Statement of the Problem:
The user is advised that changes in the data collection instrument between the 2001 and 2004 panels to the questions used to collect the data for the variables Business Income (TBMSUM1 and TBMSUM2, hereafter referred to by the general name TBMSUM#) and Profit/Loss from Business (TPRFTB1 and TPRFTB2, or TPRFTB#) likely affected the comparability of the data for these variables between the two panels. A detailed explanation follows.
(1) Differences between the 2001 and 2004 panels in the meaning and comparability of the TBMSUM# variables because of differences in the collection of data on profits from a business:
The 2004-panel collection instrument made an explicit effort to prompt each business owner (sole or joint) to report his or her share of the profits from the business. This was done primarily in question PRFTB (the “profit-share” question) that asked the person to report the share of the total profits from the business that he or she received in the reference period.1 To a much-lesser extent, it was also done in question OINCB, which was one of several questions (known collectively as the “business earnings” questions) that asked the person to report the amounts that he or she received monthly from the business as a salary, draw, or in some other way.2 In principle, there should be no overlap between the amount reported in PRFTB and the amounts reported in OINCB. Care was taken in both collecting and in processing the data for these items (see section 2 below) to make sure that no amount was included in both. In addition, the person was explicitly requested not to report profits from the business in any of the business-earning questions except OINCB. The profit-share question was the sole source of the data in the TPRFTB# variables in the 2004 panel; the business earnings questions were the sole source for the TBMSUM# variables in this panel.
What is crucial to note about TBMSUM# and TPRFTB# in the 2004 panel is: (1) each of them – though predominately TPRFTB# – may contain amounts that the person received from the profits of the business; (2) whatever profits are contained in TBMSUM# should represent monthly draws only; and (3) the two variables relate to different time periods, TBMSUM# being a monthly amount, and TPRFTB# relating to the entire time that the business was held in the 4-month reference period. These facts mean that to obtain the total income that the respondent received from the business in the reference period, it is necessary to sum over the monthly amounts in TBMSUM# and then to add this sum to TPRFTB#.3 To obtain the person’s total monthly income from the business for a particular month, the user must decide whether and how to add a portion of TPRFTB# to the TBMSUM# amount for that month.4 5
In the 2001 panel, the profit portion of a person’s total income was collected differently, with the effort being restricted entirely to the 2001 panel’s version of the business-earnings questions. The 2001-panel instrument did have a question about business profits, but it is not equivalent to the profit-share question on the instrument for the 2004 panel, even though it bears the same name, PRFTB (see section 2 below). Rather, the 2001 instrument intended that the person should report his or her share of the profits entirely in the business-earnings questions, which were the sole source of the data for the 2001 panel’s TBMSUM# variables. The extent to which respondents actually reported profits in this way in the 2001 panel is unknown. There was a concern, based on empirical evidence, that this approach was leading people to under-report their income by leaving out their share of business profits that were accumulated but not regularly drawn upon. This is why the 2001 PRFTB question was replaced by the profit-share question in the 2004 instrument, and why respondents were instructed to omit profits from the 2004 business-earnings questions other than OINCB.
These differences between the 2004 and 2001 panels in the way profits were collected means that the TBMSUM# variables in the 2004 panel are not necessarily comparable with those for the 2001 panel, for they may not (and likely do not) include profit amounts corresponding to those intended to be included in the 2001 variables.6 The effect of the 2004 approach was to separate profits (except those drawn upon monthly) from monthly earnings, which was contrary to the intent, if not the effect, of the 2001 approach. The sum of TBMSUM# and some portion of TPRFTB# for 2004 is in principle comparable with the amount in TBMSUM# for 2001, but how this theoretical comparability works out in practice is not known.(2) TPRFTB#:
As described above, in the 2001 panel, there was a question about profit/loss from the person’s business (this question was the source for the TPRFTB# data in 2001), but it applied to the total profits from the business itself and not necessarily (as did the profit-share question in the 2004 panel) to the person’s share of these profits. This means that the TPRFTB# variable in 2001 is not in principle comparable with the TPRFTB# variable in 2004.7 It also means that, for two reasons, it may not be correct when using the data for the 2001 panel to add the amount in TPRFTB# to the amount in TBMSUM# in an attempt to derive an individual’s total business income amount: (1) such profits may already be included in TBMSUM#, in which case their further addition would result in double counting; and (2) the person may not have received any share of them.
As noted, for the 2004 panel, the wording of the profit/loss question (PRFTB) was changed to ask for the person’s share of the profit/loss from the business rather than for the profit/loss of the business itself. To prevent double-counting, the 2004 instrument had a followup question (PFTBCHK) that asked whether the response to PRFTB included the amounts already reported in the business-earnings questions (such as SLRYB and OINCB) that collected the monthly amounts that the owner took from the business as a salary, draw, or in some other way. If so, any such duplicative amounts were subtracted from the amount in PRFTB during data processing.(3) THEARN , THTOTINC, TFEARN, TFTOTINC, TSFEARN, TSTOTINC, TPEARN, and TPTOTINC :
For the 2004 panel, the amounts in TPRFTB# were apportioned by month to the monthly amounts within each set of the following recodes: THEARN , THTOTINC, TFEARN, TFTOTINC, TSFEARN, TSTOTINC, TPEARN, and TPTOTINC. This process was only carried out, however, if the sum of the amounts in the TPRFTB# variables was positive. The rationale for this restriction was to prevent the creation of negative earning amounts, while avoiding undue increases in the complexity of the data processing. In previous panels (as explained above for the 2001 panel), some or all of the corresponding profit from a business may have been reported as business income (TBMSUM#), and would therefore have been included in the above recodes as part of their business-income component. The degree to which, in practice, the explicit addition of business profit to these recodes in 2004 affects their comparability with their data from previous panels is unknown.
As an additional note, there are other surveys (including CPS and ACS) that report negative net income from self-employment. With the 2004 panel data, SIPP users now have the option of creating net self-employment income with the possibility of negative values. They can do so by summing the amounts in the TPRFTB# and TBSUM# variables without restricting the sum of the TPRFTB# to positive amounts. If users choose this option, however, any summary income variables that include this alternative calculation will not match the Census Bureau’s calculated totals.
Introduction to SIPP |
Survey Content |
Technical Information |
Using & Linking Files |
SIPP Publications |
| Access SIPP Data | SIPP Users' Guide | Tutorial | User Notes/ListServe/News | SIPP Help |
Page Last Modified: 6/24/09