Recent changes at the National Center for Charitable Statistics (NCCS) in 2016 has delayed the release of the I990 Core PC data which is the main source for CVSuite’s nonprofit data. In order to avoid delays in CVSuite’s data update, we investigated other sources for nonprofit revenues data. We researched a new methodology that is equally comparable to the I990 Core PC data that uses the same data source NCCS uses to create the I990 Core PC file. Nonprofit revenues data originates from the Internal Revenue Service’s (IRS) financial files, a collection of I990 filings from all nonprofit businesses throughout the US. The CVSuite uses the IRS financial files as the foundation for the 2015 nonprofit data.


What you should know:

The nonprofit data in CVSuite for the year 2015 was collected using a different methodology than previous years. When comparing this new methodology with our past 2014 nonprofit data we have calculated a 0.16% difference in total United States “arts” revenue (an increase in $58 million). At the state level, on average a majority of states would see less than a 1% difference in 2014 nonprofit revenues using the new methodology.

Why would there be any difference at all if CVSuite’s new methodology uses the same source of data as the NCCS? We believe the difference in CVSuite’s methodology and NCCS is related to one or more factors: nonprofit organization’s NTEE codes are not correctly reported to the IRS, organization’s revenues are not correctly reported to the IRS, or changes made by NCCS are not reflected in the IRS data. We acknowledge that the NCCS applies additional processes to the data that are currently unknown and not reproducible by our research team.

The landscape of data collection and reporting changes often and we are confident in the methodology used to create the 2015 nonprofit data. The accuracy of 2014 nonprofit revenues at the national and state level using the new methodology is reasonable. Whether we continue using this methodology for future nonprofit years depends on the future of the National Center for Charitable Statistics and whether their schedule for releasing nonprofit data aligns with CVSuite’s data updates. In the meantime our research team will continue analyzing and improving this new nonprofit methodology.

When identifying percent change in nonprofit revenues from 2014 to 2015 in the CVSuite site, it is important to remember that our new nonprofit methodology is not the only factor that can cause a change in nonprofit revenues. There are many economic factors that can affect the amount of nonprofit revenues in a region. In 2015, the United States will see a 6.3% increase in nonprofit revenues from 2014. On average states will see a 4.3% increase in nonprofit revenues in 2015. We also recommend to analyze the 5 year trend for your region, in order to account for year to year fluctuations in nonprofit revenues. For example, in the past 5 years (2011-2015) the United States nonprofit revenues has increased 13.4% in nonprofit revenues, a total of $4.1 billion in sales. The trend shows that the arts nonprofit community is growing and generating more revenue every year.


Differences in CVSuite and NCCS Methodology

  1. NCCS uses a “black-box” methodology to define NTEE codes for organizations, which means their methodology to define NTEE codes is not reproducible due to unknown practices. We know some of their methods include using IRS data, 20 yrs of past record keeping, manual adjustment, and qualitative analysis of mission statements and activity codes.

    CVSuite continues to use the NCCS NTEE methodology by gathering ntee code classifications from past published nonprofit revenue data. We also retrieve NTEE codes from the IRS Business Master file and NCCS online portal. 
  2. NCCS also reviews and manually adjusts nonprofit’s revenue data when they find inconsistencies and/or inaccuracies. These adjustments can affect the year the tax filing was reported and also the amount of revenues reported.

    CVSuite does not review and make adjustments to the nonprofit revenue data. At this time we chose to report the exact data published by the IRS. 
  3. The CVSuite uses a geocoding service to find the county FIPS code. We do not know what NCCS uses for geocoding. 
  4. The NCCS adds additional data to their nonprofit Core PC files. Because CVSuite does not use this additional descriptive data, we did not pursue trying to create in our nonprofit data. 
  5. The NCCS has conflicting methodologies on whether or not they include i990pf data in their Core PC file. The i990pf data represents tax filings for nonprofit foundation organizations.

    The CVSuite chose to not use the i990pf data because 1. we are not interested in nonprofit foundation organizations and 2. the i990pf data is not collected in the same format as the i990 and i990ez forms. The variables we are interested in are either not available or not easy to calculate. 
  6. The NCCS supposedly uses the IRS Statistics of Income (SOI) files to verify and sometimes replace existing nonprofit data. Every record in the IRS Statistics of Income file are vetted by the IRS and are considered to be very accurate.The CVSuite has not developed a method for using the SOI files. The data is not in a usable format and requires extensive cleaning. We will continue to pursue this step of methodology at a later time.


The Methodology:


  1. Use IRS financials files for the current year i990 and i990ez. This is the main source of nonprofit revenue data. https://www.irs.gov/uac/soi-tax-stats-irs-tax-exempt-organization-population-data
  2. Filter the records by organizations who filed in the last 3 years. NCCS states that 3 years worth of data is good window for measuring nonprofit activity.
  3. Filter only 501c3 organizations. We only want to look at nonprofit charitable organizations. NCCS does this as well.
  4. Code each organization with a national taxonomy of exempt entity (NTEE) code that describes the type of activities the nonprofit serves. We use various sources to fill in descriptive organizational information missing from the IRS financial files. During this step we also add contact information for each record.
    1. 2014 nonprofit data from NCCS (2013 Core PC file)
    2. 2013 nonprofit data from NCCS (2012 Core PC file)
    3. The IRS business master file (BMF) – a nonprofit directory from the IRS
    4. NOTE: 4800 records are missing NTEE codes after this step. 500 of those do not have any contact information
  5. We add two additional years of nonprofit data to show 3 “complete” years of nonprofit activity. We add the 2013 and 2014 NCCS nonprofit data to the IRS financial data for 2015.
    1. 2014 nonprofit data from NCCS (2013 Core PC file)
    2. 2013 nonprofit data from NCCS (2012 Core PC file)
  6. Filter out duplicates records and keep the most recent filing. An organization usually files their taxes every year. We want to only keep their most recent revenues.
  7. Use NCCS nonprofit portal to gather the remaining 4800 records missing NTEE codes and 500 records missing contact information.
    1. http://nccsweb.urban.org/communityplatform/nccs/organization/profile/id/237255426/popup/1
  8. Filter the records to only arts nonprofit organizations using 44 NTEE codes defined as “creative”.
  9. Filter out-of-country records. The IRS Financial data contains records from Puerto Rico, Virgin Islands, Guam, American Samoa, and other countries.
  10. Geocode addresses to find county FIPS codes. We use the google api to convert addresses to lat/long, then we use the census geocoder api to convert lat/long to county FIPS code.
  11. Data cleaning
    1. Add leading zeros to EIN numbers. EIN numbers can lead with zeros and sometimes these are dropped in data processes.
    2. Add leading zeros to zip codes
    3. Check for missing values. Identify if our data has any gaps.
    4. Trim NTEE codes to only 3 digits. Sometimes NTEE codes will contain 4th and 5th digit classifiers. These are not necessary for selecting “arts” only nonprofits.


Was this article helpful to you?

Comments are closed.