Monday, October 20, 2014

Updated "A Model of Secular Stagnation"

Eggertsson and Mehrotra have released their secular stagnation paper at NBER, and since the last draft they have made a lot of progress. I will look at it more when I get some time, but given my whining I figured I should at least flag it. Excerpt:

The paper closes by extending the main model to consider investment and capital accumulation. While introducing capital does not change any of the results already derived in the baseline model, it allows us to illustrate a new mechanism for secular stagnation emphasized by both Keynes (1936) and Summers (2013). According to these authors, a decline in the relative price of investment may also put downward pressure on the natural rate of interest and thus serve as an additional trigger for secular stagnation - an insight that we verify in our model.

The price of capital is exogenous in the model, based on an argument by Summers that capital goods have gotten cheaper such that investment does not pull the interest rate up as much as it used to.

I would say that this is an important paper and is definitely worth a thorough read, which I will give it soon. If it answers my investment questions I will have to rethink things. The investment complaint is this: investment is positive in the data. The Mian & Sufi story is that investment is not linked to savings because of the ZLB. The odd thing about that is that, in the real world, the return on investment is not the Federal Funds rate. In the real world there is some sort of equity premium, etc. So even in partial equilibrium where we take the risk-free rate as exogenous (and very low/negative), it's hard to understand why returns on investment are necessarily negative as well. Then the next step of getting the low rates endogenously has its own problems, though I'll see what E&M have to say about it. Quantitatively it may be tricky.

Another thing to watch for is this from Pedro DM SerĂ´dio, whom you should follow on Twitter; in response to my post from last week he tweeted:

I'd add that the EM paper didn't have money either, so of course you'd pay a fee to consume part of your harvest tomorrow.
That's the most important channel for their 'negative real interest rate' result: you have to eat but can't produce tomorrow.
And also

The middle generation pays for storage in this model. If they held cash, negative rates would no longer be optimal.
[Updated: Was just reminded that Josh Hendrickson explained this here]. It is definitely tricky to motivate ZLB models without money. In the real world we say that the ZLB arises endogenously because cash guarantees a return of zero. You simply can't force people to pay for wealth storage, more or less. Imposing the ZLB exogenously is a nice modeling simplification, but it comes at a cost. So there is much to think about here.

Friday, October 17, 2014

Articulating my confusion

Here's a paragraph from a note by Neil Irwin, one of my favorite econ journalists (read his book):

The story goes like this: The wealthy tend to save a large proportion of their income, whereas middle and lower-income people spend almost all of what they earn. Because a rising share of income is going to the wealthy, spending — and hence aggregate demand — is rising more slowly than it would if there were more even distribution of income. Skyrocketing debt levels papered over this disconnect in the mid-2000s, but now we could be feeling its effect.

I'm honestly not picking on Irwin; he's just relaying the standard story (and seriously, his book his awesome). I'm hoping someone can help me out here. Let's go through it piece by piece.

The wealthy tend to save a large proportion of their income, whereas middle and lower-income people spend almost all of what they earn.

Ok, so far so good.

Because a rising share of income is going to the wealthy, spending — and hence aggregate demand —

Ok. Poor people have higher MPC than rich people. Rich people are getting a larger share of "income", from the income fairies or wherever. Consumption spending is aggregate demand. So he's starting to lose me; I define aggregate demand as C+I+G+NX.

and hence aggregate demand — is rising more slowly than it would if there were more even distribution of income. 

Hmm. So even over a 15-year period, output is constrained by demand (and demand=consumption). Output growth is determined entirely by consumption growth. I don't know what the mechanism is that translates consumption growth to output growth. And here is a nice counterfactual: "if there were more even distribution of income." What do you mean? Nobody is ever talking about after-tax income distribution in these discussions (well, only Scott Winship), and the higher moments of the income distribution are equilibrium outcomes, and even if you could move them exogenously I have no idea what that counterfactual looks like.

Skyrocketing debt levels papered over this disconnect in the mid-2000s, but now we could be feeling its effect.

Ok, I have no idea what this means. But I hear people say it a lot, so it must be important.

So let me see how far I can get with a model. Let income be as in Piketty world--it's exogenous, manna from Heaven, except instead of manna it's a pie, right? Because we're always talking about how "the pie" is divided. So pies fall from the sky, and they come with strict instructions about how they are to be divided. And every year, for some reason, rich people are being assigned larger shares of the pie.

And in this model there is no investment. Also there is no government spending or trade (because aggregate demand=consumption, remember?). So output growth depends on consumption growth. Wait, so output growth is not actually exogenous. Well it's still exogenous; it's an exogenously specified process. The size of the pie this year depends on how much consumption grew the previous year, and of course that was driven entirely by the pie-dividing rule and MPCs. Or maybe it's all determined simultaneously--there's no capital, so it's basically a static model. I feel like we're going in circles, but let's press on.

Rich people are saving most of their income, but it has no effect on growth because there's no investment. Don't ask me where the savings goes, I guess it's government bonds, but there's no government spending (or if there is, it's the kind where they're just dumping money in the ocean, as in some simple Ramsey models). So rich people just keep saving and saving, and the savings stock is gonna get HUGE I guess, or we're burning it, but it's not going to do anything, and rich people will just keep saving because MPC is also exogenous in this model. And consumption growth is getting slower and slower; not only do rich people have a lower MPC than poor people, but their MPC is actually something close to zero.

With me so far? This is the secular stagnation model. It's not terribly different from the Eggertson and Mehrotra model. It's also the implicit model that Mian and Sufi use. They explain the investment problem with the zero lower bound. Remember that the Fed Funds rate is near zero, so returns to investment are zero, so there's no extra investment when rich people save. Ok? Like I said, I'm just trying to articulate my confusion. In any case the ZLB wasn't binding 15 years ago. Where was the savings going back then?

Maybe someone can help me understand this, but it looks to me like the secular stagnation model is just not ready for prime time. But you wouldn't know it from the amount of attention it gets! If you repeat something enough, maybe get S&P to write a lit review on it, it becomes fact, right? It's one of those ideas that people talk about because people are talking about it. Once a few people have said it, you can cite them authoritatively as evidence. They're using it to justify a story about inequality having a significant causal impact on growth--whatever that means, since inequality is endogenous in the real world (and so is income). We're a long way from really understanding this idea or even knowing if it's plausible. And what's most interesting is that the people who talk about it the most don't seem to be doing much to seriously explain it. I think most of them have no idea what kind of assumptions are packed into their model.

Somebody, please, write down a model of this so we can understand what everyone thinks is going on. I would do it but I can't think of any clever mechanisms to deliver the result. E&M have done all the work so far, but their model doesn't have capital yet (they say they're working on it [UPDATE: here it is]). I can see that some people feel like this idea is really important. That's fair; persuade us by actually proposing some plausible mechanisms in a framework that includes things like aggregate resource constraints.

Or am I missing something obvious?

Thursday, October 16, 2014

"Few other places to turn"

Storyline is probably my favorite of the new websites this year (after Clickhole, of course; and 538 puts out awesome stuff too). Here's a great story about rent-to-own culture by Chico Harlan.

And yet low-income Americans increasingly have few other places to turn. “Congratulations, You are Pre-Approved,” Buddy’s says on its Web site, and the message plays to America’s bottom 40 percent. This is a group that makes less money than it did 20 years ago, a group increasingly likely to string together paychecks by holding multiple part-time jobs with variable hours.

And then there's the family that spent $4150 worth of monthly payments on a $1500 sofa set. They're paying $110 per week for the sofa, a smartphone, and some speakers. I think most of us know that this sort of thing is going on, yet that number is still pretty shocking. Part of the reason it's shocking is that their weekly payment exceeds the total cost of the sofa on which I'm sitting, which my wife found on Craigslist for $80 four years ago (at the time, neither of us owned a smart phone or decent speakers). And it's a pretty nice couch.

Buddy's--the rent-to-own retailer described in the article--will also sell you an "early model iPad" for $1440. This is a device that did not exist 20 years ago (or even 10). The skyrocketing price of iPads, then, isn't great support for the story about people having to get by with less in the time series. The cost of the two cell phone bills mentioned in the article isn't either.

The story reminded me of this:

Image source 
Over the last 20 years, even the lowest income group has spent less on necessities as a share of total expenditures. This is probably not the final word on the necessities/luxuries question, but it is suggestive.

So there's something that doesn't fit here. We're talking about people feeling increasingly squeezed over time (and I have no doubt that they feel that way), and we're talking about how income has fallen; yet the problems listed are that a couch has a sticker price almost 20 times as much as my couch's and costs even more when financed, and iPads are going for twice their original retail value. Meanwhile "necessities" are accounting for a smaller and smaller share of spending, or at least don't seem to be increasingly crowding other things out.

Low income is definitely a problem, but it's not the problem behind the troubling behavior in the story. I think that conclusion probably implies something about policy, though it's not immediately obvious to me what it is. Regardless, I can think of what the lessons are for me; it's all right there in Garett Jones' Piketty review.

I'm in no position to decide whether the choices being made by the people in the story are wrong for them. My point is only that something else is going on besides just stagnant incomes. 

Sunday, October 12, 2014

Charlie Munger on firm dynamics (and Piketty)

From the Daily Journal meeting:

If you take the whole history of businesses that make a fair amount of money and have a little surplus but their basic business goes to hell based on technological developments, the results are lousy. The normal result is Kodak. Imagine having a business like Kodak and having it go all the way to bankruptcy. That’s the normal occurrence: technological obsolescence. 
There are few exceptions in the history of the world. One of them is Thompson Reuters. They were a newspaper company with a few television stations added and they basically milked them as long as they could, sold them for high prices, and went into a different business – online information – and they successfully made the transition.  That is really rare. 
The other rare example, of course, is Berkshire Hathaway. Berkshire started with three failing companies: a textile business in New England that was totally doomed because textiles are congealed electricity and the power rates were way higher in New England than they were down in TVA country in Georgia. A totally doomed, certain-to-fail business.  We had one of four department stores in Baltimore [Hochschild Kohn], absolutely certain to go broke, and of course it did in due course, and a trading stamp company [Blue Chip Stamps] absolutely certain to do nothing which it eventually did. Out of those three failing businesses came Berkshire Hathaway. That’s the most successful failing business transaction in the history of the world. We didn't have one failing business – we had three. Out of that little nothing, the excess capital that we took out and put somewhere else did better than anybody’s ever done. As a matter of fact, we recently passed General Electric in terms of market capitalization, and GE was founded by Thomas Edison himself in 1892, and one of the most powerful companies in the world. 
It was a considerable stunt. But the normal result is more like Kodak. Xerox is an interesting case. They went to the brink of extinction and then came back, but they are a pale shadow of their former greatness. They actually invented most of the stuff other people made so much money out of, and they still failed. Bill Gates is a big student of this subject, and he says that the standard result is failure. Imagine General Motors who went bankrupt. Can you imagine how they towered over the economy when I was young? It was the biggest, more valuable, most admired company, and it took the shareholders to zero.

Like everyone, I have read a fair amount about Buffett (recommended) and Berkshire (recommended), and it is indeed a pretty amazing story. Buffett has said many times that the Berkshire purchase was a terrible decision. It threw off enough money to buy some insurance companies then it went under. He kept it open for some time as a favor to its employees and the New England economy. The insurance businesses generated enough float to buy some other businesses. Blue Chip Stamps didn't seem to do much after he bought it, but it also generated some float and was holding Sees when it came into Berkshire. He made some other mistakes; reading about his shoe investments in the late 80s/early 90s makes the reader cringe. But his insurance companies enabled him to purchase some big winners, like Washington Post, Geico, and Coke. Reinsurance has done great. He's a testament to the ability of markets. through all of their churning, to gradually move capital to the places where it can accomplish the most. There's a lot of trial and error, but the job gets done. A lot of it gets done by entrepreneurs, but not all:

In the whole history of Berkshire Hathaway, I can only think of one new business that we started by ourselves at headquarters, and that was the reinsurance department. Now, that is a huge business, and it’s made an enormous amount of money. Berkshire Hathaway, for all its glorious achievement, started one new business. Everything else we bought.

Could the secular decline in entrepreneurship reflect a growing propensity of existing businesses to innovate and reallocate resources appropriately without the need for entrepreneurial disruption? It's worth thinking about.

Munger also has this to say:

Mr. Piketty is a little daft. Put me down as hoping the Pikettys don’t marry into my family. It isn’t that some of his numbers aren’t correct, but he just doesn't interpret them correctly. Of course if a place as big as China gets really good at manufacturing it’s going to reduce some union jobs in the rest of the world in every trade. But they have a right to succeed. The rest of us can be mature enough to adjust instead of bitching about the fact that the world is occasionally a little tougher than we would have chosen. Of course there are going to be parts of the economy that do better or worse over a twenty- or forty-year period. It’s not some malevolent outcome. It’s a huge change, and in terms of world equality it’s enormously improved. To sit in a very rich country with a 36-hour week and complain about the fact that all the other people are coming up just doesn't strike me as a very mature or noble way to behave.

Probably Munger already knows this, but pointing out that growing inequality is a classic first-world problem isn't going to win him many friends.

Thursday, October 9, 2014

Does SBA lending reduce growth?

Salim Furth directs me to a new working paper on Small Business Administration loans and income growth:

Conventional wisdom suggests that small businesses are innovative engines of Schumpetarian growth. However, as small businesses, they are likely to face credit rationing in financial markets. If true then policies that promote lending to small businesses may yield substantial economy-wide returns. We examine the relationship between Small Business Administration (SBA) lending and local economic growth using a spatial econometric framework across a sample of 3,035 U.S. counties for the years 1980 to 2009. We find evidence that a county’s SBA lending per capita is associated with direct negative effects on its income growth. We also find evidence of indirect negative effects on the growth rates of neighboring counties. Overall, a 10% increase in SBA loans per capita is associated with a cumulative decrease in income growth rates of about 2%.

Of course, the "conventional wisdom" is wrong, regardless of how often politicians repeat it. Small businesses account for a tiny share of employment. They typically have no growth aspirations, so it's not surprising that, conditional on age, small businesses do not account for much job growth (also). The authors note this. So your prior should be that the Small Business Administration does not do much for job growth or economic growth generally. At the very least, a necessary condition for achieving this goal would be to focus on young businesses. But I'm not advocating any specific policies here.

The present study reviews some literature, which finds effects of SBA loans on job/income growth ranging from negative to slightly positive. The authors then report their own county-level regressions finding a negative effect on per-capita income growth.

A 10% increase in SBA loans per capita (which is about $3.43 for the average county in our sample) is associated with a cumulative decrease in income growth rates of about 2 percentage points. 

I would call this a large effect, but it's consistent with my priors. If the SBA is subsidizing credit for firms without growth prospects, it's just causing misallocation. But I think few people are going to be persuaded by the paper's identification strategy. The idea is to use spatial variation while controlling for a lot of stuff, and that's probably as good as anybody can do. It's definitely a worthwhile stylized fact. But SBA loans aren't random. What if the SBA gives special attention to counties with bad economic prospects? Since some entrepreneurs start businesses because their job prospects are bad (for whatever reason), areas that attract SBA attention could be areas that are heading downhill. I have also heard from some bankers that SBA paperwork can be cumbersome; one banker told me that their branch had to have an SBA specialist, while a banker at a different branch said they tried to avoid SBA loans entirely. So there could be selection effects along that dimension as well. I should read the other papers in this literature, but these authors don't provide a lot of detail about how the SBA makes decisions. 

From what I gather, this paper is an improvement on the literature primarily because it has better controls and richer spatial variation. It's useful evidence to be sure--I hope it gets some attention--but it would be great if we could find a smoking gun.

Friday, October 3, 2014

Bad arguments against Uber

This note by Catherine Rampell makes a pretty odd argument that Uber is bad for consumers. For example:

Medallions and other regulations capping the number of livery cars available are often derided as taxi cartel protectionism. But they can benefit the public, too. They limit the number of empty cars driving around looking for passengers, snarling intersections and polluting the air.

There's a lot wrong here--for example, if there is too much capacity then eventually drivers will go do something else with their time. But I want to focus on the congestion + pollution argument.

A good argument can be made that congestion and carbon are mispriced for the usual externality reasons. But Rampell has failed to argue that Uber and its customers uniquely deserve to bear the cost of correcting that market failure. I own a car; why should people who ride in Uber cars pay for their pollution and congestion while I drive around with impunity? This is a very silly way to make policy. Why don't we just put a cap on the number of cars Ford can produce every year? Or limit UPS to X trucks per city?

So yes, there is an externality associated with Uber cars driving around. But it's the same as the externality involved with everyone else driving around in cars. This is a car problem, not an Uber problem. Rampell hasn't explained why the Uber people should pay more for their externalities than everyone else.

She also uses this common argument:


In other words, for all their bellyaching about the bullies of Big Taxi, Uber and Lyft are becoming pretty big bullies themselves. Nothing about their behavior suggests the ultimate winner of the ride-sharing wars will wield its power beneficently when it controls the market and can raise consumer prices at will. Consumers will just be trading in one monopoly — loathed Big Taxi — for another, less regulated one.


Obviously the difference is that Big Taxi's market power is a result of rent seeking and regulatory capture, while Uber's is the result of being amazingly good at giving consumers something they want. If Uber and Lyft get so big that they have market power, there will exist strong incentives for new entrants to undercut them. The very existence of Uber and Lyft suggests that this market doesn't have special natural monopoly problems, unless we create them by regulating enduring market power into existence. Rampell certainly hasn't made an argument for why she thinks someone could take over this market without having to worry about new entrants.

If these are the best anti-Uber arguments out there, it's no wonder these economists can't find anything wrong with it.

Friday, September 19, 2014

"Disruption" disrupted

At The Atlantic, Justin Fox compares the popular obsession with "disruption" to findings from research on declining dynamism, including papers I've been on.

Nowadays every corporate executive wants to disrupt; the word has become a mark of forward-thinking decisiveness—though it is sometimes attached to strategies that are more about cost-cutting than game-changing. And in Silicon Valley, belief in disruption has taken on a near religious tinge. All that disrupts is good; all that stands in disruption’s way (such as, say, San Francisco taxi companies or metropolitan daily newspapers) deserves to perish. . . . 
What may be even more extraordinary, however, is the growing disjuncture between all this talk of disruption and its actual practice—at least so far as we can measure it. Thanks to data that the Census Bureau began releasing a decade ago, economists can now track what they call “business dynamism” in ways they couldn’t before. As researchers have dug into these numbers, they’ve found that most metrics of dynamism and upheaval in American business have actually been declining for decades, with the downturn steepening after 2000. Fewer new businesses are being launched in the United States, the average age of businesses is increasing, job creation and job destruction are on the wane, industries are being consolidated, and fast-growth businesses are rarer.

The article also describes results from, but does not cite or link to, unpublished work by coauthors and me on the post-2000 decline of high-growth firms.