Case Study 1 — The COVID Savings Spike and the Limits of Rational Choice

In the spring of 2020, the COVID pandemic shut down much of the U.S. economy. Restaurants closed. Travel stopped. Office buildings emptied. Conferences were canceled. For most American households, the disruption was severe and immediate.

You might expect that with so much economic uncertainty and so many people losing jobs, household saving would have fallen — people drawing down their savings to make ends meet. Instead, the opposite happened. Personal saving in the U.S. spiked dramatically. The personal saving rate, which had averaged about 7% of disposable income for years, jumped to over 33% in April 2020 — the highest level on record. It stayed above 15% through most of 2021, then gradually fell back to normal levels.

This case study walks through the COVID savings spike using the behavioral lens. The episode illustrates several behavioral mechanisms working at once: loss aversion, present bias being temporarily inverted, status quo bias being broken, and the role of choice architecture (in this case, government policy creating new defaults). It also shows what the standard rational-actor model gets right and what it misses about how households actually responded to the pandemic.

What happened

In a typical month, U.S. households spend most of their disposable income — historical average around 93% — and save the remainder. The exact rate varies with the business cycle (saving rises in recessions, falls in expansions) but the long-run pattern is steady.

In March-April 2020, the saving rate jumped from about 8% (the pre-pandemic baseline) to over 33% in a single month. That is the largest single-month increase in the personal saving rate ever recorded. By the standard explanation, several things contributed:

1. Government transfers were enormous. The CARES Act sent direct $1,200 checks to most adults plus an additional $500 per child. Unemployment benefits were significantly enhanced (an extra $600/week federal supplement on top of state benefits). The Paycheck Protection Program kept many workers on payroll. Total federal transfers in Q2 2020 were several trillion dollars — a substantial share of which ended up in household bank accounts.

2. Consumption opportunities collapsed. People could not eat at restaurants. Could not travel. Could not attend events. Could not go to gyms. Could not shop in many retail stores. The supply of consumption opportunities had fallen dramatically. Even people who would have liked to spend more couldn't find things to spend money on.

3. Income held up better than expected. Despite huge job losses, aggregate personal income actually rose in 2020 because of government transfers. People expecting income shocks often didn't experience them.

These three factors together gave households more cash than usual and fewer ways to spend it. The saving rate spiked.

This is the standard "rational" explanation, and it is mostly correct. But it is incomplete. The behavioral lens reveals what else was happening.

What the standard model missed

Behavioral mechanism 1 — Loss aversion turning into precautionary savings.

Faced with massive uncertainty about the future, households became extremely loss-averse. The thought of losing more was much worse than the thought of gaining the same amount. So they saved cash even when they had it, even when they could have spent it on the few things still available, because the prospect of being short on cash later (if the pandemic continued, if their job were lost, if a family member needed help) was more terrifying than the prospect of saving "too much."

The standard model treats saving as a smooth function of expected income and current consumption needs. The behavioral model says: when uncertainty is extreme, loss aversion dominates, and people save more than the standard model predicts as a hedge against the very large psychological cost of being short later.

Behavioral mechanism 2 — Present bias was temporarily reversed.

In normal times, present bias makes people consume too much in the present and save too little for the future. During COVID, the "present" was so unappealing — sitting at home, no restaurants, no travel, anxiety about the virus — that the usual present-bias pull was weak. People who would normally have spent on dinners out simply had no dinners out to spend on. The future-self temporarily had more weight than the present-self.

The result was that people saved at rates the present-biased version of themselves would normally have refused. When the pandemic ended and consumption opportunities returned, the present bias came back, and the saving rate fell again.

Behavioral mechanism 3 — Status quo bias was broken.

In normal times, people maintain their consumption habits even when their incomes change. They go out to dinner because that's what they do. They take vacations. They buy clothes. The status quo of consumption is sticky.

COVID broke the status quo violently. The habits people had been on autopilot about — Friday night at the bar, Sunday brunch, weekly grocery store visit, monthly gym membership — were all suddenly impossible or risky. The status quo of consumption was disrupted. And once the habits were broken, many people found that the "lower consumption" baseline was not as bad as they had feared. Some of these reduced habits persisted even after restrictions lifted.

Behavioral mechanism 4 — The government as choice architect.

The CARES Act and the Paycheck Protection Program were not framed as nudges, but they functioned partly as choice architecture. By depositing $1,200 checks directly into people's bank accounts, the government created a new "default" — money you didn't have to actively earn. People who received the checks treated them somewhat differently from money they had earned themselves; some saved them as windfalls, others spent them on debt repayment or specific purchases that felt different from regular spending.

This is one of the most-studied features of the COVID stimulus: how people actually used the cash transfers. Surveys found that roughly 30–40% of households spent the checks fairly quickly (on groceries, rent, debt), but a substantial fraction saved them. The choice architecture of "money in your account" produced different behavior than the same amount of money would have produced if it had been delivered as a tax cut, a lower price, or a conditional voucher.

What happened next

Through 2021, the saving rate gradually fell from its peak. By mid-2021, it was around 10% — still elevated but coming down. By late 2022, it had fallen to about 3% — well below the long-run average. People who had saved unusually large amounts during 2020 began drawing down those savings, which contributed to:

• The 2021–22 inflation spike (lots of money chasing not-yet-fully-recovered supply)
• The "great resignation" (some workers quit because they had financial cushion to do so)
• Increased consumer spending on travel, restaurants, and services that had been closed during the pandemic
• A boom in the housing market (people had savings for down payments)

Some economists have called the post-2021 period "the great unwinding" — the period when households spent down the savings they accumulated during 2020. The unwinding lasted into 2023 and contributed substantially to the inflation episode we discussed in earlier chapters.

What this case study illustrates

Lesson 1 — Behavioral mechanisms work in both directions. Present bias usually makes people consume too much in the present. During COVID, the present was so unappealing that the bias was effectively reversed. Loss aversion usually leads to risk-aversion and conservatism. During COVID, it led to massive precautionary saving. The mechanisms are tools that explain behavior in context, not laws that always push in one direction.

Lesson 2 — Context matters more than the standard model assumes. The same households who would normally have spent freely became aggressive savers when consumption opportunities collapsed. The change was not because they suddenly became more rational. It was because the situation was different in ways the standard model abstracts from.

Lesson 3 — Government interventions create choice architecture. The CARES Act was designed to stimulate the economy. It also functioned as a piece of choice architecture: by putting money directly into bank accounts, it created defaults that influenced how people thought about and used the money. This is true of almost any government cash transfer program. The framing matters even though the dollar amounts don't change.

Lesson 4 — The "great unwinding" was predictable. Once you know that households accumulated unusual savings during 2020-21, you can predict that they will draw those savings down once normal consumption opportunities return. The 2021-22 inflation surge was driven partly by this — too much money (accumulated savings) chasing supply that hadn't fully recovered. Behavioral economics doesn't predict the timing precisely, but the direction is clear.

Lesson 5 — Behavioral economics is essential for understanding the COVID economy. A standard supply-and-demand model would have predicted some change in saving during COVID, but not the magnitude of the spike. The behavioral lens fills in the explanation.

Discussion questions

1. The COVID savings spike reached the highest personal saving rate ever recorded (33% in April 2020). What would the standard rational-choice model predict for the saving rate during a recession with massive uncertainty? How does the actual outcome compare?

2. Some economists argue that the COVID saving spike was largely "rational" — people facing unprecedented uncertainty were prudent to save more. Others argue that the spike reflected behavioral mechanisms (loss aversion, status quo disruption) more than pure rationality. Which framing do you find more persuasive? Are they incompatible?

3. The "great unwinding" of accumulated savings contributed to the 2021-22 inflation spike. How would you have detected this risk in 2021? What would you have recommended doing about it?

4. The CARES Act stimulus checks were delivered directly to bank accounts. How might the behavior have been different if the same amount had been delivered as a tax cut or as a voucher for specific goods? What does this tell you about choice architecture in fiscal policy?

5. Looking back, do you think the COVID-era saving spike was a mistake (people saved too much) or a wise response to genuine uncertainty? Was the unwinding inevitable?