European Banking System Worries - Why I Don't Do Standard Forecasting
There have been a number of articles discussing issues within the German banking system. This is a potentially big issue, given the size of the balance sheets (and derivatives exposures) involved.From a macro perspective, there are three main scenarios.
- The Lehman disruption is repeated; developed financial markets stop functioning until global authorities interventions clear the mess. In the worst case, this leads to a European bank holiday, after which the banks reopen with new national currencies.
- Equity holders as well as other stakeholders low in the capital structure take losses, but the system muddles through.
- Nothing really happens, possibly other than some dilutive financing (which only matters for existing equity holders).
I am not the person to ask about the probability distribution of theses scenarios. Needless to say, the global authorities want to avoid scenario #1, and so a meltdown call relies on major policy errors.
From the perspective of macro theory, this situation illustrates one of the fundamental problems with mathematical models in economics. Any economic forecast produced by almost any model is going to be conditional upon which scenario we are in, as I doubt that any tractable economic model is going to incorporate credit analysis of particular banks.
Since we do not have a systematic way of determining which scenario will play out, we have no way of testing the predictive abilities of a model (keeping in mind that we are continuously facing such uncertainty). I will discuss this further in an upcoming article - "Whither Mainstream Macro?"
(The United States election - which I am otherwise ignoring in my writing - is another good source of uncertainty.)
Cheshire Cats and Money
Professor Nick Rowe recently penned the article "Cheshire Cats and New Keynesian Central Banks." (I previously wrote that I was uncomfortable with what he wrote, but it turns out that the part I questioned the most only appeared in his comments about Real Business Cycle models on Twitter. I do not wish to try to reconstruct his Twitter comments here, so I will ignore that point of debate.)
His thesis statement is:
How can money disappear from a New Keynesian model, but the Central Bank still set a nominal rate of interest and create a recession by setting it too high?During the article, he discusses how we can interpret New Keynesian models (without explicit money demand functions) as having money, even though the aggregate money balance is always zero.
Ignore what New Keynesians say about their own New Keynesian models and listen to me instead. I will tell you how it is possible.
I have become increasingly unhappy with economists' interpretations of their models. A mathematical model is a mathematical model, and is just a toy system that reflects the assumptions that you put into it. Confusing mathematical model properties for properties of the real world is obviously dangerous.
I have no problem with describing a New Keynesian model as a model of a monetary economy (although I am not particularly happy with the quality of the model). However, that is not saying very much.
The figure above shows an extremely simple model for the U-3 Unemployment Rate in the United States. I am not going to hold it out as a work of quantitative genius, but at the same time, I have seen a lot worse out-of-sample performance elsewhere. (And the "out-of-sample" performance is truly out of sample; this was used in my 2013 "Battling Straight Line" article.)
- This is mathematical model of the unemployment rate in the United States.
- That is, one aspect of the United States economy.
- The United States is a monetary economy.
- Therefore, the model is in fact a model of one aspect of a monetary economy.
I do not wish to give the top secret proprietary details of this model, but I will state that there is absolutely no use of monetary aggregates within its calculation. In other words, we do not need any monetary aggregates in a mathematical model of a monetary economy. The fact that the United States is a monetary economy is a monetary economy, rather than a barter economy, is an issue that is completely irrelevant to the model specification. The only way of determining whether it matters is to find a barter economy, and attempt to model its unemployment rate.
(This observation is partly a response to a reader who could not grasp how it was possible to eliminate "money" from economic models of monetary economies. The belief that money must appear as a model variable is a variation of the "microfoundations" fallacy -- that an economic model must be developed from the point of view of an individual within the economy.)
Helicopter Money - Sigh
I am currently launching a rhetorical attack upon Positive Money (full reserve banking) for a chapter in my upcoming book, leaving the topic of Helicopter Money for last. The more I look at "Helicopter Money," the more frustrating it is. How it is supposed to be implemented changes, as does what it is supposed to accomplish, and how it is supposed to work.
A really striking aspect of Delong’s discussion of h money here is that there is no discussion of h money. It is all about choosing up sides. You are either this kind of person or that kind of person, with us or against us. He sounds here like W trying to round up the coalition of the willing.
This fits into a pattern I have noticed. With a couple admirable exceptions (e.g. Jordi Gali) most arguments from proponents of h money don’t really discuss the mechanics of how it would (not) work. They take as a given that it would work and then immediately turn to why people don’t accept the goals that h money is meant to serve:
"What is wrong with you people? You know, benighted hard-money market fundamentalists like Mike Woodford, Olivier Blanchard or (I surmise) Larry Summers."
I will have to read the Gali article that Gerard references (I am waiting until I get to that chapter), but beyond that, I am increasingly mystified what to say about the topic.
The only alleged advantage I see of helicopter money is a reliance upon financial repression -- forcing banks to hold reserves at the central bank at below-market rates. That said, financial repression can be implemented at any time; no helicopter money needed. However, I doubt that putting banks at a competitive disadvantage versus the shadow banking system is a particularly useful policy in our current institutional structure.
The only alleged advantage I see of helicopter money is a reliance upon financial repression -- forcing banks to hold reserves at the central bank at below-market rates. That said, financial repression can be implemented at any time; no helicopter money needed. However, I doubt that putting banks at a competitive disadvantage versus the shadow banking system is a particularly useful policy in our current institutional structure.
Not sure if I am the reader who could not grasp how it would be possible to eliminate money from economic discussions of monetary economies, but I might as well be because I don't understand what you mean there. In any event, if something is repeated slowly in English often enough in different ways then I have usually found that understanding of the point being made has been possible for me, not that agreement would be reached though. But it is true that some things are beyond me. Is there an older article to read that would explain what you mean by eliminating money from economics? Because my first thought when you say that is that you are trying to discuss a barter type economy and that you want to say money is just a neutral veil over the economy's production and demand for such production. But I am hoping that is not what you mean, although if it is, would be happy to read your argument.
ReplyDeleteIt's been awhile, but I am thinking of some comments from one of the anonymous commenters.
DeleteI do not have other articles (yet), since the idea only gelled within the book. (Most of the book are articles are my exasperation with how money is used in economics.)
But my example here is as good as any - no money at all within the model, yet it fits economic data. Even of we are going to make the model more complex, there is no a priori reason to believe that the addition of monetary aggregates will help it fit data.
Why do we need "money" anyway? What matters is household or business financial assets, which includes government and private debt, and so forth. Pretending that a model needs money because it is used in non-barter economies is strange; there are a lot of other things that developed economies rely upon - energy, running water, legal systems, flush toilets, none of which are taken into account. By privileging "money," economists have to then run around and try to find a real world data series which corresponds to it. We then discover that those real world data series do not act like their imaginary concept, and the monetary aggregates have little predictive power.
I will have to read your book then. But until then it seems to me that most people engage in work or produce things or sell things to get money, at least for the most part. And unlike the other things you mention like toilets and legal systems and useful energy, money is created at basically no cost by governments, and at no cost by banks (except for risk). So it is different I think than other goods or services that we end up paying the money for.
DeleteLike I said earlier, I really don't understand what you mean or what you are arguing so I guess I really need to read more before commenting. Please let me know when the book comes out.
And I don't want to be too snarky but let me know if anyone trades a toilet or a lawsuit for the book :)
Sure, people work to get "money," or probably more accurately, an income. However, they also work to achieve social status. Why does social status not appear as a variable in economic models?
DeleteWe need to draw a firm line between the real world and mathematical models, and accept that mathematical models cannot cover everything.
As another analogy, control systems engineers have models of human pilots that are sixth-order differential equations. Nobody thinks that a sixth-order differential equation can describe the human condition as well as the works of Shakespeare, but at the same time, those models are good for identifying the risk of pilot-induced oscillation - a task that no amount of reading Shakespeare will help.
Well, I have reached the level of my understanding more quickly than I was hoping to. In all honesty, I can't even pretend to know what a sixth-order differential equation is or why it might be germane to money. But as long as they help keep those planes from landing on my house I am all for them! Are there many of them (the differential equations, not the planes) in your book?
ReplyDeleteYes I agree with you about mathematical models not covering everything. And economics, at least in my understanding, does a really poor job of accounting for things like social status. And for things like altruism, and power, and family, and even ethics. Most of that got lumped in with some kind of utility theory I think. At least back when I was in college a long time ago. But then I was a pretty poor student back then and maybe mistaken, or maybe things have changed since.
I will be quiet now and wait to read the book. Hopefully for me there are not too many differential equations in it. But stories about planes are always interesting, just in case you consider adding any.
If you fly regularly, you probably do not want to hear my stories about aeronautical engineering. (The pilot model is not germane to money, but rather shows that we can model some aspects of human behaviour as simple models. Although it might sound scary, a sixth-order linear differential equation is pretty simple.)
DeleteMy next book is almost entirely equation free, like my blog articles. (It's essentially a collection of blog articles, some new, some edited reprints.) There are "word equations" like:
Assets = liabilities + owner's equity.
(The following book is going to be a primer for understanding academic research, and it will feature equations.)
Brian, you might find something of interest in my latest comment to Nick Rowe:
ReplyDeletehttp://worthwhile.typepad.com/worthwhile_canadian_initi/2016/09/cheshire-cats-and-new-keynesian-central-banks.html?cid=6a00d83451688169e201b7c89c1b52970b#comment-6a00d83451688169e201b7c89c1b52970b
Like you, I'm trying to eliminate "money" from (macro)economics. But I don't agree that using overdrafts will lead to an aggregate money balance of zero, at least no more than "traditional bank loans" do. Do you?
If someone chooses an internally consistent accounting system for a model economy, it is difficult to argue against it. So I cannot say that the situation in Rowe's model is wrong, but one could argue that is unrealistic.
DeleteFor an overdraft to be "negative money," it has to be payable on demand (in order to be symmetric with "positive money." For example, settlement balances at the Bank of Canada can be slightly positive or negative, and banks are expected to clear them up quickly. The convention is that the total balances are close to zero. (This system broke down in the financial crisis; banks left large balances at the BoC.) However, liquidity management for banks at the central bank is quite removed from the experience of nonfinancial firms.
In North America, overdrafts might be payable on demand, but it is clear that no one would finance anything using such liabilities -- they are only there for very short-term liquidity management, and when you forget who wrote what cheque in your joint account. (Which is why I have overdraft protection...)
In the UK, overdrafts were (are?) a common form of financing for businesses. I do not think they were practically payable on demand, and so I see an interpretation of them as negative money is strained. (Nick Rowe discussed that in comments in an earlier article; the borrower is expected to periodically pay off the balance, but this was managed by the overdraft borrower, and not at the whim of the bank).
Very relevant points, Brian. Some of those I haven't thought about, so thanks!
DeleteI don't think the actual differences between countries are relevant here. Even if we assumed the overdraft is payable on demand, there is an (inherent) asymmetry here. Let me explain.
(I don't like to use Nick's language, but now that you started...) I think we need to adopt the holder's perspective here. The holder of positive money can always find goods (incl. services, incl. labor) to buy, and thus redeem all his "credits" at will. But the holder of negative money is not always able to sell goods, and thus he cannot repay his debt at will (as your BoC crisis experience testifies).
There is no symmetry, right? There must always be some notice period, and even if there isn't, nothing guarantees that the repayment/closing of the overdraft doesn't take months, even years after the bank has demanded it (and possibly the bank loses nothing as a consequence -- only gains in interest and fees). We need to look at the actual outcome. Demand is one thing; whether I can comply with the demand or not is wholly another.
You said: "the borrower is expected to periodically pay off the balance, but this was managed by the overdraft borrower, and not at the whim of the bank"
Sounds a lot like a traditional bank loan, doesn't it?
Brian: The discussion continues in Nick's blog ("Cheshire Cats.."). I re-posted my latest answer (above) to you there, so that it comes to Nick's attention.
Delete