Paul Krugman has an insightful review of Timothy Geithner’s new book, Stress Test. In it, Paul walks us through the similarities between the 2008 financial crisis and a classic bank run.
The model of a classic bank run is as follows. At any given moment, banks hold only a fraction of the deposits they receive in the form of cash; relatively illiquid assets are purchased with the rest of the deposits. The spread between the interest rate a bank pays to its depositors and the yield on the bank’s illiquid assets is what determines the bank’s profits. Banks are able to keep only a small amount of cash on hand because only a small fraction of a bank’s depositors will try to pull their money out on any given day.
But the risk of a run is always there. As Paul writes:
Suppose that for some reason many depositors do decide to demand cash at the same time. The bank won’t have that much cash on hand, and if it tries to raise more cash by selling assets, it will have to sell those assets at fire-sale prices. The result is that mass withdrawals can break a bank, even if it’s fundamentally solvent. And this in turn means that when investors fear that a bank may fail, their actions can produce the very failure they fear: depositors will rush to pull their money out if they believe that other depositors will do the same, and the bank collapses.
Paul then goes on to talk about how federal deposit insurance has been the solution to the problem of banking panics. When deposits are insured, depositors won’t all decide to demand cash at the same time, because even if the bank at which their money is stored has problems, the insurance will assure the depositors that they will still get their money back if the bank fails.
Then Paul draws an important analogy to the 2008 financial crisis, which, he notes, was essentially a classic bank run: the so-called “shadow banks” were raising money through various, and often obscure, forms of short-term borrowing, such as repo, which weren’t federally insured. As soon as a crisis of confidence emerged, the short-term lenders all simultaneously demanded their money back, forcing the shadow banks to sell their assets at fire-sale prices. The forced selling had ripple effects across the broader financial system, pushing many otherwise solvent institutions into insolvency. In turn, lending collapsed, and the economy contracted sharply.
Spelling out the analogy between a classic bank run and the 2008 financial crisis presents a teachable moment for economists. On the one hand, we have what seems like a timeless model of how bank runs emerge. (Well, timeless at least since the advent of fractional reserve banking.) This model offers a clear normative policy prescription: institute federal deposit insurance. Yet the model did not help us avoid the 2008 crisis; nobody really understood the shadow-banking sector well enough to see that a classic bank run could emerge within it. As such, we did not have an insurance mechanism in place to prevent the crisis from spreading. So what went wrong? Why was the model not helpful in real time?
To answer these questions, we need to talk about the ontology of the social world. Philosophers like John Searle and Tony Lawson have noted a peculiar thing about the social world: that it’s dynamic. And by dynamic, I don’t mean in the way most economists understand the word – as, for example, having to do with intertemporal consumption smoothing or rational expectations. No, by dynamic, I basically mean evolutionary.
In the context of the banking sector, it has evolved from consisting of traditional banks that finance themselves by taking in cash deposits to including shadow banks that finance themselves by borrowing money in the form of obscure short-term lending agreements. The key point is that our perception of how to identify a bank has changed. We used to think of banks as those places with high ceilings and polished marble floors where people go to deposit their savings. But over the years, other institutions have been formed and have evolved to do what traditional banks do, except outside the regulatory radar, as it were. Are these new institutions banks? What does the word “bank” even refer to in the model of a classic bank run?
Drawing a contrast between the biological world and the social world might be helpful. We know the biological world evolves, but only very slowly – provided that we’re talking only about large and complex organisms rather than unicellular ones. The species we see in existence today won’t evolve into fundamentally new species tomorrow, next year or even 100 years from now. Conceptually, we can therefore think of the biological world as static for modeling purposes. If climate change continues to warm the planet over the next 50 years, we can hypothesize about how that might affect the various forms of species that exist on the earth today.
The same is not true in the social world, where things evolve rapidly. Federal deposit insurance was instituted in the US in 1933. In less than a century, the whole landscape of the banking sector changed in a way that left many short-term lending arrangements uninsured, the very ones for which insurance would have prevented a collapse in confidence.
One of the biggest problems in economics is that we have a great deal static models that were created at various points in time, when the economy may have been very different than it is today. For example, and to pick on Paul Krugman, we have something called the IS-LM model, which Paul frequently references to argue for using fiscal policy to boost demand when short-term interest rates are at their zero lower bound. As the story goes, when there is a persistent excess of desired private saving over desired investment, as is ostensibly the case at the zero lower bound, stimulus puts the excess saving to work in the form of public investment, helping to equilibrate desired saving and investment at a higher level of output, preferably one that is consistent with full employment.
The IS-LM model was formulated in the 1930s, when the US was very much a closed economy. Does the model still apply today even though the US is much more open? Given the US’s role as a huge net importer, especially for infrastructure materials such as steel and concrete (see chart below), it seems plausible that a portion of any stimulus money spent on domestic infrastructure will generate jobs overseas rather than at home. If the government builds a road, the project will create many more local jobs if the concrete for the road is produced domestically rather than in China or Mexico.
That’s not to say that spending more on fiscal stimulus in 2009-10 wouldn’t have been a good idea; it’s just saying that the landscape to which the IS-LM model was originally aimed may have evolved, thereby somewhat changing the conclusions of the model. After all, we have some evidence to suggest that fiscal multipliers are much lower for open economies relative to closed ones.
To be sure, Paul Krugman has mentioned this caveat, arguing that stimulus today would probably yield more bang for the buck if it were glazed with a flavor of protectionism. But as far as I can tell, he has downplayed the caveat in other writings, even going as far as to simply assert that the US should be modeled as a closed economy.
My critique is more methodological. When we present an economic model, we should always also present a historical narrative explaining when the model was originally formulated, what world it attempted to describe and why the world today hasn’t evolved to undermine the conclusions of the model. In the context of IS-LM, globalization, NAFTA, the rise of China and the shift of the US from running trade surpluses to running large trade deficits were each evolved processes. The global supply chain for infrastructure materials is no longer located in the United States. This will not change overnight just because the US government decides to spend more money. IS-LM says nothing about these evolved processes, and if you leave them out of your analysis, you’re leaving out many key ingredients.
In the end, I probably agree with Paul on his policy prescriptions. Notwithstanding the lower fiscal multiplier due to the trade deficit, I think more stimulus following the financial crisis would have been preferable. But it’s his lack of interest in ontological investigation that annoys me the most.
Once you think deeply about the ontology of the social world, you realize that to model phenomena in it, your tools and how you apply them need to be constantly changing – nothing is timeless. As such, if the goal is to improve people’s welfare through policy, then the policies we recommend need to be constantly updated.
And, on a deeper level, even the ideals that we as a society have established and have aspired to live up to are constantly changing. This is why it’s a bit ludicrous of some to assert that if we were to just abide by the ideals outlined in the Constitution, then everything would be all well and good. Our perception of what constitutes justice has changed enormously since the 18th Century. Sure, there are ideals like freedom which seem to be universal and timeless. But the terrain to which we apply such ideals is evolving in ways that constantly complicate the application process.
For example, we in America place a high value on freedom of speech. This includes the right the support any political party of our liking. In the olden days, this meant that people were allowed to stand on street corners and hand out pamphlets praising their preferred political parties. But the economy has evolved in ways that make mass marketing more intruding, allowing those with money to express their political views more widely – in the form of TV ads, billboards and mass emails from super PACS. Does rising inequality pose a challenge to the way we apply the ideal of freedom of speech? When those with money express their freedom of speech today, does it drown out the freedom of speech of those without money?
I’m sorry, I wish the social sciences were easier. I wish our conceptualized models and ideals were timeless in application. It is truly a beautiful thing when a radically simplified static model can yield tremendous insights about the natural world, as models in the physical sciences routinely do. But the social sciences don’t work like that, because they can’t work like that.
It’s high time we get over our collective envy for physics-type modeling in economics. We should be having more discussions about ontology. The implication is that we can no longer afford to talk about models without a reference to their historical foundations and to the evolution of the social world. If we can convince Nobel laureates like Paul Krugman to employ a more reflective methodology, then we just might be able to make economics a more progressive science. If people like Paul don’t listen, well then we may need to wait until the older generation of economists dies off before real change can occur.
 It’s of course easy to use the model to explain the 2008 financial crisis after the fact. But we perhaps set the bar too low if we aspire to only explain developments after the fact in the social sciences.