Recombinant Growth

I'm going to do something I normally hate doing and that's comment on something I haven't read.

Tyler Cowen has an e-book out arguing American technological progress has plateaued. From the description:

In a figurative sense, the American economy has enjoyed lots of low-hanging fruit since at least the seventeenth century: free land; immigrant labor; and powerful new technologies. Yet during the last forty years, that low-hanging fruit started disappearing and we started pretending it was still there. We have failed to recognize that we are at a technological plateau and the trees are barer than we would like to think.

Arnold Kling notes the irony that an argument about plateaued grow is coming in the form of a digital book for only $4. Of course you could argue that this constitutes "cutting edge" is a point in favor of Cowen's argument.

Based on my educated guess, Cowen's making an increasing marginal cost claim: as you pick the low hanging fruit, you're forced to climb the tree and go after the stuff that's harder to get. I use the same analogy when I teach principles. It works great for basic stuff like picking apples. But new technology is another story.

In a great 1998 paper, Martin Weitzman argues that knowledge causes "recombinant growth," or growth that builds on itself as ideas combine with other ideas. "The paper's main theme is that the ultimate limits to growth lie not so much in our ability to generate new ideas as in our ability to process an abundance of potentially new ideas into usable form."

It's not that we make technology, enjoy the benefits, and start again where we were. If it was, then we would have fallen back into gut-wrenching poverty long ago. The ideas we gain (not to mention the immigrants who help make those ideas) stay with use, which we use to make yet more ideas. These ideas combine with other ideas and make cutting edge stuff easier to achieve than our ancestors ever thought possible. High-hanging fruit doesn't seem high hanging. It's as if the fruit we pick not only nourished us but cause us to grow larger and provided seeds for new trees to boot.

Ignore CBO Estimates

We often hear that the estimates from the Congressional Budget Office (the folks which tell us how much laws cost or save the government) are non-partisan.

The final cost estimate produced by the non-partisan CBO -- that the health care measure would cost $940 billion over 10 years, and bring down the deficit over that same time period. [Source]

The $800 billion federal stimulus bill has boosted employment by 1 million to 2.1 million and helped the economy grow about 1.5% to 3.5% larger than it would have without the stimulus, the nonpartisan Congressional Budget Office said Tuesday. [Source]

The Congressional Budget Office, a nonpartisan watchdog, forecasts that the US will post deficits in excess of a trillion dollars in each of the next 10 years. [Source]

Even the more reasonable Matthew Yglesias will cite their scoring without ever questioning them.

While the CBO is supposed to be neutral, it doesn't take much thought to realize how naive that is. Since its creation in 1975 (thanks to some 1974 legislation), the CBO director's appointed by the mutual agreement of the Senate pro tempore and the Speaker of the House after considering the recommendations of their respective budget offices (though by tradition each chamber alternate this responsibility). In addition, either chamber may fire the director by resolution. So you'd think that if you're the CBO director, you have a pretty strong incentive to be partisan.

If I'm right, then when both chambers are of one party, the director should be that party as well (or at least sympathetic to it). This is somewhat easy to check. CBO directors aren't politicians and don't wear their affiliation on their sleeves, but thanks to some Googling and checking Wikipedia, I estimated the party affiliation of each director (excluding acting directors, which are covered by the white spaces). Here's the list of past CBO directors. The size of the blocks under the CBO heading is the length of that person as the director. If the director changed mid-year, I counted the full year if that person was more than halfway through the year and not in the year at all if he or she left before the halfway mark.


Even if you complain that so-and-so isn't partisan, or so-and-so isn't the party I assigned it, the correlation's strong enough that any single change doesn't change the overall pattern.

The chart above is not to be the definitive data which proves the CBO's partisanship, just an illustration that this neutrality everyone talks about is just hopeful thinking. And I'm first to admit my categorization isn't perfect. For example, the only reason I put Douglas W. Elmendorf as a Democrat is that he was a senior fellow at Brookings, which leans left if it leans any direction at all. The more important point is that of incentives: even if Elmendorf is non-partisan, that he can be fired by Democrats (and only Democrats) gives him plenty of reason not to be.

An Example of Good Regulation

New York Times has a great article about the growing importance of electronics in the stock market. Of particular interest is high frequency trading.

They use algorithms to zip in and out of markets, often changing orders and strategies within seconds. They make a living by being the first to react to events, dashing past slower investors — a category that includes most investors — to take advantage of mispricing between stocks, for example, or differences in prices quoted across exchanges.

High-frequency traders are “the reason for the massive infrastructure,” Mr. McPartland says. “Everyone realizes you have to attract the high-speed traders.”

These trades occur mind-bogglingly fast, with speeds measuring in the milliseconds, or millionths of a second. As various trading platforms (besides the NYSE and NASDAQ, there are about two dozen smaller ones) compete for the high-frequency traders, the bill to stay in the game skyrockets.

One such project is a 428,000-square-foot data center in the western suburbs of Chicago opened by the CME Group, which owns the Chicago Mercantile Exchange. It houses the exchange’s Globex electronic futures and options trading platform and space for traders to install computers next to the exchange’s machines, a practice known as co-location — at a cost of about $25,000 a month per rack of computers.

This is a pure arms race, where value is zero sum and purely relative. At this computing level, doubling the speeds adds nothing to our wealth but costs society billions. If everyone would half their speed, we'd loss nothing (or almost nothing) as a whole AND we'd won't have to spend so much money on these damn super-super-super computers.

The SEC chairwoman, Mary L.Sharpio, has raised the idea of limiting the speeds machines can trade at and I applaud this direction. It depends on the speed that's set, of course, but the efficiency gains between 50 milliseconds and 90 milliseconds is basically zero.

Two caveats. First, it's unclear what the spillover gains from this computing technology is. Firms are expanding the limits of technology to deliver pure speed to Wall Street(s). Such computers might add little to trade efficiency but could be useful elsewhere, say medical areas, especially in the areas of genetics and nanotechnology. Getting this technology faster could save lives.

The second is the unintended consequences. These firms compete on speed: take that away (assuming there are no loopholes) and what will they compete on instead? It could encourage better customer service, but it could also encourage accounting fraud.

But in light of these two issues, I still favor a speed cap. I doubt cutting out these customers for high end computers is going to significantly reduce the investment in high speed computer technology. The second issue I'm a little bit more nervous but I suspect there's plenty of room for honest improvement to compete on.