Paycheck Protection Program spread loans widely, if not evenly

You can access a timeline tracing the many developments of the Paycheck Protection Plan here.

Congress charged the Paycheck Protection Program (PPP), enacted under the Coronavirus Aid, Relief, and Security (CARES) Act on March 27, with lending to small businesses stricken by the effects of COVID-19, and authorized private lenders to facilitate an unprecedented flow of Small Business Administration (SBA)-guaranteed loans. On this first objective, the program delivered, approving nearly 4.9 million loans with a total volume just north of $521 billion. But its performance, its efficacy, and its efficiency remain unsettled questions.

To address these concerns, the SBA released loan-level data comprising each approval during the PPP’s initial run from April 3 to June 30. This data offers a low-resolution picture of how the program performed against several objectives specified by the CARES Act, such as targeting rural and underserved markets, including single-person businesses, supporting hotel and restaurant franchises, and — its most important function — retaining jobs. 

PPP data fall into two categories: loans smaller than $150,000; and loans at least $150,000 in value. For small loans, exact dollar figures are given, but borrower information is anonymized to the zipcode, legal structure, and 6-digit NAICS code of the business. For large loans, only ranges of loan values are given, but the borrower information is precise, with names and street addresses listed. The vast majority of applicants also reported the number of jobs that would be retained should an application be approved.

The utility of the PPP data is limited by the aggregations that the SBA has chosen to perform, which it considered necessary to keep confidential the salaries of employees in smaller businesses. The decision came after a back-and-forth between Treasury, oversight bodies, and Congress. One of the most controversial situations cropped up early, after reports that large and financially sound companies took PPP loans (though more than $38 billion has been returned). Calls are still being sounded to release the names of smaller creditors and the loan amounts given to larger creditors. 

Additionally, a Bloomberg team led by Mark Niquette found that the dataset is sprinkled with errors. While no single type of error seems to be significant in the scope of 4.9 million loans, their frequency casts doubt on the precision of SBA data. But where these data are limited, we consider several studies carried out since the program’s establishment.

In short, access to loans was widely distributed across nearly 5,500 lenders. However, the pool of loan recipients looks quite similar to the pool of business owners. Though demographic data are subject to self-reporting, the available data indicate a lack of priority for women, minorities, and veterans contrary to CARES Act mandates (Fig. 2). This may have resulted from the SBA’s delay in issuing guidance prioritizing these underserved and rural markets. Rural markets, however, seem to have fared better in spite of this delay. Rural markets were overrepresented due to the ability of small lenders to ramp up loan approvals compared to large banks with specialized loan departments, though the number of loans in an area were roughly proportional to a zipcode’s number of businesses (Fig. 5).

The PPP seems to have been effective in another feature, its explicit inclusion of nonemployers, those “sole proprietors, independent contractors, and eligible self-employed individuals” who have been traditionally excluded from stimulus. Just over one million nonemployers were approved for loans. That also means the total number of loans approved is not representative of the number of jobs retained: 37% of loans saved just one or two jobs.

The hospitality industry more often reported retaining higher numbers of jobs than any other sector (Fig. 7), despite concerns about the ability of restaurateurs and hoteliers to keep employees on during lockdowns. Changes to the design of PPP, and the belief by some that blanket loan forgiveness is on its way, have prompted concerns over its efficacy. Though research has just begun, economists estimate that the program spared less than 5% of imperiled jobs.

Figure 1.

An earlier YPFS blogpost discussed the fact that more large borrowers were approved earlier.

Helping rural markets, but not the underserved

The SBA Office of Inspector General wrote in a June report that no guidance had been issued to lenders prioritizing small business concerns run by “veterans and members of the military community, small business concerns owned and controlled by socially and economically disadvantaged individuals [...], women, and businesses in operation for less than 2 years” (4). The CARES Act stipulated that the SBA issue such guidance.

As may be expected, what demographics that are available show that the pool of approved applicants closely resembles those of US business owners (Fig. 2). This finding should be taken with a grain of salt, as loan applications made demographic questions optional, resulting in only 24% of approved applicants answering in one of three fields on the application: race or ethnicity; sex; and veteran status. Information about the other groups included in those “underserved” markets was not collected during the PPP’s initial run.

Figure 2.

Approved borrowers closely resemble business owners.

A small study by the National Community Reinvestment Coalition, which has conducted several studies to investigate discrimination against Black and women loan applicants, can offer more granular information. Their method pairs white and Black small business owners with similar credit profiles applying for credit at the same banks, and compares the differences in outcomes. In line with previous studies of SBA lending, they found that PPP lenders treated white applicants better than Black applicants, and men better than women. 

To address these concerns, in late May the SBA designated $10 billion of PPP funds for community development financial institutions (CDFIs). The country’s 1,129 CDFIs are certified by the Treasury, and “provide critically important capital and technical assistance to small businesses from rural, minority and other underserved communities.” According to the press release, CDFIs had already approved $7 billion in loans during the program’s first round of funding, which was exhausted on April 16.

Though self-reported data may bias the reported statistics, CDFIs seem to have slightly increased the racial diversity of PPP borrowers, though differences in borrower gender and veteran status appear to be negligible between CDFIs and non-CDFIs. In particular, CDFIs appear to have significantly boosted loans to Black applicants. From the beginning of the program, more borrowers who applied through CDFIs reported their race as “Black or African American” than did borrowers from other lenders (Fig. 3). To some extent, the presence of CDFIs addressed the fact that, during the first round of funding, approved borrowers were more often white than in subsequent rounds.

Figure 3.

Comparison of CDFIs with rest of PPP approvals.

This “head start”, as termed by Bloomberg, may be due to the outsize impact of small lenders (Fig. 4), most of which are not CDFIs. While the four largest US banks engaged in 36% of small business lending before the PPP, they only shared 4% of loans in the first round. Smaller lenders -- more often located in rural markets, which tend to be whiter -- made up the difference. Without dedicated small-business lending departments, the smallest of these lenders were more able to redirect resources to loan approvals. Hence, the key to accessing rural markets may not have lay in the guidance, but in the type of lender authorized.

Figure 4.

Many tiny lenders.

Research using first-round SBA data argues this point, that, despite delayed SBA guidance, the wide diffusion of loans among lenders meant that more loans went to small banks and rural areas. But the paper questions the value of early relief to rural markets. In retrospect, it is clear that urban areas faced much more severe outbreaks earlier than did rural areas. The authors write that the first-round “funds flowed to areas less hard hit” by the coronavirus, a relationship which held when both the value and number of loans were considered. This research does not evaluate the PPP’s rural-markets mandate per se, but it illustrates how such guidance may not have allocated aid where it was most needed. And in general, it illustrates how speedy and equally-distributed lending (Fig. 5) may not have made for efficient allocations.

Figure 5.

Loan approvals were roughly proportional to the number of businesses.

Restaurants and hotels distribute gains widely

Different sectors also varied in their usage of the PPP. Restaurants and hotels have reported retaining comparatively more jobs than other sectors (Fig. 8). Subsections 1102(a)(2)(36)(D)(iii-iv) targeted franchisees classified under NAICS as Accommodation and Food Services to apply for PPP loans by temporarily exempting them from federal regulations prohibiting certain affiliations between businesses. Hotels and restaurants were approved for the sixth-most loans among all industries, but the most jobs retained per loan (Table 1).

Figure 8.

Restaurants and hotels report flatter job retention curves.

Academic research on restaurants and hotels have yet to be published, but the PPP received criticism early on for strict loan terms. The original terms required borrowers to spend funds within eight weeks to retain or rehire workers in order for the loan to be forgiven; this proved difficult for many restaurants that were closed or limited in the service they could offer due to government restrictions. Others waited until the term lengthened to 24 weeks before reopening, forestalling employment gains.

Table 1. Loans made, jobs retained, and jobs per loan by industry.

Industry	Loans	Jobs retained	Jobs per loan
Professional services, research, & IT	528,391	4,558,974	8.6
Other	497,728	4,425,858	8.9
Healthcare	422,974	7,052,331	16.7
Construction	387,810	4,777,952	12.3
Retail	381,133	4,466,217	11.7
Restaurants & hospitality	311,805	7,133,099	22.9
Real estate	204,003	1,379,370	6.8
Administrative support	198,557	2,851,569	14.4
Manufacturing	195,958	4,316,136	22.0
Logistics	150,690	1,561,370	10.4
Wholesale trade	141,282	2,200,202	15.6
Finance & insurance	139,727	935,968	6.7
Agriculture	120,658	982,294	8.1
NA	100,882	791,337	7.8
Entertainment & recreation	95,425	1,244,052	13.0

Nonemployers score big

Sole proprietorships, independent contractors, and self-employed individuals constitute an unknown amount of the US workforce. The conservative estimate, based on the Census Bureau’s Nonemployer Statistics series, put the number north of 25 million in 2017. However, these people have traditionally been excluded from stimulus measures since their numbers are difficult to track and their political power largely unorganized. The self-employed have traditionally been left out of the mix by US relief programs, and are usually not eligible to collect unemployment benefits. Nonemployers responded to the PPP inclusion by borrowing $15.5 billion across more than 1 million loans.

This level of uptake skews the distribution of loans heavily toward smaller amounts. But this distribution does not drop off exponentially, as the distribution of lenders does (Fig. 4). Rather, almost 177,000 loans fall between $20,800 and $20,833 — while only 3,000 approvals fall in the next $33 of loan amounts (Fig. 6). This phenomenon is sharpest for sole proprietors, contractors, and the self-employed, and reflects the maximum forgivable value of net profits. This measure is how nonemployers typically compensate owners, and is what owners of such businesses report as taxable income. Owner-operator businesses may also be approved for sums above $20,833 to cover costs such as rent and utility, so long as 60% is spent replacing owner compensation.

Due to the impact of nonemployer borrowers, many loans saved a single job. This distribution attenuates the PPP’s headline performance. Disentangling the nonemployer loans from loans to employers (such as restaurants) yields a more useful picture of the PPP. Focusing just on nonemployers exposes the tradeoff many owner-operators face between applying for the PPP and taking unemployment insurance, as individuals cannot seek both. For nonemployers with small overhead costs, that decision comes down to pre-pandemic income and state of residence. Unlike loans to businesses that face high fixed non-payroll costs, PPP loans to many nonemployers more closely resemble supports for individuals, which helped the personal savings rate touch 33% in April before dropping somewhat in May.

Figure 6.

Note that the Nonemployer Statistics series reports large numbers of S-Corporations and Partnerships as non-employers; the SBA data are unable to discriminate between nonemployers within business structures.

Overall, job retention appears low, but is consistent with loan approvals

Likely the most valuable study in terms of data quality to be released thus far is a preliminary paper published on July 22 by researchers from MIT, the Federal Reserve, and payroll software company ADP. Using several million pairs of employee-employer payroll data from ADP, authors evaluate employment levels just below and above the 500-employee general small business cutoff used by the SBA. They control for tempestuous business conditions brought by changing health regulations, as well as for variance among states. Like the above study conducted on the PPP’s first round, the authors find that the number of jobs saved is less than the number of loans made, a far cry from the 51 million jobs claimed on loan applications.

This finding would also run contrary to expectations that, at minimum, each loan saved at least one job. But that first study questioned whether the PPP had any positive employment effects at all. In contrast, this second paper estimates gains of 1.4 to 3.2 million jobs. The speed of the program was hampered in some cases by its terms, described above for restaurant owners, but applicable to many consumer-facing businesses. And, the ADP analysis cannot capture the gains of tiny firms, since it limits its analysis to 50-employee firms and larger. 96% of reporting borrowers listed fewer than 50 jobs retained. While a 60-employee business may choose to retain only 40 jobs, for instance, the ADP data are likely not fully representative of PPP borrowers.

What is clear is that nearly 4.9 million small businesses across the country (Fig. 9) received loans that they attested would help with the impacts of the coronavirus crisis. But the decoupling of loan amounts from job retention raises important questions about the efficiency of the Paycheck Protection Program with regards to protecting paychecks. Important to note is that the intent of the program has changed slightly. Legislation passed on June 3 altered the PPP’s loan forgiveness terms to include a larger portion of non-payroll costs, such as rent and utilities, crucial to basic operations.

In June, the Brookings Institution published an opinion along these lines, speculating that the PPP’s easing of loan terms — the very same that some inside the restaurant industry claimed were necessary — would help creditors of small businesses more than workers. However, these tensions, central to the oversight of the PPP as a crisis response measure, cannot be resolved before further research is conducted, and possibly more disclosures are made on the part of the SBA.

Figure 9.

On a per small-business basis, loans were well spread, though somewhat heterogeneous.