Monday, March 23, 2015

Where Is the Internet Boom?

Facebook Connections
The Internet is arguably the most important technology to emerge since 1990. Has it done anything for economic growth?

It seems it should have. Take a look at the growth of e-commerce. Or the rise of mobile banking in the developing world. Or the market capitalizations of Internet stocks in the Nasdaq. Everywhere you look, you see the Internet.

Except, as I will show, in economic data.

Now, skepticism about the economic effects of digital-age technology has quite a track record. A bad one. "You can see the computer age everywhere but in the productivity statistics," the economist Robert Solow famously said in 1987. Erik Brynjolfsson called it the "productivity paradox" in 1993.

And then, in the late 1990s, the IT-driven productivity boom finally arrived. Output per man-hour soared in the U.S. Also, economists who looked at investment in computers at the firm level began to find real, substantial productivity gains. Those findings have been supported by a later generation of research as the pace of IT gains slowed.

Just as Zvi Griliches found with the use of hybrid corn in the American Midwest, the adoption of the Internet has followed a logistic curve -- slow in the beginning, fast in the middle, and then slow at the end. The technological diffusion of the Internet, also like hybrid corn, did not happen evenly across countries. Some moved online faster than the others.

In the graph below, I've plotted the share of population with Internet access in the median country for the years 1990 to 2013, and then the 25th-percentile (i.e., low-Internet) and 75th-percentile (i.e., high-Internet) countries. The median country, as of 2013, has just over 40 percent of its population with Internet access. The gap between the 25th and 75th percentiles, called the interquartile range, is about 50 percentage points in 2013. The data come from the World Bank.

Here's the problem. Countries embraced the Internet at widely different paces, as the graph shows. If the Internet has a meaningful effect on economic growth, then the relative rates of adoption of the Internet should show up in relative rates of growth. Countries that moved online quickly should have grown more quickly than similar countries that moved online more slowly. Yet they did not, at least upon a first glance at the data.

The graph shows the average annual growth rates in GDP and Internet access by country between 1990 and 2013. There is no relationship.

There are more sophisticated ways of doing this analysis, but so far, they have ended up in the same place. (For example, if you predict GDP growth rates based on the level of GDP and past growth, the level of Internet access and the rate of Internet-access growth usually have no additional effect.) I cannot seem to find any effect of Internet access on GDP.

That's why I am making my dataset freely available here. It includes about 5,000 observations on Internet access and GDP for the roughly 200 countries for which the World Bank has data. (If you're interested in working on this, please do send me an email.)

Maybe it's the case that the Internet has an effect on productivity but not the total level of output. In a recent paper, Daron Acemoglu, David Autor, and other top economists showed that the IT boom of the 1990s resulted in higher productivity -- but only because IT-intensive manufacturing firms cut employment faster than they cut output. There was no IT boom in terms of output.

The next step, then, is to find reliable data on productivity that are consistent across countries and span the relevant period. I am currently looking at getting these from the ILO, but they will be iffy at best.

Alternatively, maybe we're just in the Internet's own "productivity paradox" moment. Those who doubted of the economic importance of personal computing in 1993 had to change their tune only a few years later. (I should note here that research at the firm and local level and for the US has found positive effects of the Internet on output.)

Yet, even if we do end up finding some positive relationship between Internet access and GDP at th country level, it would be premature to call it causal. Intuitively, a country that is growing rapidly might choose to invest in Internet access, and richer citizens may buy Internet-enabled phones. The process of investment and economic growth is importantly circular.

It will take more careful approaches to measurement to move from "countries that went online faster grew faster" to "the Internet causes economic growth." Yet, right now, we can't even say the former. So much for the Internet's effect on growth.

Thursday, March 19, 2015

Today's Links

1. Related to my post on rents, Justin Fox shows that a growing share of the value of stocks in the S&P comes from intangibles.
2. Adam Ozimek explains what should be the consensus on the effects of charter schools: "Some charter schools appear to do very well, and on average charters do better at educating poor students and black students."

3. Farhad Manjoo with spectacular photos of New York City at night.

4. Yikes: The credit-default swaps market thinks the probability of a Greek default within the next five years is nearly 80 percent.

5. Matt Rognlie, the MIT grad student who took on Piketty with convincing arguments about the elasticity of capital-labor substitution.

6. Cool 3D visualization of the yield curve.

7. A colorful visualization from the WSJ on how the Fed's forecasts have evolved:

Traveling on a Strong Dollar

"Vacation is all I ever wanted," goes the early-80s song, which I admit I know from watching Rugrats as a kid. And, as the dollar gets stronger and stronger, the newspapers are singing the tune.

The Washington Post has a list of "the best places in the world to visit while the dollar is this strong." And Matt O'Brien writes:
The good news, if you're planning on taking a trip abroad, is that the dollar is on a tear...That's made it an opportune time to take, say, that European vacation you've been thinking about, but maybe couldn't afford when the euro was worth $1.45 instead of the $1.13 it is now.
The whole world is on sale, so travel now, now, now. Or that seems to be the suggestion. Yet I was curious if Americans actually game the exchange rate in traveling abroad. In other words, does foreign travel respond to the exchange rate? Or are people going to use up their vacation days, no matter how unfavorable the currency situation is?

The U.S. government keeps surprisingly good data on travel. That's because, when foreigners come to the U.S. and clog Times Square, for example, the money they spend on plane tickets and tacky "I ❤ NY" t-shirts actually count as exports. And when you go abroad, the dollars you spend overseas count as imports.

Also, the relevant measure that determines the attractiveness of a country, at least as far as currency is concerned, is the real exchange rate. It doesn't help you if the currency is cheap but all the goods are overpriced, so real exchange rates take into account both the exchange rate and relative price levels. I've simply taken the trade-weighted version of the real exchange rate so that we can look at all travel at once.

You can try to discern patterns between inbound and outbound travel and the exchange rate in the graph below. Recall that, when the exchange rate goes up, it makes outbound travel cheaper and inbound travel more expensive. So we might expect outbound travel to rise, and inbound travel to fall, when the exchange rate rises.

It's actually not that clear to me there is a relationship. But we can look at this more systematically. The math gets a little bit involved -- if you want to know more, see here -- but, in short, what we want to know is if the exchange rate rises one percentage point, how much more will Americans travel abroad and how much less will foreigners visit the U.S.?

Using the data we discussed earlier, here's what I found.

Americans don't travel more when the dollar strengthens. But foreigners do cut back on their trips to America. In fact, for every one-percentage-point increase in the dollar's value (as measured by the real trade-weighted exchange rate), foreigners' travel spending drops by half a percentage point. And those cutbacks happen immediately after the shock to the exchange rate.

So, to disagree with Matt Klein here, the data don't support "hordes of American tourists" going to Europe. Now, the work I did here was quick-and-dirty, so a finer look at the data might find something different. I have concerns about 9/11 and the financial crisis -- which affected travel and were not directly related to the exchange rate, but happened to coincide with big moves in the dollar, being an issue here.

Nevertheless, it doesn't look like Americans spring for foreign vacations when the dollar is strong. Yet foreigners do notice a strong dollar, and they respond by traveling less or spending less when they do come. One wonders, though, why Americans don't get their acts together and game the exchange rate like the rest of the world.


Via Twitter, Université Laval professor Stephen Gordon notes that Americans do seem to travel to Canada when the dollar is strong.

Wednesday, March 18, 2015

Today's Links

1. E. Roy Weintraub's "MIT and the Transformation of American Economics" is getting some attention, with recent reviews from David Warsh and Paul Krugman.

2. Via Karan Chhabra, Yale is launching an online physician's-assistant masters program with the intention of expanding enrollment tenfold from the current on-campus base. Relatedly, see this recent Kevin Carey column, arguing that the reason MOOCs were overhyped was that they could not issue official degrees that were valued at par with traditional on-campus ones. Well then, here we go.

3. The flattening of the Phillips curve, illustrated using the 1980s expansion and present:

4. Danny Yagan finds that the 2003 dividend tax cut had no effect on corporate investment, using a comparison of C- and S-corporations. The idea was actually suggested in the final paragraph of a related 2006 paper by Raj Chetty and Emmanuel Saez, which showed that the tax cut led an increase in dividend payouts. Mike Konczal has a nice review of the Yagan paper.

The Rent Hypothesis

"Rent" is three things: an idea in economic theory, a payment for housing, and a musical. This post is, maybe unfortunately, about the first of those things.

Rent, as an economic idea, refers to some gain that, loosely, you can think of as unearned. It is, more rigorously, a payment for a resource in excess of its opportunity cost, one that instead reflects market power.

For instance, profits that a monopolist earns in excess of the profit that would exist in a competitive market are rents. So are excess profits from the production of a patented invention, or the extra compensation that an executive might get because he's buddy-buddy with the board of directors.

Rents are not always bad. Protections for intellectual property make sense not in spite of the rents they generate, but rather because of them. A patent, after all, is a temporary monopoly. These monopoly rents give incentives for innovation and allow innovators to cover the costs of research and development. (Warning: This logic can be taken too far.)

Usually, however, there is something less than endearing about rents. That's because, without some countervailing market failure, as in the case of innovation, giving away unearned gains is bad for incentives and bad for the allocation of resources. An economy where the cost of Internet access (to choose a not-accidental example) is determined by how much can be squeezed out of you and me, rather than how much it actually costs to provide it to us, is inefficient.

There has been, for the last few years, a "big idea" floating around the economics conversation that these rents are growing -- that unearned gains are eating up an larger share of income. Let's call it "the rent hypothesis." It's an appealing idea from a certain perspective. It seems to explain a lot.

Why are corporate profits so high? Shouldn't they be competed away? Why does finance make so much money? Can it really cost that much to do what financial intermediaries do? Why is executive compensation so huge? Is it really possible that, as Bloomberg put it, Larry Ellison is "still a bargain" to Oracle at $100 million a year? And why, if profits are so high, is new-business formation and business investment so low? Shouldn't those profits attract new entrants and encourage existing businesses to expand?

That last point comes from Paul Krugman, who has done some of the brainstorming about rents here and here:
So what’s really different about America in the 21st century? The most significant answer, I’d suggest, is the growing importance of monopoly rents: profits that don’t represent returns on investment, but instead reflect the value of market dominance. Sometimes that dominance seems deserved, sometimes not; but, either way, the growing importance of rents is producing a new disconnect between profits and production.
You can also find Josh Bivens and Larry Mishel discussing rents here. To Bivens and Mishel, executives would still do what they do even if they were paid less -- Larry Ellison is not about to sail off in his America's Cup yacht -- which means that executives are, in fact, being paid above their opportunity cost. (The logic here is, if their pay falls below opportunity cost, they would go do that other, next-best project which determines the opportunity cost.)

They provide evidence that executives are uniquely well-paid in comparison to other top earners, which they interpret as a sign that their pay is about their position rather than their skills or value. Similarly, the rise in financial-sector profits and incomes seems to match up well with the moment those industries were deregulated. And in a recent paper, Thomas Piketty, Emmanuel Saez, and Stefanie Stantcheva show that executives are paid more for the same performance in low-tax countries than in high-tax countries. To them, that's evidence that low taxes on the highest incomes encourage rent-seeking behavior. (Click the pictures to enlarge.)

Yet all is not well for the rent hypothesis.

There's a big problem: We can't observe rents. Corporate profits that are rents don't look different than profits that are earned in free competition. It's not as if the dollar bills are marked "rent."

What we're left with is trying to find evidence that suggests some income is a rent. To return to our earlier example, the facts of high corporate profits, low cost of capital, and low business investment seems to require some non-competitive explanation. High profits and a low cost of capital shouts "business opportunity" -- so why isn't someone taking it?

Trying to spot rents is, in this sense, a bit like trying to spot a black hole. (Or what I imagine it would be like. I am not an astronomer.) You doesn't see a black hole in itself so much as you notice an absence, a void -- and the motion of stars towards it. Similarly, the idea is to spot rents by what is missing, by the presence of a contradiction where only rents can fill the gap.

This is a fun game to play, but it ends up being pretty unconvincing. It may seem like the facts fit together, as in Krugman's account. But there are lots of other ways to get to the same result that don't involve rents. And, to be sure, most of the analyses of rents admit this. Bivens and Mishel: "The evidence on rent-shifting behavior should be viewed not as conclusive, but as highly suggestive."

If what I've given so far is a largely favorable telling of the rent hypothesis, then, let's consider a more skeptical take. The skeptic says, "This 'rents' framework amounts to a way for these economists to denounce economic phenomena that they are ideologically predisposed to dislike. Corporate profits? It's rent! CEO pay? Rent as well! See how easy this is when economists can just point at things that feel unjust or unbalanced, but never really show that they are uncompetitive? Everything they've said so far amounts to a more intellectual version of 'doesn't this look strange?' That's not a real argument."

I should emphasize that the view of "the skeptic" is not what I actually think. Some people think it -- I'm channeling John Cochrane here. I think, though, it is a critique that, if harsh, has some truth to it. Most analyses of rents, particularly on the topic of corporate profits, do an inadequate job proving that the phenomena they describe are the results of uncompetitive behavior. I am not convinced. Nor do I think you should be.

"I think we would have more intelligent economic debates if all participants were asked to state what could happen that would cause them to modify their views," Larry Summers once told Ezra Klein in an interview last year. It is a great comment. What, then, might a more convincing analysis show in support of the rent hypothesis? Here are four ideas.
  1. It should explore the relationship between industry profitability, firm entry, and investment. If the emergence of large profits in an industry once, but no longer, led to a response of an increase in new-business formation or an increase in business investment, I would be much moved in favor of the rent hypothesis.
  2. It should explore the determinants of profitability at the firm level. If the distribution of profits among firms has shifted in ways that are consistent with the rent hypothesis, I would find that evidence to be strong. For example, if the age of a firm, after accounting for firm size, was more strongly associated with profits than it had been, that would strike me as important. The idea is that the apparent return to incumbency would have risen. Relatedly, if patent count, or some measure of intellectual-property holdings, has become a more powerful predictor of profits, that would also be strong evidence.
  3. It should account for the recent increase in aggregate corporate profits in ways that would be consistent with the rent hypothesis. For example, has the increase been mostly driven by large firms? By firms in low-entry industries? By firms in industries were size or age or entry has changed most significantly in recent years? By firms in industries where government regulation has been relatively high or low, or where it has increased or decreased?
  4. It should explore the evolution of competitiveness. How has the number of potential sellers of major product categories evolved at the local level, and is there a measurable effect on profitability? For example, if the average number of telecommunications providers competing in a zip code (or county or state) have declined and total profits in the telecommunications industry have risen, and that pattern was repeated across other industries, that to me would be compelling.
The common thread among these ideas is that they test the rent hypothesis systematically. It must be forced to make predictions about the economic world that either pan out or do not. I'm writing these out in hope that someone thinks the ideas are interesting and worth giving a try, and I might myself.

The rent hypothesis has a lot to say about our economic world. If it's right, it provides a unified explanation of many important shifts. It gets to why corporate profits are so high, why the financial sector has expanded so much, why executive pay has risen, and why entrepreneurship and investment are struggling. Time to move beyond speculation and toward analyses that have a real chance of telling us new facts about how our economy is changing.

Tuesday, March 17, 2015

Today's Links

1. Atif Mian and Amir Sufi present new evidence that overstatement of income on mortgage forms helped drive the housing bubble. As you can see in the graph below, the very poorest zip codes had, in 2005, the greatest difference between income reported to the IRS and the income reported on mortgage applications. They had the greatest growth in subprime lending. Matt Klein has more here. As a disclosure, I worked as a research assistant for Mian last summer, but I was not involved in this project.
2. A remarkable piece from The Washington Post's Jonathan Capehart on why Ferguson should be a symbol of institutional racism but Michael Brown shouldn't be seen as the embodiment of one of its helpless victims.

3. Via Hal Varian, a really great and much-needed introduction on identifying causality in economics. Useful for people who want to understand how good economic research works.

4. Devi Lockwood, a friend of mine, has been maintaining a blog as she travels the world to write and collect stories of interactions with the environment, especially on water and climate change.

5. Neil Irwin on how the rapid appreciation of dollar is creating an economic shock in developing economies.