« Opportunism Knocks? | Main | Opportunism Knocks II »

On The Baker Test

Brad DeLong is doing yeoman's work knocking down people who've tried to knock down Dean Baker's No Economist Left Behind Test where we ask privatization advocates to justify their ability to combine the belief that future economic growth will be slower than past economic growth (thanks largely to slower population growth) with the belief that future stock market growth will be as fast as past stock market growth. First, he looks at a report by the Council of Economic Advisors which notes that this is mathematically possible, but fails to note that their scenario requires a steep market crash in which shares lose over 60 percent of their value. Second, he finds Andrew Samwick making the numbers add up, but it turns out to only work if corporations spend "almost 150% of their net earnings on cash flowing to shareholders, and running down their capital stocks." I'll continue outsourcing the economics on this one to Brad, since the "left behind" economists are cleverer than I am and can consistently come up with these little sleights of hand that trick me but don't trick him. Looking at this, however, I am reminded of something I do know a little bit about: Imre Lakatos' philosophy of science.

Lakatos starts out with the observation known as the "Quine-Duhem Thesis," of which I can't find a good explanation online because Google keeps giving me either course syllabi or else to term paper mills which I won't link to. The long and short of it, however, is that you can never have a decisive test of a body of scientific theory, because it's always possible to reconcile a recalcitrant data point with a theory you want to preserve by means of adding additional ad hoc auxiliary premises to your theory. The Samwick and CEA replies to the Baker Test are good examples. You can reconcile the "hard core" assumptions of pro-privatization economics with the facts about the relationship between economic growth, profit growth, and stock market growth by pulling some rabits out of your hat on an ad hoc basis. Maybe the stock market will tank at just the right time. Maybe companies will adopt insane divident policies that doom them to bankruptcy in the long run.

And, of course, maybe these things will happen. I judge them as unlikely, as does Brad, but they might happen. Certainly, if you treat the key premises of pro-privatization economics as sacrosanct elements of the hard core, something like this (or some combination of similar things, though to a lesser degree) has to happen. The hard core of premises require the auxiliary hypotheses. They may seem unlikely, but they're required, and not impossible, so why not adopt them?

Some might think that the availability of these sorts of ad hoc moves essentially doom the prospects of rational inquiry into the world. It's all a bunch of totally arbitrary choice. Theories are underdetermined by data, we can save any set of commitments we want by adding auxiliary hypotheses, and it's all a kind of scam. Lakatos thinks, however, that we can save science by distinguishing between "progressive" and "degenerative" research programs. The issue here has to do with the nature of our auxiliary hypotheses. When you're engaging in a progressive research program, your auxiliary hypotheses are useful in generating new knowledge about the world. Hence, even though experimental anomalies may persist, and even though the ad hoc addition of new premises may look a little dodgy, your program is progressive science, making the sort of valuable contributions to human life that we expect from science.

A degenerative research program's auxiliary hypotheses aren't like this. The only function they serve is to shore up the hard core of premises you started with. They reconcile the recalcitrant data with the old theory, but don't produce new knowledge about the world, or even new question we should ask that might increase our knowledge.

The Samwick/CEA approaches, especially the CEA's more microeconomically plausible auxiliary hypothesis, strike me as pretty clearly degenerative. If Andrew Samwick really thinks that corporations will start giving 150% of net earnings to shareholders, this would be a very interesting new piece of knowledge about the business environment with implications far beyond the narrow issue of Social Security privatization. Something or other important ought to follow from it. We might want to revise our theory of the firm, or say something new about the operation of the market in general. Similarly, if the CEA really thinks the stock market is about to tank, they might want to ask why this is going to happen and what, if anything, we can do about it. The consequences of such an event would be severe, and the CEA is supposed to be producing policy-relevant research to guide national economic policy. Can we engineer a "soft landing" of some sort? What does it say about the past several years that the market is so grossly overvalued.

One could go on. It's clear, however, from the privatizers' lack of interest in such arguments that they don't really think their auxiliary hypotheses should be incorporated into our overall understanding of the American economy. This is not to accuse either Samwick or the CEA of doing anything as crude as lying. Rather, they are toying with a mathematical problem posed to them by Dean Baker and other Social Security supporters. They're looking at the numbers, looking at their hard core of premises, and showing that, yes, it's possible to save the premises in light of the data. But this is, as I was saying above, always possible in some sense. At some point, you need to stop playing the game ("how can I save the premises...") and start asking if the moves you need to make are really sensible moves. A good test is whether or not you're willing to pursue your new auxiliary hypotheses and stand by them as useful principles for purposes other than saving the hard core. Are CEA members advising their friends and families to base their investment decisions on this forecasted crash? One is inclined to guess that they are not.

February 6, 2005 | Permalink

TrackBack

TrackBack URL for this entry:
http://www.typepad.com/services/trackback/6a00d8345160fd69e200d8342256d153ef

Listed below are links to weblogs that reference On The Baker Test:

» Okay, I'm convinced from Indefinite Articles
Given the following fact: The SS "default" date has perpetually been 30 years in the future for the last 15 years. And given my belief that we are just now starting to enter an era of compound-interest GDP growth, I... [Read More]

Tracked on Feb 6, 2005 7:13:41 PM

» SUNDAY MORNING
YGLESIAS
from MaxSpeak, You Listen!
Atrios is right: these two posts by Matt are great. I had missed this one, which is also good. I especially liked this bit, which Matt found in reading of all things the George Bush Budget's section on the Railroad... [Read More]

Tracked on Feb 13, 2005 11:05:56 AM

» Probability and Projection from The 80/20 Club
The probability of me agreeing with Alex Tabarrok on a policy issue is slight. But not null. For instance he and I both agree with Michael Schrage on the importance of assigning probabilities to their reports. It's time to require... [Read More]

Tracked on Feb 21, 2005 1:41:26 PM

Comments

http://www.j-bradford-delong.net/movable_type/2005-3_archives/000288.html#comments

Brad DeLong's example does not make a case for a severe bear market or for not investing in stock in our current retirement or taxable accounts. Actually the example gives us reason to be confident about long term stock market returns. The example simply suggests that if long term economic growth is about 1.9% as the Council of Economic Advisors projects, we should not expect nominal stock market returns of 9.5% [6.5% real returns].

We are assuming the p/e ratio stays a constant. Growth is assumed to be 1.9%. Then growth in earnings can be 0.9% and stock buybacks can be 1.0% which gives 1.9% stock returns. Add in current dividends which are 1.6% for the S&P Index after Vanguard costs. That gives us 3.5% real stock returns [6.5% nominal returns].

But, the Council of Economic Advisors assumes real stock returns can be 6.5%. There is a gulf between 3.5% and 6.5% that can only be made up if stocks fall in price raising the dividend or if the current p/e actually rises to remarkable levels or if economic growth is faster than 1.9%.

A 6.5% nominal long term return for the S&P Index would still make for a better return than long term bonds. This is a bullish argument for stocks, just not as bullish as the Council of Economic Advisors argument.

What I did that was different than Brad DeLong is to add the current S&P dividend of 1.6% to the projected return. Should long term economic growth be more than 1.9%, stock market returns can be more than 6.5% with a constant price earning ratio of 20. But, then there is no reason to borrow massively to set aside Social Security for private accounts. Social Security is fine and we may hope the economy grows at least 1.9% over the years so that the stock market will give a reasonable return in our other investments.

Posted by: anne | Feb 6, 2005 2:08:05 PM

"since the "left behind" economists are cleverer than I am and can consistently come up with these little sleights of hand that trick me but don't trick him."
Implying that they know every word they write is bullshit. Hold the front page!

Posted by: John Isbell | Feb 6, 2005 2:12:49 PM

This is off-topic, but my reputation is ruined by now anyway. I've been waiting for a Rawls reference, and Lakatos will have to do:

"Although John Rawls' A Theory of Justice (1971) could be construed as alluding to or reflecting, or in some way speaking to or about, politics, it was a distinctly contextless work written by a professional philosopher which lifted the perennial debates about liberalism and the ground of values to a new level of abstraction while apparently allowing academic commentators to believe that they were actually saying something about politics."

John Gunnell, The Descent of Political Theory, Chicago, 1993, pp. 272-3. (Citation slightly adapted in order to leave Nozick out of it.)

John Gunnell, before becoming a political theorist, reportedly served as an aide to Big Daddy Unruh of the California Legislature -- one of the most practical of practical politicians.

Posted by: John Emerson | Feb 6, 2005 2:12:57 PM

I am unaccountably reminded of Star Trek: specifically, how Geordi (Next Generation was far and away the worst offender, though Voyager had its run, too) would invent or adapt a piece of technology to solve the Problem of the Week. Next week, however, the implications of Geordi's discovery would be forgotten. Harmonize the deflector array with the impulse drive to generate a tractor beam capable of shifting the doomsday asteroid from its trajectory? Great! But harnessing such force ought to allow Federation starships to do all sorts of other things, like sling Romulan ships across the sky the next time

Oh, gosh. I'm geeking out, aren't I. Um. Anyway, that's what degenerative research programs made me think of. Trekian technobabble.

Posted by: Kip Manley | Feb 6, 2005 2:14:12 PM

Possibly the same sort of criterion could be used for analyzing legal arguments--constructive when you can willingly use them in other situations, and degenerative when they are used to create a specific desired result, and then disappear, never to be seen again. For some reason, Bush v. Gore springs to mind.

Posted by: matthew wilbert | Feb 6, 2005 4:14:03 PM

I agree, classical liberalism is a degenerative research program. Hayek, Friedman, Posner and Buchanun are frauds.

Posted by: zx | Feb 6, 2005 6:28:55 PM

Just for the record, you've presented enough information about this, with a reasonably sober voice and style, and I am at the same time optimistic enough about the future growth of our country's GDP that I have decided that you're probably right - 40 years is a long time to pre-solve a problem that seems unlikely to be as severe as it seems right now.

And I say this as a unabashed free-market capitalist. I'd rather focus on (as you say) things that will help our GDP to grow rather than fixing nebulous future accounting forecasts.

More here if anyone cares to read.

Posted by: john brothers | Feb 6, 2005 7:26:27 PM

Warren Buffett said it all in the fall of 2000. Either growth has to be faster than we know is possible, or the share of capital has to rise faster than we know is possible, to legitimate the kind of projections that people are making about the mean return to investment in equities at this point in time.

The United States was built on this kind of fantasy, and it worked for two centuries. But our luck has run out. You can dip into that pool only so many times and come up with a pearl. There are no pearls left down there.

Posted by: knut wicksell | Feb 6, 2005 9:16:33 PM

If DeLong says it, then you know it HAS to be wrong. What a moron.

Posted by: Al | Feb 7, 2005 1:09:10 AM

I had always heard it called 'preserving the appearances', from Ptolemy's epicycle upon epicycle of spherical orbits to fit planetary observations to the geocentric theory.

Posted by: tib | Feb 7, 2005 8:59:34 AM

One huge problem with the Baker idea and I was shocked to see an international economist like Krugman pick it up.
Why is everyone stating that profits (and thus dividends and stock prices) of companies listed in the US be directly limited by growth in the US economy? Influenced, for sure, but limited by? You think Mocrosoft makes most of its profits in the US? Intel? Exxon? There’s a global market in capital. US companies invest overseas and overseas companies list in New York.
There is no reason whatsoever for future growth in stock prices and or dividends to be limited to the future gowth of the US economy. Limited to the future growth of the world economy (which is expected to be a great deal higher than that in the US), and to the relationship between US capital invested in it, yes, (whether by US companies investing abroad or outside companies listing in the US) but the 1.9% expected growth in USA GDP? Nonsense and Baker knows it.
The whole set up of the question is fraudulent becuase it ignores the effect of a global market in capital. If Baker didn’t know it then Krugman most certainly should.

Posted by: Tim Worstallt | Feb 7, 2005 11:11:29 AM

Problem with this argument, Tim, is that US is running a current account deficit. A current account surplus would be necessary to finance an increase in foreign assets.

In fact, the claims of the rest of the world on the US are considerably greater than ours on them -- that's what it means to say the US is a net debtor, which it is. So if we look at the US in a global context rather than as a clsoed economy, the stock-return propsects for private accounts get WORSE, not better. An increasing fraction of of US corporate profits flow to foreign investors and aren't available for private accounts or other domestic investors.

You're right, tho, that it would be very strange if Baker and Krugman had overlooked the existence of the rest of the world. Next time you find yourself thinking, "either I'm wrong or Paul Krugman has made an elementary economis mistake," I'd bet on option A.

Posted by: lemuel pitkin | Feb 13, 2005 6:19:09 PM

John Emerson,
Thanks for the reminder about Big Daddy. He was the guy that said: If you can't take their money, drink their booze and screw their women and stil vote against them, you don't belong in politics. AH! for the good old days.

Posted by: dilbert dogbert | Feb 13, 2005 8:05:46 PM

John Emerson,
Thanks for the reminder about Big Daddy. He was the guy that said: If you can't take their money, drink their booze and screw their women and stil vote against them, you don't belong in politics. AH! for the good old days.

Posted by: dilbert dogbert | Feb 13, 2005 8:11:00 PM

"If Andrew Samwick really thinks that corporations will start giving 150% of net earnings to shareholders, this would be a very interesting new piece of knowledge about the business environment with implications far beyond the narrow issue of Social Security privatization."

This reminds me of the Boskin Commission's price index adjustments downward to gain savings in calculating COLA increases, extrapolated to measure inflation and productivity in a way which would have effectively wiped out the 1973-95 productivity slowdown from economic history.

Posted by: Russ Hicks | Feb 15, 2005 12:40:39 AM

The comments to this entry are closed.