Mis(under?)diagnosing Financial Crises
on missing the forest for the trees and taking the long, long view
At first, there had been a frenzy for real estate. Eventually, credit conditions tightened, and liquidity in the market began to dry up. Creditors began to pull back, shortening the duration of their lending. In an attempt to meet redemptions, debtors began dumping assets in fire sales, resulting in systemically depressed asset prices. In the context of a systemic contraction, credit could only be obtained at punitively high rates as creditors pulled back from as much counterparty risk as they could. The crisis hit the real economy, real estate in particular; evictions rose and real estate values plunged. Eventually, the government stepped in and, in classic lender-of-last-resort fashion, provided huge amounts of liquidity at low interest rates through secured (collateralized) loans to chartered banks in order to quell the panic and protect balance sheets. This is not a description of the US in 2007–2009, but of Rome in 33 AD.
Is this time really different?
Before I took my job at the Yale Program on Financial Stability (YPFS), I was invited to a dinner with a retired family friend, during which this particular host was kind enough to ask about my new job. I explained that we’d be looking at case studies of financial crises reaching back to the fourteenth century to hopefully draw conclusions about the best ways to fight them, and to identify commonalities between crises across time and space. He more or less scoffed at this idea, and said something to the effect of, “you’re not going to learn anything from the Dutch Tulip Mania that’s applicable to today’s world.” He was mostly wrong (but correct that the Dutch Tulip Mania wasn’t really a financial crisis, though I’m not sure he realized that). In separate interactions over the years, I have heard various forms of, “you can’t understand 2008 by reading a book or studying old crises; you have to understand how unique it was,”’ or “real estate never lost money,” or “it was a one-of-a-kind event.”
Related to this view is the view that crises can be explained by micro factors (e.g., bad interest rate management, lazy credit rating agencies, real estate prices, etc.). This is somewhat akin to saying that the assassination of Archduke Ferdinand caused World War I: not really wrong, but also not really right, and clearly missing the causal dynamics at play.
This note is an attempt to refute these notions of financial crises as being rare, unique, unprecedented, or the result of random micro events. It is also an attempt to argue for a view—that financial crises are fundamentally macroeconomic events, and that it is neither possible nor desirable to reduce their probability of occurrence to zero.
Entire books—very famous ones—have been written on this subject. I won’t attempt to outdo them, but I hope to bring the message to you, my readers, in a distilled form.
First, two recent-ish case studies to illustrate the point, and then a broader body of work.
The Global Financial Crisis and the Global Savings Glut
So, there are lots of narratives about the Global Financial Crisis (GFC) of 2007–2009 floating around. The entire story is not going to fit in a blog post. But I’ll try to talk here to explain the macro story that is now widely accepted as the likely fundamental cause of the dislocations that resulted ultimately in the GFC.
Dr. Andrew Metrick (disclosure: founder and head of YPFS, my employer) does this best, and I urge you to watch his Coursera class (with Tim Geithner) on this topic, so I won’t try to summarize it here. Needless to say, there is a lot more to the story than “but no one thought real estate would crash,” or “those corrupt bankers tried to bankrupt vulnerable people.”
But weren’t real estate prices crazy high? Didn’t the ratings houses fail at their jobs? Weren’t banks taking irresponsible risks? Yes, yes, yes, of course. But why? You don’t just wake up one day, have a kick-in-the-gut cup of coffee, and decide to become irrationally exuberant. Yes, financial firms were taking excessive risk, and yes, they wrongly thought they had it managed. But why? Because they were idiots? The financial engineers with so many PhDs that they could wallpaper a room in Ivy league diplomas—those people were idiots? Like all financial crises, there is an underlying macro story.
The abridged version of the global financial crisis origin story
Once upon a time in the US of A, a man presided over a fiscal surplus. No really. That man was Bill Clinton. This was fine and dandy, and although he never got much credit for it (somehow people think the surplus was Reagan?), this wasn’t ideal for the Chinese (or Saudis for that matter).
Let’s switch gears for a minute and take a trip to China. When you get paid, you can do one of two things: spend your money or save (invest) your money.1 In China, there were lots of savings since Deng Xiaoping’s industrial reforms and the birth of the manufacturing boom. There are lots of complicated reasons for this, but the main one is that China (1) did lots of exporting; and (2) managed its exchange rate.2 Basically, when you export a lot and import a little, normally your exchange rate would appreciate, and your workers’ wages would rise, which would balance the economy more to consumption. But that’s not what happened in China. Exporters brought dollars back, but instead of passing those dollars on to shareholders or employees, they swapped them for renminbi at the People’s Bank of China (PBOC).
The PBOC in essence absorbed huge dollar inflows and provided Chinese companies and households short-term renminbi deposits in exchange, while managing their exchange rate. The PBOC, being a central bank, needed to then do something with those dollars. If it did nothing, it would lose money to inflation—like any asset manager. But, naturally, it needed risk-free investments (because they’re careful public stewards, not gamblers). Risk-free dollar investments, you say? Obvious answer: Treasury securities. Same thing’s going on in other massive surplus countries, like Saudi Arabia: they exported a bunch of stuff (oil); got dollars back; the central bank mopped up those dollars; they needed somewhere to put those dollars.
Back to Clinton. Yeah, so, if you’re the Chinese or the Saudis, you really like American debt (America’s fiscal deficit = the investment opportunity for the surplus nations of the world). Problem is, Bill’s fiscal prudence is kind of raining on your parade: remember, you need risk-free dollar assets, and the Americans aren’t creating as much now. Not to worry, another solution: buy the debt of the Agencies—Fannie Mae and Freddie Mac. They’re basically government-backed anyways, right? They trade hardly any pricier than Treasuries. Problem solved. But what happens when the surplus nations of the world have insatiable hunger for safe “risk-free” American assets? The demand is through the roof: yields on Treasuries and Agencies are at all-time lows.3 This means that Fannie and Freddie can pass those savings off to aspiring American homeowners: mortgages get cheap. As the cost of borrowing money (mortgages) get cheap, asset valuations (home prices) rise. Cheap financing for ever-appreciating assets? Sounds like the deal of the century! You can see the cycle here.
Stateside, though, pension managers in the US need to invest in something and get some yield. Naturally, they reach for yield,4 and start trying to find riskier assets that will yield more so they can at least get 4% or something respectable. The financial system did what it will always do: it met the demand for “safe” assets by manufacturing them, in this case via securitization. Like all asset bubbles, though, the music had to stop, and we’re now in the familiar territory of the housing crash. Brad Setser explains this succinctly:
Think of it this way: when China bought a Treasury bond from an American insurance company or bank, if provided the pension fund or bank with funds to invest in riskier assets that offered a higher yield than Treasury bonds. Wall Street proved more than capable of churning out ever more complex kinds of mortgage backed securities – and securities composed of parts of other mortgage backed securities – to meet this demand.
To recap: Chinese, Saudi, and other surplus nations’ households were saving too much and consuming too little (savings glut), while American households were savings too little and consuming too much (consumption glut)—economists like to call this situation a global imbalance. China exported consumer goods to the US, got paid in dollars for them, and then turned around and used those dollars to fund the US deficit. As this happened, yields on US assets plunged, and the world’s investors started seeking riskier and riskier assets to invest in, as all the usual safe ones had been mopped up (sitting in central bank accounts in China, Saudi Arabia, Germany, and some other places). The world needed safe investible dollar assets and surplus nations’ hunger for them coincided with a shortage (when the US was in surplus). As a result, the financial system had to try to “safe-ify” riskier and riskier things to meet that demand (via securitization). The cycle fed on itself, with cheap financing from the global savings glut (Ben Bernanke’s term) driving up home prices, which fed the frenzy, until the music stopped.
This isn’t that new a story; this is capital flows, the bugbear of Southeast Asia since 1997, and Latin America during the Latin American Debt Crisis of the 1980s. Capital flowing across borders is only helpful insofar as it can be productively absorbed. This point is well understood by emerging market central bankers, who will tell you exactly this point (in fact, a former governor of the Reserve Bank of India put it to me almost verbatim this way): money flowing from a foreign country into your country for investment is only productive to the extent that your domestic markets can absorb and channel that investment productively. Once you’ve reached the point of saturation where your economy is operating at its optimal investment needs, then each marginal dollar of investment risks driving up asset prices speculatively. This is almost exactly what happened in Thailand in 1997 (although their exchange rate policy had more to do with attracting the investment in the first place): more capital flowed into the country than could be productively put to use, and eventually the investors became speculators, bidding up prices of real estate. And then, of course, it crashed. This type of thing has been happening for a very, very long time.5 Probably the first recorded case of a financial crisis resulting from capital flows building up unsustainable asset prices as investors reached for yield was the Panic of 1825, in which UK investors poured money into Latin American mining and infrastructure projects, which, as the name suggests, ended in tears.6
For more on this (the Global Savings Glut story), see e.g., Bernanke 2005; Setser 2009.
The Panic of 2023 and why it’s safer to buy a house in Paris, Texas than Paris, France
As many readers will recall, in the spring of 2023, the US had some banking drama. Silicon Valley Bank (SVB) collapsed after failing to obtain Fed emergency funding. Silvergate and Signature also failed and First Republic was scooped up by JPMorgan (who else). This all prompted a pretty extraordinary Fed facility (the Bank Term Funding Program, or BTFP), and the Fed’s discount window got some heavy use. (There was also a seemingly related but not all that related failure of Credit Suisse in following weeks, which we won’t discuss here.)
One answer to “why did this happen” is that certain banks (you know who you are) failed to appropriately hedge their interest rate risk. This is the received wisdom and, on the facts, looks pretty compelling. The interest rate risk story goes something like this: you buy some assets when rates are low, and you put them on your books as worth something ($100, say). Then, for whatever reason (like inflation or something), interest rates rise. As a result, those assets you have are worth less7 ($90, say). Now, you have lost money ($10, say)—sad! So, this is a thing, and obviously can take down a bank. For an intuitive explanation (with some figures I may or may not have had something to do with) of how this played out at SVB, see this paper.
If you like this explanation, though (and we’ll see there are some other convincing narratives), then you have to answer the question of why US banks do such a bad job hedging their interest rate risk—rates went up all around the world in 2023, but with the exception of Credit Suisse (whose failure was not attributable to interest rate risk), the only notable bank failures were in the US. This was an America-specific interest rate crisis in a literal world of high and rising interest rates, which doesn’t immediately make sense. It turns out that the proximate answer to this apparent conundrum is pretty simple: because in the US we don’t regulate interest rate risk in the banking book (IRRBB for those who like the lingo) the way the rest of the world does. Interest rate risk hurts both the asset and the liability8 sides of banks’ balance sheets and other jurisdictions (like Canada and the Eurozone), in accordance with the Basel standard, regulate that risk to banks. In, for example, the Eurozone, if a bank has seriously high interest rate risk, its supervisors would make it go get more capital to survive potential losses.
So, Europe’s looking a lot safer, right? Well, before we get there, we have to remember that risk doesn’t disappear, it just gets moved around. European banks, on net, carry less interest rate risk than American ones typically do, as a result of this regulation. So, who bears the risk? European homeowners! They mostly have floating-rate mortgages, so when interest rates rise, their payments rise too. Balance sheets get impaired in Europe, too. It’s just personal/family balance sheets instead of bank balance sheets. In the end, once you peel back the layers of the onion, it’s a macro/policy design question: interest rate changes entail risk to the aggregate financial system; America prefers that risk to live on bank balance sheets; banks exist on the standard normal distribution just like most things do; so, when net risk increases and that risk lives in banks, some banks in the left tail of that distribution will fare poorly. This is a pretty standard story of macro events hurting weaker parts of an industry and—systemic impacts of the financial industry in particular set aside—isn’t all that novel or unique.
But what if we don’t buy this whole interest rate risk narrative? As Steven Kelly has pointed out, plenty of other banks, like mainstay Bank of America, had a lot of interest rate risk and didn’t fail. Now, to be fair, this is somewhat counterfactual given that after SVB the system had the BTFP facility, which took away a lot of that risk. But still, other weak banks failed (First Republic, I’m looking at you). As Mr. Kelly puts it, the run on SVB was in effect a run on its business model and the issues weren’t all on the asset side (i.e., losses on long-dated Treasuries), but also included a geographically and sectorally concentrated liability side (i.e., tech firms and VCs in Silicon Valley running down deposit balances during COVID).
Either way you look at it, though, it’s still a macro event: inflation → rate rises → losses on bond securities and/or losses to the business model of Silicon Valley. This particular story isn’t even new: inflation and rate hikes causing bank failures by causing asset-side losses and undermining business models was very much the story of the 1980s Savings and Loans Crisis.
The upshot is: yes, some people in risk management at some banks sucked, but some people in risk management at some banks have always sucked and will continue to suck, and yet there aren’t systemic financial crises every day. Financial crises are, ultimately, macro events. So . . . people should probably study this?
The literature: a sweepingly reductive overview
They have. There is a whole field of financial history and economics about exactly this phenomenon. The canonical books are probably Lombard Street by Walter Bagehot (1873), Panics, Manias, and Crashes by Charles Kindleberger (1978), and This Time Is Different by Reinhart and Rogoff (2009).
Look, financial crises have been happening for a very, very long time. Of course, taxonomically, there are debates about what counts as a financial crisis and what doesn’t, and those are debates worth having. But for the time being, let’s go with the most expansive overview, which draws on all the major existing databases and therefore is likely the most exhaustive: the Metrick-Schmelzing database.
I’m not going to attempt to cover and categorize the financial crises of world history because (1) you don’t want me to; (2) I don’t have the word count for that (some self-discipline remains on this site); and (3) other people have done that. But I will attempt to impart here a mere taste of the richness and diversity that is the study of financial crises, and to highlight that none of this is remotely new.
First, just for fun, here are a few examples of financial crises from the database, for your viewing pleasure:
In June 1360 in Barcelona, major municipal banker Jaume dez Vilar defaults. Fearing contagion, the city government offers capital to his competitors to take over Vilar’s accounts.
In April 1465 in Florence, Salviati Bank is on the brink of bankruptcy. To prevent a disorderly collapse, the de facto central bank—Medici Bank (yeah, you’ve heard of them)—provides an ad hoc emergency loan to Salviati.
In September 1720, when the South Sea bubble bursts, the Bank of England lends freely and eventually assumes some of the South Sea Company’s assets.
In September 1805, depositors run on the Banque de France in response to rumors that Mr. Napoleon had exhausted the bank’s silver reserves in his war (joke’s on them, he hadn’t even invaded Russia yet). The Banque de France suspends withdrawals and eventually Napoleon nationalizes it (the Banque de France is currently France’s central bank).
In 1857, in response to the international crisis of 1857 (which originated at an Ohio-based insurance company), the Bank of Finland begins broad-based lending and approves some debt restructurings (mostly lengthening repayment durations).
In 1927–28, the Bank of Japan lends to the Bank of Taiwan (Taiwan was then a Japanese colonial possession). In addition, the Japanese government provides as JPY 200 million guarantee for the Bank of Taiwan and eases collateral requirements.
In June 1984, the central bank of the Philippines provides nearly a billion pesos of ad hoc lending after thirty savings banks fail and a run begins on the largest savings bank in the Philippines.
For the even longer view, here’s a nice visual from the Metrick-Schmelzing paper:
If you want to go way back, you can look at Rome in AD 33, as we’ve done. Beyond that, records get trickier, but certainly there was recorded financial volatility in response to macroeconomic/geopolitical shocks, such as commodity price jumps in Babylon in response to pending invasions (Alexander the Great, literally). Those could have been economic and not financial crises, but the wide-scale presence of debt contracts (and government mandated debt forgiveness in times of strife) indicates that a protean financial system was vulnerable to crises of confidence then as well.
No, it’s not just the mortgage-backed securities
An important note is that this isn’t an asset-specific issue. First of all, asset price crashes are not synonymous with financial crises (lots of stock market crashes end in nothing at all except for stock market crashes). That said, often asset bubbles cause financial crises, but there’s no one specific asset that’s the problem. Of course, a consolidated standardized database of every asset price in history doesn’t quite exist, but peak-to-trough drawdowns in the US since 1880 have surpassed –60% for commodities, gold, and the S&P (not so bad for Treasuries, but that certainly wouldn’t be true of sovereign debt writ large):
Courtesy of Jack French, here
You don’t have to look far beyond sovereign debt crises (i.e., sovereign defaults) and hyperinflation (e.g., Zimbabwe 2007–09, Germany in the 1920s) to see that supposedly safe government securities and even cash are not truly safe. They can (and often do) suffer crushing peak-to-trough drawdowns.
Suter 1992, 89
Finally, people (characteristically my American compatriots) tend to think that real estate only ever “goes up.” First of all, no. Second, definitely not true for the world writ large (Japan suffered a two decade-long real estate slump). Data is also hard to find here. I imagine if we had a reliable chart of real estate valuations in the US circa 1812 (War of 1812), 1861–1865 (Civil War), 1873 (Panic and Depression of 1873), or 1929–1939 (Great Depression), the US picture would be clearer, but for now, the Dallas Fed has put together a fascinating historical time series of real (i.e., inflation-adjusted) global house price data. Here’s a nice chart:
Federal Reserve Bank of Dallas.
From the Dallas Fed: “NOTE: Colored bars indicate contraction (peak to trough), determined using the Bry and Boschan (1971) classical cycle dating approach implemented in Harding and Pagan (2002). Black indicates contractions of the all-country aggregate; red, the U.S.; and blue, all other countries in the database. Restrictions are imposed to avoid spurious cycles, so each phase of expansion and contraction must last at least three quarters for real house prices and two quarters for real personal disposable income.”
Losses in real estate happen, sadly, all the time. (No, I don’t know what was going on in Germany—DE in the chart—and yes, it looks bad.) And this is only back to 1975!
If I had data on hand (or the time to do the digging), you’d see the same types of things for railway and canal bonds in the 19th century, etc. All asset classes can be the subject of a bubble.
Fire-surfing through history
Anyways, financial crises happen a lot.
A common analogy you’ll hear for fighting financial crises if you hang around the space long enough is that of firefighting. This is a more apt (if not worn) analogy than it might at first appear.
Fires happen all the time and for various different unique reasons (e.g., someone smoking in bed, a campfire gone wrong, etc.). Those reasons are often pretty unique (aunt Betsy forgot the towel was near the stove) and often change as technologies change (gasoline production really changed the game for fire risk). Fire risk is, therefore, different over time and place (Arizona is riskier than Finland; 1850 Chicago was riskier than modern Chicago). This is similar for financial crises: collateralized loan obligations are pretty unique formats of financial risk. And some places (that in the aggregate do more maturity transformation) are riskier than others: Switzerland does lots of maturity transformation relative to the size of its economy, and so is a high risk for financial crises; Mozambique does very little, and is a lower risk for financial crises.
Nonetheless, there are common risk factors and fundamental causes of fires that just don’t change all that much (flammable stuff, heat, dryness). Financial crises are similar. Obviously, the cause of the 2008 global financial crisis was very different from the AD 33 Roman credit crunch. But at the end of the day, they were both credit bubbles that burst in the context of heightened demand for safe assets that outstripped their supply. Things can only change so much. Basically, the ingredients of a financial crisis are runnable short-term debt (e.g., bank deposits, money market fund shares, repo), asset bubbles (e.g., real estate, government debt, railroad stock in the nineteenth century), and macroeconomic events/variables (e.g., inflationary shocks, capital flows, wars). Almost all financial crises are just some rejiggering of those things.
While at first the initial fire might be seen as punishment for the irresponsible (your house burns down when you smoke in bed—duh), once it becomes a forest fire, it destroys indiscriminately: it will equally engulf the arsonist and the innocent stander-by. There is always a moral hazard impulse: if the fire department comes to the rescue every time someone smokes in bed, what will prevent people from smoking in bed? On the other hand, when someone smokes in bed, we don’t shame them and let their house burn down; we put the fire out (in part to protect them, in part to protect their neighbors) and then fine them later. Taken too far, the moral hazard impulse for poetic justice tends to put the innocent adjacent homeowners at risk and allows the fire to spread. But, at the same time, if you never let the underbrush burn and always suppress all forest fires, you tend to create even more dangerous ones in the future.
Finally, as Dr. Metrick is wont to point out, while the various causes of fires are sundry, you put them out pretty much the same way: with water (and maybe chemicals for oil fires). Likewise, there aren’t that many different ways you fix a financial crisis, regardless of its cause. This is why plenty of extremely smart people, who know a lot about this stuff, think it’s a good investment of time and money to study the history of financial crises and responses to them. To quote Reinhart and Rogoff, this time isn’t different.
You can follow me on Twitter (X) @ArnoldVincient, or connect with me on LinkedIn.
This work is independent from and not endorsed by the Yale Program on Financial Stability or Yale University; all views are my own.
You might not think of saving as synonymous with investing, and that’s understandable, because popping your money in a 1% yield savings account doesn’t really feel like Wolf of Wall Street stuff, but remember that those savings are funding bank investment, so, in the aggregate, savings = investment.
In a great twist of historical irony, another driving factor for emerging market demand for Treasury securities was their own prior scarring experiences with financial crises in the 1990s, which led them to build up strategic stockpiles of dollar reserves. For more, see Bernanke 2005.
Readers will be familiar by now with the mechanics, but again: yields move opposite to prices; prices are low because demand is high (for a given amount of supply), just like any other thing.
The term isn’t new, or unique to the GFC. The idea is that some folks in an economy (most folks actually), have some future costs they need to fund (e.g., retirement). When current interest rates are low, say 1%, but they need 3% in order to meet future obligations, they’re forced to buy up riskier assets to get the same yield.
However, typically capital also abruptly turns and gets pulled from the recipient country, directly causing a liquidity crisis; notably this didn’t happen in the US. In fact, if anything, foreign countries turned more to dollar assets (driving the dollar exchange rate up) during the GFC. Again, this is an artifact of the US being the issuer of the world’s reserve currency at the time, so is somewhat idiosyncratic. But the fundamental dynamic of foreign capital driving domestic asset prices up in unsustainable ways still holds.
This crisis is a feast of intrigue. Perhaps one of the greatest financial swindles ever pulled off happened as the crisis was building, when a swashbuckling Scotsman entirely fabricated a make-believe country called Poyais, sold its sovereign debt in the UK (at hardly a spread over real countries), and then absconded with the proceeds. People later died trying to settle in Poyais, which was in reality a “parcel of malaria-infested jungle,” the ‘Mosquito Coast’ of modern-day Nicaragua and Honduras.
For those unfamiliar with this concept, the underlying simple logic is the following: if I buy a 2% bond when rates are 2%, that’s fine; but when rates are 5%, who wants a 2% bond? In a world of 5% interest rates, how much would you pay for a 2% bond that otherwise has the same characteristics? Yeah, not nearly as much as you did when rates were 2%.
We earlier talked about why higher rates might hurt the asset side of the balance sheet. The intuitive explanation for higher rates hurting the liability side (read: deposits) is that if you pay 1% on your checking accounts, that’s fine; but in a world where rates are 5%, you probably have to pay more than 1% on your checking accounts. (Or you can just ignore that and do whatever you want, but you’ll probably lose depositors, who will likely go search for higher interest rates at other banks.)