Matt Rognlie on Piketty, net capital returns, and housing

Brookings emails me:

Capital income is not growing unboundedly at the expense of labor, and further accumulation of capital in fact most likely means a fall in capital’s share of total income – refuting one of the main theories of economist Thomas Piketty’s popular book Capital in the 21st Century — according to a paper presented today at the Spring 2015 Conference on the Brookings Papers on Economic Activity (BPEA).

Existing studies that show an increase in capital’s share of income miss the growing role of depreciation in short-lived capital, in items such as software, says MIT’s Matthew Rognlie in “Deciphering the Fall and Rise in the Net Capital Share.”  Rognlie subtracts depreciation in seven large developed economies (the US, Japan, Germany, France, the UK, Italy, and Canada) to get net capital income, and finds that the only long-term rise in capital’s share of income is in housing. Capital income elsewhere in the economy has grown moderately, but it is only recovering from a large fall that lasted from 1948 through the 1970s.

Piketty’s Capital argues that the role of capital in the economy, after falling during the Depression and two world wars, is set to recover to the high levels of the 19th and early 20th centuries. According to Piketty, wealth will accumulate amid slowing economic growth to push up the capital-to-GDP ratio in the economy, which will then cause an increase in capital’s share of income — and growing inequality.

In contrast, Rognlie finds that a rising capital-to-GDP ratio is most likely to result in a fall in capital’s share of income, since the net rate of return on capital will fall by an even larger proportion than the capital-to-GDP ratio rises. Outside of housing, postwar changes in the value of the capital stock have not led to parallel changes in capital’s share of income. In fact, the value of the capital stock relative to private income reached its highs in the late 1970s and early 1980s, when capital’s share of income was near a low.

Rognlie shows that the share of net income generated by housing has risen in all seven large developed economies since data became available. “Housing’s central role in the long-term behavior of the aggregate net capital share has… not been emphasized elsewhere…Observers concerned about the distribution of income should keep an eye on housing costs,” he writes.

Brad DeLong offers comment.

Here Jim Tankersley has a superb profile of Rognlie and the story behind his comment, MR plays a role too.  Recommended.


Cars made housing cheap in the postwar era by putting a huge amount of land within reasonable commuting difference. But the failure of the flying car to arrive has left us commuting at about the same speed as the 1960s.

The failure of teleportation to develop is even more responsible for the lack of increase in commuting speed.

Clearly the failure of time travel is ultimately responsible.

But joking aside, Steve is basically right. Presumably driverless cars will do the same thing by allowing longer commutes and removing the cost/constraint of parking at the passenger's destination?

It's quite possible that autonomous cars will have a greater effect than even the flying car would have. You still have to find a parking spot for a flying car and traffic congestion would probably always limit flying over major cities. But autonomous cars can drive off to park themselves and can drive in far denser packs on the roads.

Of course, flying autonomous cars trump both autonomous and flying cars.

The last point is exactly right. Autonomous driving cars are a necessary step before flying cars. No way would hundreds of micro and mini planes be allowed to drive within a few metres of each other by human control alone.

New roads make more land available and can shorten commute times. High rise buildings substitute for land. I blame slow growth policies rather than some natural.


New roads are effectively making new land.

In Southern California, which comprises a sizable chunk of the home values for entire country, the road grid system on areas of non-mountainous land is close to pervasive. How many roads are left to build within 40 miles of LAX?

A vastly expensive tunnel under Mt. Wilson to more conveniently connect the high desert suburbs to downtown Los Angeles could be imagined, but those suburbs are still stuck out in the high desert with its less desirable climate.

Maybe we should replace local and sate slow growth policies with a program to collect a bigger percent of our taxes through property taxes and maybe even a Georgian land tax.

Speed is only a first-order approximation of the controlling variable. The controlling variable is "awfulness." Commutes today of a given distance are roughly the same awfulness as commutes of that distance in the 60's.

I mention this because there are a few technologies that might not literally speed up commutes, but could make them less awful, and if you believe in these technologies and that housing cost is related mostly to how far people are willing to live from work, it should affect your sense of where housing prices are going.


1. A genuinely driverless car at a price point affordable to humans would probably make most people comfortable with a significantly longer commute. If you can watch TV or play computer games or read on your commute, in a comfortable, relatively quiet, private car, then my guess is that at least 80% of everyone could deal with another 20-30 minutes each way.

2. Also adding to car technologies, if electrical or other next-gen fuel technologies significantly reduce cost per mile, then presumably that creates marginal pressure upward on commute distances.

3. More speculatively, perhaps the Oculus Rift or other virtual-reality-ish stuff can finally deliver on the process of a remote work force where your commute is just to your home office every day.

I note that I'm not here to flog the idea that any of these technologies are guaranteed -- but we'd be fools to totally discount them, either.

Just having a smartphone connected to the car is a big improvement for me. I can consume interesting downloaded content, rather than rely on radio.

There's still definitve hard points. For instance, if I need to get my kids to and from school, there's only a certain amount of "commute" that my wife and I will tolerate. What I think Steve overlooks is the growth of so-called Edge Cities and a gradual shift of work towards more distant locales. For instance, in the Chicagoland area, the 294 corridor has all sorts of high-profile companies: Kraft, Allstate, Abbott, CVS, Takeda, Walgreens. For any of these companies, you can live in Wisconsin and have an easier commute than living in Chicago proper. Expanding the highway system and more outward growth of companies will allow lower densities, and yes, we're still expanding the road system dramatically 'round these parts. 294 just got done with a significant expansion in the Milwaukee-Chicago corrdior.

You used to be allowed to slap an apartment over the garage for your ne'er-do-well son. No more of that though I do not why.

Correctamundo! What's cooking, Mrs. C.?

Nothing prevents you from putting living space over a car garage, other than conservative policies designed to keep out non-whites and white trash by limiting 2 or 1 acres of land to one family unit. By putting an apartment over your garage, you are creating a two family unit structure, which means you can rent to some poor brown family who will burden the schools without paying the taxes on 2 acres plus 2000 sq-ft finished structure.

The mechanical issues are well understood and not all that expensive, unless the existing structure was built to house only 4 people with a small septic, 100 amp electrical, etc. The siting of the house might make the added parking a problem to accommodate the four cars needed by a two family structure.

After all, how can you create an "in-law" apartment for your sister-in-law and kids after her spouse (your brother) died that is not suitable for a black woman on welfare with kids? Maybe your white brother married a black woman.

I live in NH where the public debate has switched to the lack of young people for employers, fewer than there are children, caused by the lack of affordable housing for young families. Mass, VT, Maine have income taxes and sales taxes in addition to property taxes, and all three of those taxes are used to fund schools, while in NH the property tax is the only serious way to fund schools. If a town lets high density family friendly housing to be built, then a 1200 sq-ft house which is a large Levittown home in typical town density means the schools need to be funded by taxes on houses that is sending at least 1 and maybe 2 kids to public school - say $10,000 taxes on a 1200 sq-ft starter home that costs $80,000 for land and construction.

The next town over with zoning to limit the number of poor kids in school has a 4 acre minimum which means the house has 3000 sq-ft and a three car garage, etc and it costs $500,000 plus for land and construction. The number of people who can afford it are older and likely to have fewer kids, kids going off to college soon, so $10,000 in taxes is going to do more school funding per kid.

Libertarian/conservative economic policies have increasingly made NH unfriendly to people starting families, and that limits business to the border where people can commute in to NH and then pay income taxes to their home State while the business owner with adult kids living outside NH pays less in taxes than being in Mass where he would need to pay for the schools that his employees need to work for him.

TANSTAAFL, but politicians since Reagan are required to promise free lunches to get elected. And when the policies are bad for the voters, the voters look for new free lunch pols. Unable to afford a house with good schools, they look for tax cuts and tax credits and public private partnerships that will provide free education without paying taxes, etc.

In NH the voters want lower property taxes, smaller lot sizes for starter homes for families, better schools with better science and vocational labs, better teachers, no income taxes, lower property taxes, more parks and programs for their kids like in the old days, no sales taxes, and lower property taxes, more kids starting families to pay taxes to support all the upper middle class to rich boomers who moved to NH to dodge taxes as they get old.

This comment is like "War and Peace", by which I mean that I didn't read it.

Only in your fevered dreams are zoning and building restrictions pushed only by conservatives or libertarians. Talk about mood affiliation.

We now have two obsessive regular posters.

telecommuting and driverless cars will fix that.

Why hasn't telecommuting done so already? 15 years ago I figured you might see a real push away from the cities, particularly among young, tech-savvy people. But though the urban influx is mostly hype, I'm pretty sure there's been no broad push away from expensive and congested cities.

Most of the drawbacks of cities don't apply if you are young and single.

Fair point. But is more-than-occasional telecommuting still a growing thing among the 30+ set?

Maybe the labor market just isn't tight enough yet for things to lean the employee's way.

It's increasing, revolutions don't happen overnight: "Citing Census Bureau statistics, the New York Times reported earlier this year that telework in one form or another grew 79 percent between 2005 and 2012, with 3.2 million workers — 2.6 percent of the workforce — now working from out of the office at least half-time"

"Fair point. But is more-than-occasional telecommuting still a growing thing among the 30+ set?"

I'm an engineer in my mid 40's. I don't generally telecommute from home, but I routinely (as in at least 1/3rd of my work days) access some remote plant from my office and perform work. At least half of those days would have involved travel to a site, even 10 years ago.

And during the recent bad weather, most of the engineering staff worked from home.

Do you date and have sex over the Internet?

Where do you think a 25 year old engineer who happens to earn $250K so he can live in the suburb with $500,000 houses with other professionals almost all married go to socialize to meet people, date, have sex, get married and start a family?

If the solution to living in the suburbs and telecommuting for dating is to drive into the city every night to go to pubs and clubs and events to meet people, why not live in the city and telecommute.

I remember Silicon Valley gurus in 1993 telling us the Information Superhighway would mean you could work from anywhere, which presumably meant land prices in Silicon Valley weren't going to go up so fast anymore.

Funny how that worked out ...

In my experience, MBAs value face time and schmoozing more than they value good work product. Management is heavy with MBAs. The rest is pretty much self-explanatory.

'Existing studies that show an increase in capital’s share of income miss the growing role of depreciation in short-lived capital, in items such as software'

Which, considering that Google's Android is GPL software, means that Google is depreciating software technically worth $0 from a strict accounting perspective (recognizing there are a lot of caveats to that assertion - Android certainly represents an investment worth considerably more than $0). Facebook, Amazon, et al are pretty much in the same framework - it is almost disingenous to talk about 'depreciating software' in a world where 4 out of 5 all smart phones run on free software, for example -

And sure, SAP software has a 'short' life span - which is why they have been selling pretty much the same core software for over two decades, though the name tends to wander (yes, I do work sometimes for a Systemhaus, that traces its roots in ERP software back to the time when IBM was fostering multiple proto-ERP companies, of which SAP became by far the most well known). Depreciation was originally intended to cover the idea that something is actually used up in production - sharpening a lathe blade removes some of the blade's metal, for example, until not enough of the blade is left to be useful. In the modern world where physical production is scorned as having minimal profit margins and requiring actual effort, depreciation tends to be another term for 'tax avoidance strategy.'

Just check out the German term 'Pflegevertrag' for how that works in practice - paying a monthly fee, a company is provided all updates and program fixes - and newer versions. Nonetheless, even though the software in essentially kept up to date in terms of operating systems, legal requirements, and new ideas, the original software purchase is fully depreciated - on software that technically is in better shape, two version later, than it was when originally acquired. (Again, plenty of caveats - and the details here are definitely related to the home field of the world's largest business software company.)

If one has a clever enough tax department, you might even be able to depreciate each new version, using a standard depreciation table based on how much it cost, as represented by that Pflegevertrag's monthly payments. Not a game for amateurs, of course - but then, software as a capital cost tends to imply the sort of company that often has a clever tax department.

Clever enough to convince people that modern ERP software like SAP's 'depreciates' while it is being maintained at the latest level.

There may be some tax advantages involved, but the beneficiaries are the software manufacturers. Instead of an income stream depending on upgrade purchases, they have monthly billings that smooth out the cash flow.

There was a serious flaw in Piketty's view where he imagined capital growing organically. He described the l'Oreal family fortune as growing by extraordinary amounts, but what it represents is a successful business selling products into the developing markets. If the products didn't sell, the value of that capital would shrink quickly. Same with the Walmart family fortune.

Of course software depreciates. Maintenance and licensing annual fees are another form of depreciation, both of which attribute cost to the temporal periods in which the benefit of the asset is derived. The form is slightly different from hard machinery, welcome to reality.

I think Rognlie's point is that the value of the original code to the developer depreciates rapidly (not necessarily to the user). Think about the business of developing iPhone apps. Your investment in capital is writing the code and updating it (and the purchase of computers to do so). What would happen to value of that code over the course of 2 years if you didn't invest real resources to update it? As new competitors show up and new versions of iOS are released it quickly goes to zero.

Say what you like about German there is no denying it has wonderful compound words. Who wouldn't prefer Pflegevertrag to software maintenance agree.

The WaPo article by Tankersley is clear and on point about Rognlie's 'capital deepening' argument. Also I am surprised to find Rognilie was the first to make the argument that housing is the main driver of capital increasing, I thought it was others, though that's not unusual given the rising role of housing to the nation's banks over the last 100 years. The Brad DeLong post is all over the place, and apparently is making the point that skilled labor is also contributing to inequality ('Bill Gates working' vs Bill Gates as rentier). Also of interest was DeLong comes from a rich family, as he quotes his relatives being in the top 0.1% and gives figures for 1968. What is very interesting to me is Piketty concedes--and this is damaging to his case--that inequality may be temporary after all- WaPo article: " [Piketty] said he never predicted inequality would “rise forever” -- only that it could reach “higher levels than what we have today, and that this is sufficiently important to be concerned.” WOW! Why give up this important point? I don't think Piketty thought this off-the-cuff remark though. For maximum impact Piketty should not concede this point, especially when his book finds inequality only reduces during wartime.

The rate of return on capital has been falling since the 1970s, a point made by Cowen in his interview by the NYT last Fall, so it's not surprising that capital's share of income would fall. Of course, this is what Marx predicted would be the fate of capitalism, so it's more of a confirmation of Marxism than a refutation of Piketty. But the beauty of capitalism (markets) is that excesses are self-correcting. In this case, the falling rate of return on capital (r in Piketty's famous thesis) induces owners of capital to seek a higher return by taking on more risk through speculation, which in turn produces bubbles that burst, financial crisis, and a collapsing value of financial assets. The collapsing value of financial assets falls most heavily on the owners of capital, which is to say the wealthy, thereby causing inequality to fall and, in time, restoring a more sustainable economic growth as the rewards are spread more evenly. There is one potential flaw in this analysis: intervention by government and the Fed to prevent (stop) the value of financial assets from collapsing. For those who have read Piketty's book, that is his prediction: governments and central banks will always intervene in a financial crisis, as they did in 2007-09. While many economists have obsessed over Piketty's thesis (r > g), his more important insight, or error, is his prediction that governments and central banks will always intervene and prevent the market from self-correcting the excesses. I should add that the Austrians prefer the path of self-correction, for governments and central banks not to intervene and for markets to self-correct the excesses. And the influence of the Austrians is growing, and may have a place in the next Republican administration. These are interesting times.

@rayward, nice word picture, but it's flawed: you say "in time, restoring a more sustainable economic growth as the rewards are spread more evenly" - supposing that the bursting of bubbles helps the middle class and the poor. I don't see it that way. As a member of the 1%, we bought DC real estate at the top of the bubble in 2006, it went down about 20% at most, but now it's back up to past the bubble highs. The people who really suffered are the bottom 40% who bought at the peak and lost everything, since typically these folks don't buy premier locations like we do, as they can't afford it. Premier real estate in DC, LA, SF, NY is bubble proof, whereas Miami, Phoenix, Las Vegas are where the middle class invests during bubbles.

So Marx (capital deepening) and Piketty (r > g) are still right. My solution to inequality? Razor wire, taking a page out of the South American playbook.

You are correct: intervention (primarily monetary stimulus) "saved" the owners of financial assets and those who have the good fortune to reside and own houses in the nation's financial hubs, NYC, the West Coast, and DC, the latter which makes it all happen. The rest, well, what's their view of intervention? Not exactly positive. So next time, and there will be a next time, will the Fed have the political will to intervene, will the Fed chair appointed by a libertarian President have the desire to intervene. My view is that self-correction is worse than intervention (does any sane person want to experience another great depression), and that a balanced intervention, with both monetary stimulus to stop the financial collapse and fiscal stimulus to create a more balanced recovery, is far better. Unfortunately, there doesn't appear to be much support for a balanced intervention, so I fear that self-correction may be in our future. Of course, anyone reading this comment should be aware of where the comment is being made. As for the detailed and arcane analysis written by the experts, including DeLong's comment linked by Cowen (is it even possible today to distinguish labor and capital income given the financialization of the economy), I'm reminded of what Tim Geithner wrote in his memoir: that like his one-time mentor, Dr. Kissinger, he wasn't burdened with economic theory since he's no economist and approached the financial crisis as a crisis manager not a theoretician. Like Geithner, I'm not burdened by theory, and my comments are simply the product of my observation.

"“Housing’s central role in the long-term behavior of the aggregate net capital share has… not been emphasized elsewhere…Observers concerned about the distribution of income should keep an eye on housing costs,” he writes."

Tyler, I think Kevin Erdmann is writing some of the most interesting stuff on the macro economy, and housing's crucial role, right now.

I'm not gonna lie, I haven't read all 21 parts of the series yet, but you should be able to knock this out over a cup of coffee.

@BD - you keep shilling Erdman, this is at least the second mention of him. I found nothing radical in his series, but then again I am more versed in economics than you. Erdman apparently gets your approval since he's a monetarist and Target NGDP advocate (Erdman: "This Wikipedia article on the "Causes of the Great Recession" would be funny if it wasn't so sad. Among dozens of causes discussed, monetary policy is barely mentioned..."). What neither he nor you realize is that money is neutral even superneutral in most instances short of hyperinflation, and the Fed has little or no influence over the economy. As I told you in another thread, during the Volcker Fed, the Fed lowered interest rates and inflation *fell*, and it raised interest rates and inflation *rose*, the opposite of conventional wisdom. While a 'never reason from a price change' economist or clever statistician who incorporates artificial lags into Fed policy can 'explain' these counterintuitive examples, Occam's Razor is the simplest explanation: the Fed has no real power beyond a few hours or days, and whatever they do is dwarfed by the market. Fed actions are like a press release is to a stock: rarely worth more than a pop in the price for a day or two.

Good post. Always interested in your views on chess, IP, and young Filipinas.

What is remarkable is that both Erdman and Lopez are fundamentally wrong. And stupid.

I disagree. Your comment is unremarkable. And stupid.

And you want to be my latex salesman?

Matt Yglesias pointed out a little over a year ago (in his better Slate days) that Picketty actually documents this in one of his charts on France. Picketty just wasn't that good at reading his own charts:

Since February < April, seems Yglesias -- which means church in both Greek and Spanish--beat Rognlie to the punch by one month on the second point of Rognlie's two-prong critique, namely, "Rognlie found the trend to be almost entirely isolated to the housing sector.". But the hard work of the analysis belongs to Rognlie not Yglesias. So if this was a patent priority contest, Rognlie wins for the combination of housing + capital deepening (vs housing alone or capital deepening alone, which 'belong' to Yglesias and Karl Marx, respectively). Case closed.

This is an impressive essay.

Given that Delong's grandfather was a one-hundreth-of-one-percenter, should I read that to mean that Delong is also insanely wealthy, and if not, why should anyone care about his anecdote or "inequality" generally if it can't even survive two generations?

Only if you ignore the effects of primogeniture.

“Housing’s central role in the long-term behavior of the aggregate net capital share has… not been emphasized elsewhere…Observers concerned about the distribution of income should keep an eye on housing costs,” he writes.

I couldn't agree more.
This is just anecdotal, but from the perspective of a middle-income earner, the high cost for an average single-family home is a major impediment to mobility.
This is the one asset that traditionally launches people from the lower to middle class. Own a home, and you don't have to pay rent on it, and suddenly, you have all sorts of extra cash to do things like put your kids through college. On a traditional 15 year mortage, if you buy the house right around the time of your first child that would give you just enough time to pay it off. Secondly, rising housing prices disproportionately benefit those who already owned property at the expense of those who didn't, thus contributing to wealth inequality. Renters end up playing more in rent and have less money left to save for an even larger down payment, so it takes longer to even get into a mortgage. Property owners earn more in rents. The lax banking policies designed to make bigger loans easier to get are unhelpful. You're still borrowing more money, just taking longer to pay it back and you end up paying more in interest, so you're basically spending about 3 times as much, relative to income, as people in the past did. It's almost impossible to but a house without violating basic financial common sense, like don't borrow more than 3x your income.

I agree that there are tax benefits to homeowners that need to be erased. But, this is tricky. First, more universal access to home ownership will cause prices to rise. There is no way around it. Making access to ownership of any asset more difficult is not generally the way to create an equitable economy.
Second, the added capital income to homeowners is partly due to tax favors that pull investment into housing. But, besides that, the rise in housing income is a product of the housing shortage that has been exacerbated by the crisis. Rents are rising, so that all wage earners have lower real compensation, and homeowners are earning that back in the form of imputed rental income, which is tracked as part of GDP, but not as part of household income.

One of Rognlie's point is to use the capital share net of depreciation. This in turn eliminates the increase in the capital share during the 1970s-2000s period, which in fact remained pretty constant, just with a U-shape. (See his working paper for the figures and the details.)

This also goes against the Piketty's findings, which instead point to an increase in the capital share during the same period. But does Piketty really use the gross share?

Here, I am a bit puzzled. If you look at Figure XII in Piketty and Zucman (2014), "Capital is back ...", they clearly state that they are using the net capital share, i.e. net of depreciation (see note 64.) This is also the figure (6.5) presented in the book, "Capital in the twenty-first century", where again Piketty states clearly that he is using the net share, not the gross (see chapter 1, page 43).

What's going on? Is Rognlie double counting depreciation? Am I missing something?

I forgot to mention one thing: Piketty's figure is usually reported from 1970s on, while Rognlie's from 1950s on. This might create a bit of confusion, but the point about depreciation remains: is Rognlie double counting it?

Comments for this post are closed