Step 4: Reviewed and Refined the PRI dataset

Once the initial dataset of all reported PRIs was constructed, our team began an intense process of reviewing and revising it. This included five sub-steps (A-E), which are described below.

A. Looked for Expected PRIs
First, we searched the dataset for PRIs that we expected to find based on our team’s prior knowledge of PRI activity by Minnesota private foundations. Most of these expected PRIs were in fact present in the dataset, but some were not. The most significant examples were six investments totaling over $24 million (unadjusted) made by The McKnight Foundation from 2009 to 2015 in the areas of affordable housing, economic development, and environmental preservation. In speaking with the foundation’s staff, we learned that while these investments were primarily made to advance its charitable purpose and had below-market financial terms, McKnight’s leadership, with the advice of counsel, chose not to include them in its qualifying distribution. As a result, the foundation did not officially classify these investments as PRIs to the IRS in Part IX-B of the Form 990-PF. Because of this, we decided together not to include these particular investments in the dataset for this report, but it is important for us to recognize here the historical leadership role that The McKnight Foundation has played with regard to charitable, below-market-rate investing both in Minnesota and beyond.

B. Reviewed Reported PRIs with Tax Experts
Originally, our team planned to keep every reported PRI in the dataset in order to avoid introducing subjectivity into the methodology. However, as we reviewed the initial dataset, it quickly became obvious that many PRIs were incorrectly reported and that including all of these incorrect PRIs would dramatically skew our findings. As a result, we engaged two tax experts at Minnesota-based CliftonLarsonAllen, the ninth largest accounting firm in the US, to help us revise the initial dataset. Together, we ended up removing 303 (35%) of the reported PRIs.

Review Decisions for Reported PRIs


We decided to omit reported PRIs for a variety of reasons. The most common was the misclassification of grant distributions as PRIs followed by the improper reporting of traditional endowment investments like normal savings accounts and market-rate mutual funds as PRIs. In our review, we always took into account context clues such as who was completing the form and the foundation’s pattern of reporting over multiple years.

C. Invited private foundations to review and comment on their data
After we reviewed all of the reported PRIs and made decisions about which to include in our dataset, we then reached out proactively to each of the 39 foundations that we had identified as having made at least one PRI during the research period. We shared with them the PRI data we had found in their public tax forms and invited them to give us feedback on that data. This process mostly confirmed our data and provided us with a couple of clarifications. It also gave us more information for fiscal year 2016 than was available at the time in GuideStar.

D. Completed the dataset
For each PRI included in the dataset, our team ultimately entered in over 90 columns of information. Much of this information was drawn from other fields of the Form 990-PF tax form in which the PRI was reported. We also did research on the reported recipients of each PRI. Additionally, we categorized each PRI by issue area, first using the taxonomy promulgated by Foundation Center and then adjusting those categories to best represent our usage statistics. (Endnote 10)

E.  Compared Dataset with Foundation Center PRI Database
Because the Foundation Center’s PRI Database has been the most commonly used resource for studies on past PRI activity, we wanted see how our dataset compared with it. To do so, on March 22, 2017, we purchased the rights to use Foundation Center’s PRI Database for this project. Holding constant the treatment of The McKnight Foundation’s investments mentioned above, the Foundation Center (FC) identified 191 PRIs greater than or equal to $10,000 made by 20 Minnesota private foundations totaling ~$74 million (unadjusted) for the period 1998-2014. This compared to 455 PRIs greater than or equal to $10,000 made by 34 Minnesota private foundations totaling $118 million (unadjusted) in our final dataset over the same period. Only three of the PRIs reported in the Foundation Center’s PRI Database were not also found by us in our research of the Form 990-PFs, and only one of the PRIs reported in the Foundation Center’s PRI Database was found during our research but then removed from our final dataset.

Comparison of Venn's PRI Dataset
to Foundation Center's PRI Dataset