top of page

How exactly could a recession unfold?

Moving forward—propagation mechanisms:

I recently posted on the current economic impact of the coronavirus, arguing that it was still too early to have a good grasp of the situation. Given the orders for social distancing, we would expect large increases in unemployment and decreases in consumer spending and GDP, so it is hardly surprising that we are indeed seeing those changes in the data. However, the real question is whether these initial shocks will transform into a full negative economic cycle. For that, the big question is on what structural economic relationships will prevail in the months to come.

The truth is that no one knows what’s going to happen when people go back to work. Don’t believe anyone who tells you otherwise. This is an unprecedented and historically significant event, and it will undoubtedly change some of the economic structure that economists use to forecast. What we can do, however, is discuss how economic cycles usually come about and which economic relationships are likely to come into play.

In economic terms, the structural economic relationships that cause shocks to reverberate through the economy, potentially lengthening and worsening an economic crisis, are called propagation mechanisms. There are known mechanisms that affect economic swings all the time, and there are mechanisms that can be unique to individual crises, such as the effect of the subprime mortgage market on insurance companies in the 2008 crisis. Let’s take these in turn.


Traditionally, a big transition mechanism for business cycles is changes in investment.[1] In the standard DSGE and New-Keynesian models used in economic simulations, risk-averse consumers want relatively constant consumption over time, and they save and borrow to maintain this constant consumption. This fluctuation in savings causes fluctuations in investment, and these investment fluctuations are the primary mechanism for economic cycles. In addition, there are relatively standard extensions to these models that include producer risk-aversion, too.

I would imagine both consumer uncertainty and producer uncertainty are large factors driving investment decisions at the moment, and the recent declines in the stock markets reflect that uncertainty. However, this is actually a blessing, as the uncertainty is something easily reversed in time. Without any other mechanisms making things worse, investment could potentially return to pre-crisis levels. In the past weeks, I have seen many professional forecasters predict a quick recovery... I believe this is due to the expectation that investment could quickly recover.

In terms of policy, the response to investment shortages is straightforward and well-established. The central bank’s decrease of the baseline interest rate and fiscal spending are standard practices that address this issue. Now that we have reached the zero lower bound, some of the non-standard fiscal policy used in 2008 such as quantitative easing are also potentially useful. To be clear, these policies are very useful for getting investment started again once the shock has ended, but it is limited use during the period of reduced consumer demand from health restrictions.


Unemployment is also a potential transmission mechanism for the current crisis. Unemployment is usually considered a lag variable in economic models, as it follows the behavior of other variables due to the time it takes to hire good people and the subsequent reluctance to fire people as a result. Unemployment was a primary concern for Keynes when he wrote about the Great Depression in his General Theory. In it, he described a cycle where involuntary unemployment lead to reduced demand for goods among the unemployed, which in turn lead to unemployment.

We have already seen a significant spike in unemployment claims during the first half of March. What remains to be seen is whether these newly-unemployed people are able to return to their old jobs once demand starts to recover. One key element to labor markets is “search and matching,” meaning that it is difficult to find a single appropriate job for a candidate and a single appropriate candidate for a job. However, if each unemployed worker can simply go back to his or her previous job, then this problem is significantly reduced.[2] This does require, however, that those previous jobs (and previous employers) still exist. Thus, while unemployment is usually slow to respond to recessions and recoveries, we have already seen it respond quickly to the current crisis, and it could conceivably respond quickly to a recovery, mitigating its impact as a propagation mechanism.

There are a range of potential policies that the government can institute to aid in this, such as encouraging employers to temporarily furlough employees rather than lay them off entirely and provide bridging loans to employers who wish to meet payroll during this time.[3] I hope that the US stimulus package is designed well for this purpose, though the details are still relatively uncertain.


I have written in a previous post about the stagnant inflation leading into this crisis. Price shocks can have a few effects on the economy: sudden price increases make consumers relatively less affluent, investment goods such as machines and tools more expensive, and the overseas demand for exports lower. In addition, there is evidence that high inflation is more volatile, meaning higher risk for producers that the price they expect when they produce will be different from the one they can charge. Finally, in order to reduce inflation, the standard practice is for the central bank to increase interest rates, thus reducing overall investment levels.

Luckily, we have had low inflation over the past decade despite conditions in which one might expect the opposite, and we are currently experiencing a negative demand shock, which is likely to cause prices to decrease rather than increase. This raises its own problems, as decreasing prices (or deflation) can cause money to be relatively more valuable than investment, leading investors (and consumers) to just hold on to money. Furthermore, price decreases make assets in short supply relatively more valuable, potentially causing commodity bubbles.

As we are arguably seeing both a supply and demand shock due to the current health situation, the overall direction of price movements is highly uncertain. It will largely depend on how quickly demand recovers and whether any of the other mechanisms change the relative rate of recovery between supply and demand. In this respect, one benefit of the current situation is that our third largest trading partner, China, preceded us in this crisis, so these supply lines could be strong again by the time demand recovers.

Thus, we have a lot of uncertainty on the potential response of prices to the current situation. With both supply and demand shocks, inflation could go either way, and any of the other propagation mechanisms discussed here could potentially alter the relative recovery speeds of supply and demand. Likewise, the structural elements that led to abnormally low inflation over the past decade are likely still in place, limiting our understanding further.

To compound that uncertainty, the US government is now engaging in the largest fiscal stimulus in history. Coupled with the current fiscal deficit, we are potentially adding 14% to our national debt this year alone. While the size and the scope of this package is probably appropriate for the size of the potential decline, the point here is that this stimulus adds to the uncertainty over price changes. I am sure the Federal Reserve Board is watching inflation indicators closely.

Sovereign Debt:

As just mentioned, overall fiscal cost on the government for the stimulus packages currently proposed is in excess of $2.2 trillion dollars. In context, the American Reinvestment and Recovery Act of 2009 cost the government $800 billion, and it was criticized by Republicans as being too costly for the national debt. The budget deficit leading into this crisis was roughly $1.1 trillion, so overall national debt might increase by 3.3 trillion this year alone, or roughly $9,000 per person.[4]

Again, the magnitude of this stimulus is probably necessary for the size of the economic challenge currently facing us. The question is how we will pay for that later. We can’t pay for it this year, as the sum is just too large and increasing taxes right now would be a bad idea. We therefore have two options: either just paying the interest with more debt, essentially doubling the deficit each year moving forward, or some sort of austerity measures such as reducing public spending or increasing taxes to make up the difference. Currently, we are paying 2% on 10-year treasury bonds, so this will cost $60 billion every year just to pay the interest. This 2% is historically very low—if the interest rate jumps to a moderate 5%, that $60 billion will become $150 billion.[5] Thus, even if we decide to just add this to our debt in and never pay it down, the cost of servicing the debt would be a significant cost.

As a percent of GDP, the US non-defense discretionary spending is at historically low levels. The dollar amount is only $850 billion, 18% of the total government spending.[6] If we were to reduce that by $150 billion, that would be more than the current budget of Health and Human Services and Housing and Urban Development combined. It is entirely possible that the CDC’s budget will be cut heavily as a result of this epidemic.

The worry is that the current fiscal stimulus will lead to further fiscal contraction relatively soon. Classical economists describe an effect called Ricardian Equivalence, in which a fiscal stimulus such as this won’t have an effect as the consumers receiving the stimulus are likely to save the money in anticipation of future taxes. This is not my concern (though I do wonder about the $1,200 direct payment in the package for this reason). Instead, my concern is that the current package will lead to a fiscal cycle, in which the current stimulus will require a contraction later. If the downturn lasts longer than we expect, then that contraction could come before the recovery is fully stable. Now, more than ever, we need the federal government to institute a fiscal rule requiring revenues to increase when GDP is high and unemployment is low.

Corporate debt and bankruptcy:

For mechanisms particular to this crisis, a significant issue surrounds corporate debt. Debt exhibits a key threshold effect: if you can’t meet the market rate of interest, you risk default. Debt has played a key role in severe recessions before, and Ben Bernanke famously argued that it was a central element in the severity and length in the Great Depression. There are a series of issues with bankruptcies in the economy, from the liquidity effect of bankruptcies making it harder for banks to lend to other companies to the supply side effects of reduced production capacity and people out of jobs.

I am generally less alarmist about the current state of corporate debt than some other economists,[7] but even in a strong economy most businesses cannot sustain 2 months of lost revenue with labor costs remaining constant. This is where the current stimulus package is most important, supporting commercial paper markets, providing payroll relief to small companies, and bringing back a direct instrument for the central bank to lend to small businesses.[8]

However, allowing businesses to borrow to meet overheads only addresses liquidity, not solvency. Even with perfect access to loans at low interest rates, the liabilities of many small businesses might be too high to take on more debt and still remain solvent. The government’s response to this has been to make the debt forgivable after a period of time, but I’m not sure how that potential relief transfers onto the balance sheet… can one record it as an asset or discount the debt somehow? If not, solvency is not addressed through this package.


As supply chains have been interrupted by the health crisis, many businesses have had to find alternative sources for potential suppliers. Bankruptcies and continuing hotspots of the outbreak will further compound this issue. Commodity price fluctuations and subsequent changes in raw material production will make some sources less profitable than they were previously. The current price of oil is one example of this.

Anticipating the effect of the coronavirus on trade is very difficult. Anticipating this effect on the overall economy is even harder, as our trade networks are complex and strongly integrated into our economy. Exports are a significant aspect of aggregate demand, comprising 13% of overall production in the US. However, imports are a 18% of supply, as well, and many imported products are integral to the supply chain, magnifying this effect. These numbers, if anything, understate the effect of trade, as international trade is concentrated in a small percentage of firms that are both larger and more efficient than their competitors. Furthermore, the largest export industries in the US include some key investment goods, such as industrial machinery, computers, medical devices, and aircraft. Without the economies of scale in these areas, access to these investment goods and the efficiency of production in the US would likely decrease. However, one caveat to this is that the discrepancies between these large internationally focused firms and the domestic markets likely precipitated growing income inequality that has led to political turmoil across the western world.

The international nature of the coronavirus means our economic vulnerability is not limited to our domestic markets. Here, policy and diplomacy have large potential. If we are able to respond to the crisis in the international community through increased trade and multilateralism, then our export industries will likely benefit and our consumers will likely face lower prices for the goods they require. However, policymakers should recognize the multilateral approach would probably aggravate economic inequality at a time with potentially high unemployment. Thus, support for multilateralism should be contingent on heavy social restrictions on the businesses being supported, including high labor standards and moderate executive pay.

Financial Markets: (a bit technical)

The recent declines in stock market values themselves could represent a wealth shock. More importantly, the underlying lack of demand in the current situation could fundamentally reduce the values of financial assets to the point that household consumption remains depressed for some time. This is distinct from the standard investment argument for a rather technical reason—the standard new-Keynesian models are based off models in which the return on capital is tied closely to the marginal productivity of capital. As a result, it is virtually impossible for consumers to lose money on financial panics in these models. Instead, we have to allow for another shock to household wealth.

The stock market is not the economy, yet the value of housing and retirement funds in the US comprise the majority most households’ assets. A decrease in the value of these assets would lead to a readjustment of expected earnings and potentially of spending. A standard New-Keynesian/DSGE model is insightful here, as it discusses consumption-investment decisions in light of expectations. Interestingly, if the shock to wealth is perceived as permanent, then people’s investment habits would probably not change much, as there would be a new expectation of normal, and both savings and consumption would decrease in proportion, keeping the relative balance constant. However, if people anticipate the market to rebound, they are less likely to reduce their current consumption, leading to a significant drawdown in savings and investment, potentially causing a negative supply shock. If people anticipate the market to decline further, consumption could be low for some time, and we could have a lack of aggregate demand, effectively a reversal of the situation over the past 20 years. Current low interest rates and the potential for quantative easing are good policies, and the central bank should hold steady. In the case of a negative supply shock, however, this could require keeping interest rates low in the face of increasing inflation.


The current economic outlook is highly uncertain, but economic theory is very useful for understanding potential risks to the economy and appropriate policy responses. All of the mechanisms discussed above are issues regularly discussed in macroeconomic theory, and each has a potential to cause a powerful economic cycle in the near future. More importantly, however, is the potential relationships between these effects: a series of bankruptcies could lead to high inflation and high unemployment, or a fiscal contraction could lead to higher interest rates, lower investment, an exchange rate appreciation, and lower export levels.

We really won’t know which of these effects are taking hold until people start going back to work. Once that happens, public policy will also play a huge role, as limiting the magnitude of one mechanism could prevent other mechanisms from becoming problematic. If I were the government, I would be doing everything I possibly could to help companies avoid bankruptcy, promote multilateral trade, and keep social programs in place. I would pay particular attention to business investment and bankruptcy data. Consumer price indices and unemployment insurance claims will also be useful indicators, but only in a month or two.

Footnotes: [1] My understanding is that this is the primary mechanism in the models discussed yesterday on the impact of a bad influenza outbreak leading to the conclusion of a 6% decline in GDP. The papers cited above remain unclear about what is included in their modeling. [2] There might arguably be a potential efficiency benefit of this process, as employers who have laid off large portions of their workforce can potentially only hire back the workers they like, giving rise to a “spring cleaning” effect. [3] The stimulus just passed in the US does this, but it goes one step further and potentially forgives these loans after a period of time. I believe that to be a bad idea as it would make the policy more expensive and therefore limit the potential scope of the assistance. A better option would be to offer a negative interest rate on the loan under strict conditions. [4] This is not even taking into account the changes to non-discretionary spending spending such as welfare on the budget. [5] or over $400 per person just in interest payments on the current stimulus [6] 850 billion compared to 4,700 billion [7] For one, commercial and industrial loans held by commercial banks is 11% of GDP, which is historically high, but still lower than it was for most of the 1980s. [8] I personally think that this part of the package didn’t go far enough… we should have one month freeze on all long-term debt contracts. No one should have to pay any payments on any loans this month, interest should not accrue, and the term of the loan should be extended by one month to compensate.

50 views0 comments

Recent Posts

See All


bottom of page