• Become an SSTI Member

    As the most comprehensive resource available for those involved in technology-based economic development, SSTI offers the services that are needed to help build tech-based economies.  Learn more about membership...

  • Subscribe to the SSTI Weekly Digest

    Each week, the SSTI Weekly Digest delivers the latest breaking news and expert analysis of critical issues affecting the tech-based economic development community. Subscribe today!

Recent Research: Did PPP actually save businesses or jobs?

February 10, 2022
By: Jason Rittenberg

A research team including members from MIT and the Federal Reserve Board assessed the Paycheck Protection Program (PPP) to determine if the initiative was able to keep businesses from closing and people from becoming unemployed. The authors present the highlight finding, which has been covered in multiple publications, as indicating that only about 23-34 percent of PPP funds went to workers who would have lost their jobs otherwise. This rate of effectiveness implies a cost of $170,000-$257,000 per job-year[1] of employment. The outcome seems surprising, given the program’s requirement that at least 60 percent of funds be spent on payroll. A dive into the results and policy implications bears lessons for future emergency program design.

PPP was a platypus of a federal small business financing program. Like SBA’s loan guarantees, businesses accessed the funding through a lender, such as a bank. However, PPP was designed to be forgivable, with requirements more similar to grants or tax incentives than loans.

The authors estimate that 94 percent of employer businesses with fewer than 500 employees received at least an initial PPP loan. This breadth is indicative of the program’s primary challenge, according to the authors, as the near-universal use rate suggests that most eligible companies received PPP support, whether the business needed the assistance to maintain its employment levels or not.

The other problem the authors identify in the program design relates to the employment requirements for PPP loan forgiveness. Businesses could use PPP funds to pay for employment costs — at least 60 percent — as well as utilities and rent or mortgage. Recipients also needed to maintain average full-time equivalent employment and three-quarters of their wages, but Congress eventually added some flexibility to how this could be met. As of Jan. 23, 2022, 83 percent of PPP recipients have requested forgiveness.

Despite the program’s 60 percent payroll expense requirement, the authors find a rate of 23-34 percent of funds being spent on employees who would have lost their jobs without the assistance. This finding does not represent ineligible use, but is instead derived by comparing the performance of PPP recipients to other companies.

The study leverages two comparison groups to produce these estimates. The first is companies with 401-500 employees (likely eligible for PPP) versus those with 501-600 employees (likely ineligible for the loans). The second comparison is limited to companies with fewer than 50 employees and looks at those that received a PPP loan early in the program with those receiving one later.[2]

Small businesses with 401-500 employees saw employment decreases that were about 4 percent better than those with 501-600 employees after about a month. This difference decreased to less than 2 percent by July 2020, with no difference remaining by September. The authors say this implies that the program saved about 1.98 million job-years — the lower limit of their estimate. Dividing the $510 billion in available PPP at the time (more on this, below) by the job-years saved yields an average cost per job-year saved of nearly $258,000. Further, the authors assume average employee compensation of only $58,200,[3] implying that only 23 percent of PPP funds went to the employees themselves.

The results for smaller businesses look slightly better. For firms with fewer than 50 employees that accessed PPP early, employment was about 12 percent higher than for other firms of this size. The authors use this information to estimate that the program may have saved as many as 3.0 million job-years, which yields an average cost of approximately $169,000. This cost-per-job implies that 34 percent of PPP funds went to the employees.

Notably, the authors derive these estimates of cost-per-job on the basis of the total amount of PPP financing available at the time. The result arguably tells the story of how many PPP dollars were spent for each job that may have otherwise been lost, but, because PPP funds could be used for payroll or certain hard costs, the cost-per-job figure used in the study (and, therefore, cited in the news media) does not represent what companies spent per job. More specifically, if companies spent 40 percent of their PPP funds on utilities and rent/mortgage, as allowed, the average cost per job-year saved would be $102,000-$155,000 instead of the estimates of $169,000-$258,000 provided above.

Speaking of PPP’s use for a company’s non-employment expenses, the authors attempt to determine how well PPP did at helping businesses remain open. The data suggest that, at least for companies with fewer than 50 employees, about two-thirds of the jobs saved by PPP loans were among companies that may have closed without the assistance (i.e. 8 percentage points of the employment difference between early adopters of PPP loans and later adopters can be attributed to companies that would have closed).

The authors are not impressed with this impact, however, for two reasons. First, they say there is evidence that could indicate PPP kept businesses open that would have closed without the pandemic (or assistance). Second, the comparison of businesses with 401-500 employees versus those with 501-600 did not yield a measurable difference in closure activity.

Ultimately, then, how well did PPP work? The evidence suggests that the program did save millions of jobs, but did so with unnecessary expense.

Such excess could, perhaps, be forgiven as part of the rush to provide urgent help in the face of a unique emergency. Congress was uncertain how broad or long the pandemic economic downturn would be and was trying to provide aid quickly. The depth of this uncertainty is evident throughout 2020 emergency legislation: in addition to authorizing PPP and several other loan programs and tax benefits to keep businesses employing citizens, the additional benefits for unemployment compensation meant that many Americans could make more money if they were laid off than retained by their employer.

Unfortunately, the rush to put PPP assistance in place quickly contributed to uneven access to the first round of loans by businesses in under-served geographies and to a fraud rate that has been estimated to be as high as 15 percent.

More significantly, PPP’s lack of a needs assessment — something likely considered a feature, as automatic eligibility contributed to rapid approval — may be its biggest failing. The authors’ finding that 94 percent of employing small businesses received a loan indicates that a rigorous test would have provided funds to far fewer firms. The design decision not to tie PPP funding to need likely violated one of the maxims of good economic development policy: that public funds should be spent if and only if the private investment (i.e. continued payroll expenses) will not happen but for the public expense (i.e. PPP loans).

The COVID-19 pandemic was a very different driver of economic harm than a natural disaster or the collapse of the housing market, and it is not surprising to learn that the design of an emergency response program may have been sub-optimal. However, we must learn from the experience and be better-prepared to implement emergency programs that can direct assistance to companies and employees that need the aid.


[1] A “job-year” is a unit representing full-time equivalent employment for one year.

[2] The process used for this second comparison is far more complicated than described. The authors use an event-study model estimating an average impact for businesses during the first 11 weeks and then compare this experience against those in later weeks. See the full article, particularly footnote 4, for additional details.

[3] This is calculated by using Bureau of Labor Statistics weekly wage data ($786 average, including truncation for pay above $100,000 per year) multiplied by a 42 percent increase for benefits.

coronavirus, sba, small business, debt, economic impact