A cost-effective stimulus to mend the effects of a 24 percent drop in GDP would cost no more than $1.3 trillion over a 6-month period. The bill that Congress just approved is much bigger because it allocates resources to people who are not necessarily affected and rescuing businesses, like Boeing, that are in trouble for pre-existing reasons.

President Trump just signed into law the $2 trillion Coronavirus Aid, Relief, and Economic Security (CARES) Act. Everybody agrees that some intervention was needed and, while there are many shortcomings, most people agree that this act is better than no act. Yet, how suboptimal is this plan?

To measure the degree of suboptimality, we need to establish a benchmark. At the beginning of 2008, Larry Summers, then only a professor at Harvard, advocated for a fiscal stimulus that was targeted, timely, and temporary (TTT).

A fortiori, his principles should apply to the fiscal stimulus needed to fight the Covid-19 pandemic. At the cost of breaking the alliteration, we dare to add another important principle: cost-effectiveness.

We fear the need for intervention will extend over time and we want to have the resources to sustain the economy, without incurring political fatigue or debt sustainability constraints.

We think there are three main goals in responding to the Covid-19 shock: to support the fight against the virus, to redistribute the burden of the coronavirus shock and to preserve the production capacity of the U.S. economy.

In this piece, we try to calculate how expensive it is to reach the second and third goals. We ignore the first not because it is not important (it of the utmost importance), but because it lies outside our expertise.

The most targeted, timely (and temporary) way to achieve the second and third goals is to subsidize firms to pay their workers and keep their current structure in place (i.e., a pause) until the end of the Covid-19 emergency.

When a firm is not operating, it still incurs three major sources of costs: salaries, financial costs, and rents.  To ensure that firms can weather this storm, we need three programs aimed at directly paying for each of these costs: a wage replacement program modeled on what the UK Parliament has approved, an interest replacement program, and a rent replacement program. In what follows, we try to estimate how much each of these programs could cost per month, depending on what fraction of the economy is shutdown.

The wage replacement program (WRP) targets all firms that see their revenues decline during this period. Since it is cumbersome and slow to establish causality, we envision that this program assumes that any decline in sales incurred in March 2020 over the sales made in March 2019 is due to Covid-19. WRP will pay 80% of the monthly wage of each idle worker in proportion of the drop in sales experienced by her company.

In this way, we can easily estimate the cost of the WRP from the average hourly cost for employees (equal to $34.7). Assuming 8 hours per day, 22 working days per month, the cost of replacing 80% of the monthly compensation equals $4,890 per person. In January 2020 the private sector employment in the United States was 136 million people (obtained by subtracting 23 million government and state employees from a total employment of 159 million).  Thus, if 100% of firms in the private sector were completely idle, the total wage bill under WRP would be $665 billion.

If we assume on average a loss of 24% (as predicted by Goldman Sachs), we arrive at $160 billion per month for the WRP. Assuming that employees will forgo the remaining 20% of their salary, the WRP achieves two goals: support the people affected and maintain their jobs without burdening their employers.

The interest replacement program (IRP) targets the same set of firms, with the same set of principles. To estimate its cost, we start from the total of nonfinancial corporate business debt securities and loans equal to $10.1 trillion.

Assuming an average interest charge of 5%, we find that the total financial obligation of the business sector amounts to $43bn per month. If we go with the same 24% drop in GDP as before, the cost of IRP amounts to $10.3 bn per month.

To calculate the rent replacement program (RRP), we need to determine the total amount   commercial leases. We start from the revenues of Real Estate and Rental and Leasing in the Census (NAICS code 53) and subtract the Lessors of Residential Buildings and Dwellings (NAICS code 5311), to obtain a total of $46bn per month. If we have to indemnify only the firms that lost revenues and in proportion of the revenues lost (24%), the cost amounts to $11bn per month.

Thus, a cost-effective TTT stimulus (which we will relabel TTTC) to mend the effects of a 24% drop in GDP would cost “only” $181billion per month (or $1,085bn over a 6-month period): half the bill just approved by Congress. Even if we add the $100billion for hospitals and the $150bn for the states (assuming they are spent in Medicare reimbursements), we arrive to a $1.3 trillion package, one third less than the bill just approved. Why is the approved bill so much more expensive?

Because it allocates a lot of resources handing out money to people who are not necessarily affected (like many who will get $1,200 check per person) and rescuing businesses, like Boeing, that are in trouble for reasons that precede the crisis and are using the opportunity to get subsidized. The stimulus might be timely and temporary, but it is certainly not targeted and even less cost-effective.

Yet, the most important difference is that our $1.3 trillion TTTC program will cover the U.S. economy for 6 months. After the $2 trillion CARES Act how many months (weeks?) of lockdown would we need before Congress will vote a new fiscal package?

Most certainly less than 6 months.

Note: We have taken Goldman’s projection of 24% (subsequently revised to 34%), which is annualized, to be applicable quarter on quarter. The quarterly estimate of 6.6% (subsequently revised to 9.9%), was in our opinion too small.