Two Pointed Questions Posed as Tweets

Posted on by

1. From Josh Hendrickson:

I don’t get it. Everyone has a model; whether they use math/graphs/words. Why are only models w/math denigrated?

Other things equal, it is harder to understand what is going on in a math presentation. Other things equal, insisting on math restricts the sort of assumptions you can work with to those that you find tractable.

That would suggest that verbal arguments dominate mathematical arguments. I am not going to insist that this is always the case, but I think it does create a presumption in favor of verbal arguments. Yes verbal arguments can be vague. But a lot of hand-waving goes on in mathematical papers as well.

So the way I would put it is that today there is a strong presumption in favor of expressing models (or, to use my preferred term, interpretive frameworks) in mathematical terms. I would like to see the presumption go the other way.

Pete Boettke has an essay/post that is pertinent and aligns with my views. Strongly recommended.

2. From someone with the Twitter handle “representative agent’:

I’m thinking about PSST as a business cycle theory. what are its most distinctive implications?

One important implication is that unemployed workers will not be hired back into the same jobs they had before. I believe that in the 1950s, there were recessions that were primarily inventory corrections, so that after you went through a couple of quarters with automobile manufacturers and their suppliers laying off workers, those workers got recalled. Those examples run counter to PSST.

A related implication is that just “boosting demand” in general will not do much to deal with unemployment. The adjustments that are needed are specific to workers located in specific parts of geographic/industry/skill space. It predicts that just throwing money at, say, the green energy industry, will not necessarily increase employment.

Another implication is that shocks to sectors that are closely connected to other sectors (as might be shown by a network graph) will have more effects than shocks to sectors that are more isolated. That may explain why the crash of the dotcom bubble did very little, but shocks to the energy sector in the 1970s and to the banking sector at other times have had severe impacts.

15 thoughts on “Two Pointed Questions Posed as Tweets”

  1. One important implication is that unemployed workers will not be hired back into the same jobs they had before. I believe that in the 1950s, there were recessions that were primarily inventory corrections, so that after you went through a couple of quarters with automobile manufacturers and their suppliers laying off workers, those workers got recalled. Those examples run counter to PSST.

    That is extremely true for the recessions from 1946 – 1982 when most recessions were Fed/Bond Investors interest rate induced. Most jobs had a last hired/first fired/first hired realities and now it is you are laid off for good.

    Again, this is one reason why the S&L recession stung the economy in 1990 a lot more than history assumes because the unemployment was still rising 2+ years later.

  2. It is an interesting observation, that there were workers who could be held ‘in inventory’ – basically unpaid leave – and recalled. These were largely interchangeable and the jobs didn’t evolve much, nor skills apparently degrade materially; the locations were fixed or centralized and the demand varied by percent, not wholesale.

    This is in strong contrast to, for example, big Pharma; in which 8,000 organic chemists could be let go in a single job action. Those workers are dispersed to the four winds and the jobs they were doing do not reappear in some obvious form at the same location.

  4. Math models can contain implausible assumptions, but they can’t contain outright contradictions. Verbal models can contain both.

    Here’s an SAT analogy question: demagoguery is to verbal models as ________ is to math models.

  5. Math can be clarifying or obfuscating, it takes effort and shouldn’t be an excuse to avoid the data or the understanding behind it but as part of a program to advance them.

    Profitable businesses expand and losing ones shrink all the time so it really takes miscalculation/divergent collapsing expectations on a large scale to result in recession, whether from large long lead time investment disappointments, bad accounting, exuberance and despair. More interesting than searches for new patterns is why growth evolves to large imbalances and disappointments lead to retrenchment rather than redoubling of other efforts.

  6. “Other things equal, it is harder to understand what is going on in a math presentation. Other things equal, insisting on math restricts the sort of assumptions you can work with to those that you find tractable.”

    Are you kidding? Math may present barriers to those who aren’t used to math, which is unfortunate, but verbal arguments about economics tend to be far more convoluted and hard to follow than math. And don’t even start with assumptions: The key to narrative arguments is that you’re not forced to make your assumptions explicit and transparent. In math, the assumptions jump off the page. It’s a far more intellectually honest way to proceed, though I admit you are right that assumptions tend to be made for analytical tractability.

    Encouraging everyone to learn enough math to follow a mathematical model would be far more productive than putting the stigma on math and leaving us all to fight out economic problems with paragraph after paragraph of internally inconsistent verbosity without clear discussion of assumptions. A few equations can replace multiple pages of words.

  7. The problem with a mathematical model is that it is going to ignore any variable that can’t be adequately quantified. Things like political uncertainty, economic uncertainty and good old “animal spirits.”

  8. There are a lot of times when I’m reading a long piece of prose in some journal somewhere, and thinking: man, I wish he’d put some of this into an equation or model somewhere.

    I rarely read a bunch of models with descriptions of the major variables, and think: man, I wish he’d written this in prose form instead.

    Now there are times when I think the equation is too complicated for its own good, but it’s easy for me to come up with a rough approximation that’s much simpler. There are also times when I don’t understand part of the math, but as long as I know what the variables refer to, I can approximate an idea of what they’re saying.

    I’m okay with some variables or sections of the problem being incomplete or ill specified because the author feels that part of the problem can’t be adequately addressed mathematically just yet, and would prefer to write about it instead. But if they can’t organize the problem well enough for it to be understood, at least in part, mathematically, then that strongly suggests the problem can’t really be scientifically tested.

    I realize this favors the mathematically inclined over the verbally inclined. And I think that a lot of this math stuff is just signaling, but without math signaling we’d have prose signaling which is just a headache to read. I do agree with what you’re saying that having a precise mathematical equation to specify everything is not as important as having the main idea, and knowing a good way to test it and then apply it. But I think at least an imprecise math equation is often necessary.

  9. One important implication is that unemployed workers will not be hired back into the same jobs they had before. I believe that in the 1950s, there were recessions that were primarily inventory corrections, so that after you went through a couple of quarters with automobile manufacturers and their suppliers laying off workers, those workers got recalled. Those examples run counter to PSST.

    But maybe the post-war experiences don’t really run counter to PSST after all.

    Actually, let me step back a little first and propose a hunch of mine: There are perhaps both ‘Samuelson Eras’ and ‘Kling Eras’, in which disruptions (or the character of business cycles) are either mostly quantitative (in terms of inputs) or qualitative (in terms of technological and others big social changes), respectively. As such, different interpretive frameworks may be more apt at different times.

    If I had to put some rough estimates on on, I’d guess that pre-WWI was a ‘Samuelson Era’, 1920-WWII was a ‘Kling Era’, and immediately after the war reached a high water mark for quantitative, Samuelson-like character, which has been declining ever since. Sometime around the early 80’s the economy transitioned to having a predominantly Kling-like character, which has been increasing ever since.

    My guess is that the post-war era, especially in the US, was special and different, kind of like a short version of Hanson’s ‘dreamtime’. It took a long time to establish new patterns of specialization and trade per Alexander Field’s narrative of a very disruptive burst of innovation in the early 20th century, especially those relying on internal combustion engines and electric motors.

    After these new realities had been ‘socially absorbed’, and just after the war, technological development has progressed to the point where mass prosperity – not just in commodities but also new manufactured items – was now possible, but capital was still mostly complementary to low-skill labor, and so demand for people with average amounts of human capital rose fast. It was the famous ‘era of Everyman’.

    But as technology continued to improve and capital because more of a substitute than a complement for labor, more and more individuals shifted out of widget-creation and into building organizational capital or providing other kinds of specialized services in an increasingly complex economic ecosystem.

    Now that capital-based substitution seems to be accelerating in those ‘quantitative sectors’, every recession takes longer and longer to work out, because the qualitative changes to PSST must be discovered and implemented. And notice how recessions since 1981 keep taking longer and longer to recover, and how The Great Depression was perhaps the first real taste of a true qualitative-adjustment recession.

    This way of seeing things goes slightly easier on Samuelson than a story in which those sorts of ideas were completely misconceived and useless from the very beginning and even in their own time, when a PSST approach would be have better.

    Instead, that ‘interpretive framework’ was perhaps ‘good enough’ for the post-War era, precisely when PSST predictions seem to be weakest. It’s just that Samuelsonianism has outlived its usefulness. Times change, the nature of the economy changes, and so different interpretive frameworks are called for.

    But the trouble is that Samuelson’s way of looking at things was subject to social phenomena which entrenched that perspective in the most prestigious posts in the academy and in government, such that the prevailing viewpoint is much more inflexible that the character of the economy itself, which has moved on. So now it’s long past time for Economics to move on too.

  10. Economic variables of motives and human behavior are hard to quantify. A good example is the women are paid less than men argument. The answer is not math.

  11. @Handle has an excellent point about “Samuelson” or “Kling” eras.
    And there is even a hint of a metric:
    “One important implication is that unemployed workers will not be hired back into the same jobs they had before.”
    Well, where is the data? It should be known how many previously unemployed folk were hired back: into the same firm (laid off, rehired), into the same industry role (assembly line worker), into a totally different role.

    Part of good economics should be asking important questions and gathering data about the possible insight.

    Since 2006, I’d guess a large number of home builders who were laid off/ fired, have not returned to constructing houses.

Comments are closed.