I Don’t See Any Method At All

I’ve felt for a while that most critiques of economics miss the mark. They start from the premise that economics is a systematic effort to understand the concrete social phenomena we call “the economy,” an effort that has gone wrong in some way.

I don’t think that’s the right way to think about it. I think McCloskey was right to say that economics is just what economists do. Economic theory is essentially closed formal system; it’s a historical accident that there is some overlap between its technical vocabulary and the language used to describe concrete economic phenomena. Economics the discipline is to the economy the sphere of social reality as chess theory is to medieval history: The statement, say, that “queens are most effective when supported by strong bishops” might be reasonable in both domains, but studying its application in the one case will not help at all in applying it in in the other. A few years ago Richard Posner said that he used to think economics meant the study of “rational” behavior in whatever domain, but after the financial crisis he decided it should mean the study of the behavior of the economy using whatever methodologies. (I can’t find the exact quote.) Descriptively, he was right the first time; but the point is, these are two different activities. Or to steal a line from my friend Suresh, the best way to think about what most economists do is as a kind of constrained-maximization poetry. Makes no more sense to ask “is it true” than of a haiku.

One consequence of this is, as I say, that radical criticism of the realism or logical consistency of orthodox economics do nothing to get us closer to a positive understanding of the economy. How is a raven unlike a writing desk? An endless number of ways, and enumerating them will leave you no wiser about either corvids or carpentry. Another consequence, the topic of the remainder of this post, is that when we turn to concrete economic questions there isn’t really a “mainstream” at all. Left critics want to take academic orthodoxy, a right-wing political vision, and the economic policy preferred by the established authorities, and roll them into a coherent package. But I don’t think you can. I think there is a mix of common-sense opinions, political prejudices, conventional business practice, and pragmatic rules of thumb, supported in an ad hoc, opportunistic way by bits and pieces of economic theory. It’s not possible to deduce the whole tottering pile from a few foundational texts.

More concretely: An economics education trains you to think in terms of real exchange — in terms of agents who (somehow or other) have come into possession of a bundle of goods, which they trade with each other. You can only use this framework to make statements about real economic phenomena if they are understood in terms of the supply side — if economic outcomes are understood in terms of different endowments of goods, or different real uses for them. Unless you’re in a position to self-consciously take another perspective, fitting your understanding of economic phenomena into a broader framework is going to mean expressing it as this kind of story, about the limited supply of real resources available, and the unlimited demands on them to meet real human needs. But there may be no sensible story of that kind to tell.

More concretely: What are the major macroeconomic developments of the past ten to twenty years, compared, say, with the previous fifty? For the US and most other developed countries, the list might look like:

– low and falling inflation

– low and falling interest rates

– slower growth of output

– slower growth of employment

– low business investment

– slower growth of labor productivity growth

– a declining share of wages in income

If you pick up an economics textbook and try to apply it to the world around you, these are some of the main phenomena you’d want to explain. What does the orthodox, supply-side theory tell us?

The textbook says that lower inflation is normally the result of a positive supply shock — an increase in real resources or an improvement in technology. OK. But then what do we make of the slowdown in output and productivity?

The textbook says that, over the long run interest rates must reflect the marginal product of capital — the central bank (and monetary factors in general) can only change interest rates in the short run, not over a decade or more. In the Walrasian world, the interest rate and the return on investment are the same thing. So a sustained decline in interest rates must mean a decline in the marginal product of capital.

OK. So in combination with the slowdown in output growth, that suggests a negative technological shock. But that should mean higher inflation. Didn’t we just say that lower inflation implies a positive technological shock?

Employment growth in this framework is normally determined by demographics, or perhaps by structural changes in labor markets that change the effective labor supply. Slower employment growth means a falling labor supply — but that should, again, be inflationary. And it should be associated with higher wagess: If labor is becoming relatively scarce, its price should rise. Yes, the textbook combines a bargaining mode of wage determination for the short run with a marginal product story for the long run, without ever explaining how they hook up, but in this case it doesn’t matter, the two stories agree. A fall in the labor supply will result in a rise in the marginal product of labor as it’s withdrawn from the least productive activities — that’s what “marginal” means! So either way the demographic story of falling employment is inconsistent with low inflation, with a falling wage share, and with the showdown in productivity growth.

Slower growth of labor productivity could be explained by an increase in labor supply  — but then why has employment decelerated so sharply? More often it’s taken as technologically determined. Slower productivity growth then implies a slowdown in innovation — which at least is consistent with low interest rates and low investment. But this “negative technology shock” should again, be inflationary. And it should be associated with a fall in the return to capital, not a rise.

On the other hand, the decline in the labor share is supposed to reflect a change in productive technology that encourages substitution of capital for labor, robots and all that. But how is this reconciled with the fall in interest rates, in investment and in labor productivity? To replace workers with robots, someone has to make the robots, and someone has to buy them. And by definition this raises the productivity of the remaining workers.

Which subset of these mutually incompatible stories does the “mainstream” actually believe? I don’t know that they consistently believe any of them. My impression is that people adopt one or another based on the question at hand, while avoiding any systematic analysis through violent abuse of the ceteris paribus condition.

To paraphrase Leijonhufvud, on Mondays and Wednesdays wages are low because technological progress has slowed down, holding down labor productivity. On Tuesdays and Thursdays wages are low because technological progress has sped up, substituting capital for labor. Students may come away a bit confused but the main takeaway is clear: Low wages are the result of inexorable, exogenous technological change, and not of any kind of political choice. And certainly not of weak aggregate demand.

Larry Summers in this actually quite good Washington Post piece, at least is no longer talking about robots. But he can’t completely resist the supply-side lure: “The situation is worse in other countries with more structural issues and slower labor-force growth.” Wait, why would they be worse? As he himself says, “our problem today is insufficient inflation,” so what’s needed “is to convince people that prices will rise at target rates in the future,” which will “require … very tight markets.” If that’s true, then restrictions on labor supply are a good thing — they make it easier to generate wage and price increases. But that is still an unthought.

I admit, Summers does go on to say:

In the presence of chronic excess supply, structural reform has the risk of spurring disinflation rather than contributing to a necessary increase in inflation.  There is, in fact, a case for strengthening entitlement benefits so as to promote current demand. The key point is that the traditional OECD-type recommendations cannot be right as both a response to inflationary pressures and deflationary pressures. They were more right historically than they are today.

That’s progress, for sure — “less right” is a step toward “completely wrong”. The next step will be to say what his argument logically requires. If the problem is as he describes it then structural “problems” are part of the solution.

19 thoughts on “I Don’t See Any Method At All”

  1. In the intro economics texts I dimly remember, the circular model of economic activity was held together at 12 and 6 o’clock by the financial system, banks. Since those dim, dark days, the 12 and 6 o’clock positions have morphed from bilge pump functionality to large, some would say disproportionate, contributors to the numbers underlying your analysis. Isn’t it possible that the contradictions you’re describing result from essentially non-productive activity, i.e., finance for finance’s sake, in the mix?

    1. Finance isn’t in the textbook model. Or at most it’s an intermediary between the “savers” and the “investors” of real goods.

      I’m somewhat averse to the idea that we were in the world of Walras and Marshall until big finance screwed things up. I think credit and finance has always been central to capitalism, and it’s never looked like a network of trades in goods. Graeber gets the big picture right IMO.

      1. JWM: “I think credit and finance has always been central to capitalism, and it’s never looked like a network of trades in goods.”

        Okay. But don’t forget relative size. “Finance” and “excessive finance” are two different things.

  2. As an amateur economist and fervent blog reader, this is how I see it.

    “Low wages are the result of inexorable, exogenous technological change, and not of any kind of political choice. And certainly not of weak aggregate demand.”

    “If that’s true, then restrictions on labor supply are a good thing — they make it easier to generate wage and price increases. But that is an unthought.”

    With a good jobs report it will be interesting to see what the Fed does. Will the hawks rally to pressure Yellen to hit the brakes? Seems the doves can point to financial conditions and record low yields on safe assets in order to hold off hikes.

    Today it seems we’re nearing full employment and wage gains, a time analogous to the late 90s when wages shared in productivity only to be cut short by the bursting of the tech-stock bubble and jobless recovery of the Bush years.

    1. I agree it will be interesting. But we should also be a little skeptical that normal monetary policy has much effect on real activity one way or another.

  3. yes, economic theory reminds me a lot of 12-tone music, with its own arcane rules of composition with marginal overlap with music that people can listen to. Economics is its own art form, not useful for understanding the world.

  4. In the 60’s it seemed that Economics indicated what should be done and Political Science explained why it was not done and what occurred.

    Today I would say both have little to say about what should, will, or has been done.

  5. “The textbook says that, over the long run interest rates must reflect the marginal product of capital — the central bank (and monetary factors in general) can only change interest rates in the short run, not over a decade or more.”

    And it says this despite it being settled that there can be no such thing as a well behaved aggregate production function which has the kind of aggregate capital for which the idea of a general marginal product of capital in real terms would make any sort of sense. This was hashed out in the 60’s Cambridge Controversies.

    “Ah, but the aggregate production function is really just a first approximation of what is really going on in General Equilibrium.”

    Except then in the 70’s it was rigorously established that not only can general equilibria exist, but they can exist in an such an arbitrarily large number with arbitrarily bad dynamics that their existence is of absolutely no use for talking about a real marginal product of capital. And that is even given information premises for Walrasian General Equilibrium that are egregiously bad, involving perfect information to some permanent and immovable information horizon, after which there is total blackness.

    Thorstein Veblen argued that the Neoclassical approach to economics was not a scientific approach, and given the way that its descendants behave, it seems that the apples do not fall far from the tree.

    1. Hi Bruce, thanks for the comment.

      Yes, this is part of the weird doublethink here. When things like the aggregate production function are criticized on theoretical grounds, we’re told not to take them so literally, they’re just useful shorthand for describing empirical data. But when it turns out they don’t contribute anything to empirical work, we’re told you have to include them because otherwise your models lack microfoundations.

      given information premises for Walrasian General Equilibrium that are egregiously bad, involving perfect information to some permanent and immovable information horizon, after which there is total blackness.

      Yes. One thing that’s always struck me as weird about rational expectations is that a “shock” just means a change that wasn’t expected.

  6. So if you wrote a textbook that had no theory at all, what would be in it? (I would really like to read that book.)

    1. I wouldn’t write it with no theory. Rather, as Peter T. suggests below, with a bunch of different theories for specific problems, as opposed to one big overarching theory.

      This is a question I think about a lot, as you can imagine, since my actual day job is teaching economics. Right now I’m leaning toward what I guess you’d call a system dynamics approach — focusing on causal links between observable variables. Check out the “business cycles”, “causal relations in the macroeconomy” and “open-economy macroeconomics” documents on the teaching section of this site for some examples. Basically I’d like to use flowcharts rather than equilibrium conditions as the main analytic tool.

      1. “I wouldn’t write it with no theory. Rather, as Peter T. suggest, with a bunch of different theories for specific problems, as opposed to one big overarching theory. ”

        In my opinion, the problem is exactly the opposite: the “neoclassical approach” has some circular definitions; these definitions make the theory very slippery when applied to the real world; a slippery theory cannot be tested empirically. Thus the problem is exactly a lack in the “overarching theory”.

        With the usual disclaimer that I’m not actually an economist, my opinion on what are the logical problems in “mainstream economics” that prevent the construction of a non-circular (and thus testable) model. (mostly it’s the Cambridge Controversy stuff)

        1) The concept of “real value” is defined in two different ways, that shouldnt be used contemporaneously, but usually are.
        The first definition is:

        (a) The “real value” of an object is determined by supply and demand.
        If we take (a) seriously, there isn’t really something like “real value”, the words “real value” are only a shorthand for “the price that the market is currently giving to this object” (discounting nominal factors).

        But there is also another, usually implicit definiton, that is very difficult to pin, but would probably be something like:
        (b) The “real value” of an object is a representation of its standardized social utility.

        The definition in (b) is necessarious in many arguments, such as the idea that a perfect market approximates the real values of the traded objects. For example, thake the idea that someone’s wage in a perfect market is equal to his/her “marginal productivity”.
        If we use the definition in (a), this claim is completely meaningless: “someone’s wage in a perfect market approximates the price that the market sets for his/her work”. It’s obvious that the claim about “marginal productivity” implies the idea that:
        – someone produces some stuff, that has a “phlogiston real value”;
        – the market gets the phlogiston real value correctly, and thus the wage reflects this phlogiston real value.

        The definitions in (a) and (b) are used interchangeably, but are completely opposite one another in terms of causation: in definition (a), the market causes the real values; in definiton (b), the real values (that are implicitly supposed to be preexistent from the market) “cause” the market’s behaviour.

        Compare this tho 19th century “labor theory of value”: according to the LTV, prices approximatively reflect the cost of production of stuff, expressed in terms of needed human labour.
        The LTV is, in theory, testable: one could calculate the “needed labour” for varius goods, check the market price of said goods, and see wether the prices are proportional to labor or not, and if not either change the theory or posit some explanation for this or that empirical phenomenon (for example, that for this or that good there are “market imperfections”).
        If we take the mainstream concept of “real value”, on the other hand, it is totally impossible to define it, so we cannot see wether prices correspond to these “real values” or not, so for example it is impossible to give a definition of a bubble that is not self referential (“a bubble is a situation where the market overestimates the value of some assets relative to the value that it should give to the same assets if it wasn’t overestimating them”).
        Also the concept of “market imperfections” is quite dubious, since the concept of imperfection imples a benchmark against wich to compare actual prices, but we don’t have that benchmark (and we can’t even define it logically).

        2) Wich leads to the Cambridge capital controversy: If I understand Sraffa correctly, his point is that there isn’t one market equilibrium, but a family of equilibria for every possible wage share between “subsistence wages” and 100%. The “neoclassical” theory on the other hand says that the wage share is determined by the “production function”, wich is bogus for various reasons, not least the fact the amount of capital is measured in terms of “real value”, so we cannot even distinguish price effects from quantity effects when we speak of the quantity of capital.
        If we take the “neoclassical” side, we can’t have distributional effects on demand; if we take Sraffa’s side, it is true that the market can work at various levels of wage share, but the output of the system can be either in consumption goods or in capital goods (actually base goods), and the base goods are used to expand the economy, so distributional effects are key to expansions and retractions of the economy (although this requires some addons to Sraffa’s theorier, that are basically supply side).
        Thus in my opinion the correct approach would be that of two different concepts of equilibrium, that actually refer to two totally unrelated phenomena:

        2.1) a supply side “Sraffian” equilibrium, where “the market” reaches the correct relative prices of goods, and alkso the correct mix of produced goods, but not a “correct wage share” (that doesn’t exist in this context), and as a consequence neither a “correct” profit rate, nor full employment or stuff like it, and
        2.2) a “postkeynesian” (for lack of a better word) equilibrium between supply and demand, that depends on cycles of increasing/falling investment, wage shares, financial variables etc. (Steve Keen has a model that works adding a Minsky cycle to a Goodwin cycle that looks interesting to me, but has the bad charachteristic of being defined in “real” terms, whereas I would like one that calculates these cycles in some relative measure).

        The fact that mainstream economics takes the “neoclassical” side of the argument means that we cannot divide the two problems, that causes the whole confusion. This confusion means that we cannot isolate “demand side” effects (2.2) from “supply side effects” (2.1), because we have an amorphous “real growth” that sums price effects, productivity effects, demographic effects, and unemployment effects, so how can we have an empirical discussion of the data?

        3) In addition, if I understand Sraffa correctly, while the relative price of each “capital good” (actually of all goods) depends on the wage share, the aggregate price of all capital goods relative to the value of income should be constant for all wage shares, so the profit rate should depend on the wage share and on some technical capital to income constant. But Sraffa’s theory don’t allow for finance, only for “real” capital (meaning: produced means of production). If we allow for “financial” capital, whose value is just the capitalised value of the profit stream through some assumed interest rate, the “interest rate” doesn’t anymore correspond to the sraffian profit rate: for example, the low interest rates of today might represent an high real profit rate with a financial capital that is overvalued, and thus depresses the interest rate. We can’t see this because the concept of “real value”, as used in the definition (a), cannot distinguish between “real” capital and “virtual” capital. Obviously changes in the value of capital assets (an increase of “virtual” capital) can have big effects on demand through credit channels, but we can’t calculate this if we can’t distinguish between “real” capital and “virtual” capital.

    2. I’d also spend a lot more time on data, measurement, definitions and accounting than most textbooks do.

  7. The post strikes me as right, but what would a better economics look like? I think it would look more like history or climate science – lots of study of local conditions and the details of particular categories of interest, coupled with syntheses across various areas and studies of cross-connections. In short, lots of field work and no simple models. Hard to see the current profession moving that way, though (for one thing, they mostly lack the skills needed).

    1. Incidentally, at UMass people used to refer a bit dismissively to “8th floor dissertations”, that being where our computer lab was. The alternative — quite rare in the profession but the norm there — being a dissertation that involved fieldwork, surveys, or otherwise going out into the wider world.

  8. I believe a better economics starts from the principle:
    Jane Goodall is a scientist. She never tells the chimps: “You’re doing it wrong.”

    [By the by, thank you, on behalf of humanity, for rejecting your disinclination to discuss methodology.]

  9. Long term, I would say this is mostly a China story of labor expansion in world markets. Greater employment and investment there, less here, and lower inflation and interest rates as their investment was largely self funded and recycled here. Shorter term it is a lack of profitable domestic investment opportunities, or more accurately, that such investment would come at the cost of profitability. These imbalances are lessening and eventually innovation will bring new investment and profits.

  10. The Lord (7/13 6:45) says Salvation is at hand if we only wait a long long time. (We’re all dead by the way by then – see JMK)

    I see your Act of Faith and refute you with TR Malthus, 1798, who also played the long long game.

Comments are closed.