Financial Knowledge and “Don't Know” Response

• Published date: 01 December 2019
• Author: Namhoon Kim,Travis P. Mountain
• Date: 01 December 2019
• DOI: http://doi.org/10.1111/joca.12275

1948 THE JOURNAL OF CONSUMER AFFAIRS

NAMHOON KIM AND TRAVIS P. MOUNTAIN

Financial Knowledge and “Don’t Know” Response

Weinvestigate the effect of group characteristics and educational inter-

ventions on young respondents’ objectivenancial knowledge level. We

examine six questions about personal nance and covariates selected

from the 2015 National Financial Capability Study. Because these

questions include “Don’t Know” or “Refuse” (DK/RF) responses, a

simple regression model could cause researchers to reach misleading

conclusions if DK/RF responses are not random. Thus, we suggest

a binomial-latent regression model to evaluate the effect of educa-

tional interventions and group differences that are hidden in DK/RF

responses. The estimation result shows that rejecting nancial edu-

cation opportunities is disadvantageous to obtaining proper nancial

knowledge. In addition, both formal and informal nancial education

are less effective in improving objectivenancial knowledge in our pre-

ferred model. Wealso nd few or no gender, income, and age differences

in young adults’ objectivenancial knowledge level after controlling for

nancial education interventions.

Many studies have argued that sufcient nancial knowledge improves

nancial decisions and behaviors, particularly in the case of young adults

(Hilgert, Hogarth, and Beverly 2003; Lusardi and Mitchelli 2007; Robb

and Sharpe 2009; Robb and Woodyard 2011). However, people in the

United States appear to lack satisfactory knowledge about personal nance

and are less prepared for proper nancial behaviors, such as appropriate

ways to use credit cards, manage debt, and engage in long-term nan-

cial planning, compared to those in other Organisation for Economic

Co-operation and Development (OECD) countries (Klapper, Lusardi, and

Van Oudheusden 2015; OECD 2016; OECD 2017).

One popular survey of nancial knowledge in the United States is the

state-by-state National Financial Capability Study (NFCS) conducted by

the Financial Industry Regulatory Authority (FINRA) Investor Education

Foundation (FINRA Investor Education Foundation 2016). Although the

NFCS has been used for many studies, there is a concern about the survey.

The factual questions that assess respondents’ nancial knowledge include

Namhoon Kim (hoonkim@krei.re.kr) is a Research Fellow at Korea Rural Economic Institute.

Travis P. Mountain (travis35@vt.edu) is an Assistant Professor and Extension Specialist in the

Department of Agricultural and Applied Economics at Virginia Tech. This research is provided in

part by the Virginia Agricultural Experiment Station and the Hatch Program of the National Institute

of Food and Agriculture, U.S. Department of Agriculture.

The Journal of Consumer Affairs, Winter 2019: 1948–1969

DOI: 10.1111/joca.12275

Copyright 2019 by The American Council on Consumer Interests

WINTER 2019 VOLUME 53, NUMBER 4 1949

“Don’t Know” or “Refuse” (DK/RF) responses along with substantive

responses. One of the methods for handling DK/RF responses is to regard

them as incorrect answers. However, this procedure likely neglects the

effect of DK/RF responses, which in turn can cause researchers to reach

misleading conclusions if the DK/RF responses are nonrandom choices.

The effect of including DK/RF responses in factual questions used to

determine a respondent’s knowledge level has been analyzed primarily in

political science (Delli Carpini and Keeter 1993; Mondak 2000; Mondak

2001; Mondak and Davis 2001). The main argument is that when a respon-

dent is not fully knowledgeable, s/he has two possible options: choosing a

DK/RF response or selecting a substantive answer based on his/her knowl-

edge level. Either choice decreases the validity of using the number of

correct answers as a dependent variable because it ignores the unobserved

difference between a substantive answer and a DK/RF response.

Although DK/RF responses would matter in evaluating the true effectof

an intervention, there is no econometric model that successfully addresses

the misleading results obtained from ignoring DK/RF responses. Thus, in

this study, we suggest a newregression model that prevents distortion from

DK/RF responses. In proposing our model, we focus on the unobserved

true number of correct answers rather than the observed true number of

correct answers. This identies each individual’s actual nancial knowl-

edge and estimates the true parameters hidden in DK/RF responses. The

second purpose of this study is to investigate group differences in, and the

effect of educational interventions on, the objective nancial knowledge

level. We estimate (1) gender, age, and income differences; (2) the effect

of rejecting nancial education; and (3) the effectiveness of formal and

informal nancial education. We also apply both a binomial regression

model and our preferred model and then compare the different conclusions

drawn from the two models.

In the following section, we review the literature related to the denition

of nancial knowledge, the appropriate measure of knowledge, and meth-

ods used to control DK/RF responses. In the subsequent two sections, we

describe the survey data and econometric models used in this study. In the

results and discussion sections, we present the results of the estimations and

discuss the advantages of our preferred model. The nal section provides

conclusions, limitations, and policy implications derived from this study.

LITERATURE REVIEW

Financial knowledge, literacy, and education have been dened

separately with respect to human capital. Financial knowledge is the stock