This blog post analyzes the 45 reported cases of allegedly fraudulent PPP transactions involving approval and disbursement of one or more loans larger than $150,000.[1] The 45 cases involved 38 lending institutions. Since these cases often involved more than one fraudulent loan application associated with different businesses, submitted through one or more lending institutions, they are associated with 115 originated loans larger than $150,000.[2]
The analysis shows that in the great majority of cases, the allegedly fraudulent loan was obtained from an internet bank or a fintech, or via an online transaction with a community bank that apparently had no prior relationship with the loan applicant or was non-local. Our analysis corroborates reports from earlier this month, from Bloomberg and from the Project on Government Oversight (POGO), which found that PPP loans being investigated by the Department of Justice (DOJ) disproportionately were handled by fintech companies.[3]
Like these other studies, we rely on DOJ news releases and criminal complaint affidavits cross-referenced with loan level data from the U.S. Small Business Administration (SBA). Among our findings:
- The 14 large banks on the SBA’s list of top 15 PPP lenders (which additionally includes one fintech-related lender) originated approximately 75,000 loans of at least $350,000 in size, of which just 10 are allegedly fraudulent.
- The percentage of allegedly fraudulent loans among the PPP loans of the major internet banks and fintech lenders is nearly 48 times the percentage among the PPP loans of these 14 large banks.
While our analysis focuses on quantifiable data, we also reviewed the DOJ reports and other web postings for background information. The information strongly suggests that most allegedly fraudulent loans at community and midsize banks were to borrowers with no prior relationship with the bank, including loans originated through fintech-related marketplace platforms.[4] Thus, another important takeaway from our analysis is the importance of a prior relationship with a prospective borrower for mitigating fraud risk in online, small business lending.
Of course, more cases of PPP fraud may yet be uncovered as the DOJ continues to review potentially duplicitous transactions. Depending upon the outcome of these reviews, not only might the overall incidence of fraudulent PPP borrowing be substantially greater than assessed to date, but the patterns highlighted in our analysis might change.
About Three out of Every Four Alleged Cases of PPP Fraud Had Some Fintech Involvement
The point of departure for our analysis was POGO’s list of 97 allegedly fraudulent, approved PPP loans, corresponding to 42 DOJ cases, along with the entities that received the loans and the originating lenders, as of September 30, 2020.[5] We validated POGO’s list by matching to the corresponding DOJ news release or criminal complaint affidavit and cross-referencing the name of the borrowing entity with the SBA’s loan-level data to confirm the identity of the lender.[6]
We then conducted a further search of the DOJ website and found additional loans satisfying our criteria, and some additional cases. Our final sample consists of 45 cases involving 115 loans and 38 lending institutions, with an average loan amount of just over $1 million.
Figures 1 through 3 summarize some characteristics of these cases. As indicated in Figure 1, close to a third of the cases involve three or more funded loans, and almost half involve two or more different lenders.
Figure 1: Distribution of Cases by Number of Loans Obtained and Number of Lenders Involved
As indicated in Figure 2, in about two-thirds of the cases, the total approved loan amounts were between $350,000 and $2,350,000. Only four cases had a total loan amount less than $350,000 (each of which involved a single loan).
Figure 2: Distribution of Cases by Total Approved Loan Amount
Figure 3 shows the distribution of cases characterized by type of lender or lenders involved. The Fintech or Internet Bank category consists of firms with substantial PPP lending volume that are readily identifiable as fintech or online lending specialists. Community Banks are depository institutions with less than $10 billion, and Midsize Banks those with at least $10 billion but less than $50 billion in total assets, that are not identifiable as fintech or online lending specialists. The Large Bank category is defined as depository institutions with at least $50 billion in total assets (which is mutually exclusive from the fintech or online lending category.)
About three fourths of the cases had some involvement of a fintech or internet bank, either as the sole lender or lenders (the majority of cases that involved a fintech or internet bank) or in combination with other types of lenders. The rest of the cases are about evenly divided between large banks (in one instance a large bank plus a credit union) and community or midsize banks or credit unions.
Figure 3: Distribution of Cases by Lender Type
The Frequency of Fraud at Large Banks Was Minimal Compared to Fintech Lenders
Next, we consider the incidence of individual loans associated with alleged fraud. For this analysis we limit attention to loans larger than $350,000, to mitigate a potential selection problem affecting loans of smaller size, since the data suggest that DOJ has, to date, mostly focused on potential fraud involving large loan amounts.[7] Findings are qualitatively similar for the full sample of allegedly fraudulent loans.
There are 81 loans larger than $350,000 associated with the alleged cases of PPP fraud. Table 1 shows the breakdown of these loans by lender category. Note that here we split the Community or Midsize Bank or Credit Union category into two groups: those that actively originated PPP loans via online, fintech-related marketplace platforms such as Lendio or Nymbus, and those that (as best we could determine) relied on traditional channels.
Lenders in the fintech or internet bank category originated 60 percent of the allegedly fraudulent loans. This category combined with the group of banks that actively participated in online marketplace platforms originated nearly three out of four of these loans.[8]
Notably, in cases of alleged PPP fraud involving a community or midsize banks with online marketplace participation, each fraudulent loan was to a business that was non-local or did not have a pre-existing account relationship with the bank.[9] However, even in the case of community or midsize banks that were not online marketplace participants, at least six of the 13 alleged fraudulent loans were to a business that was non-local or did not have a pre-existing account relationship with the bank. For example, in one particularly surprising instance, a resident of New Jersey received three loans, each approximately $3 million, from small banks located in Florida, Utah, and California.[10]
Table 1: Number of alleged fraudulent loans by lender type
Note: Restricted to loan sizes greater than $350,000
Table 2 shows the frequency of allegedly fraudulent loan transactions for the fintech and internet bank category of lenders; the 14 large banks from the SBA’s list of top 15 PPP lenders; and overall. The rate of allegedly fraudulent loans within the fintech and internet bank category is nearly 0.6 percent, 48 times larger than the rate for the large banks.
Table 2: Frequency of alleged fraudulent loans compared to total originations by lender type
Note: Restricted to loan sizes greater than $350,000
Concluding Thoughts
The PPP was an enormous undertaking that rapidly provided more than $525 billion to more than 5.2 million small businesses to help them maintain employees on their payroll and meet other financial obligations when their cash flows were disrupted by the Covid-19 pandemic. Based on information available to date, despite the rapidity with which applications were processed, the minimal documentation requirements and the reliance on borrowers’ self-certifications, fraudulent loan applications were exceedingly rare.
The fraud that did occur was much more pronounced at fintech firms and internet banks than at the nation’s largest banks, which had very few fraud cases. We also find that most allegedly fraudulent loans at community and midsize banks were to borrowers with no prior relationship with the bank, including loans originated through fintech-related marketplace platforms. This finding highlights the importance of a prior relationship with a prospective borrower for mitigating fraud risk in online, small business lending.
Again, an important caveat is that more cases of PPP fraud may yet be uncovered as the DOJ continues its reviews of suspect transactions. Depending upon the outcome of these reviews, not only might the overall incidence of fraudulent PPP borrowing be substantially greater than assessed to date, but the relative incidence across lender groupings may shift as well.
_____________________
[1]The identity of the lender generally is available only by matching the defendants’ names to SBA loan level data, which provide the lender name only for loans larger than $150,000. There appear to be only a couple of cases consisting solely of allegedly fraudulent, approved and funded loans less than or equal to $150,000 among the DOJ’s reported cases, and a similarly negligible number of loans of this size associated with the cases involving larger loans.
[2]Sometimes these businesses were real but the number of employees grossly exaggerated, and sometimes the businesses themselves were fictitious.
[3]See Michelle F. Davis, “PPP Scammers Made Fintech Companies their Lenders of Choice,” Bloomberg BusinessWeek(October 8, 2020), at https://www.bloomberg.com/news/articles/2020-10-07/ppp-loans-scammers-used-fintech-companies-to-carry-out-fraud, and Ryan Summers and Nick Schwellenbach, “Lamborghinis, Strip Clubs, Bogus Companies, and Lies,” POGO.org (October 8, 2020), at https://www.pogo.org/investigation/2020/10/lamborghinis-strip-clubs-bogus-companies-and-lies/
[4]We are unable to determine with certainty the channel by which a loan was originated or whether the borrower had a prior relationship with the lender, but we can assess the circumstances based on relevant factors such as geographic locations of the borrower and lender; the lenders’ reported reliance on online fintech platforms, and whether the loan was deposited at another bank.
[5]We much appreciate POGO’s online sharing of this information (which obviously reflected many hours of work); at https://docs.google.com/spreadsheets/d/1KprGhgHt23fJAsyUAw5csHorW_CVgtQCVgFLdGCYZqo/edit#gid=0
[6]We applied the following two filters to this list, which resulted in the exclusion of a few loans from the original POGO’s list. First, we required that the loan amount exceed $150,000, because the entity name is not provided in the SBA data for smaller loans and therefore it is not possible to identify the type of bank that disbursed such loan. Second, we required that the loan amounts associated with an individual lender in an alleged fraud case were reported to be at least partially tapped by the borrower. As such, loans that are immediately placed on hold or reversed are considered to be equivalent to denied loans.
[7]The alleged cases of fraud disproportionately consist of large loans: in the $150 – 350,000 size range, there are 34 allegedly fraudulent loans among a total population of 662,516, which compares to 81 out of a total population of 281,779 in the larger than $350,000 range. Moreover, allegedly fraudulent loans smaller than $350,000 generally were associated with cases involving both large and small loans. These facts suggest that individual small loans have not (yet) drawn the attention of investigators.
[8]Of course, we cannot determine whether a specific loan was originated via the online marketplace channel.
[9]Information in the DOJ reports indicated that the business was non-local or that the PPP funds were deposited at another bank or at a new account opened to receive the funds.
[10]The loan applications, that were submitted on behalf of three different business entities, allegedly fabricated employees, manipulated bank records, and falsified a driver’s license.
