Starting with the 2019 SIPP, six variables will use new disclosure methods to better align the SIPP’s published poverty rates and internal poverty rates—rates that are calculated without disclosure methods applied. Since public data users can only observe published poverty rates, this user note aims for transparency in how disclosure avoidance methods affected the quality of published poverty data prior to the 2019 survey year and how the quality of that data changed beginning with the 2019 survey year. While most public users associate disclosure avoidance with topcoding that affects individuals in the right tail of the income distribution, selected variables are censored to impact individuals potentially in the left tail of the income distribution (a.k.a. “bottomcoded”). The six variables that changed disclosure avoidance methods with the 2019 data are business profits (TJB(n)_PRFTB), hours worked (TJB(n)_JOBHRS(i)), net income from rental properties (TJSRPNETINC, TJORPNETINC, and TORPNETINC), and income from other financial investments (TOINVINC). 

For business profits (TJB(n)_PRFTB), net income from rental properties (TJSRPNETINC, TJORPNETINC, TORPNETINC), and income from other financial investments (TOINVINC), a new bottomcode procedure replaces all values at and below the bottomcode threshold. In prior panels, the bottomcoded values were replaced with the mean of all values at and below the bottomcode threshold. For example, if a respondent reported -$10,000 of net income from rental properties and -$10,000 was below the bottomcode threshold, the value output to the public-use data could be more negative, (e.g., -$20,000). The replacement of reported values with the mean below the bottomcode threshold pulled a small number of respondents into poverty by reducing their total monthly earnings and income. Replacing large negative values with the bottomcode threshold is intended to reduce the number of such cases. 

Users wishing to implement prior panels’ bottomcode procedures to consistently estimate means for business profits, net income from rental properties, and income from other financial investments over time can use the variables TPRFTB_BMEAN, TJSNET_BMEAN, TJONET_BMEAN, TONET_BMEAN, and TOINV_BMEAN. See the separate user notes for “2019 SIPP: Changes to Business Profits Bottomcoding” and “2019 SIPP: Changes to the Bottomcoding of Negative Asset Income” for a more detailed discussion of the changes and alternative bottomcodes.

In the case of hours worked (TJB(n)_JOBHRS(i), the topcode procedure has been changed to use the mean at and above the topcode rather than the topcode threshold itself. Workers who earned an hourly wage were occasionally pulled into poverty by this topcode methodology. The workers that were not in poverty in internal calculations but in poverty in public-use calculations often were paid relatively low hourly wages but worked many hours per week. For example, in the public-use data prior to the change in topcoding, a worker earning $10 per hour and working 80 hours per week could have their hours worked per week topcoded as 60 hours, implying their income in the public-use data would fall by $200 per week. Users that want to learn more about the change to the hours topcode methodology can learn more in the “2019 SIPP: Changes to the Hours Worked Topcode” user note. 

It is worth noting that the topcode procedures for TJB(n)_PRFTB, TJSRPNETINC, TJORPNETINC, TORPNETINC, and TOINVINC are unchanged — values at and above the topcode threshold are replaced with the mean of those values above the threshold. See the “2019 SIPP: Change to Business Profits Topcoding” user note to learn more about business profit variables.