Friday, January 21, 2011

Cowen, Murphy, and Hayek

Tyler Cowen challenges Bob Murphy and ABCT on the grounds of what he considers to be an embarassing co-movement of investment, capital maintenance, and consumption during the boom. Murphy had originally claimed that investment and consumption can move together because capital maintenance is neglected.

A few obvious questions emerge from this explanation from Murphy. The first one that comes to my mind is also mentioned by Cowen - why would monetary expansion encourage investment but not capital maintenance? Wouldn't those same low interest rates that encouraged malinvestments also encourage capital maintenance? It seems to be an awfully convenient thing for Murphy to grab for.

But the whole discussion also reminded me of something that Hayek said in Prices and Production (lecture 2) that has bothered me since I first read it. It solves the co-movement problem without resorting to Murphy's capital maintenance explanation. The problem is, I'm not sure why we should believe this "solution". Hayek writes:

"The raison d'etre of this way of organizing production is, of course, that by lengthening the production process we are able to obtain a greater quantity of consumer' goods out of a given quantity of original means of production. It is not necessary for my present purpose to enter at any length into an explanation of this increase of productivity by roundabout methods of production."

So Hayek says that you don't even need to resort to the neglect of capital maintenance. Investment and consumption move together because the lengthening of the capital structure makes investment more productive.

I wish he had explained it more because I see absolutely no reason to think this is true. He's not talking about technological progress that happens to elongate the capital structure. He's simply talking about increased roundaboutness itself. Why does that - in and of itself - make things more efficient? Does anybody believe this?

This line always bothered me for precisely the reason that I think Murphy's explanation bothers Cowen - it seems like a very convenient solution to the problem that is also conveniently short on details, citations, or empirical support. Take this assumption away and (it seems to me) you still have a theory of the business cycle that makes some sense (unsustainable changes in the capital structure - malinvestments that might need to be liquidated later on), but one that seems less binding theoretically (why can't we just grow into these malinvestments - clearly some roundabout production is necessary, and regular growth rates should allow us to eventually make use of investments that, just a few years ago, where malinvestments) and less convincing empirically (Tyler's co-movement problem reemerges).

UPDATE: Peter Boettke has thoughts here.


  1. That quote doesn't explain why there can be co-movements. Remember that production takes time. That quote explains why a lengthening of the structure of production NOW, will lead to greater production of consumer goods LATER. It's not talking about the simultaneous increase in consumption and investment that Cowen is talking about.

    I have class, but to answer your question in the post: the lengthening of the structure of production leads to the creation of capital goods meant to increase production in later stages, and so these capital goods themselves increase the productive intensity of the next stage (otherwise, why would anybody buy them?). It can be based on existing technology or on newly developed technology (technology is a corollary of capital accumulation, in any case).

  2. Co-movement is not the same as contemporaneous, though. Both are pro-cyclical. That is the point, and its not the same as saying that one doesn't precede the other.

  3. To clarify - I agree with you that the comovement isn't "embarassing" for Austrians at all. I just think:

    1. Some of the replies aren't always the greatest ways of dealing with it, and

    2. Austrians need to recognize that one of the primary implications of the comovement of this data is that we operate below full employment. Some Austrians are willing to accept this - many aren't. If we are below full employment, there is no conflict at all between basic ABCT and co-movement of the data.

  4. The other option for satisfying observed co-movement is to do what you and Garrison do - say "credit allows you to push out both to an unsustainable level".

    That's all well and good, but now suddently you have people talking about two different things... are we talking about a production frontier or are we talking about a sustainability frontier? Are we talking about full employment or aren't we? Garrison is not wrong he's just talking about something slightly different - and if we don't realize that it can confuse the discussion.

  5. Austrians always operate under the assumption of under employment. Austrians don't believe in optimized economic preformance, so there is always unutilized economic resources (Hayek made this clear in P&P).

    I don't think you can push consumption to an "unsustainable" level. Consumption is what decides the degree of capital intensiveness. I think investment is the specific part of expenditure which may be unsustainable.

  6. Jonathan, I think you're clearly wrong on both counts.

    Let's start with the second one first. Would you say consuming 100% of output is "sustainable"? Consumption and investment can be done unsustainably, and sustainability is something to think about - I just wanted to highlight that Garrison frames things slightly differently from how they're usually framed. Where is full employment in Garrison's models - that's a good way to think about it. You don't know. You can't know.

    As for the first one - some do and some don't. More than enough defend even the straw-man version of Say's law and dispute the prospect of general gluts. Some, of course, are fine with the idea. But this gets back to my problem with Garrison's exposition of Keynes on pages 158-159 of Time and Money. I know Garrison doesn't consciously think about being at full employment all the time - but he slips into writing as if we do. There is a big difference between the prospect of underemployment (anyone who knows the basic economics of the minimum wage believes that), and prospect of a natural below full employment equilibrium. Some Austrians believe in a natural below full employment equilibrium - many do not.

  7. "Would you say consuming 100% of output is "sustainable"?"

    Sustainable towards what end? The end is what matters. In investment, the end is production. That you can't accomplish that end makes it unsustainable.

    In any case, with an increase in consumption you're not talking about the same type of insustainability. You aren't necessarily consuming more than you're producing; the problem is in the structure of production, not in the level of consumption. Investment isn't matching consumer preferences - that's what creates intertemporal discoordination.

    "As for the first one - some do and some don't."

    All should. Most Austrians do not believe in optimization. If you did, you'd have to also believe that people can't make mistakes, and people always choose the best option and therefore necessarily hold all knowledge. It's just not a realistic series of assumptions.

    Garrison is trying to model different approaches to macroeconomics. I do think that Garrison's exposition is poor. But, Garrison mentions several times that his exposition is over simplistic.

  8. Daniel, you said,

    "I wish he had explained it more because I see absolutely no reason to think this is true. He's not talking about technological progress that happens to elongate the capital structure. He's simply talking about increased roundaboutness itself. Why does that - in and of itself - make things more efficient? Does anybody believe this?"

    Believe it? I think it rivals in importance any other supposed factor which accounts for economic growth - most notably technological progress and the division of labor. In fact, it is best to understand the benefits of roundaboutness as similar in nature to the benefits of the division of labor.

    Given that you have read some Austrian literature, I would imagine you've encountered the arguments for why roundaboutness is more productive, but your comment seems to indicate that either you did not understand it or were not convinced. In any case, here's a brief explanation of the principle.

    First, the division of labor. If there are N people each producing on their own with their own "production functions" the total product is given by f1L1)+f2(L2)+.....fN(LN)=TP. Each person is like their own factory - like multiple self-sufficient farmers. Alternatively, if the same N people compose a group rich in the division of labor, each person is instead some connected part of a much larger whole. Rather than many small and disparate "wholes" there is now one super whole factory. As a consequence, the number of possibilities available are increased non-linearly. The same N people, using the same amount of labor - L1+L2+....LN - under the division of labor can produce a total product that is non-linearly greater in magnitude than it would be under autarky. It is just an example of the idea of "emergence" in any complex system - for example consciousness emerging from the interaction of a web of neurons (alone those neurons can do little).

    Roundaboutness is beneficial for the same reason.

    Consider a fixed amount of time T (e.g. 1 year). Over the course of one year there are many options available for how the product is produced. But for simplicity, suppose the options given to you are: 1. Use a process that finishes in 1 month; 2.One that finishes in 3 months; 3. 6 months; 4. The full year.

    Over the course of the year each process will generate a certain amount of total product. In process 1 it will simply be whatever is produced each month times 12; for process 2 what ever is produced after 3 months times 4; process 3 the mid-year product times 2; and process 4 the product at the end of the year.

    The same amount of labor is employed over the course of a year in every process, yet it should be clear that the longer you go without accessing the fruits of your labor the greater can be the product in a non-linear way. It is the same as the division of labor. If I take on a process that takes a full year without interruption the upper bound for total product is non-linearly greater than if I break it up into a process that churns out a product every month. By going without interruption you can form sub-parts which ultimately combine to form some super-whole.

    And of course, uninterrupted production in and of itself is not productive; just as the division of labor does not necessarily out-produce autarky. A man can certainly take on a process that is more roundabout yet entirely unproductive. The whole point is that the upper bound for production is non-linearly greater. Whether or not the upper-bound is reached or approached is an entirely different argument. But put shortly, under a relatively competitive system with a reasonable amount of variation (and hence experimentation), there is reason to believe that it's benefits (roundaboutness) are certainly reaped (plus the plethora of empirical evidence).

  9. I should also point out one more distinction. The main point above was that greater roundaboutness is so fruitful because it allows for subroutines to be collected into some greater whole - thus allowing for types of processes that are impossible when done in shorter periods-and hence for non-linearly greater product.

    But there is also the very mundane gain in product that will arise because of roundaboutness even if the production technique used is identical for both the uninterrupted process and the more frequently-interrupted process. It is just the principle of compound interest. If I roll a snow ball for 2 hours and set it aside it will be larger than a snow ball that's created by rolling two separate snow balls (each created in 1 hour) and then sticking them together (ignoring that the bigger one will be heavier - so harder to push). The same labor time produces different output by virtue of the fact that the uninterrupted snowball accumulates at a faster rate as it gets larger and larger. And this using benefit is reaped even when an identical technique is used in each process.

    What would make the example more like the benefits described in the other post would be if, instead of just rolling a snowball around, I walked for 1.5 hours to a steep hill, packed a snowball, and let it roll down to accumulate into a far, far bigger snow ball than I could ever have rolled without it. But If I take on output intervals that only mature every hour it would be impossible to reach the hill. The gain is thus non-linearly greater.

  10. Jonathan -
    I would have assumed "sustainable" is "sustainable" towards the end of maintaining the current level of activity. That's what sustaining something is. Being able to do what I'm doing now tomorrow and the next day and the next day. Is 100% consumption "sustainable"? Absolutely not. This is how Garrison uses it with his PPF - not with some discussion of "ends" except for maintaining those levels of activity. Garrison has a sustainable PPF on two axes: investment and consumption. It sure sounds like consumption comes into the picture.

    RE: "Investment isn't matching consumer preferences - that's what creates intertemporal discoordination."

    Right. But you could presumably have a stable and an unstable intertemporal discoordination. If you are well below full employment you could have a capital structure that is discoordinated with intertemporal preferences but is sustainable by virtue of the excess factors of production.

    re: "If you did, you'd have to also believe that people can't make mistakes, and people always choose the best option and therefore necessarily hold all knowledge. It's just not a realistic series of assumptions."

    Although you can certainly use optimization as a modeling assumption without believing it always happens. You can also model mistakes as well to make the optimization less problematic. I've never seen the big difference between the action axiom and marginalist conceptualizations of optimization. They both say exactly the same thing: man will act in accordance with his preferences.

    I wouldn't say Garrison's exposition was poor - but I agree there were problems. I've documented those problems here before.

  11. EdP -
    First, I certainly understand the division of labor. I just want you to know I'm not that dense (although some days I wonder).

    I'm not sure you've made any case for roundaboutness though. You've simply reasserted the point when you say "it should be clear that the longer you go without accessing the fruits of your labor the greater can be the product in a non-linear way".

    Now, clearly some production processes simply take longer than others, and rushing complicated production processes can lower the value of the product. Entrepreneurs will balance these benefits against the cost of employing labor and capital longer. At that optimum length of time, it is less efficient to take longer to produce the product and less efficient to take shorter to produce the product. Let's say you're making a widget and it takes ten weeks to make it optimally - to balance the marginal benefit of the sales revenue of the widget with the marginal cost of the factors of production. If you take nine weeks to produce that widget instead the process you're describing kicks in - the reduction in revenue is not going to be compensated by the increase in cost. If you take eleven weeks the opposite problem is going to kick in. But given this product and it's optimization time, there's no reason to think that - holding labor applied and capital applied constant, stretching out the time is going to make the process more efficient.

    You allude to increasing returns and the division of labor, but clearly this has limits as well. Smith wrote, "the division of labor is limited by the extent of the market". I see no reason to adopt an assumption that increasing the division of labor always makes things more efficient. But it certainly can. Hayek does not seem to be saying about roundaboutness what Smith is saying about the division of labor.

  12. Daniel, what you're describing has nothing to do with roundaboutnes. I'm clearly not conveying the the difference between it and the concept you have in mind. You seem to be thinking of something like eating a meal. If I rush eating too much than I won't enjoy it as much as if I slowed down; but if I eat to slow it'll be too tempting. Of course, everyone has their optimal rate at which they eat their meal so that, by definition, any deviation from that rate will lower their utility. This has nothing to do with roundaboutness.

    Again, roundaboutness has to do with the ability to exploit recipes that are impossible in a shorter amount of time.

    Bohm-Bawerk gave this example. A man walks to the stream with a bucket everyday for his water. Over the course of N days he collects N buckets of water. Now suppose the man, by saving up some of his water, embarks on a project in which he constructs a pipe that draws water from the reservoir. This takes N/2 days. Over the course of N days then, he is able to draw N^2 buckets now instead of N.

    This is possible simply because he is willing to do something that will not yield fruit immediately but only in the more distant future. It has nothing to do with the widget example you gave above. What you seem to have in mind is thinking solely in terms of one technique and then discerning whether it is optimal to hasten or slow the pace of that technique. For example, if we're considering the process of the man walking to the reservoir, you would only ask, "should he walk faster or slower?" Slower will reduce the time he can do other things, but less water will be spilled. Faster walking will increase the time he can do other things but it will result in more water dropped. Obviously somewhere in the middle will be reached. Again, THIS HAS NOTHING TO DO WITH ROUNDABOUTNESS.

    Hayek does discuss it quite extensively in the Pure Theory - see Page 71 - it is even referred to as the "vertical division of labor". Lachmann in "Capital and Its Structure" also devotes a large section to the same principle.

    And I agree with the Smith comment. You can't have much of a DOL without a market. You need some effective mechanism to unite disparate individuals for the DOL. But once the group is integrated and exploiting the DOL, I would argue the returns will be non-linearly greater than the total return if each were operating as it's own factory. But I do not mean to say that this product is unbounded (as population and resources are not infinite)- but for the practical case of say, just 20 billion individuals, all integrated in a rich evolving vertical and horizontal division of labor, I see no reason why income couldn't be 20, 30, 100,.....times larger per capita than what income is now. I know this sounds completely nuts - but it's because people still slightly think like Malthus, even though they don't admit to it.

  13. Right, but that is a completely new project that is viable with a long period of production time available. It's not holding the factors of production constant. Perhaps I didn't make myself clear before. My concern is not that lower interest rates afford the opportunity of additional, more roundabout, production techniques which are more efficient. My concern is Hayek's claim that the application of the same factors of production in a more roundabout way is more efficient. Your Bohm-Bawerk example is not what Hayek describes. Your Bohm-Bawerk example is what I think makes sense. It is Hayek that I think has misspoken w.r.t. roundaboutness. The fact that you think my example of the use of precisely the same factors of production has nothing to do with roundaboutness is encouraging, actually - because that's precisely what I'm criticizing in Hayek.

  14. It is true that the technique is completely different-that is the principal reason why output/head has increased. But my interpretation has always been that what's being considered as constant is the total amount of factor input-or what Hayek would call the original means of production.

    In the BB example the same amount of input - say 30 days of labor - yields only 30 buckets of water when the man walks each day or 900 buckets of water if he constructs a pipe. That's 1 bucket/labor day or 30 buckets/labor day, using the same amount of factor input. The pipe, for example, is really just uninterrupted labor time.

    If you could refer me to where you have gotten this impression from Hayek I'd appreciate it. All of my interpretations come from the Pure Theory - specifically pages 70-80 - in which I think he is very explicit about why more roundabout processes are more productive. It's also interesting because he discusses the problem of the average period of production.

  15. The pipe is not a factor of production?!!?!?!? We seem to have deeper differences of understanding if that's the case.

    I quoted the second lecture of Prices and Production - page 227 of the version.

    I think what makes sense is to say that an artificial lowering of the interest rate makes certain more elongated production strategies plausible when they previously weren't. This is my understanding of Bohm-Bawerk. This the version of Bohm-Bawerk that Keynes discusses in the GT, accepts, but considers to be relatively minor. It is much less tenable to talk about an implicit efficiency in longer production processes.

  16. It is clear that we're not on the same page. I thought that you had gotten your interpretation from more than the quote you posted, but that doesn't seem to be the case. Hayek's comment is precisely in line with what Bohm-Bawerk wrote and all of the examples I have presented. It does not suppose anything less.

    I feel like I'm beating a dead animal.

    If we are to proclaim one thing more efficient than another what can we mean? We mean that given the SAME AMOUNT OF LABOR HOURS process A produces more output than process B. That is, the ORIGINAL MEANS OF PRODUCTION are assumed to be the same in both processes. This must be the case for comparison. In the bucket example the original means of production are the same in each case - 30 days of labor. That in the roundabout process the labor hours take an indirect route in constructing intermediate products does not change the fact that 30 hours of labor is still ultimately used-both cases use the same amount of original means of production. Thus, the amount product/original means of production increases as consequence of a more roundabout route. There is nothing about this example that vitiates Hayek's claim.

    What about his statement leads you to think he has something else in mind?


All anonymous comments will be deleted. Consistent pseudonyms are fine.