The original design of the analysis of factors affecting the level of credit card debt for US families was outlined by Chien and Devaney (2001). It is further preserved by retaining the three main categories of the factors driving the debt: demographic, economic, and attitudinal. However, the original model is substantially altered to accommodate the new variables which appeared as the result of the continuous transformation of the survey.
The credit card debt is measured by the total value of credit card balances held by households in 2016 USD, denoted as a variable ccbal, which includes the total outstanding amount on all credit cards remaining after the last transaction. (FED Staff 2017).
6.1 Demographic Variables
The first demographic variable used in the analysis is age, which defines the age of head of the household.
To control for the level of education, educ variable is employed. This variable represents the highest achieved grade of the head of the household ranging from -1 (less than 1st grade) to 14 (doctorate or professional school degree) (FED Staff 2017, p. 728 - 729).
Dummy variable married defines whether the head of the household is either married or living with a partner (denoted as 1) or otherwise (denoted as 0).
The occupation of the head of the family is also included to the analysis. The SCF contains a categorical variable which classifies the occupation of the head of the family by four main categories: 1 – professional or managerial; 2 – technical, sales or services;
3 – production, craft, repair workers, operators, farmers, foresters or fishers; 4 – not employed. This categorical variable is further transformed into dummy prof which is denoted as 1 for the professional or managerial area, 0 otherwise, alike to Chien and
Devaney (2001). It is a best practice to avoid using categorical variables in the regression model by transforming them into a set of dummy variables. (Wooldridge 2012)
Another demographic factor is the ethnicity of the correspondent. The original variable in the SCF categorized heads of the families into 4 groups: white non-Hispanic, black or African American, Hispanic, and others. Similarly to Chien and Devaney (2001), we control for white non-Hispanic families by creating a dummy variable white which takes the value of 1 for families with white non-Hispanic head, 0 otherwise.
6.2 Economic Factors
The main variable highlighting the economic performance of families employed in the analysis is the total real income of the household for the last calendar year in 2016 USD.
A log of income is utilized to allow for relative comparisons, thus the variable utilized in the model is log_income.
Additionally, we control for real estate possession. Dummy variable housecl takes a value of 1 if families own a ranch, farm, mobile house, house, condo, etc., 0 otherwise. In this context, the value of 0 can mean that households are either renting the real estate or they are homeless.
6.3 Attitudinal Factors
The first factor which aims to explain families’ behavior and preferences indicates their attitude towards credit. Credit attitude information was obtained by asking correspondents the following question: “Do you think it is a good idea or a bad idea for people to buy things by borrowing or on credit?” (FED Staff 2017, p. 94) Based on their answer, correspondents were bucketed into three groups (1 = good idea; 2 = good in some way, bad in others; 3 = bad idea). This question produced three dummy variables.
Households who answered “good idea” are marked as a group with a favorable attitude towards credit by variable bshoprdl (1 if responded “good idea”, 0 otherwise). Variable bshopmodr takes the value of 1 if the answer was “good in some way, bad in others”
and this group is characterized by a neutral attitude towards credit. Variable bshopnone indicates the group with a negative attitude towards credit (1 if responded “bad idea”, 0 otherwise) Variable indicating negative attitude is left out from the regression model as a reference point to avoid multicollinearity. (Wooldridge 2012).
The main value-added to the original research is the integration of two new variables which highlight the financial literacy of the households. The SCF evaluates correspondent’s financial literacy by addressing three questions, originally formulated in the work of Mitchell and Lusardi (2007). These questions evaluate households’ financial literacy in these main fields: stocks, interest rate compounding, and inflation in the context of saving. (Bricker and et al. 2017) The exact questions are provided by Fed Staff (2017, p. 389):
1) “Do you think that the following statement is true or false: buying a single company's stock usually provides a safer return than a stock mutual fund?”
2) “Suppose you had $100 in a savings account and the interest rate was 2% per year. After 5 years, how much do you think you would have in the account if you left the money to grow: more than $102, exactly $102, or less than $102?”
And having answered at least one of two questions from above, correspondents were asked the last question:
3) Imagine that the interest rate on your savings account was 1% per year and inflation was 2% per year. After 1 year, would you be able to buy more than today, exactly the same as today, or less than today with the money in this account?”
Thus, we evaluate the correspondent’s objective financial literacy by the total score of correct answers from 0 to 3. Since this is also a categorical variable, it was transferred into a series of dummy variables in the following fashion: variable finlit0 takes the value of 1 if correspondent answered zero questions correctly, 0 otherwise; finlit1 is denoted as 1 if only one questions were responded correctly, 0 otherwise; and so on until variable finlit3.
The SCF provided another dimension to the financial literacy by including the question about the perceived or subjective financial knowledge of families. The questions go as follows (Fed Staff 2017, p. 89):
“Some people are very knowledgeable about personal finances, while others are less knowledgeable about personal finances. On a scale from zero to ten, where zero is not at all knowledgeable about personal finance and ten is very knowledgeable about personal finance, what number would you (and your {husband/wife/partner}) be on the scale?”
Thus, this categorical variable takes values in the range from 0 (not at all knowledgeable about personal finance) to 10 (very knowledgeable about personal finances). Similar to the previous modifications, this variable is transformed into a vector of dummy variables.
Dummy variable knowl10 takes value 1 for families who believe they are very knowledgeable about personal finances, 0 otherwise. Dummies to knowl0 are created in the same manner, which takes value 1 for families who are not at all knowledgeable about personal finance, 0 otherwise2.