More on the Late Paul Samuelson

*Augean* stables?

Samuelson is the poster child for what’s wrong with economics. bad assumptions shrouded in intimidating language (math). what has been the efficacy of economics over the last 100 years other than perhaps “confirming” the obvious fact that people acting voluntarily maximize happiness.

thanks, keep up the good work…(sarcasm off)

Mathematics is a useful tool, but economics is far too complex to solve much more than simple problems. Have you ever tried to solve a rate of change equation (like heat or wave motion) with three variables? Solvable, but these are well defined functions with known relationships.

Not only does economics have many more (often very complex) variables, I would argue that economics does not even follow logical rules. There is too much emotion, too much politics, too much illogic (where people do not even do things in their own self interest, particularly in the long run).

Differential equations must have boundary conditions, and the (solvable) solutions follow only a particular form: generally e to the i x.

I am not a math expert, but to try to shoehorn economics into what is solvable does not seem very useful to me at all. To rely on a formula when you do not know all the variables, or how those variables interact in the real world, sounds to me like madness, which could very likely result in a complete breakdown of our economic system.

Oh.

Never mind.

When comparing Samuelson to Mises, the first thing that comes to mind is that Mises predicted in the 1920s that the Soviet Union would be a failure, while Samuelson, as late as 1985, claimed that “The Soviet model has surely demonstrated that a command economy is capable of mobilizing resources for rapid growth.”

Both by training and by natural inclination, I tend to put problems into a mathematical framework; I am not satisfied unless I can do so. But, as a non-economist, I have to ask: what good has mathematics done to economics, if it could not even be used to figure out that the Soviet Union was going to collapse?

Apologies to economists for my being blunt.

Change will come from the fringes of the profession, not from the center. If it comes at all.

But Samuelson’s own contribution refute this. Samuelson changed the economics professions in a certain way and he was definitely an insider.

If one really wants to fancy themselves among the technocratic elite of Economics, they would assert that Comp Sci > Math when it comes to languages that best model Economies.

You can simply do much more with computer simulations than a linear regression will ever get you; of course, comp sci does require math.

”
zefreak writes:

David:

Austrians do borrow insights from Behavioral and Experimental economics (except perhaps those at LvMI, but they would if they bothered understanding it). Behavioral Economics actually fits very well within the Austrian paradigm, and even substantiates both the complexity and subjectivity of utility scales, an important aspect of the Austrian case against intervention.

Austrians mainly focus on the normative philosophy of the major Behavioral Economists (Ackerman, Sunstein et al). While they are often rigorous in their research, these economist often offer naive policy prescriptions that don’t even consider their own research on bounded rationality.

It seems that their goal is to ‘fix’ these cognitive bias and maximize rationality (in the ‘driven by thought’ sense of the term, not the driven by incentives usage which is found in Austrian literature), even though their very research shows that frames and forms matter.

I find that their naive ethical views drive their policy advice, as much of their research lends credence to the Austrian view of the impossibility of top down utility maximizing.”

This is very interesting – I thank you for the thought here because I do think it has an interesting validity.

But, you do not mention institutions – which, although Austrians, in the way you point out, do not focus on the complexity of individual’s thinking in their “policy prescriptions” or their theoretical results that comes from their assumptions–their assumption tend to implicitly ignore those things–because they base their models often on just an institutional framework.

On the point of methods and methodology: the non-mathematical may sometimes allow even more lack of declared assumptions though its hard to get worse than most mathematical economists. However, when they do make assumptions explicit (which they often do) Austrian assumptions are institutional and relatively straightforward.

Although they do not focus on a complexity of mindset, many times, they do not rule it out either: they assume only institutional structures, however they assume that these institutional structures dominate actions at least to a significant degree that all-else-equal they may alter outcomes toward the ends that result from the policy that effects them.

In other words: if a policy changes the incentives that result from the institutions in the society (e.g., soften a budget constraint) it is assumed that all-else-equal this will tend toward a certain outcome (e.g., the actors will spend more and take more risk) which in itself says nothing about the other more complex behaviors of the mind–what would behavioral economics say about what these actors would do given softer budget constraints, all-else-equal?

Would behavioral economics suggest that, all else equal, a softer budget constraint would not tend to produce a firm that tends to worry less about costs, take more risks if given chance for risk, and possibly spend more? Or how would this be affected by taking bounded rationality even more into consideration (I think bounded rationality is certainly to be assumed)?

Thanks for engaging this topic. 🙂

David, Marty and other math-negative or math-neutral commentators:

The value of mathematics in economics isn’t that mathematics will make more accurate or more correct descriptive or normative comments, per se. The value lies in the fact that mathematics forces a logical and process oriented approach to issues, thus making it easier to test for and identify invalid assumptions, spurious methodology, or outright distortions. Mathematics also provides a common frame of reference and common “language” that is generally much less prone to subjective interpretation by the originator and the recipient.

Can mathematics lead to oversimplification of problems and issues? Certainly, but it’s still a better approach than, “People sometimes act complexly or irrationally, so models of their behavior will always be wrong.” Additionally, mathematical economics can and has gone beyond simplistic regression analysis, three variable equations, and even calculus. Some of the professors at my university are doing economics research involving feedback loops, hysteresis effects, and multidimensional vector and array analysis. While we may never achieve modelling perfection, advances in economics, mathematics, and computational power should be able to provide better normative and predictive models.

I just read this post after reading another blog using plenty of economics jargon to explain how pulling this lever or that lever will improve the economy (i.e. jobs and spending). The problem is very few people sit down and do the kind of math economists do. Maybe the government and large corporations do it. But not the small businesses that employ most Americans and not the households that do most of the spending. Economists are constructing formulas after the fact to explain why people did what they did.

liberty:

Thank you for your response, I found much of what you say about institutions interesting and I agree that Austrian research is heavily rooted in institutional analysis and the incentives they create (and rightfully so). This means of course that the underlying motivation of the actor is unimportant, as it is the incentive structure that is questioned rather than the particular end.

As per the rest of your post, if I may clarify a few things and eliminate a potential misunderstanding.

My last three paragraphs were not directed at Austrian economists but the ‘libertarian paternalism’ of pop-behaviorist literature such as Nudge. Many of the well known behaviorist researchers seem to share these views, and their policy proposals betray an incomplete understanding of the implications of bounded rationality.

As an example, Behavioral Economists criticize neoclassical economics for ignoring what is known as ‘anchoring bias’. This is partially a strawman, as neoclassical economics ‘rightly understood’ (Bryan’s phraseology) does not really make the mistake, but many in the field do fall victim to it. Anchoring bias is the cognitive shortcut of using relative rather than absolute measurements.

For example, a naive rational choice model would submit that $1 million dollars is better than $500 thousand dollars, and that a rational person would choose the former rather than the latter. However, Behaviorists (and Austrians) argue that such simple comparison of magnitudes doesn’t take into account the unique psychological ‘frame’ of the individual, such as accepting lower absolute magnitudes of compensation in exchange for a greater magnitude relative to other people.

So obviously, frames matter. However, the Behaviorists that subscribe to libertarian paternalism seem to forget this fact. They focus on the fact that cognitive bias affects choices in a way that, to a (naive) rational being, is sub-optimal. Their policy proposals generally ignore the way that even interventions that seem harmless can affect action. Think of a child wanting to clean his room of his own accord and then being told that if he doesn’t clean his room he will be punished. That sure dampened my cleaning spirit as a child!

Ultimately, I find behavioral and experimental research to substantiate many Austrian and subjectivist theories. I think a synthesis of behavioral research is both possible and necessary.

Carl, you’ve obviously not read any DSGE papers in the last five years or so. I’ve seen multiple articles using every single aspect of reality you assert must be assumed away.

While I agree that there should be a better balance between intuitive discussion and mathematical precision, blatantly misrepresenting the opposition is a guaranteed way to be sure your views aren’t taken seriously. If you’re going to attack straw men, you have to at least be more cunning about it.

I think Marty is right that Samuelson’s great contribution was as a technician, which I think was a step in the right direction. I’m sure that few here would complain about all the math that meteorologists and climatologists use, their often faulty predictions and erroneous interpretations of data notwithstanding.

Mathematical econ allows us to not only know exactly what an author means, but why their beliefs (assumptions) lead to the conclusions they do and in many cases when the empirical implications are not found in evidence, which assumptions failed. As we’re all aware, though, the discipline has gone astray in viewing technical prowess and good ideas as substitutes rather than complements.

“Those mocking Samuelson for his statements on the Soviet Union would be wise to remember that most Sovietologists – both economists and political scientists – badly overestimated the strength of the USSR as late as 1991.”

Many Sovietologists badly overestimated their *military* strength, and *political strength* but those studying the Soviet economy (e.g., Nove, Davies, Harrison, Wheatcroft, Kornai, Rutland, etc.) could see how badly it was mangled.

“Samuelson’s opinion was correct based on the apparent figures of Soviet GDP growth provided then, which turned out to be overestimates. You can hardly fault someone for having the wrong conclusions when given the wrong data.”

Except that (1) all those named above could tell you those numbers were fakes, (2) the whole problem with mathematical economics is the static models and the aggregate numbers–even if GDP growth *was* 10% or whatever, does this mean the economy was efficient? Does it mean that the people’s demands were being supplied? It turns out NO, it does not. And Mises and Hayek (and all the economists that came before them) could tell you that, because it does not take a lot of economic theory–just a basic understanding of supply and demand–to see that. However, Samuelson could not see that. He also failed to see that year after year, decade after decade, despite these figures of 10% GDP growth and better, the Soviet economy remained the same relative size compared to the US. That, sorry to say, requires some serious blinders.

“To be crass, Hayek also predicted that socialism would rapidly lead to totalitarianism; for all that, Western Europe remains resolutely democratic.”

Hayek argued that *socialism* requires totalitarianism. How did he define *socialism*? Hayek defined socialism as “common ownership of the means of production.” Despite what people call Western Europe today, it does not have common ownership of the means of production–does it? Does the state own all the manufacturing facilities, banks, transport, supermarkets, etc?

Liberty:
Hayek argued that *socialism* requires totalitarianism. How did he define *socialism*? Hayek defined socialism as “common ownership of the means of production.”

Actually, I don’t think that this is accurate: Hayek recognized that nazi totalitarianism had its roots in the “collectivist” cultural climate of Germany — where “collectivist”, or “socialist”, does not imply common ownership of the means of production.

Hayek also feared that the “socialism” of the 1945 Labour Party would lead Britain in a totalitarian direction. Few people seem to remember today that Britain had less economic freedom than most of Western Europe between ww2 and Thatcher. Since Britain did not become totalitarian, Hayek might seem to have been wrong — but actually I think that post-ww2 Britain gives the strongest evidence for the Hayek/Friedman hypothesis, that democracy requires economic freedom. When British socialism faced collapse in 1979, the Labour Party could have saved socialism by abolishing democracy, but they chose to save democracy instead. That they had to make the choice shows that Hayek was right.

Other European countries that had a British level of central planning were less democratic; e.g. in Sweden the Social “Democratic” Party had a near-monopoly of political and media power.

Liberty:

Thank you for your thoughtful responses. You have picked up on a very interesting idea, that of analyzing policy and its system of incentives similarly to those of institutions. At a fundamental level, I do not think there is a difference. Private property is itself a politically crafted institution (albeit developed via evolutionary process. Hume and Hayek are both insightful on this point) and so can’t be strictly differentiated from ‘softer’ regulations like rent control or reserve limits. In the case of private property and free trade of property titles, the institutions created did have meaningful incentives, such as internalizing cost and allowing such complex processes as the division of labor.

(I just want to be clear that when I say private property is political policy, I do not mean that it was crafted top down by some form of government. Rather I mean that, while developed through centuries by those societies whose benefits it discovered, it requires a territorial monopoly via a legal system that protects its claims and is therefore political in nature)

The problem I have with libertarian paternalism is similar to the knowledge problem with some public choice thrown in: firstly, their regulatory nudges can affect behavior in unpredictable ways, and (according to their own research) can diminish the agents satisfaction even if the stated goal is achieved if it does not adequately and completely take into account the motivations and frames of the individuals.

Secondly, there are public choice arguments against such ‘soft’ regulation. I won’t go into details, but I probably shouldn’t have to on this website 🙂

Ultimately, these are largely practical and not theoretical objections. It is certainly possible, perhaps probable that there are policies that would shape incentives in such a way as to increase on average the wellbeing and happiness of the average individual. In the absence of complete knowledge however, including the unobservable knowledge of preference and frames (except through action, I am definately Misenian in that regard), I am skeptical of additional regulation and default to the regulatory system (basic private property) that best supports the discovery process.

I also didn’t mean to solely pick on Behavioral Economists for their policy proposals. Austrians of the Rothbard/natural law variety can definitely share the blame of naive normativity 🙂