A few years ago, a bail bondsman in Hawaii named Nick Lindblad e-mailed me. He had just read my book on financial economics and noted that he used some of the same data-driven strategies when deciding who should, or shouldn’t, get a bail bond. “Deciding who gets a bond is just like picking a stock,” he later said over the phone. The bondsman faces a tough risk decision, he explained: he must put up thousands of dollars to post bail for the accused. If the defendant doesn’t show up for the court date, Lindblad loses that money.
First, Lindblad does a risk assessment, based on available data. Those data may include the defendant’s past criminal record, or who his family is and how involved they are with his case (a mother showing up is a good sign), or other aspects of his behavior—like whether he has a phone and bothered to set up a voicemail. Lindblad weighs this information with his years of experience before deciding who gets a bond.
Then Lindblad tries to manage his risk. He might ask for a cosigner, usually a relative—again, mothers are a good option. He stays on top of the defendants, reminding them of their court date. And he updates his strategy as new information emerges. If a relative tells him that the defendant is back in trouble and may miss a court date, Lindblad or a bounty hunter—“some muscle,” as he calls it—will grab the defendant and bring him in. As technology progresses, moreover, so do the tools at Lindblad’s disposal. He can use social media to track his customers, for example—even if they skip their court date. This rarely happens, Lindblad says, largely because he usually makes the right risk decision to start with, and then does all he can to manage his risk.
This is just one example of the power of new tools and technologies that make it easier to analyze data for the purpose of assessing risk. One of the most important of these is financial economics. It didn’t surprise me that Lindblad is a big fan of financial news and podcasts. Many developments in risk measurement and risk science happen first in financial markets, simply because lots of money is at stake. But Lindblad’s story shows that we can rely on the same tools to manage risk in other domains of life, too—though we too often do not.
Humans have always struggled with risk management because life has always been uncertain. Medieval and Renaissance Europeans followed many superstitions and rituals, for example, that promised to keep them safe during plague times. They avoided meat, sex, and bathing. Renaissance Venetians drank their own urine, as both a cure and preventive measure. Officials would ask astrologers to study the stars to assess how long a plague would last. Ethnic groups, often Jews, found themselves scapegoated and persecuted as plague-bearers.
Some ideas worked better than others. Separating and quarantining the infected had some success. But mostly, plagues ripped through communities and killed lots of people. Urine-drinking provided no help. Other proposed remedies, like tying the rump of a plucked chicken to the patient, may actually have exposed the victim to more bacteria. But anything that gave people a sense of control was attractive, even if it sounds crazy today.
As an economist who studies risk in markets, I’ve often read about such behavior and marveled at how far we’ve come. We live in an age of reason—not only for hard science but also for risk management. At the start of the Covid-19 pandemic, I felt that pandemics might still pose a threat to our health and safety but that we had ways to make them less perilous. We had more knowledge of infectious disease and of how to develop effective treatments and vaccines. We knew the importance of hygiene and clean drinking water. We knew much more about risk management, overall. We had the tools to make sense of the data as they came in, so we could weigh costs and benefits and choose the best strategies. It all was an enormous improvement over asking the stars.
I felt confident that public-health authorities would use the available tools and come up with reasonable policies that adhered to the basic risk-management principles. Then, a few months later, the government mandated that I wear a facial covering every time I entered a restaurant; I could take it off once I sat down, but then had to put it on again if I got up. For months, I was told to wear a mask outside if I couldn’t “socially distance,” and was kept from buying a drink in a bar if I didn’t order “substantial” food to go with it. Not only did these rules defy what we know about preventing airborne viruses; they also violated everything that finance teaches us about risk management (to say nothing of common sense). Some days, it seems that we know no better than our superstitious ancestors.
Risk management accepts that life is full of trade-offs, since reducing risk poses costs: you give up upside to make the world less dangerous. This is true with regard not only to our portfolios but also to bail bonds, public health, and much else. Financial economics provides a powerful framework to collect and use data to determine what costs are worth paying, and how. It’s our best defense from panic, superstition, and the politicization of public policy. Provided we understand how to use these tools and their limitations, we should apply them more broadly.
Risk management does not presume that risk can be eliminated; the world is too unpredictable, and the benefits of some risk-taking are too valuable. The purpose of finance is to move money into the future and make it grow. Money tends to grow faster if you invest in riskier assets. But the potential for greater riches always comes with the possibility of loss. Generally, the more you chance loss, the more potential you have for gains. Risk management aims to strike the right balance: allow investors to see their money grow, while reducing the odds of a lousy outcome, or if a bad outcome does occur, make it less bad. Risk management in finance goes poorly when investors (or their advisors) forget these principles and think that they can get all upside and no downside.
In financial markets, two ways exist to manage risk. You can hedge it, which means taking a little less of it and giving up some potential upside—so you balance your investment portfolio of stocks, which tend to be volatile, with safer assets, like bonds. Lindblad hedges by being picky about who gets a bond. If he issued bonds for everyone needing one, and they all showed up for court, he’d make more money. But since some defendants typically won’t show, Lindblad hedges by giving bonds only to about half the people coming to him.
Insurance is the second strategy. You pay a premium and get in exchange a promise of payment if something bad happens—your house burns down, say. Or you can pay for the option to buy or sell a stock at a set price. Insurance leaves all the upside risk (minus what you pay for the insurance) but limits potential losses. Lindblad’s network of bounty hunters are his insurance. If people skip bail, Lindblad can usually haul them in.
The trick is knowing what risk-management technique makes sense for you, and how much to pay for it. In finance, risk gets priced, bought, and sold in markets; this provides data to measure risk. It has gone on for thousands of years. The Code of Hammurabi in ancient Mesopotamia established rules for contracts on goods and services that set prices before the transaction took place. This meant that the ancients could create futures contracts on olive and grain harvests. Buyers and sellers could lock in a price before the harvest took place and a market price emerged. The Mesopotamians could smooth out the business cycle and ensure stable payments, which, back then, were largely determined by the whims of nature.
Risk management as a science in finance took off in the 1970s. Technology—advances in computing power—helped make it possible. And following the end of the Bretton Woods international monetary system and the rise of globalization, demand for new financial solutions to risk exploded. Like many jumps in innovation, risk management also developed because smart people came together and benefited from synergy. At MIT and the University of Chicago in the 1960s and 1970s, scholars applied cutting-edge tools in statistics and mathematics to measuring and pricing risk.
A famous innovation was the Black-Scholes formula, an options-pricing model that enables people in finance quickly and consistently to put a price on risk for stocks, based on observable data. Black-Scholes was derived using stochastic calculus. But you don’t need to know the exact techniques to deploy it. In fact, one can derive an options price with just a calculator. The ability to put a price on risk, which then can easily be observed and replicated, transformed financial markets. Before Black-Scholes, options were traded, but pricing was unreliable and inconsistent, so the market for most such securities remained small, if it existed at all. But once derivatives contracts could be priced and sold on exchanges, the market expanded rapidly; by the end of last year, outstanding derivatives nationally totaled more than $661 trillion.
“Risk management as a science in finance took off in the 1970s. Advances in computing power helped make it possible.”
What happened in the 1970s enabled academics and investors to take a more data-driven approach to risk. Investors could survey the latest data, put a price on risk and risk reduction, and use this information to make decisions. Investors could better understand the trade-offs—what risks needed to be avoided, which should be hedged (and by how much), and which should be insured. In some ways, the innovations made financial markets less risky. But they also allowed investors to take on more risk. Analogously, consider Lindblad: if he knew that he could get a good bounty hunter for cheap, and that the bounty hunter would always be available, he’d probably issue more bonds.
Taking on more risk, even if insured, can be dangerous because things happen that we don’t expect. Risk management tends to work well when it comes to risks that occur 68 percent of the time. But maybe we’re unlucky, or maybe our data were misleading or misjudged. Maybe what we thought unlikely was more probable than we anticipated. During the financial crisis, many discovered that they had underestimated the odds that house prices would fall (this hadn’t happened in decades), that markets would crater, and that a deep, prolonged recession could happen. Similarly, the Covid-19 pandemic took many by surprise. A major pandemic had not hit America and Europe since 1957.
In hindsight, one can say that we should have known that we were in a housing bubble that was destined to pop—just think about the run-up of prices and the poor credit quality of many borrowers. And if we had looked more carefully at SARS and H1N1 and had had more transparent data from the Chinese government, we might have acted differently in February 2020. But pandemics and financial crises occur so infrequently that it’s always hard to predict them. They’re what finance calls “tail risks”: events considered so unlikely that they’re out in the tail of the probability distribution.
The tail risks that loom large in our minds tend to be existential—events that will kill many people, like a lethal pandemic, climate catastrophe, nuclear war, or an asteroid smashing the planet. An existential risk need not be a tail risk: famines and infectious-disease outbreaks once occurred with some regularity. They were not tail risks, in other words, even if they were catastrophic, because the odds of them happening were relatively high each year. We had to live with such risks, lacking the power to control them. But in modern times, technology and know-how have made them much rarer, moving them to the tail of a probability distribution.
Because tail risks are so unexpected, when they do happen, we often panic at first, and even turn to superstition, so that we feel as though we’re doing something. But such events can also be instructive. The financial crisis forced a reckoning in how we use models when dealing with tail risks. We can apply the lessons learned to tail risks in public health and other areas, too.
The risk-management approaches from finance have faced increased scrutiny for their failure to predict, and then manage, tail risks. Some argue that they actually can make tail risks worse because they give us a false sense of security, making us less resilient and vigilant for the unexpected. But this criticism misunderstands the nature of risk management, and where it has gone wrong. The existence of tail risks doesn’t mean that we shouldn’t use the tools from finance; it means that we need to use them better.
Measurements are based on data, and they are particularly challenging when it comes to putting a value on tail risks, which have, by definition, fewer data points—how many modern pandemics are available to study? The lack of data makes it harder to predict how bad things will get, how long the effects will last, what risk-management technique is best, and how much it should cost. Because tail-risk events are less frequent and less predictable, the market for insurance or assets that can hedge the risk is limited. We can safely assume, though, that existential events are expensive to manage because they affect many people at roughly the same time.
Another way to reduce risk in finance is through diversification, or holding many different assets, so if one goes down, another might go up, and you don’t lose as much money. Insurance companies can afford to take on our risk by insuring lots of people, most of whom normally won’t need to file insurance claims simultaneously. But insuring against existential events—which are simultaneous (or often so) and widespread—can’t rely on diversification, and that makes insurance costlier or unavailable.
Critics of Black-Scholes also note that the model assumes that risk is normally distributed, or that tail risk is extremely unlikely—when it happens more often than we think. If we don’t fully account for tail risk, they say, we might misprice it or underinsure for it and expose ourselves more than we realize. But this isn’t an argument against using financial tools; again, it shows how we need to apply them with greater care, and then, if the unexpected happens, update our estimates and adjust our strategies. We can manage tail risks better if we hedge and insure, that is, while retaining flexibility. Finance doesn’t guarantee that no one will lose money, but it does offer a methodical way to make sense of markets and reduce the risk of participating in them, even when the unlikely occurs.
Financial models are built to manage risk in markets. It can seem inappropriate to turn to them when life and death are at stake—in a public-health emergency, say. Further, pandemics involve risks where one person’s choices can have a significant impact on others. It’s a more complicated situation, in other words, than managing an investment portfolio. Yet financial risk-management models still offer something valuable in such circumstances: a systematic method to understand the risk faced— including systemic risk—and weigh the trade-offs in reducing it.
If we had used the tools from finance, how would things have differed with our pandemic response? The tools would not have prevented the health emergency, but they could have helped make a terrible situation less horrendous. First, a clear chain of accountability should have been established. In financial markets, the actors might be an individual investor, responsible only to himself, or a trader reporting to his boss at a bank, both ultimately accountable to their clients, whose money they manage—but an accountability chain exists. For a pandemic, public-health authorities make recommendations to politicians, whose job it is to weigh the trade-offs—lowering infection rates versus social isolation and economic disruption from the restrictions potentially imposed. These politicians are accountable to voters. This is not an easy decision-making process, needless to say, given that everyone has different preferences and risk tolerance. The principles from finance can help politicians weigh these complex trade-offs. For example: evidence soon emerged from previous pandemics and wars that prolonged school closures reduce the future earnings of children from low-income families. This reality could have been weighed against the evidence that schools were not a high-risk pandemic environment. A simple cost-benefit analysis that makes projections about the future would have demonstrated, even to risk-averse policymakers, that many schools could have been reopened earlier. Instead, politicians tried to offload decisions onto the health authorities, “following the science,” as the saying goes. The accountability chain wasn’t so clear.
The other risk-management failure during Covid-19 was in risk estimates. We went into the pandemic seemingly better prepared than any previous generation. Technology today offers improved data collection and the computing power to make sense of it. The initial hope was that we could use our data capabilities to do contact-tracing, figuring out who had the virus and keeping them from spreading it. But that quickly proved futile—from lack of testing data, lags in testing results, and the challenges of constraining people’s behavior in a free society. A more rational use of data would have been to determine earlier exactly what situations posed the biggest health risks. It took months longer than it should have to declare outdoors relatively safe, given the information we already had from Asian outbreaks, and what we know about how diseases spread. Instead, we insisted that people never leave their homes. Good financial risk principles would have avoided this approach by defining the safety problem early, and not pretending that it could be completely alleviated, but only reduced.
True, data analysis is hard in such situations. But in the future, the government can fund programs dedicated to estimating extreme risks and hire people up to speed on the latest techniques. Such a program once existed in America under USAID, but it needs more funding and modernization. The government can also avoid groupthink, which can make even the smartest people blind to tail risks, by encouraging prediction markets on catastrophic events. It’s notable that, during the pandemic, some of the most accurate risk estimates came not from the government but from Youyang Gu, a 26-year-old data scientist. Gu had no previous experience with infectious disease-modeling but did some Internet sleuthing and used his experience in data science to produce better estimates of deaths than many experts. “If you can’t make an accurate model in finance, you won’t have a job anymore,” he told MIT Technology Review. He was also early to recognize that reaching “herd immunity” was an unrealistic standard for a return to normalcy.
We need more Youyang Gus. In finance, the presence of many investors in the market means that contrarians are always about, hunting profit, and this creates more estimates, and better ones. A similar dynamic would help in the nonfinancial world, too.
When it comes to risk management, financial models are highly precise, though they often don’t directly generalize to public health. But the principles are the same. No Black-Scholes exists for bounty hunters, the World Health Organization, or the president, but the model does offer a logical framework to understand what makes one situation riskier than another and to determine when insurance will be more valuable. For example, the longer the options contract is for, the more it’s worth; a longer time frame means extra risk. Longer-term risks are costlier to insure, and shortening the time you are exposed to risk is valuable—if it is possible. Or the more variability there is in a stock’s price, the more valuable an option becomes. Or if there is more uncertainty about how bad and infectious a virus might be, the more valuable insurance is; for a pandemic, that insurance might mean redundant numbers of hospital beds or antivirals. (As for inflation, it has an impact here: the higher the rate of interest, the more insurance costs. This would make it more expensive to finance lockdowns, bailing out businesses and paying unemployment.)
It’s important to learn not only from risk-management failures but also from successes. Vaccine development turned out to be the best insurance strategy against the effects of exposure to the virus, and policymakers played a key role. Operation Warp Speed offered financial guarantees to pharmaceutical companies developing vaccines, reducing their considerable financial risk if those vaccines did not work. This accelerated the production of the new vaccines and proved a risk-management triumph.
A better future requires a more rational approach to tail risks. Climate change, for example, may pose risks to our way of life and our future prosperity. But how it will play out is unpredictable. We don’t know how much temperatures will rise, or what the effect on the environment will be, or what technologies could come along and help us adapt. It’s not surprising that, when it comes to climate, many people are again turning to superstition, with many environmentalists sounding like doomsday cultists.
“Good financial risk models force decision-makers to be accountable, use logic, and access the best data available.”
Financial economics once again offers an alternative. The University of Chicago’s Lars Hansen, Nobel prizewinner for his work on risk, has focused on how to measure the risk that climate poses and the economic trade-offs that we face. MIT financial economist Robert Pindyck’s research also considers climate risk. He advocates reducing emissions, but he projects that lowering the global temperature is unlikely, no matter what we do. He offers a framework to weigh the costs of different forms of adaptation and how to reduce the odds of an existential tail risk.
Climate poses many unknowns and the possibility of extremely bad outcomes; the work that these economists are doing acknowledges the uncertainty involved and the weaknesses of the data. But rather than turning to superstition and demanding unrealistic and costly interventions like phasing out fossil fuels in the next ten years or promising everyone an electric car, the risk analysis from financial economics offers a logical framework to weigh costs and benefits that can be updated as we learn more and face the surprises that are surely in store.
Odds are the next disaster will be something we haven’t taken seriously, or even imagined. We exit the pandemic in a more precarious place because we have much less insurance for such an event. Insurance comes from the ability to spend, and issue debt, when the next costly risk is realized. Overspending today can mean that we have less space to spend in the future—and trying to pay for a natural disaster, another pandemic, or a deep recession will result in spiking interest rates or a falling dollar. Reckless spending in good times is akin to forgoing payment of insurance premiums when you’re healthy and young.
Financial economics got a bad reputation during the financial crisis for creating more risk instead of reducing it. But that view largely reflected a failure of the industry to adhere to the very principles that we all should be using. The beauty of good financial risk models is consistency and transparency. They force decision-makers to be accountable, use logic, and access the best data available. And the principles are useful in any risk problem, from everyday risks that are well-defined, like Lindblad issuing a bail bond, to tail risks that are hard to predict, like environmental disaster.
True, financial models sometimes fail or are used incorrectly. But what is the alternative? We saw it during the pandemic: chaos, distrust, and superstition guiding policy and decision-making.
Top Photo: Bail bondsmen hedge their bets by using bounty hunters (above)—or “muscle”—as a form of insurance. (EVAN HURD/CORBIS/GETTY IMAGES)
