Time is probably the trickiest problem relating to the IRS 990 data. To understand how mySidewalk reports on this data source’s time, it’s helpful to begin with how we collect this data from the IRS.

Once the data has been collected from the IRS, mySidewalk’s Data Team carefully combines each table into a single data set. For 990, 990-EZ, and 990-PF calendar year return extracts, a single tax year is collected. To collect a full sample of 990 submissions for a specific tax year, three calendar year tables are required. This is a requirement for two reasons:

Collecting a full sample of tax year data from calendar year extracts, published by the IRS, means that data summarized by geography is always an “apples-to-apples” comparison. Data for each time element in the mySidewalk data library is always a snapshot of the state of the nonprofit sector at a given point in time. In short, each time element refers to nonprofit organizations reporting on their activities for the same period of time. This makes it possible to do year-over-year comparisons, and to make time-series charts.

Typically, around 12% of 990 returns are submitted in a matching calendar year (e.g. 2023 tax year data submitted in the calendar year 2023). Between 80-85% of returns are submitted in the following year, and less than 10% are submitted two years later. Because of this pattern, reporting on a given tax year before three calendar years’ worth of data is published would mean that data is incomplete.

Practically speaking, publishing data in mySidewalk’s library before three calendar years have been released by the IRS would produce misleading charts. Time series charts, for example, would always show a ~85% reduction in critical data for the most recent year.

Other data vendors, like Guidestar, publish data differently than mySidewalk. These vendors’ value proposition is to provide a directory of nonprofit organizations in a given area. In this way, they function almost like a phone book for a specific economic sector. Their publicly-available data has several advantages and disadvantages.

These vendors often purport to offer the most recent data available. The source of their data is the exact same as ours (specifically the EO BMF). The main difference is that these vendors do not appear to distinguish their data by tax year. Instead, they provide a “rolling” list of most-recent nonprofit data. This approach has the advantage of capturing more recent entrants into the nonprofit sector, but forces “apples-to-oranges” comparisons. For example, the most recent data of Nonprofit A may be from 2023, but Nonprofit B may not have filed their 990 since 2021. These returns are treated as equivalent in their data. But economic conditions shift rapidly year-over-year, making these comparisons imprecise and introducing error. mySidewalk’s data avoids these accuracy and precision errors by providing data by tax year.

The various elements of the 990 data set are released throughout the year. By the first quarter of each year, the IRS should have fully released the previous calendar year’s data. Given this release schedule, mySidewalk anticipates releasing new tax year data in February of each year.

Each calendar year extract released by the IRS is two years behind the current year. For example, the 2023 calendar year data was released in 2025. Additionally, given the requirements detailed above, tax year data is incomplete until three calendar year extracts contain data relating to that tax year. That means that in 2025, mySidewalk released 2021 tax year data. This pattern will continue yearly into the future.