How Much Stimulus Do We Need?

March 19, 2020

If the experience of the last recession is a guide, avoiding a severe downturn will take far more stimulus spending than is currently being discussed—as much as $3 trillion.


The coronavirus is, first and foremost, a public health crisis. The most immediate questions it poses are how to keep it from spreading, and how to ensure that health care gets delivered to those who need it—both those with severe cases of COVID-19 and those with other health care needs.

That’s the biggest and most immediate problem. But after that come the economic problems. From a macroeconomic standpoint, the coronavirus is primarily a negative demand shock—that is, a decline in spending. It’s important to make this point clear: Whatever disruptions the virus creates in supply chains, the reason businesses are closing and people are being laid off is because no one is buying what they sell. Not because they can’t get the inputs they need to make it. Airlines are in trouble because no one wants to fly, not because they can’t buy jet fuel. Restaurant workers are being laid off because people aren’t eating out, not because meat and vegetables can’t be obtained.

This point might seem obvious, but it’s important to stress. Because it tells us, at the broad-brush level, what the policy response needs to look like. The response to a demand shortfall—a lack of spending—is for the government to step up and spend money, or to find ways to get others to spend. To stop the coronavirus downturn from becoming the coronavirus depression, what’s needed is to get a lot of money into the economy, fast.

How much money? How big should a fiscal stimulus be?

Conceptually, the question is straightforward. First, we make a guess about how large the output gap will be—that is, how much GDP over the next year will fall short of the economy’s full potential. (For purposes of this exercise, we can assume the economy prior to the crisis was at potential.) Then we divide that by our estimate of the fiscal multiplier.

If we frame the problem in terms of unemployment instead of output, there’s an additional step. Okun’s law says that there is a predictable relationship between GDP growth and the change in the unemployment rate. So now we take our expected rise in unemployment and convert it to the expected shortfall in growth, and then divide that by the multiplier.

What kinds of numbers does that give us?

Orthodox economics long held that fiscal multipliers are small. But the professional consensus has shifted in recent years; today, it is widely accepted that in recessions, fiscal multipliers are much larger than in normal times. Today, a conventional estimate of the multiplier in a recession would be around 1.5. In other words, for every dollar of additional spending, GDP increases by $1.50, as the initial spending generates incomes that are in turn spent.

Now we ask, what would happen to GDP in the absence of stimulus?

This is harder to answer, but we can plug in some guesses. Minneapolis Fed president Narayana Kocherlakota suggests a gap of 6 percent of GDP for two years. This translates into a required stimulus of 4 percent of GDP for two years, or a total package of around $1.8 trillion. Analysts at Goldman Sachs are suggesting a 5 percent fall in GDP. Given a baseline of positive growth, this implies a similar number.

What if we use unemployment, rather than output? The forecasters at the UCLA Anderson School of Management project that the unemployment rate will rise to 6.6 percent in 2021. A standard formulation of Okun’s law says that for each additional point of GDP growth, the unemployment rate falls by about 0.4 points. So an increase in the unemployment rate of 3 points, as predicted by the UCLA team, implies a growth shortfall of 7.5 points. Using the multiplier of 1.5, as before, we find that preventing this requires a stimulus of 5 percent of GDP, or about $1.1 trillion, over the next year.

This range from $1.1 to $1.8 trillion seems reasonable as a first cut. We can take this as a low end of what’s required, which is still bigger than what’s been considered so far. But there is reason to think that the stimulus required to avoid a deep downturn might in fact be much larger.

When Kocherlakota suggests a 6 percent fall in GDP in the absence of stimulus, he points to the 2008–2009 recession, when output did indeed fall about 6 percent. But there’s a problem: 2008 is not an example of absent stimulus. That recession saw a major stimulus package—the American Recovery and Reinvestment Act—passed in the opening days of the Obama administration, following a smaller but not insignificant stimulus (about 1 percent of GDP) passed under the Bush administration. Not only that, it saw the Fed cut rates from 5.25 percent to zero. The fall in demand in the absence of these interventions would have been much larger than the 6 percent we actually experienced. If we are going to use 2008–2009 to guess what a major downturn could look like without stimulus—which we need to, to estimate how much stimulus we need—then we have to back out the effects of these policy interventions.

Fortunately, we can do that calculation. It’s a bit tricky because one should also take account of the shift in the trade balance (lower GDP means less imports, all else equal), as well as the effects of inflation on the real interest rate as perceived by borrowers. But the bottom line is, using the standard multiplier of 1.5 and assuming a 1 percent increase in GDP for each 1 percent reduction in interest rates (also standard), the fall in private demand between 2007 and 2009 was more than 20 percent of GDP. Not 6 percent.

In other words: From 2007 to 2009, GDP fell by about 6 percent. But this was after a massive stimulus bill, and a massive cut in interest rates. Interest rates were cut by 5 percent in those two years, and the federal deficit increased by 9 percent. The 2008 crisis also saw a fall in the trade deficit of 2 points, which is the equivalent of further stimulus spending. Given the global nature of today’s crisis, a similar improvement in the trade balance seems unlikely. If we use our standard estimates of the effects of public spending and interest rates to estimate what GDP would have done in the absence of these responses, we have to conclude that it would have fallen by more than 20 percent.

Some of the 9 point increase in the deficit between 2007 and 2009 wasn’t explicit stimulus; tax revenues always fall in a downturn, and some categories of social spending increase. But the majority of it was stimulus spending and tax cuts.

Is today’s crisis on the scale of 2008’s? It seems very possible that it is. And this time, of course, there is no question of a 5 point fall in interest rates. The federal funds rate coming into the crisis was only 1.5 percent.

So if we look back to 2008 as a model of the kinds of demand shortfalls our economy can suffer in the face of a severe shock, we should be thinking in terms of a 20 percent fall in private demand, not Kocherlakota’s 6 percent. And that implies a required stimulus on the order of 13 percent of GDP, or $3 trillion—perhaps sustained over multiple years.

Stimulus isn’t all that’s needed, of course. The current crisis, unlike past economic crises, involves significant disruptions on the supply side, which may get bigger if the outbreak gets worse. People under quarantine can’t go to work, whether there are jobs or not. In cities where bars and restaurants have been closed by the authorities, a lack of paying customers is no longer their problem. But these interruptions to production don’t cancel out the demand shortfall, which will come back with a vengeance once the virus is contained. And well before then, there will be an increasing number of unemployed people without money to meet their basic needs. Financial obligations also continue to bind. Every past economic downturn has been amplified, if not sparked, by a wave of defaults ripping apart the web of credit through which modern economies are organized. Unless we see much more comprehensive measures to forgive debts and suspend rent payments, it’s essential to stabilize incomes so that people can keep making these payments.

Another question is the form a stimulus should take. Current proposals emphasize direct, flat payments to individual households. There is a lot to recommend this approach: To avoid a deep downturn, getting money out quickly is important, and flat payments are quicker than something more targeted. And while not all recipients of stimulus checks will need them, or spend them, there’s a converse danger that a targeted program could miss people who do need support—especially in a fast-changing situation where those in need tomorrow may be hard to predict today. Other proposals, including tax cuts and no-strings-attached bailouts for selected industries, are less defensible; a better option might be generous aid to state and local governments, which are already bearing the brunt of the public health crisis and soon may be facing a collapse in tax revenue. In the medium term, we should be using federal dollars for investment in urgent social needs, like health care and decarbonization, and not just sending out checks. (For a broader economic program in response to the crisis, see here.)

But whatever form stimulus takes, to avoid a depression, the federal government is going to have to ramp up spending over the next year by much more than is now being contemplated. I don’t claim that $3 trillion is the right number; if someone says they are confident about economic developments over the next year, you can safely ignore anything they say after that. But if you look back to the last recession for a sense of what kind of stimulus we might need today, then $3 trillion is the rough number you will come up with.