No, please don't show me your model

Unsurprisingly, on my post "The Art of Economics", which attempted to put the mathematical models beloved of mainstream economics firmly in their place, is a comment defending mainstream mathematical models. Here it is, in part:
Secondly, you definitely don't need obscure heterodox models to predict a financial crisis. I've cited it before, but for instance Kiyotaki-Moore basically sketches out how a crisis like this can occur. There are actually plenty of examples of perfectly fine mainstream papers on this topic. And it wasn't just heterodox economists that predicted it. People like Dean Baker, Roubini or even Krugman didn't exactly rely on post-Keynesian or Minskyian economics, their logic was fairly straight forward. Stiglitz has some great models on bank failure, which are essentially mainstream info-asymmetry economics. I also think Minsky is useful but not that useful, and it's not especially scientific. He doesn't really have any kind of model, he just essentially asserts that banks will turn to speculators (and also made a lot of mistakes with regards to importance of credit cards, diminishing importance of large infrastructure loans etc..) The mechanisms aren't adequately explained. At least the Austrians, who I definitely oppose, have a mechanism for how banks turn to unstable speculators - aggressive monetary policy. 
This is bad science of the "show me your model" variety. A mathematical model may give apparently accurate results, but that does not mean it has the right theoretical foundations. Valuing the model over the theory was lampooned by the great physicist Richard Feynman, in this lovely metaphor:
[Feynman] imagines a Mayan astronomer who had a mathematical model that perfectly predicted full moons and eclipses, but with no concept of space, spheres or orbits. Feynman then supposes that a young man says to the astronomer, “I have an idea – maybe those things are going around and they’re balls of rock out there, and we can calculate how they move.” The astronomer asks the young man how accurately can his theory predict eclipses. The young man said his theory wasn’t developed sufficiently to predict that yet. The astronomer boasts, “we can calculate eclipses more accurately than you can with your model, so you must not pay any attention to your idea because obviously the mathematical scheme is better.”
In dismissing Hyman Minsky's hypothesis because the model was incomplete, my commenter has behaved like Feynman's Mayan astronomer. Never mind the theory, show me your model....

So, let's look at the mainstream model recommended by my commenter. Like all (yes, I mean all) pre-crisis economic models, Kiyotaki-Moore does not model the financial sector accurately - in fact it does not model it at all. And because of this, it models a financial crisis as starting with some kind of exogenous shock coming out of the blue, in this case a temporary shock to productivity. The model is a farming model, so a productivity shock of this kind might be an adverse weather event, perhaps.

Now, there may indeed be a shock that triggers a financial collapse, but it is not necessarily exogenous. In 2008, it was the fall of Lehman Brothers, which was by any reasonable standards an endogenous shock: similarly in 2007, BNP Paribas's announcement that it could not value subprime MBS, was an endogenous shock. Do endogenous shocks have different effects from exogenous ones? We do not know, and the model does not tell us.

But in the absence of any model explaining how debtors become fragile, we can have no reason to assume that ANY shock, exogenous or endogenous, would have destructive effects. Indeed Kiyotaki-Moore themselves say this is a weakness in their model:
A weakness of our model is that it provides no analysis of who becomes credit constrained, and when. We merely rely on the assumption that different agents have different technologies. 
"Different technologies". What a get-out line. But if you exclude the financial sector from your model, that's the kind of blanket excuse you end up with.

If your model cannot explain how people and corporations become over-leveraged and therefore fragile, it can have no predictive power whatsoever. All it can do is say "IF people/businesses are over-leveraged when a shock hits, THEN this is likely to be the effect". So Kiyotaki-Moore can neither explain nor predict a financial crisis. It merely describes how an (unexplained) shock propagates itself through an (unexplained) over-leveraged population, with long-lasting negative effects. That is useful, of course - in fact I think this model does a pretty good job of explaining the amplifying effect of collateral price falls in debt deflationary collapses. But that isn't what my commenter claimed it did.

Let me be clear. I don't have a problem with mathematical models, as long as they use appropriate mathematics and have a sound theoretical basis. But we have to respect their limitations. They don't necessarily adequately explain economic events, let alone reliably predict them. And they are never a substitute for logical thought. .

So I don't want you to show me your model. I want you to explain your thinking. What is your theory, and how have you defined it? What thought processes brought you to this point? What are your assumptions, and how have you justified them? If you cannot explain these in words, then however clever your mathematics, your model is devoid of substance. Throw away your Greek dictionary, and write me an essay in plain English.

Related reading

Photo of Richard Feynman from


  1. This comment has been removed by the author.

  2. Financial crashed and disruptions are one my interests in history. I can see and understand that there may be some common factors or key features that enable a judgement to be matter later on, but when I look at the likes of John Law, Arnold Nesbitt and Montague Norman I do not really see how modelling can explain or could predict what happened. Especially Arnold Nesbitt.

  3. "A mathematical model may give apparently accurate results, but that does not mean it has the right theoretical foundations."

    You're strawmanning me. In that paragraph at least, I did not mention mathematics. A 'model' does not *need* to be mathematical - this is clearly the case because I contrasted Minsky with the austrians, who are *clearly* not mathematical. My point is there needs to be a thorough break down, or as some people like to call 'concrete steps', of how A gets to B. My argument is that (and I admit I haven't ready *that* much Minsky, so I may have missed it somewhere) he never provided such a good mechanism as to how banks become the speculator type, it was all a bit of a black box - in contrast to Austrians, who have a mechanism that starts with monetary policy. I also didn't dismiss it, I said it was useful but not *that* useful - that's hardly a dismissal.

    "(yes, I mean all) "

    Have you actually looked at 'all' models? There are thousands and thousands of models in all kinds of journals over a century, why are you so confident that 'all' do this? You could not have possibly read all published papers.

    "In 2008, it was the fall of Lehman Brothers, which was by any reasonable standards an endogenous shock"

    I don't understand, why was it an endogenous shock? Macroeconomic models aren't bank sector models, they are aggregated models of the entire economy - they might include things like aggregate consumption, investment, government spending. Or even simpler, just interest rates, output and inflation. It's not clear why Lehman Brothers balance sheet is endogenous to those variables. Which of those would cause Lehman to collapse? Why is it any more endogenous than a technology shock?

    "All it can do is say "IF people/businesses are over-leveraged when a shock hits, THEN this is likely to be the effect".

    But that clearly gives it predictive power. For instance, using that model in the mid 2000s, given that we would have known that many borrowers were hugely leveraged, and that we would have known that much of this is secured against a specific collateral as in the model (housing), that at least allows the model to give us a very good conditional prediction: if house prices fall for any reason *even by a small amount*, that could lead to a severe financial crisis. That is predictive power. It might not predict the timing of a drop in house prices, but you're giving it a highly unreasonable standard that practically no model can do, given how much these prices are guided by animal spirits. What it does do is provide two useful things: a conditional prediction of a crisis, should a house price decline occur, and a prediction of the persistence and severity of this effect (also clearly very useful!).

    I think the problem seems to be you're asking mainstream economics to be a theory of everything, able to predict evolution in all possible economic variables in the economy. That's an unreasonable standard, economics is better served as conditionally predictive models coupled with statistical data.

    "What is your theory, and how have you defined it? What thought processes brought you to this point? What are your assumptions, and how have you justified them? "

    This is exactly what people mean when they say "show me your model". It doesn't have to be mathematical, but the problem with not using maths is that it makes it much easier to use hidden or implicit assumptions - whereas using a mathematical model makes these explicit. Also, this of course should be accompanied by a verbal description - if you cannot come up with a plausible story to explain your maths, it's useless, I agree.

    1. Britonomist,

      Since my earlier post was specifically about mathematical models, and I am quoting your comment on that post, it is disingenuous now to claim that you were not talking about mathematical models. Clearly, I don't have any problem at all with a theoretical model logically defined, whether or not that model uses mathematics. I have a problem with the separation of model from theory, or the confusion of model with theory. Hence my Feynman quote.

      No, clearly I have not looked at "all" models. I am, however, talking about mainstream economic models, not "all" models. And of course, models build on models. How many mainstream models are built on the incorrect "money multiplier" model of credit intermediation? I do not need to look at all of them to know that they are inadequate. That doesn't mean they have nothing to say - I quite like Diamond & Dybvig on bank runs, for example - but they don't adequately predict or explain the tendency of the financial system to generate bubbles and crashes. For example, because D&D ignores the money destruction aspect of bank runs (in their model money is not destroyed, it simply moves, which in a leveraged system is incorrect), it understates the negative effect of bank runs and fails to explain why economic production is adversely affected.

      Disturbingly, many post-crisis mainstream models STILL do not properly model the financial sector. They make money look as if it comes from Mars. Until we model this accurately we will not understand how financial crises form and why their effects are so devastating.

      On exogenous shocks. I think it reasonable to regard the collapse of a financial firm under enormous deleveraging pressure due to ongoing stresses in the financial system as an endogenous shock to the financial system. If Lehman had been blown up by terrorists or demolished in a plane crash, I would regard that as exogenous.

      On the predictive power of THIS model. Yes, if prior to the crisis we had had other models that showed that households were becoming dangerously leveraged because of expectation of ever-rising house prices, and that the house price trend would reach a turning point, this model would have had predictive power regarding the effects of falling house prices. But we didn't, so it didn't.

      I did say that the model was useful in forecasting how collateral price falls would amplify a debt deflationary spiral, with long-lasting negative effects. But by itself, it could not predict or explain the whole crisis - which is what you said it did. Personally I would not expect it to.

      I don't think it is unreasonable to expect the profession of economics to develop theories to explain the main economic variables in the economy. The omission of an accurate model of finance is to my mind a serious defect. So too is the assumption of linearity. I only had to translate one model into endogenous money terms to show how flawed the mainstream (linear) view of the financial economy prior to the financial crisis is.

      In my experience, when people say "show me your model", they don't want logic, they want maths. The math-worship going on is way over the top. I've a damn good mind to translate my "financial storm" model into musical notation, just to annoy the math-worshippers. It could make a good symphony.

    2. "How many mainstream models are built on the incorrect "money multiplier""

      Honestly? Not that many, that's a textbook thing. But most models simply don't care - they focus on interest rates generally, many (most?) modern macro models don't consider base money at all.

      "I think it reasonable to regard the collapse of a financial firm under enormous deleveraging pressure due to ongoing stresses in the financial system as an endogenous shock to the financial system."

      I think we might be using endogenous differently. Macroeconomic models generally aren't models of the financial system, so Lehman would generally not reasonably be endogenous to a typical macro model with the variables it looks at (unless you're an Austrian and think low interest rates = apocalypse). Now, in other aspects, Lehman is endogenous to the economy in general - but basically everything is, including technology and productivity. I think that's a bit of a weak criticism. I think it's fair to criticize macro models for not having a financial sector, but criticizing a model that *doesn't have a financial sector* for modelling a financial sector shock as exogenous is unfair - unless you can clearly show me that the specifically modeled variables *cause* that financial shock in the first place?

      "If Lehman had been blown up by terrorists or demolished in a plane crash, I would regard that as exogenous. "

      Why isn't a binge of irrational exuberance and possible fraud not exogenous?

      "Yes, if prior to the crisis we had had other models that showed that households were becoming dangerously leveraged because of expectation of ever-rising house prices"

      I'm not sure you need a model for that, you just need to look at the data - although the 'dangerous' part is where you can use Kiyotaki Moore.

      "we didn't"

      We had data.

      " The omission of an accurate model of finance is to my mind a serious defect."

      This was something I conceded is a problem in the majority of macro. There have been attempts though.

      "So too is the assumption of linearity. "

      What do you mean here?

    3. Let's get this straight. Mainstream models that purport to model financial crises without modelling, or building on some coherent model of, the financial system are by definition flawed. The model you cite falls into that category.

      So, the model you cite excludes the financial system. In such a model, severe disturbances in the financial system are indeed exogenous. But it is a definitional absurdity to claim that a financial shock is exogenous simply because this particular model excludes the financial system. The shock is endogenous to the system, not to the model.

      A binge on irrational exuberance and possible fraud is endogenous to the financial system. The financial system itself facilitates this.

      On dangerous household leverage - no, you need some benchmark model of fragility. Remember that this is a feedback loop (see Mian & Sufi): households' leverage increases as house prices fall, because their net worth is reduced. You can't just look at the data and arbitrarily decide that households are borrowing too much. You need to know the expected trajectory of house prices, incomes and inflation, as well as household debt. Data were there, but there were no models to interpret it.

      On the assumption of linearity, I refer you to Olivier Blanchard's post "Where Dangers Lurk". Link is in the "Related Reading" section of my previous post.

    4. "Let's get this straight. Mainstream models that purport to model financial crises without modelling, or building on some coherent model of, the financial system are by definition flawed. The model you cite falls into that category. "

      Non comprehensive is not the same as flawed. It's a useful model, it's extremely hard to model everything. I also reject that it has no coherent model - it has a perfectly coherent system, it has debtors, it has credit constraints, it has collateral - it may not be a model of *the* modern financial system, but it certainly contains a model of *a* coherent financial system. I know you understand this, but inexperienced onlookers might read a paragraph like that and use their confirmation bias to conclude adheres to the stereotype of economists ignoring debt, borrowing, money, assets etc.. I just think it helps to be more clear about what you're referring to when you say financial system.

      "In such a model, severe disturbances in the financial system are indeed exogenous. "

      It's not a severe disturbance to the financial system, what is modeled *is* the severe disturbance, what's exogenous is just a drop in the value of collateral.

      " But it is a definitional absurdity to claim that a financial shock is exogenous simply because this particular model excludes the financial system."

      It is a definitional absurdity to claim that a terrorist shock is exogenous simply because this particular model excludes international relations. It is a defitional absurdity to claim that an energy crisis is exogenous simply because this particular model excludes the energy sector. See where I'm going with this? There's no such thing as categorical exogeneity, it depends on the situation and what is being modeled.

      "A binge on irrational exuberance and possible fraud is endogenous to the financial system. The financial system itself facilitates this. "

      And capitalism might facilitate radical terrorist attacks. Therefore, it's also endogenous.

      "You need to know the expected trajectory of house prices, incomes and inflation, as well as household debt."

      I agree with this, in fact I basically said the same myself. But again consider 2005. With the data on house prices & leverage, a good model like Kiyotaki-Moore, the only other thing you need to know is that there is at least a small possibility house prices might decline in the future - with that any regulator or policy maker with that knowledge would have good reason to be concerned about a future crisis. Are you saying nobody in government/central banking back then had any idea house prices could decline? A few had their head in the clouds maybe, but I know some were definitely nervous. It's not a neutral assumption to assume house prices won't decline. At the very least, you would know that the economy is very exposed to house prices - that's definitely useful knowledge.

    5. Actually, in Kiyotaki-Moore, the decline in land value is caused endogenously (by too much debt + borrowing constraints), thus Kiyotaki-Moore is an endogenous model of credit cycles!

    6. Haha, that's true!

      With regard to the house price trend, it is abundantly clear that US policymakers did NOT expect house prices to fall, and did not even recognise the possibility of adverse effects if they did. I've read through the minutes of every FOMC meeting from 2004-2008. The principal concerns, all the time, are inflation and the "normalisation" of interest rates. Rising house prices are regarded with approval. The tailing-off of house price rises from 2006 onwards was regarded with some concern, but not because of the possibility of a sharp fall in domestic demand and a debt deflationary spiral, which is what actually happened. No, the concern was simply that normalisation of interest rates might have to slow down. Seriously, they were asleep at the wheel.

      I can sort of forgive US policymakers for this, given that the US had not experienced a serious fall in house prices since the 1930s. But UK policymakers were just as bad, with much less excuse. The UK suffered the worst housing crisis in living memory in 1990 - yet only 25 years later, the Bank of England was ignoring everything except inflation, the FSA was letting lenders do whatever they liked and Gordon Brown was celebrating "the end of Tory boom & bust".

      To be fair, there were some warnings. The BIS persistently warned about the derivatives bubble from 2005 onwards, as did Raghuram Rajan at the IMF. Why they were ignored I don't know. Behavioural economics might have some insights on herding effects and the madness of crowds as applied to central bankers and government policymakers, I suppose. Lemmings, anyway.

    7. "and did not even recognise the possibility of adverse effects if they did."

      I'm not sure this is so true, I know they were at least aware of many other people warning them, because there were repeated attempts from Greenspan & later Bernanke to reassure people about housing ("prices are regional" etc...). In fact, look at this article:

      "[Greenspan] said any fall in home prices was unlikely to have a broad impact on the economy."

      Hah! If only Greenspan was aware of and thought properly about the implications of models like Kiyotaki Moore, I certainly don't think he would be so confident about it not having broad impacts then (of course, he may have been fully aware and just didn't want to talk down the market).

      Anyway, I think we need to set the record straight on this model, because until now it has been mischaracterized. It's not a simple Bernanke style model of a financial shock being amplified. Section 3 is a full model of oscillating credit cycles, making the economy go from boom to bust persistently. The bust is not exogenous, but it's caused by the boom + leverage! Sure, you might need a small positive shock to get it going, but once you do you get a persistent credit cycle oscillation that looks awfully familiar. In this model, it generates a boom in land values coupled with rising leverage, eventually the leverage gets too high and you get a collapse (this part is all endogenous). It looks a lot like a housing boom & bust to me.

      So really, the model is a lot more powerful than you give it credit for in the OP, I think it should at least be updated a little as I think you rather rudely dismissed it there - it's not the collapse is not a simple shock that comes out of the blue.

    8. I still don't think you are quite getting my "endogenous shock" point.

      I dismiss completely the idea that an essentially endogenous shock can be treated as "exogenous" for the purposes of a model, for reasons I shall explain shortly. So that leaves two possible ways of looking at this. Lehman is an endogenous shock under either approach.

      Firstly, the financial crisis was - initially - a crisis of the financial system, not of the whole economy. There were severe spillover effects to the rest of the economy, of course. But the trigger for the financial collapse was a shock to the financial system which came from WITHIN the financial system. Terrorism is endogenous to the political system, yes, but it wasn't the political system that collapsed. To the financial system, terrorism is exogenous.

      That is something of a micro definition, I agree. So now let's take the economy as a whole. What would be a genuinely exogenous shock? I suppose if the earth got hit by an asteroid you might regard that as exogenous. Maybe the Icelandic volcanic eruption that caused air traffic to be grounded for three weeks in 2010 was an exogenous shock. And perhaps the Japanese tsunami, and Hurricane Katrina. Acts of God are exogenous - would you agree?

      And that brings me to the essential difference that I pointed out in the post between Kiyotaki-Moore's shock to productivity and the financial crisis. A productivity shock in farming can be caused by adverse weather. If it's weather, then it's exogenous to the economy, since humans don't cause or control weather. But the fall of Lehman was ENTIRELY caused by the activities of humans. Defined in this way, therefore, the fall of Lehman can only be regarded as endogenous, whereas Kiyotaki-Moore's shock might not be.

      I think the majority of economic shocks are endogenous. And I think economists define endogenous shocks as exogenous in order to avoid having to include in their models the human systems that give rise to them. It's understandable, in order to keep the models tractable, but we should really distinguish between shocks that are endogenous to some aspect of the economy that is not being directly modelled, and shocks that are genuinely exogenous.

      Of course, you might decide that as humans are part of the natural order, so is anything we create, including our economic system. In this case, then, the only genuinely exogenous shock would be an alien invasion, since even an asteroid hit is a natural hazard. Evolutionary economics are great fun, but seriously - how far do you want to take this?

    9. "Defined in this way, therefore, the fall of Lehman can only be regarded as endogenous, whereas Kiyotaki-Moore's shock might not be. "

      Again, I think you're simply mischaracterizing Kiyotaki-Moore at this point. The initial 'shock' is fairly irrelevant to the story (and remember, a shock in economics can be both positive and negative, a shock is just an unanticipated change in a variable). The 'shock' in this paper in section 3 is simply a small positive shock which starts the cycle off. It is not the crisis. The crisis in this model is fully endogenous, it is caused by high leverage, booming land values and credit constraints. The cycle does not go from boom to bust due to any exogenous shock.

    10. I'm not mischaracterising anything. All I did was point out that Kiyotaki-Moore does not define whether the trigger shock is exogenous or endogenous. And I asked whether endogenous and exogenous shocks had different effects. That's a reasonable question, surely?

      Kiyotaki-Moore does not model the whole cycle. It does not attempt to explain the buildup of credit, and it relies on an undefined "shock" as the trigger for the crisis. Both of these omissions stem from the same cause, namely the complete exclusion of the financial sector from the model. In my view any model which purports to model a financial crisis but does not include, or build upon, a coherent and accurate model of the financial system is fundamentally flawed. Kiyotaki-Moore offers some useful insights regarding the way a crisis propagates itself. But simply is not a predictive model of a financial crisis. Not by a long way.

    11. Let me give you an example of why an endogenous shock might have a completely different effect from an exogenous one.

      A few days ago, my cat broke her leg. We don't know how she did it, but the vet thinks she got her paw trapped in something and broke her leg freeing it. My cat is young, fit and healthy. The broken leg is an exogenous shock - it does not stem from her state of health. The animal hospital pinned her leg back together and she is already back on her feet and (hopefully) on the road to recovery.

      But suppose that she broke her leg not because of an extreme incident, but because she had some illness that made her bones fragile and liable to break under ordinary stress. In this case, it would be quite wrong to model the ordinary stress that broke her leg as an "exogenous shock" of a similar order to the one that I described above. The outcome is the same, but there isn't really a "shock" at all. The cause of the break is the illness. And because the problem is the illness, the treatment needs to be different and the recovery is likely to be much slower.

      Kiyotaki-Moore clearly intends to model the second of these. But actually all it does is state that the cat is ill, then apply a shock of some kind to break the cat's leg. This is why the type of shock matters. If it applied the first kind of shock, it would break the leg even if there were nothing wrong with the cat. But the break could then be taken as proof of illness. The result would be misdiagnosis and wrong treatment.

      An incomplete model like Kiyotaki-Moore can have useful insights, but it is not by itself an adequate diagnostic tool.

    12. Francis said,

      "You can't just look at the data and arbitrarily decide that households are borrowing too much. You need to know the expected trajectory of house prices, incomes and inflation, as well as household debt. Data were there, but there were no models to interpret it. "

      I think accountants using accounting ratios picked up on it early. It was in the accounting data.

      Just off hand ratios that can be looked up in accounting books:

      Interest coverage ratio
      Time to cover
      Leverage ratios
      Margin of Safety - Benjamin Grahm

      Why is it that accountants are not interviewed on TV about the economy?

      Point 2: If you offshore 3% per year of jobs of 30 year mortgage debtors would that cause a subprime problem? If the house prices are run up with available credit does that encourage employers to look for employees with a lower cost basis abroad? If people borrow to escape loss of purchasing power of money did that run up demand in over priced houses?

    13. Liquidity ratios

    14. More on ratios:
      Part of a free book/course on accounting.

      Accounting is so much more valuable to learn than economics. If one can't bring them selves to learn the basics of accounting maybe the first part of this book might be encouragement.

      "Double Entry: How the merchants of Venice shaped the modern world - and how their invention could make or break the planet" by Jane Glesson-White

    15. Oh, and one further point. Section 3 effectively sets up a perpetual-motion model. But they don't exist in reality. Even in economics, we would expect frictions to slow and eventually stop the oscillation. For example, leverage buildup might diminish over subsequent cycles due to learning. So if the oscillation DOESN'T slow, something must be maintaining it. K-M don't include the financial sector, so in their model creditors lend directly to debtors. They assume that debtors (farmers) will always make themselves fragile by borrowing up to the maximum. And they assume that creditors will always lend to borrowers, even if doing so vastly increases the likelihood of loss. If the Austrians looked at this, they would be screaming "misallocation of capital due to central bank low interest rate policy". Whether or not you agree with this, at least they have an explanation. K-M does not. You criticised Minsky for exactly the same thing.

      I would also point out that because K-M is a farming model, output directly relates to landholding. Leveraging up in order to buy land therefore increases output (and, in Section III, investment). That is not true in a residential property market.

    16. "Even in economics, we would expect frictions to slow and eventually stop the oscillation."

      I need to work through the maths and it's really quite complex, but I think it does slow actually but I'd need to really go through the maths which I have not much desire to do. Anyway, obviously it's just a toy model - the point is it IS a model of credit *cycles* - you kept acting like its a financial accelerator model where the crash is exogenous, but in section three both the buildup in debt and asset price boom, plus the subsequent crash, are entirely endogenous. The shock this model uses in section 3 is just a positive productivity shock to get it going, like turning the ignition key on for the model. I just want to make this clear. And I do think the boom and bust dynamics are at least somewhat corollary to the housing crisis and various other crises - obviously it's nowhere near a 1 to 1 mapping, it's just a toy model (not a farming model either, that's just for illustrative purposes).

      And generalized toy models are useful. For instance, there are actually a huge amount of papers with detailed models of the financial sector, I could have just googled 'financial fragility' and got the first result that came up, e.g.

      In fact even my masters thesis was just a VAR model relating non performing loan ratios of all the top US banks on their balance sheets to various macroeconomic indicators. This model can actually predict a banking collapse in theory, if banks have a given level of capital and a recession hits (which I find it SO annoying when people keep insisting economists ignore banks, when I spent half of my 4th year doing nothing but examining bank balance sheets).

      But the thing is, with these highly detailed models - a regulator can just make a few tweaks and they instantly become invalid. But a highly generalized model - if even in a super simple economy with just farmers can have a tendency towards debt crises, that's fairly telling no? That points out that debt crises is a problem more fundamental than modern banking idiosyncrasies.

    17. Yes, clearly it's a toy model, and I don't have a problem with it from that perspective. I merely wished to point out its limitations as a predictive model in an advanced economy with powerful financial intermediaries and interventionist government/central bank. You have misinterpreted me, by the way: I did not assume that the crash was exogenous - I merely pointed out that the model did not specify whether the crash was exogenous or endogenous, and I said that this could make a considerable difference to the outcome.

      I do criticise some of the assumptions, though. For example, how can the agents have "perfect foresight" if they ignore the possibility of collateral value falls after the first oscillation (in which, of course, collateral values fall)? Surely we might expect that credit constraints would be tightened in each oscillation? I didn't notice any recognition of such learning effects in the maths. The idea that farmers will spend ALL of their output on servicing debt and investing in more trees is a bit far-fetched, too.

      But I agree, it does show that there can be debt crises in a peer-to-peer lending model - and that is an extremely important insight.

      I think we should distinguish between pre- and post-crisis models. Unsurprisingly, financial stability has been all the rage as a research topic since the crisis. Sadly it was not so fashionable before the crisis. Had it been, the path of history might have been very different. The Bundesbank paper you cite dates from 2013.

    18. I'm not sure what an endogenous shock is. Exogenous determined outside the model, endogenous determined inside. But how do you determine a shock? This may seem like semantic quibbling and perhaps it is, but if you write down some sort of model with deterministic cycles (such as these? ) is 'shock' really the right word? I suppose if you don't know exactly when it's going to happen because you don't know relevant parameters (or perhaps model is sufficiently complex). I wonder how many of the current crop of endogenous financial crises models have predictable (not a shock) crises, and how many have unpredictable.

      you might find this presentation about pre and post crises economics models of crises interesting, covers much of what you are talking about

  4. This comment has been removed by the author.

  5. LM can't be a complete model since its begin with rational and sure expectations. It doesn't take account of the main contribution of Keynes: the impossibility of probabilistic estimation of future, and the inevitable use of subjectivity. Keynes thought that all the relations between S and I are not functionally stable, because a change in other variables than interest rate are determinant. In IS-LM it is too much static, no possible accumulative reaction of all the economy to rise or fall.
    And financial problems as endogenous money and definite irrationality of financial markets are not contemplated.
    But perhaps is a pedagogic utility to learning some basic relation under the condition of a big "If".

  6. Frances:

    I am a mathematician who taught math and then later used math in my business of designing computer hardware and software for large enterprises. They bought my systems because they thought they could make more money. They thought my systems would help them make better decisions. So, as a result of all this exposure I became familiar with the business context in which business decisions were made. I saw a system that did not work for the common good. So, the debate between you and Noah is the wrong debate. Our current economic system is not worth modeling because it is designed to enslave ordinary people. I saw no executives make decisions that took into account the common good. They gave it lip service, but they did not mean it.

    I have tried to understand economics. I have taken a class, I have read a few books, including Paul Krugman’s 940-page textbook, and it just does not make sense to me. Max Tegmark is from Denmark, and like many of us he made a stab at picking a career when he graduated from high school. Here is what he wrote:

    "When the time came to apply for college, I decided against physics and other technical fields, and ended up at the Stockholm School of Economics, focusing on environmental issues. I wanted to do my small part to make our planet a better place, and felt that the main problem wasn’t that we lacked technical solutions, but that we didn’t properly use the technology we had. I figured that the best way to affect people’s behavior was through their wallets, and was intrigued by the idea of creating economic incentives that aligned individual egoism with the common good.

    "Alas, I soon grew disillusioned, concluding that economics was largely a form of intellectual prostitution where you got rewarded for saying what the powers that be wanted to hear. Whatever a politician wanted to do, he or she could find an economist as advisor who had argued for doing precisely that. Franklin D. Roosevelt wanted to increase government spending, so he listened to John Maynard Keynes, whereas Ronald Reagan wanted to decrease government spending, so he listened to Milton Friedman."

    Tegmark decided to become a physicist and is author or coauthor of more than two hundred technical papers, twelve of which have been cited more than five hundred times. He holds a Ph.D. from the University of California, Berkeley, and is a physics professor at MIT. I applaud Tegmark’s wish to make “our planet a better place,” and I agree with his description of economics as being “a form of intellectual prostitution.” I wish I had thought of it.

    Another scientist, more famous than Tegmark, was Albert Einstein. He shared Tegmark’s wish to make the world a better place, and he believed that the economic scourge of capitalism produced more evil than good. He was inclined toward socialism which is a dirty word in America today, but based on my reading of his views I think he was more inclined toward any system that worked for the common good. In any case, he wondered if the field of economics would be useful in designing a government of the future. In 1949 he wrote:

    "Let us first consider the question from the point of view of scientific knowledge. It might appear that there are no essential methodological differences between astronomy and economics: scientists in both fields attempt to discover laws of general acceptability for a circumscribed group of phenomena in order to make the interconnection of these phenomena as clearly understandable as possible. But in reality such methodological differences do exist. The discovery of general laws in the field of economics is made difficult by the circumstance that observed economic phenomena are often affected by many factors which are very hard to evaluate separately. Economic science in its present state can throw little light on the socialist society of the future.

    There is nothing more to say. Economics is a con, and like all cons it hurts the innocent and the gullible.

  7. Now, if the stick in the mud economists say we have theories that are not vague and the theories actually specifically predict things contrary to the real world data, then what?

    Specifically wrong is obviously wrong!

  8. I should have changed the word "predict" to "demonstrate".

    This would allow looking at things that have already occurred and not require the often impossible prediction of the future.

  9. Question:

    Are economics students usually required to learn double entry bookkeeping or accounting?

    If not why? If so, in what countries?

  10. Why do most economists seem to have a personal issue in including the financial system in their theories and models?

    1. Accountants include the financial system. And, the financial system includes accountants.

    2. Accountants are not economists.

      Accountants only count existing beans.

      Economists create inaccurate models that do not include the bean creation process.

    3. Back to economists. I can only guess not beeing an economist.

      If it is so then: Perhaps, for some economists it could be ignorance on how to include the financial system. Or an attempt to not enlighten. Yet really educated people might not take ignorace personally and might be curious to learn. Also, the financial system like the economy is dynamic rather than the beginning assumption of equilibium.

      My pet theory is that knowledge of accounting is so imoportant that economists with out it have less of an understanding, less ability to quantify in that way, and also model.

      You see accounting measures transactions or movement of ownership of all things. That is a powerful measuring tool and modeling frame work.

      What I don´t understand the premise that econmists are actually ignorant of these things because the existance of National "Accounts" with seprate sectors that include the financial sector. But, on the internet the premice aperars to be predominantly correct. I mean they collect the data, some of them MUST be using it? I never saw anything about the national accounts in my macro text book. Wynne Godley used it and modeled with it.

      "Sir John James Cowperthwaite, KBE, CMG was a British civil servant and the Financial Secretary of Hong Kong from 1961 to 1971. ... He refused to collect economic statistics to avoid officials meddling in the economy. According to Catherine R. Schenk, Cowperthwaite's policies helped it to develop from one of the poorest places on earth to one of the wealthiest and most prosperous: "Low taxes, lax employment laws, absence of government debt, and free trade are all pillars of the Hong Kong experience of economic development."[2]

      1. first 2 pages are a summery exapmple, the financial sector is there.

  11. Your point is well-made that economists (or arm chair commentators) should be able to explain the logic of their models. One thing you mention briefly that deserves more attention is the difference between prescriptive and descriptive models. I think that it's impossible to have a good descriptive model that isn't based in some type of theory which can be explained by the modeler. Descriptive models, by definition should help you understand what is going on in the economy. I do, however, think that you can have a high-performing predictive model absent theory. This isn't the kind of model I would necessarily want, but I have seen many cases in industry of "black box models" that consistently predict well, but no one (sometimes not even the modeler in the case of neural nets and such) can explain the theory behind why it's doing what it's doing. It depends on the use case for the model. To be fair, these types of models are also what can lead to edge cases where an algorithm starts a downward spiral of trades. My point is that we should consider what the model is meant to accomplish before saying that there can only be theory-based models.

    1. Hi Drew,

      I instinctively feel uncomfortable with models that have no theoretical grounding. "It works, but we don't know why it works" is far too close to cargo cult economics for me.

      Of course, the users of a model don't need to know how it works, any more than I need to have more than a passing acquaintance with how my car's engine works. If it goes wrong, I employ an engineer to sort it out. But if the creators of the model don't know why it predicts so accurately, we no longer have control of technology.

  12. Tweets with good point on data:

    Danny Blanchflower ‏@D_Blanchflower Aug 20 Wyoming, USA

    @Noahpinion @Frances_Coppola @ProfSteveKeen @JoMicheII @ari1601 heterodox or orthodox one of these days you will both have to look at data

    Danny Blanchflower ‏@D_Blanchflower Aug 20 Wyoming, USA

    @ProfSteveKeen @politybooks @Noahpinion @Frances_Coppola @JoMicheII @ari1601 debate missed need both to have harsh confrontation with data.

    (((FrancesCoppola))) ‏@Frances_Coppola Aug 20

    @D_Blanchflower @ProfSteveKeen @politybooks @Noahpinion @JoMicheII @ari1601 good point

  13. "What is your theory, and how have you defined it? What thought processes brought you to this point? What are your assumptions, and how have you justified them?"

    Glad you asked:-), giving me a chance to expound on my theory.

    My theory is scarcity of anything ensures proper usage. Let us take a few examples:

    1. In a place where water is scarce people are more likely to be careful in its usage than in a place where water is plentiful.
    2. In a place where electricity is in shortage, electricity cost has to be higher in order to ensure it is not wasted or used injudiciously. You are likely to switch off the lights and fans when not using it as it pinches your pocket.
    3. Assume you run a bar, if you want only sophisticated people coming to it, you cannot offer it free because beggars might flock to it.

    There can be many take on this but the main point is mispricing a product leads to misuse (also leads to higher demand, a topic for another day about Unicorns).

    While everything is based on money (the price you pay for it), the price of money itself is the interest you pay for it.

    Thus when you misprice money it leads to misuse, which happens to be speculation or malinvestment because speculation or malinvestment will yield better results in the short term (as long as the speculative instruments keep rising -- it can be stocks or tulips or real estate - it does not matter) but at some point it tends to break because nothing can rise for ever and at some point it will be out of reach of most people and this lack of demand will start bringing price down and a downward spiral will start in right earnest). Additionally given the nature of people when you misprice they tend to want to buy it today rather than save and buy it tomorrow (again leading to malinvestment).

    Thus my theory is simple--mispricing money (low interest rates) is the problem. Now you may ask what is the right interest rate?

    I feel the right interest rate is the rate below which savers start saving more (the interest rate at which they start showing discomfort). Incidentally I belong to the camp (probably a solitary camp) that feels savers as a group happen to be the Hercules holding up capitalism. Imagine destroying the very people who are holding up capitalism. This is what central bankers are doing and hence I hold them singularly responsible for the destruction they are causing across the world. As an aside are savers introverts? (no bearing on the topic but a thought that struck me)

    Thus, IMO, if you want to save capitalism you have to either close the central bank or ensure they cannot manipulate interest rates.


Post a Comment

Popular posts from this blog

WASPI Campaign's legal action is morally wrong

The foolish Samaritan