4.22 Income

This section describes the income questions asked to NLSY79 respondents since the cohort’s inception. The NLSY79 cohort is a unique source of income information. Because the original NLSY79 panel contained a supplemental sample of 5,295 economically disadvantaged, non-black youths, researchers are able to precisely measure income of low-income and minority households. Moreover, because in-depth income questions have been asked since 1979, detailed age/income profiles can be traced over time.

Most NLSY79 income questions refer to the previous calendar year. For example, if the survey is being fielded in 1992, most questions ask the respondent to report how much they earned during the 1991 calendar year. A summary of the questions asked in the early surveys is shown in Tables 4.22.1 and 4.22.2. During each of the first four surveys (1979–82) NLSY79 respondents were examined to see if they met one of following five criteria:

• 18 years old or greater
• Had a child
• Enrolled in college
• Married
• Living outside their parents' home

If they did not meet any of these five criteria, respondents were asked the simple income section outlined in Table 4.22.1. However, if respondents fell into one of these five categories, the interviewer asked a more detailed set of questions outlined in Table 4.22.2.

Table 4.22.1 Short Form NLSY79 Income Questions: 1979-82

Beginning in 1983, the NLSY79 questionnaire used the same set of income questions for all respondents, since all respondents would have been at least 18 years old and thus sent through the long series of questions. In this set of questions, all respondents are asked about income from a variety of sources, as shown in Table 4.22.3. First, they report how much money they received from working; questions are asked about their military income, wages, salaries, tips, farm income, and business income. Then respondents provide information about transfers from the government through programs such as unemployment compensation, AFDC payments, Food Stamps, SSI, and other welfare payments.

Table 4.22.2 Long Form NLSY79 Income Questions: 1979-82

Respondents are then queried about transfers from non-government sources such as child support, alimony, and parental payments. Finally, respondents report income from other sources such as scholarships, V.A. benefits, interest, dividends, and rent.

Nonresponse: One major concern when asking individuals about their income holdings is nonresponse bias. While it is outside the scope of this section to fully investigate nonresponse bias in the NLSY79, this section briefly describes nonresponse in 1992 as an example. Researchers interested in a fuller discussion of nonresponse should consult Chapter 5 of this guide. There are two primary types of questions on income: general questions asking whether the respondent received income from a particular source and specific questions on the amount of income. Factors that are likely to contribute to nonresponse are suspicion, uncertainty, shared responsibility for family finances, and complex financial arrangements.

Table 4.22.3 Detailed NLSY79 Income Questions 1983-2004

Table 4.22.4 provides information on response rates to income questions in 1992. The table is divided into three sections. The first section shows the response rates for questions asked about the respondent. The second and third sections show the response rates for the spouse and partner, respectively. The average response rates (99.9 percent) in the receipt column show that almost every NLSY79 respondent will tell the interviewer if they received income from a particular source. Additionally, the amount column, which is calculated based only on individuals who received a particular type of income, also shows high response rates. Except for alimony payments (54 percent), more than 90 percent of all NLSY79 respondents knew and were willing to divulge how much they earned from various sources.

Examining the sections labeled “Spouse’s Income” and “Partner’s Income” shows a very different picture. Spouse’s income is known with less certainty and partner’s income with much less certainty than the respondent’s income. For example, 91.9 percent of the respondents reported how much they earned from their own business or farm. However, these same individuals were able to report only 85.6 percent of the earnings of spouses and only 59.7 percent of their partner’s income from the same source.

Top Coding: Because the NLSY79 is a public use data set distributed widely throughout the research and public policy communities, the survey takes extensive measures to protect the confidentiality of respondents. One method of ensuring confidentiality is to “top code” unusually high income values. The NLSY79 top code values were originally designed to prevent identification of the top two percent of respondents.

Table 4.22.4 Response Rates to Questions on Income: NLSY79 1992

The NLSY79 has used four top coding algorithms for income. (1) From 1979 to 1984, every NLSY79 income question that elicited a response above $75,000 was truncated to $75,001. (2) From 1985 to 1988, the values were increased to $100,000 and $100,001 respectively. Unfortunately, this algorithm results in a sharp downward bias in the mean value of NLSY79 income holdings since the entire right hand tail is truncated. (3) To fix this problem, a new algorithm was introduced beginning in 1989. The new top code algorithm replaced all values above the cutoff with the average of all outlying values. (4) Beginning in 1996, another new algorithm was used. This algorithm takes the top two percent of respondents with valid values and averages them. That averaged value replaces the values for all cases in the top range. Top coding primarily affects seven of the NLSY79 income variables. The seven variables that are top coded include the income from respondent’s wages, respondent’s business, spouse’s wages, spouse’s business, partner’s wages, rest of the family, and other sources such as rents, interest, and dividends.

A second issue with NLSY79 data concerns individuals living outside the U.S. Living outside the U.S. does not preclude a respondent from being interviewed. For example, in 1992, 125 respondents lived abroad. Between 1989 and 1992, for people who hold assets denominated in foreign currency, little effort was made to transform these assets into dollar figures. Instead, such values are classified as “invalid skips” in the data. Beginning in 1993, an effort was made to convert these currencies whenever the unit in which the response was made could be determined. While researchers are warned that this occurs, relatively few individuals live outside the U.S.

Created Variables and Summary Statistics: Each year, the CHRR staff creates an NLSY79 variable entitled "Total Net Family Income." This variable is designed to provide researchers with a summary variable of all the income received in the household. The actual computer code used to create this variable is found in Appendix 2 of the NLSY79 Codebook Supplement. Appendix 2 shows that each year a slightly different program is used to compute net family income. However, these programs all have a similar structure. For example, the 1992 net family income variable is created by:

• Initializing 19 different income categories to valid skip (-4).
• Examining various demographic variables such as military status to see which of the 19 categories can be automatically classified as zero.
• Counting the number of months a respondent or partner participated in a government aid program, such as AFDC or Food Stamps. Multiplying the number of months of participation by the average weekly/monthly benefit to create total yearly income from that source.
• Summing all 19 components to form total net family income. If any component is MISSING, REFUSED, or UNKNOWN, then set total net family income to MISSING, REFUSED, or UNKNOWN. Hence, if the respondent knows how much they received from all sources except for one, the net family income variable will be coded as missing.
• No items are subtracted from the 'Net Family Income' variables to account for taxes or other adjustments.

Researchers should note that in the survey years from 1979 to 1986, total net family income was created a little differently than from 1987 to present. In the early years when many of the NLSY79 respondents were younger and living in the parental household, the parent was given the Household Interview (Version A). These interviews obtained income from all household members related by blood or marriage. If Version A was used, then the total net family income program picked up income from the Household Interview, and the component income variables from the "Income" section of the questionnaire were ignored. Usually, if the parent completed the Household Interview, then the youth respondent went through a limited set of income questions that would not allow for the creation of total net family income. If Versions B or C of the Household Interview were given, then the respondent went through the "Income" section and the program picked up the component income variables. Beginning in 1987, the three versions of the Household Interview were reduced to one (Version C) and all respondents go through the income section regardless of whether or not they are living in the parental household.

Survey Instruments: The income variables are found in the following sections of the questionnaires: Section 11 (1991, 1993), Section 12 (1981, 1987, 1989, 1990, 1992), Section 13 (1983, 1986, 1994–2002), Section 14 (1982, 1985, 2004), Section 15 (1984, 1988), Section 17 (1980), and Section 21 (1979).

Data Files: Interested users should examine the "Income" area of interest for these variables.

Comparison to Other NLS Cohorts: A small amount of income information is gathered for the NLSY79 children. More detailed income information is gathered from young adults and is patterned after the main NLSY79 section. Information on income is regularly collected from the other cohorts. Users should note, however, that the income sources included have varied widely over time and among cohorts. For more information, refer to the BLS website at http://www.bls.gov/nls or the appropriate cohort’s User’s Guide.