Transparency and Careful Scholarly Work Required to Estimate Giving

To the Editor:

Your article “Group Estimates Philanthropy Rose 13% in 2013, Clashing With ‘Giving USA’ “(January 17) looks at differences in approaches to estimating charitable giving.

Such conversations are healthy for the sector and for strengthening estimates so nonprofits get an ever-more-accurate picture of giving. “Giving USA” welcomes these discussions and offers insights to help inform the debate.

First, to be fair, accurately estimating charitable giving is difficult; it takes a lot of rigorous work and careful analysis. Producing accurate estimates requires commitment to the scientific method, including transparency, validation of results, testing and refining methodology, and peer review by objective scholars.

Developed by leading researchers in philanthropy and the nonprofit sector, including those at the Indiana University Lilly Family School of Philanthropy and other universities and research institutions, “Giving USA’s” estimates are scientifically rigorous.

“Giving USA” is transparent with its methodology and the data it uses to create charitable-giving estimates. Its peer-reviewed methods are evaluated and refined by top outside scholars of nonprofits, philanthropy, economics, and other fields as well as leading sector professionals, through both an advisory committee on methodology and an editorial review board. Its methods are published academically and in each issue of the report. Its transparency makes “Giving USA” replicable and verifiable.

Second, contrary to recent misconceptions, “Giving USA’s” estimates do not incorporate a survey, nor do they rely solely on IRS tax data on itemized deductions for charitable giving. Rather, they are based on complex econometric analyses, tabulations of tax data, economic indicators, demographics, and relationships between giving and these and other factors. Its scholars have tested more than 700 variables and model specifications, incorporating into the methodology only those scientifically proven to predict charitable giving.

While IRS data on charitable deductions are and should be a foundational piece in generating any estimate of charitable giving (each year two-thirds or more of all household charitable dollars come from itemizing households, and recently it has been four-fifths of household giving), it is only one component of the methodology of “Giving USA’s” estimates, which include individual, bequest, corporate, foundation, and total giving.

For example, in estimating giving by individuals, “Giving USA’s” econometric process incorporates the most recent IRS data available on household charitable deductions, along with U.S. household giving data from the Philanthropy Panel Study, contemporaneous economic variables, and other current data that our research has found to affect the amounts donated. This process generates estimates for giving by all American households—including those who itemize and those who do not.

Third, as the sector explores various approaches, it’s important to understand what data can and can’t tell us, and when. Ideally, there would be a source or a method that could provide reliable, verifiable estimates of monthly and state-level giving amounts more frequently than annually, but currently this is simply impossible.

No actual charitable giving data exist that can be used to derive monthly or state-by-state data on either total giving amounts or each of the sources of giving, making it impossible for researchers and nonprofit professionals to validate estimates of monthly or aggregate statewide giving amounts.

Fourth, like other responsible, reliable, and rigorous research organizations, to ensure accuracy “Giving USA” annually revises its data as updated data become available from the IRS, Bureau of Economic Analysis, and other sources. The models for “Giving USA” have proven very accurate over time. In fact, for the years 2004 to 2011, the mean difference between “Giving USA’s” original estimates and final estimates was 1.4% percent; the median difference was 1.1 percent.

Finally, we understand that fundraising and nonprofit practitioners are seeking keener insights into giving behaviors, trends, and data that will help inform and strengthen their work. That’s why we applaud the growth in research around giving and fundraising. For such research to be meaningful and useful, various models and methods should be shared, tested, and replicated widely. Those that are not made available for collaborative analysis and refinement represent a missed opportunity for the sector. The need for reliable data is too important for any group to claim that its methodology is a secret sauce.

Transparent, replicable, and verifiable methods let all of us in the philanthropic sector work together to develop approaches that result in the most accurate information possible. We welcome all ideas and invite our colleagues across the sector and the research community to join the conversation about the best way to enrich giving data.

Gregg Carlson
Chair
Giving USA Foundation

Patrick M. Rooney
Associate Dean for Research and Academic Affairs
Indiana University Lilly Family School of Philanthropy