PJ5. SBA Office of Inspector General, Report 21-07, January 14, 2021, Finding 2 page 10 (PDF page 14). Exact selected passages:
"We determined that SBA’s publicly reported and loan-level PPP data was inaccurate and incomplete. Job statistics were inaccurate and incomplete, industry classification codes were incomplete, and underserved market data was incomplete. Without accurate and complete data, SBA cannot reliably and accurately inform SBA management and Congress about program effectiveness and measures needed to inform program decisions."
"We found that SBA’s loan-level data for job statistics was inaccurate and incomplete. In our review of data extracted from SBA’s mainframe loan accounting system as of June 30, 2020, we found that"
"191,003 loans totaling approximately $11 billion did not include employment information in the required job field for the number of current employees"
"148,019 loans totaling approximately $6.7 billion had no employment information for the number of current employees and the number of jobs retained"
"845,778 loans did not contain data in the jobs retained field"
"SBA’s system has only one required job field, “number of current employees,” which is also on the PPP loan application. However, a “jobs retained” field, which SBA used for public reporting on PPP loans, was not included on the borrower application for PPP loans. SBA officials told us that only fields on the borrower application were mandatory data entry fields to obtain a PPP loan number."
"SBA did not require borrowers to provide data on jobs retained and did not require lenders to input data for jobs retained at the time of loan application. As a result, SBA officials and national leaders do not have enough information to make informed decisions or determine to what extent the PPP met national program objectives. Additionally, SBA cannot accurately report jobs retained by PPP borrowers, which was a statistic used in public reporting on PPP loans."
The three missing-data sets above overlap and cannot simply be added. These are OIG findings about data quality, not fraud convictions or a count of actual jobs lost.

PJ6. NBER Working Paper 29972 issued April 2022; authors David Autor and coauthors. Selected abstract excerpts (minimal quotation for commentary):
"We use high-frequency administrative payroll data from ADP--one of the world’s largest payroll processing firms--to estimate the causal effect of the PPP on the evolution of employment at PPP-eligible firms relative to PPP-ineligible firms, where eligibility is determined by industry-specific firm-size cutoffs."
"Employers retained an estimated additional 3.6 million jobs due to the PPP as of mid-May 2020, and 1.4 million jobs at the end of 2020."
These are research estimates, not directly observed saved jobs, and the dates differ from the loan-data snapshot.
