Assets

Data collected: From the first survey year, NLSY79 respondents have been asked about their savings, home, and vehicle ownership.  Over the course of the survey, these questions provide information on when saving begins, how savings habits are formed, and how persistent savings habits are.

Each of the first four surveys (1979-1982) contain identical sets of questions asking if the respondent or their spouse had any money set aside for savings, owned a vehicle, or owned their own home.  Unfortunately, the respondent was never asked how much savings were held, the value or number of vehicles, or the value of, and mortgage on, their home.  Additionally, respondents were only asked questions on assets 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

This selection process eliminated many respondents from these questions.  Early NLSY79 data show that few individuals answered the questions until they turned 18 years old.  For example, in 1979 only five percent of those interviewed under age 18 answered the asset questions.  Except for the question on home ownership, asset questions were dropped during 1983 and 1984.  Beginning in 1985, when all respondents had turned 18, NLSY79 respondents were administered a much larger wealth section.  As Table 1 shows, respondents were given the opportunity to answer approximately 20 questions about a variety of asset and debt holdings.  In most years respondents estimated how much their home, cash savings, stock and bond portfolio, estate, business, and automobile were worth. Additionally, respondents estimated how much mortgage debt, property debt, and other debt they had accumulated.  Together these variables provide a rough overview of the net worth of each respondent.  As the cohort has aged, the wealth section has grown in length and detail.

The only major change in the wealth series occurred in 1991, 2002, and 2006 when budgetary restrictions resulted in the elimination of wealth questions for these rounds.

Table 1. NLSY79 Asset Questions 1985 to 2008¹

Nonresponse: Financial questions often elicit from respondents either a "refusal" or a "don't know" response.  From 1979 to 2000 the NLSY79 interviewers faced with a "refusal" or a "don't know" simply went on to the next question.  Starting in 2002 the NLSY79 began handling these cases of item non-response three different ways. Some respondents: 

1. are asked for a range of values
• The range of values questions gives respondents a method of providing a lower and upper bound answer.  Examining the data suggest that some respondents give quite precise bounds.  When a respondent gives a range, researchers should note that the upper and lower bound variables are not always in the correct order.  The survey always asks for the upper bound first and the lower bound second.  However, if a respondent gives the smaller number first, that number will be typed into the upper bound variable and the range values will be reversed.
2. were asked if they could provide an answer within $10,000
• Within $10,000 is designed to tell the respondent that a rough figure is an acceptable answer.  Extreme precision is not needed for these answers.
3. are asked simple above or below questions which bracketed the true value
• Bracketing lets respondents answer if the amount is above or below a particular value.  These initial values are called the variable's entry point.  The list of all entry points used by the NLSY79 survey in 2002 and 2004 is found in table 2 and for 2006 and 2008 in table 3. For example, the entry point for the person's vehicle is $20,000.  If the answer to the entry point question is neither a refusal nor a don't know then the respondent is asked one more "is the value above or below question."  This results in a series of four brackets that provide researchers with a rough idea the range the item's value falls into.  One caution for researchers is that while three of the brackets have a defined range, the top range is open-ended.

Example: An example of how item non-response to financial questions is handled is seen in the questions that ask about income from the military in the 2002 survey. Question Q13-16_TRUNC asks respondents "About how much total income did {spouse name}receive during 2001 from the military before taxes and other deductions?"  If the respondent is unable to provide a specific amount the survey in 2002 decides which of the three alternative methods will be used in question Q13-16_EXP.

In question Q13-16_E~000001 some interviewers probe for the answer by asking if they could provide "an approximate range for that amount?"  If the respondent can provide a range both the upper and lower number is captured. In question Q13-16_D some interviewers asked "To the nearest $10,000, can you tell me about how much your spouse or partner received during 2001, from the military before taxes and other deductions?"

Table 2. Entry Points for NLSY79 Income and Assets Section in 2002 and 2004

Table 3. Entry Points for NLSY79 Income and Assets Section in 2006 and 2008

Entry Amount Selection: Respondents were randomly assigned to get one of the three methods (unfolding bracket, nearest $10k, range) discussed above when they stated don't know or refused in an income question.  Additionally, among respondents who were given unfolding brackets, some respondents are given relatively high starting points (called in some of the codebook pages “the 60 Percent group”) or relatively low starting points (called “the 40 percent group”). In 2002 researchers interested in knowing which type of question a respondent was selected to receive should look at question HH_INC_3_EXP, which has the title "Type of question R assigned for household interview income recall experiment."
In later years, like 2008, there is no specific question available that marks if the respondent gets a low or high entry point.  Researchers can determine if a respondent is in the low or high group by looking at one of the entry symbols.  For example, the public assistance and welfare entry question has the Qname “symbol_entry_asst.”  Respondents who have a $5,000 value (see table 3) for symbol_entry_asst are part of the low entry amount group, while respondents who have a $7,500 value (see table 3) for this item are part of the high entry amount group.

Table 4. Number of Respondents Assigned to High and Low Entry Amounts

Top Coding: Because the NLSY79 is a public use data set that is 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 values.

The NLSY79 has used three top coding algorithms for assets.  From 1979 to 1988, every NLSY79 asset question that elicited a response above a specified cutoff value, such as $100,000 for some income variables, was recoded to the truncation value plus one dollar, such as $100,001.  Unfortunately this algorithm results in a sharp downward bias in the mean value of NLSY79 asset holdings since the entire right hand tail is truncated.  To address this problem, beginning in 1989, a new algorithm was implemented.  The new top code algorithm replaces all values above the cutoff with the average of all outlying values.

Beginning in 1996, the top two percent of respondents with valid values were identified.  Values within that top range were averaged and that averaged value replaced all values in the top range.

The extent of top coding for NLSY79 asset questions varies greatly.  For example, in 1993 there were only two individuals whose money assets exceeded the cut-off value of $500,000, while 581 individuals gave a market value for their residence above the cut-off value of $150,001.  While top-coding presents problems in analysis of individual observations and alters some statistical properties, the new algorithm does not affect the estimates of mean and median holdings.  Table 5 shows the number of people shielded by top codes in both 1985 and 1993.

Table 5. Number and Percentage of Respondents Whose Assets Were Top Coded in 1985 and 1993

Respondents Living Abroad: A second out-of-range issue with NLSY79 data concerns individuals living outside the United States.  Residing outside the United States 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 of the response could be determined.  While researchers are warned that this occurs, relatively few respondents live outside the United States and only a small number of individuals in this group cannot report their wealth in U.S. dollars.

Revisions to resolve asset and debt issues: In the spring of 2008 a revised set of asset and debt variables were released to the public.  These revised asset and debt variables fixed a number of problems with the NLSY79 data by eliminating some implausible outliers, generating uniform topcodes for all rounds, and constructing a total net worth variable.  The following provides details on revision process.

What Users See: Prior to the spring 2008 release users saw a single asset or debt question for each item in the wealth section of the questionnaire.  For example in 1987 the questionnaire asked each respondent who owned a home or apartment the market value of their residential property.  The questionnaire asked respondents "About how much do you think this property would sell for on today's market?"  Until the spring of 2008 the respondent answers were found in a single 1987 variable that had the following R and Q numbers:

R23627.00    [Q1947] (TRUNC)

After the revision was done, two more asset variables were added to the data set based on the same underlying property responses.  The two new variables are:

• R23627.01    [*Created] (TRUNC) (REVISED)
• R23627.02    [*Created] (TRUNC) (IMPUTED)

The variable that ends in (.00) R23627.00 is the original variable in the dataset and is left so that researchers can reproduce previous results.  The variables that ends in (.01) R23627.01 is a new variable which uses a revised topcoding algorithm.  By revising the variable researchers are now provided with some extra information that was not available before.  The variable that ends in (.02) R23627.02 is a new variable which imputes missing and unknown responses if possible as well as using the revised topcoding algorithm.

There are two new variables because some users will not want to use imputed data.  The (.01) variables are cleaned and re-topcoded but do not have any imputed values.  The (.02) variables have as many missing or unknown values imputed as possible.  In general the survey staff recommends that users without a strong preference should use the (.02) asset or debt value that ends in the label "(TRUNC) (IMPUTED)."

Comparison to Other NLS Cohorts

Information on assets has been regularly collected from each cohort (except for NLSY79 children under age 15).  Users should note, however, that the assets included have varied widely over time and among cohorts.  Data on the respondent's debts have been collected from each cohort on a less regular basis.  For more details, including specific years, consult the BLS website at www.bls.gov/nls or the appropriate cohort's User's Guide.

References

Zagorsky, Jay L.  "The NLSY79 Wealth Data Evaluation."  Columbus, OH:  CHRR, The Ohio State University, 1997.