it wasn't, and that inequality grows except when bad things happen", does he make no attempt to correlate his time series of (r - g) with inequality data? If such a correlation was found, that would obviously represent evidence to support his hypothesis.

His claim that I quoted is a causal one, however, as I am sure you are aware, it is very standard practice in all fields of inquiry to propose causal relationships from careful and qualified analyses of correlations. You are of course correct that demonstrating causality would require more than time series analysis, however a lack of causal analysis cannot be taken of itself to invalidate a well-identified correlation (I am of course assuming good faith and rigour on his part, which I am not in a position to attest to at the moment). Contrary data, or an explanation of a mechanism that would induce the identified correlation without the proposed causal relationship should now be sought to invalidate his claims, simply saying that he has only shown correlation is probably not enough.

Finally, could you just explain this a bit more, I didn't quite get it: "If you simply have r > g (ignoring volatility, as he's done so far), and treat capital as homogeneous (as he's mostly done so far), then this will explicitly NOT happen: ..."

Re:Piketty’s Capital in a Nutshell by yummyfajitas on Monday April 28, 2014 @07:46AM (#7659310)
There is no careful analysis. The graphs have basically 3 features: high inequality until the great depression, then low inequality, then inequality rises again. He doesn't have much in the way of graphs of r-g.

Rather than explaining why you need more than just a correlation, I'll link to the Lucas Critique: http://en.wikipedia.org/wiki/Lucas_critique

Finally, could you just explain this a bit more

Not sure I can explain better than the comment I linked to. If your capital and mine grow at the same rate, we'll have the same ratio of capital forever. I.e., my wealth will be X exp(rt), yours will be Y exp(rt).

This seems to be Piketty's implicit model in most of the book, though once or twice he calls it unrealistic. But he never actually states explicitly what his model is, so it's a bit hard for me to be sure.

Re:Piketty’s Capital in a Nutshell by mrow84 on Monday April 28, 2014 @09:02AM (#7659577)
I hadn't heard of the Lucas Critique before, but it seems to be constrained to policy advice that operates on economic variables over which market actors have control, and I'm not sure that applies here - I'll have to think some more.

> Not sure I can explain better than the comment I linked to. ...

Thanks for the clarification, I'm not feeling very switched on at the moment. I would question your analysis though. It is certainly correct on its own terms, however it conflates wealth with capital (which is something I've seen people saying Piketty does - does he?). To me it seems to be vital to include both major sources of income, capital and labour. If you do that, then you can calculate the quantity

(Xl exp(gt) + Xc exp(rt)) / ((Xl + Yl) exp(gt) + (Xc + Yc) exp(rt))

Where Xl and Yl are the labour "holding" for X and Y, and Xc and Yc are the capital holding. Here I'm taking r as the return on capital, and g as the return on labour, which seems abusive but I think it captures the point. This quantity tends to the capital ratio where r > g, and the labour ratio where r < g. Whilst it ends up at the same limit as your model (for r > g), it shows that there is a shift of wealth from labour to capital over time, so the inequality between capital holders and labour "holders" increases over time, as the returns on labour are dominated by the returns on capital.

edit: Just thinking about it a bit more, this implies that we definitely do want g > r, because then in the limit wealth is proportional to labour, i.e. productivity, not to capital bequeathment. One can still make returns on capital, but in the limit your wealth will be proportional to your effort. Surely that is the goal of a meritocratic society?

edit edit: To clarify, I have broader views on the value of pure meritocracy, in my first edit I was just trying to relate the implications of the wealth ratio given above to meritocratic goals.

Re:Piketty’s Capital in a Nutshell by yummyfajitas on Monday April 28, 2014 @10:58AM (#7660333)
The Lucas Critique basically just says that if you want to use broad correlations, you should make sure that the microfoundations which drive those correlations still apply.

Unfortunately the relation of r to g tells us very little about meritocratic goals. Suppose that the growth rate r is driven entirely by startup founders building gigantic companies (or lotteries, which would be mathematically equivalent), which cease growing after the founder dies. Then you could have r be huge, inequality would be huge, and everything would be entirely meritocratic.

Re:Piketty’s Capital in a Nutshell by mrow84 on Monday April 28, 2014 @11:35AM (#7660631)
I think that the quantity I offered suggests why there might be a correlation between inequality growth and (r - g), independent of the mechanisms that underlie the returns on capital and labour, and inequality. In words, when r > g, the income distribution tends towards the capital distribution, and when r < g the income distribution tends towards the labour distribution.

The relationship between this point and meritocracy is that average returns on capital are approximately "unearned" - you can get close to the market rate of return with minimal effort, whereas returns on labour are positively earned, by your labour. So when r < g the limiting case is for income to match labour, i.e. effort. Note that this by no means inhibits the possibility of greater returns to intelligent capital investment (something that you definitely want to keep) - the point is that you control average returns to stop capital domination fostering inequality.

Re:Piketty’s Capital in a Nutshell by foobarqux on Monday April 28, 2014 @08:56AM (#7659546)
Putting aside the fact that limiting processes thousands of years into the future are probably less interesting from an economic perspective than a mathematical one, can you explain why you think r and g must converge?
Re:Piketty’s Capital in a Nutshell by newblahbl4hblah on Monday April 28, 2014 @07:40AM (#7659284)
It's clear that you dislike the implications of "r > g"...maybe you could explain why?
Re:Piketty’s Capital in a Nutshell by yummyfajitas on Monday April 28, 2014 @07:51AM (#7659325)
Not sure what to explain. What implications do you believe I dislike?