Financial literacy, financial advice, and financial behavior

The term financial literacy was introduced in the U.S. by the Jump$tart Coalition for Personal Financial Literacy in 1997, defining the concept as “the ability to use knowledge and skills to manage one’s financial resources effectively for lifetime financial security”. Later, also in the U.S., this characterization was adopted in a universal definition provided by the President’s Advisory Council on Financial Literacy (PACFL 2008). However, Hung et al. (2009), in their review of competing financial literacy concepts, find that the literature has proposed several definitions and lacks a universally accepted notion of what financial literacy really means. They document a large variety of conceptual definitions and show that each of them stresses different dimensions of financial literacy, i.e. actual and perceived knowledge of financial matters as well as the ability to apply that knowledge, but also individual financial experience and even sound financial behavior. In another extensive review of financial literacy operationalizations, Huston (2010) surveys 71 studies using 52 different data sets and corroborates that there is no such thing as a standardized conceptualization of financial literacy. 72% of studies did not even include an explicit definition. Moreover, there was no universally accepted meaning of financial literacy among those studies which did propose financial literacy definitions. Finally, the terms financial literacy and financial knowledge were used interchangeably by almost half of all studies under review. Like other standardized concepts of literacy such as computer literacy or health literacy, however, Huston (2010) stresses that “financial literacy should be conceptualized as having two dimensions‐understanding (personal finance knowledge) and use (personal finance application)” (p. 306). A similar understanding is given in Hung et al. (2009), who consolidate the various definitions they review and propose an overarching conceptualization specifying financial literacy as the “knowledge of basic economic and financial concepts, as well as the ability to use that knowledge and other financial skills to manage financial resources effectively for a lifetime of financial well-being” (p. 12). A recent definition employed in the 2012 Program for International Student Assessment (PISA) has been provided by the Organization for Economic Cooperation and Development (OECD 2014) and includes both the knowledge and the application domain: “Financial literacy is knowledge and understanding of financial concepts and risks, and the skills, motivation and confidence to apply such knowledge and understanding in order to make effective decisions across a range of financial contexts, to improve the financial well-being of individuals and society, and to enable participation in economic life” (p. 33).

There is an ongoing debate as to how financial literacy is distinct from related concepts like numeracy and cognitive abilities. Hastings et al. (2013) document that respondents with higher cognitive abilities and more comfortable with numerical calculations on average exhibit higher levels of financial literacy. They also review a number of studies which find a positive relationship between cognitive abilities and numeracy on the one hand and sound financial behavior on the other hand (e.g., Banks and Oldfield 2007; Grinblatt et al. 2009; Christelis et al. 2010). Thus, Hung et al. (2009) argue that, for designing effective programs to improve financial literacy, it is important to differentiate general cognitive abilities from core aspects of financial literacy.⁵ Lusardi et al. (2010) address this point of criticism by analyzing both a measure of financial literacy and a proxy for cognitive ability obtained from the National Longitudinal Survey of Youth (NLSY). On the one hand, the authors confirm a positive correlation between financial literacy and cognitive ability. However, they also show that cognitive factors cannot account for the entire variation in measured financial literacy levels, thereby leaving room for other dimensions of financial literacy.

Hung et al. (2009) show that the various test-based measures they review are generally highly correlated with each other and when the questions are worded identically, answers feature high test–retest reliability across different survey waves. Thus, in the following, we will focus on three specific test questions introduced by Lusardi and Mitchell (2008) in a special module of the 2004 Health and Retirement Study (HRS).⁶ These questions have been widely adopted in the U.S. and elsewhere and have become known as the Big Three.⁷ The first one of this parsimonious set of questions addresses individuals’ numeracy and their ability to do simple calculations and is worded as follows⁸:

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; less than $102; do not know; refuse to answer.]

The second question refers to inflation and money illusion:

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, exactly the same as, or less than today with the money in this account; do not know; refuse to answer.]

Finally, the third question tests if respondents are familiar with the concept of risk diversification:

Do you think that the following statement is true or false? ‘Buying a single company stock usually provides a safer return than a stock mutual fund.’ [true; false ; do not know; refuse to answer.]

Although the Big Three generally do not demand advanced financial knowledge, only 34% of respondents in the original survey were able to answer all three questions correctly (Lusardi and Mitchell 2014). Straightforwardly, individuals who fail to correctly answer the first two questions will likely experience difficulties when facing even basic financial decisions characterized by an investment today and returns in the future. Providing the correct answer to the third question requires some knowledge about stocks and stock mutual funds as well as about the concept of risk diversification and thus indicates if respondents are able to effectively manage their financial assets.

In subsequent analyses, surveys were extended by additional questions beyond the Big Three so as to capture other dimensions of financial literacy. Specifically, Lusardi and Mitchell added two more items measuring knowledge in asset pricing and mortgages to the 2009 National Financial Capability Study (NFCS). In a recent large scale survey of financial literacy levels in more than 140 national economies, Klapper et al. (2015) elicit four quiz questions closely related to the Big Three. Thereby, the survey addresses four fundamental concepts for financial decision-making: risk diversification⁹ (related to the third question of the Big Three), inflation¹⁰ (very similar to the second question of the Big Three), numeracy¹¹ (new question, previously not included in the Big Three) and compound interest¹² (an advancement of the first question of the Big Three as this question actually refers to the concept of interest compounding).

Recently, the OECD has adopted an even broader perspective on the measurement of financial literacy for the 2012 PISA assessment (OECD 2014). An expert group consisting of regulators, academics and practitioners from different countries designed questions in three dimensions: knowledge and understanding (content), approaches and mental strategies (processes) and financial situations (contexts), reflecting real-life situations of 15-year-old students. The assessment consists of 40 finance-related questions as well as questions in the areas of mathematics and reading abilities. Students were asked to analyze simple graphs, calculate interest rates or evaluate payment checks and invoices. Thus, the financial literacy measure designed for the PISA assessment differs from the Big Three along two dimensions. First, the questions asked cover a much larger array of financial issues and place emphasis on financial decisions faced by 15-year-old students. Second, due to a more extensive battery of questions, financial literacy levels elicited in the PISA assessment capture more nuances of knowledge and abilities related to personal finance matters.

Although test-based approaches towards measuring financial capabilities—most prominently the Big Three—have now become the international benchmark for the assessment of financial literacy, there is little evidence on whether this set of questions is indeed a superior approach to capturing financial literacy. Hastings et al. (2013) emphasize that it is generally unclear if questions are a suitable means for measuring financial capability and, if so, which questions lend themselves most effectively for identifying it. Specifically, the authors criticize that surveys eliciting financial literacy levels do not incentivize respondents to provide carefully considered answers reflecting their actual knowledge. Besides, study designs usually do not permit participants to tap into other sources of information in order to prepare their decisions. Yet, accessing the internet, talking to friends and family, or consulting with a financial advisor¹³ are important channels of expertise used by many consumers to compensate for their individual lack of financial literacy when making real-life financial decisions. Ignoring them likely biases the observed impact of financial literacy on financial behavior.

Another shortcoming of test-based measures of financial literacy is their sensitivity to framing. Specifically, Lusardi and Mitchell (2011a, b) and van Rooij et al. (2011b) document that the answers of survey participants differ significantly based on the wording of the test questions. In fact, the percentage of correct answers doubled in the latter study when the wording for the third question of the Big Three was “buying company stock usually provides a safer return than a stock mutual fund” as compared to phrasing the question reversely, i.e. “buying a stock mutual fund usually provides a safer return than a company stock”. Hence, Lusardi and Mitchell (2014) conclude that some answers classified as “correct” might instead reflect simple guessing of respondents and highlight that measurement error might be an issue when eliciting financial ability based on test questions.¹⁴

Finally, Meyer et al. (2015) document that the quality of data obtained from household surveys has declined in recent years. This development owes to increasing non-participation of households (unit nonresponse), a tendency of not answering certain questions (item nonresponse), and, finally, increased measurement error due to greater inaccuracy of the participating households when answering the survey questions. Although this trend is not limited to items related to financial literacy, it constitutes an additional issue regarding the reliability of recent surveys on financial literacy.

Absent a survey-based measure of financial literacy, a number of studies have turned to different proxies for subjects’ financial literacy levels. Given the robust correlations between several socio-demographic characteristics and measured financial literacy (see Sect. 4.1), several contributions lacking an observable measure of financial capabilities exploit this evidence and use respondents’ demographics in order to capture their financial literacy. Corresponding proxies for financial sophistication used in the literature include (disposable) income and wealth (Dhar and Zhu 2006; Vissing-Jorgensen 2003; Calvet et al. 2007, 2009) as well as age (Calvet et al. 2007, 2009; Georgarakos and Pasini 2011), educational attainment (Christiansen et al. 2008; Calvet et al. 2007, 2009), professional status (Calvet et al. 2009), and even IQ (Grinblatt et al. 2011, 2012). Likewise, both Chalmers and Reuter (2012) and Hackethal et al. (2012) use subsets of these demographics to proxy for financial literacy in their analyses.

A methodologically advanced approach to infer a demographics-based proxy of financial literacy recently applied in Stolper (2016) combines individuals’ demographic characteristics as well as directly measured levels of financial literacy by drawing on two different datasets each of which contains an identical set of demographics for a given cohort. The difference between the two datasets, however, is that only one of them contains the explanatory variable of interest, i.e. a direct measure of financial literacy, while the other features the financial outcome of interest but lacks a financial literacy measure. To overcome this data limitation, the author resorts to the imputation method proposed by Browning and Leth-Petersen (2003) and designed to link datasets featuring the above-mentioned properties. Following Graham et al. (2009) and Korniotis and Kumar (2013), who apply this approach to infer individuals’ perceived competence and smartness, respectively, the author proceeds in two steps to obtain a demographics-based financial literacy variable. First, he estimates an empirical model of financial literacy using the first wave of the PHF survey which contains both a direct measure of financial literacy and several of the demographic characteristics that have been shown to explain a significant proportion of the cross-sectional variation in people’s financial literacy levels (see Sect. 4.1). In a second step, the author then employs this model to predict the financial literacy levels of the households sampled in the primary dataset which only contains the respective demographics. Specifically, he uses the coefficient estimates obtained from the PHF-based model and plugs in the relevant demographic characteristics available in the primary dataset in order to construct a variable capturing predicted financial literacy.

Given the drawbacks of test-based measures, some have proposed alternative approaches towards capturing financial literacy. Instead of surveying households, a natural way to infer financial literacy levels from other available information is to look at individuals’ readily observable financial behavior and use the quality of their financial decisions as a proxy for their financial literacy. Such outcomes-based proxies for financial literacy comprise observed risk diversification in equity portfolios (Goetzmann and Kumar 2008; Grinblatt and Keloharju 2001), prior investment experience (Goetzmann and Kumar 2008; Nicolosi et al. 2009; Seru et al. 2010), and the propensity to invest in complex financial instruments (Genesove and Mayer 2001; Goetzmann and Kumar 2008). Following this idea, Calvet et al. (2009) draw on the security accounts of a large-scale panel comprising 4.8 million Swedish households and construct an index of financial sophistication based on the sampled households’ (observable) ability to avoid poor financial decisions such as holding underdiversified portfolios, displaying inertia in risk taking, and exhibiting a disposition effect. Clearly, such an outcomes-based strategy crucially hinges on increased data availability. Nevertheless, as Hastings et al. (2013) point out, using consumers’ actual financial behavior as an indicator may be a more effective approach to predict future outcomes than the existing standard which, as described above, relies on more general proxies of financial literacy such as the Big Three.

A robust finding across countries is that financial literacy levels are lowest among the young and the old (e.g., Lusardi and Mitchell 2011a, c). Thus, we generally observe a hump-shaped distribution of financial literacy with respect to age. Low literacy among the young might be problematic since this group faces financial decisions that influence their (financial) well-being for decades to come. This is one reason why the OECD included a battery of financial literacy questions in the 2012 PISA assessment for 15-year old students as increasing financial literacy for this group seems to be particularly promising.²⁷ Low levels of financial literacy among the old is also problematic as individuals aged 60 and older hold about 50% of the wealth in the U.S. (Finke et al. 2016). With respect to cognitive changes associated with aging, Gamble et al. (2015) show that a decrease in episodic memory is associated with decreasing abilities in numeracy. In addition, a decrease in semantic memory associated with aging comes along with a decrease in financial knowledge. In consequence, a decrease in cognitive abilities is associated with decreasing financial literacy for the elderly. With respect to the magnitude of the effect of aging, Finke et al. (2016) find that financial literacy scores decline by about 1% a year for people older than 60. As already mentioned in Sect. 2.3, there is a wide discrepancy between test-based and self-assessed financial literacy for the elderly as this group shows high levels of overconfidence: Gamble et al. (2015) and Finke et al. (2016) show that confidence in financial abilities does not decline with age, making the elderly particularly vulnerable to financial scams and fraud (Deevy et al. 2012). Analyzing the SAVE survey, Bucher-Koenen and Lusardi (2011) underscore this evidence for Germans. Specifically, they also document a hump-shaped distribution of financial literacy levels with respect to respondents’ age and find that the least financially literate are individuals aged 65 and above. Admittedly, however, the negative correlation between age and financial literacy documented in the above-mentioned studies might as well be interpreted as a cohort effect: for instance, older people arguably have less investing experience in the pre-401(k) era and the proportion of individuals with higher educational attainment is lower among older cohorts.

Another robust finding across many countries is a gender gap with respect to financial literacy (Lusardi and Mitchell 2009; Lusardi and Tufano 2009, 2015; Lusardi et al. 2010; Hung et al. 2009; Mottola 2013; Bucher-Koenen et al. 2016; Agnew and Harrison 2015, Klapper et al. 2015): men usually score higher on measured financial literacy than women. Two channels have been found to drive this result. On the one hand, women give fewer correct answers in test questions. Lusardi and Mitchell (2014), for instance, document that in the U.S. the fraction of men having all Big Three questions right is 38.3%, while the respective number for women is 22.5%. On the other hand, women seem less confident regarding their financial capabilities as they are more likely to choose the “do not know” category. According to Lusardi and Mitchell (2014) 50.0% of women in the U.S. indicate that they do not know the answer to at least one of the Big Three questions, while the respective fraction for men is 34.3%. A number of studies try to explain this finding arguing with traditional role models (Hsu 2011) suggesting that women only have an incentive to invest in financial literacy late in their lives (Fonseca et al. 2012), differing levels of confidence (Bucher-Koenen et al. 2016), and diverging interests in financial matters (Brown and Graf 2013). However, none of the approaches can entirely explain the gender gap, thus making the issue a promising avenue for further research.

Again, we use the PHF survey to investigate if the gender gap can be observed for recently collected German data, too. As can been seen in Fig. 3b, we indeed observe a moderate, yet statistically significant gender gap in the PHF survey for Germany. The fraction of participants answering all Big Three questions correctly is 64% for male and 57% for female respondents, respectively. Note that previous evidence obtained from the 2009 wave of the SAVE survey (Bucher-Koenen and Lusardi 2011) documented a larger gender gap among German consumers. Specifically, the authors find that almost 60% of male respondents give correct answers to all Big Three questions as opposed to only roughly 48% of surveyed females.

Furthermore, the majority of contributions to the literature document a positive correlation between formal education and financial literacy (Lusardi and Mitchell 2011c; Christelis et al. 2010; Lusardi 2012). For example, Lusardi and Mitchell (2014) report that in the Netherlands 69.8% of individuals with a university degree answer all Big Three questions correctly, whereas among the least educated, the respective percentage amounts to only 28.0%. Of course, it is important to analyze whether the positive correlation might be driven by cognitive abilities of respondents rather than by formal education. However, few studies try to separate cognition from the effect of formal education. Lusardi et al. (2010), e.g., find that formal education is a relevant factor even after controlling for cognitive abilities.

Figure 3c reports the evidence we draw from the PHF data and turns out consistent with the general finding that formal education correlates positively with financial literacy. Among those respondents featuring the highest educational attainment, an overwhelming majority of almost 90% manage to answer all Big Three questions correctly, whereas for the group of least educated individuals, the respective fraction comes to only 53%.²⁸ At this, the novel data we analyze broadly confirm earlier evidence for Germany provided by Bucher-Koenen and Lusardi (2011) who also document a positive relationship between education and financial literacy based on the SAVE survey.

Furthermore, the PHF survey collects information about participants’ professional status. Figure 3d shows that the self-employed are significantly more likely to answer the Big Three questions correctly, a finding which has also been documented in Bucher-Koenen and Lusardi (2011). Finally, a number of contributions have found a positive association between individuals’ income and wealth levels and their levels of financial literacy (e.g., Hung et al. 2009; Lusardi and Tufano 2015; Lusardi and Mitchell 2011c; Klapper et al. 2015). As can be seen in Fig. 3e, f, we again confirm this finding for the PHF survey. With respect to income, 76% of the individuals in the highest income quintile managed to answer all Big Three questions correctly. The respective fraction for the lowest income quintile turns out significantly lower (50%). We find a similar result with respect to wealth; 73% of the individuals in the highest wealth quintile answer all Big Three questions correctly, whereas the respective fraction for the lowest wealth quintile is 51%.²⁹

As mentioned above, the relevance of financial literacy crucially depends on its impact with regard to sound financial behavior. Consequently, a voluminous literature analyzes the question whether high levels of financial literacy trigger superior financial decision making. As we will review shortly, the majority of papers document a positive correlation between measures of financial literacy and sound financial behavior in various domains.³¹ However, absent true randomized control experiments allowing for direct causal inference, the effect of financial literacy on the quality of individual financial decisions is difficult to pin down. Since most evidence on the impact of financial literacy stems from non-experimental research, endogeneity presents a pervasive issue which should be considered carefully when examining the role of financial literacy for financial outcomes. While endogeneity does not rule out the possibility that financial literacy improves individuals’ financial decision making per se, it complicates interpreting the magnitudes of the estimated effects as they are almost surely upwardly biased in magnitude (Hastings et al. 2013).

What causes endogeneity and how does it impact inference? Omitted variables are one of the three sources of endogeneity and refer to those explanatory variables that should be included in the model but in fact are not. If the positive correlation between financial literacy and good financial decisions observed in a given setting likely owes to some underlying third factor which contributes to both higher levels of financial literacy and better financial outcomes, endogeneity enters the model by way of one or more omitted variables. In statistical terms, the inability to explicitly include these determinants in the regression equation means that instead of appearing among the explanatory variables, the impact of these omitted variables appears in the error term, thus distorting estimators and making reliable inference virtually impossible.

Indeed, financial literacy might not be distributed randomly and those individuals exhibiting high levels of financial literacy might share certain characteristics like superior numerical abilities, intelligence, motivation to deal with personal finances, or patience. The literature documents several instances of such hard-to-capture factors likely to influence both financial literacy and financial behavior. Meier and Sprenger (2010) show that those who voluntarily participate in financial education programs are more future-oriented. Hastings and Mitchell (2011) find that those who show patience in an experiment also have a higher propensity to save additional amounts for retirement in their pension accounts. In the same vein, Bucher-Koenen and Lusardi (2011) hypothesize that there might exist an omitted variable bias stemming from missing information on individuals’ ability or motivation to deal with financial matters.

Additionally, a positive correlation between financial literacy and sound financial decision making could stem from reverse causality. Specifically, does financial literacy improve financial behavior or does being involved in certain financial activities instead lead to greater financial literacy? Again, the literature provides a number of examples of potential endogeneity due to a reverse causation channel. Disney et al. (2015) investigate the effect of financial literacy on the decision to seek credit counseling and argue that financial literacy may develop endogenously with the receipt of credit counseling. Bucher-Koenen and Lusardi (2011) hypothesize that individuals with higher levels of financial literacy might better recognize the need and the financial benefits of saving for retirement and thus be more inclined to enroll in a savings plan. They acknowledge, however, that it may as well be retirement planning which affects financial literacy rather than the other way around: Those who have planned for retirement have acquired some level of financial literacy simply by virtue of their savings plan participation. Likewise, the finding of Hilgert et al. (2003) that most individuals cite personal experience as the most important source of their financial learning suggests that some element of reverse causality is likely.

Finally, endogeneity may also arise from measurement error when it comes to the financial literacy variables, e.g. the possibility that answers to test-based financial literacy measures might not measure “true” financial knowledge. As mentioned above, Lusardi and Mitchell (2009), for instance, show that the Big Three are sensitive to framing, i.e. implying that some answers judged to be “correct” are likely attributable to guessing rather than skill.

The standard approach to address endogeneity is finding an instrumental variable (IV) for the endogenous regressor and use this IV in a two-stage least squares (2SLS) regression in order to produce consistent parameter estimates. Generally, a clear understanding of the economics governing the question of interest is key to identifying a valid instrument. A good example of instrument choice is given in van Rooij et al. (2011b) who document a positive relationship between stock market participation and financial literacy which is not only consistent with financially savvy investors having knowledge about expected excess returns of stocks, but also with shareholders learning from their investment experience.³² To establish causality, the authors resort to IV estimation and instrument financial literacy with information regarding the personal finances of their siblings and parents, respectively. Specifically, they asked respondents whether the financial situation of their oldest sibling is either worse or the same or better than their own financial situation and also collected information on how they assess the level of financial knowledge of their parents. Why do these items make particularly good IVs? For one, these instruments for financial literacy are exogenous with respect to respondents’ stock market participation since, arguably, the financial experience of others is beyond their control. At the same time, however, respondents likely learn from their families, thereby increasing their own literacy.³³ Hence, the instruments affect the outcome (the respondent’s propensity to participate in the stock market) only via their effect on the endogenous variable (the respondent’s financial literacy level), i.e. satisfy both the relevance and the exclusion condition required to hold for a valid IV.

Observational studies using carefully chosen IVs are sometimes regarded as being equivalent to quasi experiments regarding their power to support causal claims (Angrist and Krueger 2001). However, a few general comments are in order when discussing the IV regression approach. First, since the error term is unobservable, one cannot empirically test the exclusion condition, i.e. whether or not an instrument is correlated with the regression error term. Consequently, there is no way to statistically ensure that an endogeneity problem has been solved. Moreover, Roberts and Whited (2013) stress that truly exogenous instruments are difficult to find and, in particular, that it should be rigorous economic arguments rather than formal falsification tests that eventually decide over the instrument’s validity.

Evidence on the question of whether or not there is an endogeneity bias caused by omitted variables, reverse causality, or measurement error in studies examining links between measured financial literacy and financial behavior is rather inconsistent. In their review of 11 studies estimating both OLS and IV specifications from their data, Lusardi and Mitchell (2014) find that the IV financial literacy estimates always prove to be larger than the ordinary least squares estimates and conclude that, if anything, non-instrumented estimates of financial literacy underestimate the true effect. This evidence is a strong case for substantial measurement error biasing the OLS estimates, since the magnitude of the coefficients should be upwardly biased if omitted variables and reverse causality were the only sources of endogeneity. By contrast, Fernandes et al. (2014), who conduct a meta-analysis of the relationship between financial literacy and financial outcomes and—unlike Lusardi and Mitchell (2014)—consider standardized coefficients, find significantly smaller effects for 24 studies using both instrumental variables and OLS estimation lacking those controls. They conjecture that non-instrumented regression models in fact overestimate the effect of financial literacy, which reflects endogeneity bias predominantly owing to omitted variables and reverse causality in the OLS designs. Additionally, they test the proposition that designs using instruments for financial literacy and 2SLS are similar to quasi experiments with regards to their ability to support causal inferences—in which case effect sizes should be comparable to what one finds in intervention studies that manipulate financial literacy by means of providing financial education to the treatment group. However, rejecting this claim, they find that intervention studies on average show much smaller effects than econometric studies with instrumental variables and question the validity of instruments used for financial literacy in the studies they review in their meta-analysis. Based on these findings, the authors conclude that past work supporting a causal role for financial literacy might need revisiting on methodological grounds.

Taken together, both the relevance of endogeneity concerns and the tools to remedy potential endogeneity bias are discussed rather controversially in the literature. Outside of controlled experiments, there is no way to ensure that endogeneity problems are eliminated or sufficiently mitigated to allow for reliable causal inferences. Thus, addressing endogeneity concerns by way of IVs should always rest on the strength of the researcher’s arguments supporting the identification strategy.

In what follows, we survey the literature on the impact of financial literacy on financial decision making in various different domains. While we will not explicitly discuss potential endogeneity concerns in the contributions under review, the reader should keep in mind that these issues might still apply.

With respect to investment and savings decisions, arguably most research has been conducted on whether financial sophistication has a positive impact on retirement planning (e.g., Lusardi and Mitchell 2007, 2008; van Rooij et al. 2011a). Analyzing German survey data, Bucher-Koenen and Lusardi (2011) provide evidence for a strong correlation between financial literacy and retirement planning. Regarding the Big Three, the authors show that about 70% of the households who have planned for retirement give correct answers to all Big Three questions, whereas the respective fraction is only 54% for non-planners. Studies analyzing financial behavior in the U.S. also find that individuals with low levels of financial literacy are less likely to plan for their retirement (e.g., Lusardi and Mitchell 2007, 2011b). In a recent study, Clark et al. (2015), using a dataset that links administrative data on investment success with financial literacy, document a positive relationship between individuals’ financial literacy and their propensity to participate in a 401(k) plan as well as the profitability of the respective investments. A related strand of literature has also documented a positive relation between financial literacy and savings behavior (e.g., Lusardi and Mitchell 2011c; Chan and Stevens 2008; Behrman et al. 2012), i.e. providing additional evidence that financially literate individuals exhibit a greater tendency to plan ahead.

Another robust finding in the literature is a positive correlation between stock market participation and financial literacy (e.g., Kimball and Shumway 2006; Christelis et al. 2010; van Rooij et al. 2011b; Balloch et al. 2015; Clark et al. 2015). Specifically, van Rooij et al. (2011b) document that financial sophistication is positively related to stock market participation of retail investors in the Netherlands. Analyzing U.S. survey data, Yoong (2011) confirms that financially sophisticated individuals are more likely to hold stocks and mutual funds. In a related study among retail investors in the U.S., Balloch et al. (2015) find that, besides trust, stock market literacy positively correlates with their likelihood of stock market participation. In addition, they show a positive association between financial sophistication and the conditional magnitude of investing in stocks for those households who do hold stocks in their portfolios. Finally, Jappelli and Padula (2015) present an intertemporal choice model in which individuals can invest in financial literacy. Drawing on cross-country data, the authors find empirical support for the model’s main prediction, i.e. that stock market participation and financial literacy are positively correlated.

A related strand of literature analyzes the association between financial literacy and trading behavior (e.g., Feng and Seasholes 2005; Bilias et al. 2010; Hoffmann et al. 2013; Bucher-Koenen and Ziegelmeyer 2014; Guiso and Viviano 2015) and the corresponding studies generally document a positive impact of financial literacy as financially sophisticated investors tend to commit less investment mistakes.³⁴ In a recent contribution, Bucher-Koenen and Ziegelmeyer (2014) use the financial crisis as a natural experiment to examine individual investors’ ability to cope with sudden economic shocks and document that low literate households are significantly more likely to sell off assets that have lost in value, thereby making paper losses permanent. Shunning stock markets altogether is also associated with a decrease in expected returns on investments (Bucher-Koenen and Ziegelmeyer 2014). Other research in the field has documented that financial literacy is associated with smart choices when it comes to the selection of financial products. Müller and Weber (2010), for instance, investigate the role of financial literacy for mutual fund selection and show that financially sophisticated German retail investors pay lower front-end loads, are less biased in their past return estimates, and are more likely to correctly assess the risk profile of their fund investments. In a related laboratory study, Choi et al. (2009) show that many U.S. investors—even those with high self-assessed financial literacy levels—fail to choose a fee minimizing portfolio even in a setting where fees are the only relevant distinguishing characteristic of the investments and differences in fees are considerable. Moreover, a number of studies has documented a positive link between financial literacy and portfolio diversification: highly literate investors tend to manage their risks significantly better than the group of low literate individuals (e.g., Calvet et al. 2007; Goetzmann and Kumar 2008; Guiso and Jappelli 2008; von Gaudecker 2015; Clark et al. 2015).

The literature also documents a positive link between financial literacy and sound investment decisions. Calvet et al. (2007, 2009) show for investors in Sweden a positive correlation between financial sophistication and account performance and conclude that richer and financially more sophisticated individuals invest more efficiently. Likewise, Clark et al. (2015) document a positive correlation between financial sophistication and excess stock returns among U.S. individuals, while von Gaudecker (2015) finds that this group of retail investors is more likely to hold well-diversified portfolios. Deuflhard et al. (2014) analyze interest rate levels for savings accounts of Dutch consumers. They find that financial literacy is positively associated with higher returns on these accounts. By contrast, Bodnaruk and Simonov (2015) provide evidence against the common finding of a positive relation between financial sophistication and investment performance. In particular, the authors have access to the private portfolios of Swedish mutual fund managers—arguably highly sophisticated market participants—and show that this unique group of individual investors neither outperform, nor diversify their risks more effectively as compared to similar investors in terms of age, gender, education, income, and wealth.

Finally, the literature documents a positive role of financial capabilities in a variety of other domains. For example, Shen et al. (2016) document for Taiwan that individuals with higher levels of financial literacy are less likely to engage in financial disputes. Banks et al. (2015) find for the U.K. that financial literacy and numeracy are significantly positively related to individuals’ propensity to shop around for an annuity when receiving funds from their defined contribution pensions. In an early contribution to the literature, Hilgert et al. (2003) find a strong correlation between financial sophistication and day-to-day financial management skills such as cash-flow and credit management. Finally, de Bassa Scheresberg (2013) documents a positive relation between consumers’ financial literacy and their individual likelihood to hold precautionary savings.

Compared to the large body of literature linking financial literacy to saving and investment decisions, evidence on the role of consumers’ financial capabilities for their financing behavior is scarce. Not surprisingly, the literature typically documents a negative correlation between financial literacy and mistakes in financing decisions: the less financially literate individuals are, the more likely they are to make poor financing decisions. Most prominently, there is solid evidence that low levels of financial literacy are associated with high-cost borrowing and suboptimal mortgage choices (e.g., Moore 2003; Lusardi and Tufano 2015; Lusardi and de Bassa Scheresberg 2013; Disney and Gathergood 2013). Lusardi and Tufano (2015) show for the U.S. that individuals exhibiting low levels of financial literacy use high-cost borrowing and pay higher transaction costs and fees. Lusardi and de Bassa Scheresberg (2013) also examine high-cost borrowing in the U.S., including payday loans, pawn shops and auto title insurance. They also confirm that low literate individuals are substantially exposed to high-cost methods of borrowing. Disney and Gathergood (2013) confirm this finding for the U.K. by showing that low levels of financial literacy are associated with an excessive use of high-cost credit like payday loans or mail order catalogue debt.

Other recent studies document that individuals with low levels of financial literacy are significantly less likely to use their credit cards efficiently (e.g., Lusardi and Tufano 2015; Mottola 2013; Allgood and Walstad 2013). Analyzing U.S. adults, Allgood and Walstad (2013) find a robust negative relation between financial literacy and costly credit card practices. The authors also show that the influence of self-assessed financial literacy on costly credit card practices is greater than that of test-based financial literacy, providing evidence that the two concepts are distinct from each other.³⁵ Analyzing the same dataset, Mottola (2013) also confirms that the financially literate respondents are less often found to exhibit costly credit card behaviors such as being charged a late fee for late payment or not paying down credit card debt in full. In addition, some studies show that a lack of financial literacy is associated with excessive debt accumulation (e.g., Stango and Zinman 2009; Lusardi and Tufano 2015). Analyzing U.S. consumers, Lusardi and Tufano (2015) show that the least sophisticated with respect to debt literacy are exposed to excessive debt loads and the authors also find that this group is not able to judge their debt positions. Finally, Gerardi et al. (2013) find that numerical abilities—a skill set which is positively associated with financial sophistication—are strongly negatively correlated with mortgage defaults.

The arguments in favor of financial education are straightforward. Common sense suggests that introducing financial education initiatives will increase financial literacy, and improved financial literacy, in turn, relates to better financial decision making (Alsemgeest 2015). Accordingly, governments around the world have identified financial education as an intuitive remedy in order to help individuals mastering their personal financial affairs (Fernandes et al. 2014; Willis 2011). For example, policy makers in the U.S. have embraced financial literacy as a means to avoid poor financial decision making and launched a number of financial education initiatives in the wake of the financial crisis. Most prominently, the Office for Financial Education, a subdivision of the Consumer Financial Protection Bureau (CFPB) which was established in the wake of the financial crisis, has an explicit mandate to develop a strategy to increase the financial literacy of U.S. consumers as well as to make recommendations for the launch of programs to improve financial education outcomes.

Although great effort is put in financial education, the question whether educating individuals in the financial domain is beneficial remains controversial both from a theoretical and an empirical perspective. Accordingly, Willis (2008, 2011) provides a number of arguments against financial education. First, she questions whether financial literacy programs can improve consumers’ financial knowledge to an extent that truly qualifies them for the complexity of novel financial products. In particular, she suggests that “the predicate belief in the effectiveness of financial literacy education lacks empirical support. Moreover, the belief is implausible, given the velocity of change in the financial marketplace, the gulf between current consumer skills and those needed to understand today’s complex nonstandardized financial products, the persistence of biases in financial decision making, and the disparity between educators and financial-services firms in resources with which to reach consumers.” (p. 197). Second, she is concerned that individuals’ perceived confidence might increase due to financial education while their actual abilities have not significantly improved which might lead to even poorer financial decision making. Third, she suspects financial education to weaken the position of consumers as related initiatives might come along with a “regulation-through-education model” which blames individuals for bad financial outcomes and thus prevents effective market regulation (Willis 2008). Finally, she pledges for a division of labor as consumers usually do not serve as their own doctors and lawyers and, following this notion, should not serve as their own financial experts, either.

Although the systematic conceptualization of financial literacy is a rather recent development, financial education programs have a long tradition, at least in the U.S. These programs have been initiated by either policymakers, the financial services industry, employees, or nonprofit organizations.³⁶ Hastings et al. (2013) report that financial education interventions in the U.S. date back as long as to the 1950s. Tang and Peter (2015) document that the number of U.S. states in which a personal finance course is required for high school graduation has risen from 13 in 2009 to 17 in 2013, thus highlighting the increased relevance of the topic. The authors also report that the financial services industry is very active in encouraging financial education as, e.g., 98% of U.S. community banks sponsor financial literacy programs and 72% provide an individual program for customers.

Table 4 provides a list of selected financial education programs and we include this list for scholars interested in researching specific programs with respect to their effectiveness. We compiled the list of initiatives by searching the literature on financial education programs and by searching the web for respective keywords. Since we cannot do justice to the large number of financial education programs initiated around the world, we focus on initiatives carried out in German-speaking countries (Panel A) and extend this sample by selected programs in other countries arguably most relevant for previous research on financial literacy (Panel B).

As can be inferred from the table, the initiatives differ in terms of the initiator (e.g. banks, endowments, supranational organizations), the target group (e.g. adults, teenagers, low-income individuals), the educational approach (e.g. in-class teaching or online courses), the intensity of the intervention, and, finally, whether the effectiveness of the respective program has received an academic evaluation.

In what follows, we will review the literature assessing the effectiveness of the different financial education programs. Clearly, the implementation of such programs comes at a cost and, furthermore, their specific content might be biased by initiators’ incentives as well as political agendas. With regard to costs, Fernandes et al. (2014) highlight that educational interventions are not only associated with real costs but might create much larger opportunity costs. Taking high school education as an example, introducing personal finance courses is most likely associated with replacing other important elective courses. With respect to politically differing views of the world, left-wing governments might want to implement other curricula as opposed to right-wing or market-liberal governments. As far as the financial education initiatives of the financial services industry are concerned, these programs will most likely omit important topics like, e.g., fees. So, these programs would hardly teach participants to buy index funds instead of actively managed funds although refraining from actively managed funds is generally regarded as good financial behavior.

A voluminous literature evaluates the association of financial education and test-based financial literacy (e.g., Mandell 2008; Walstad et al. 2010; Heinberg et al. 2010; Lührmann et al. 2015) as well as financial behavior (e.g., Bernheim et al. 2001; Servon and Kaestner 2008; Clark et al. 2015; Lührmann et al. 2015). The literature usually approaches the topic as follows: A particular financial education intervention is analyzed with respect to its impact on measured financial literacy and—in most assessments—on financial behavior. One way to elicit potential changes in financial literacy levels involves measuring test-based literacy scores prior to educating subjects about the financial matters of interest and at some point in time after they have received the respective manipulation (e.g., Walstad et al. 2010). Alternatively, the financial literacy scores of a treatment and a control group are contrasted after a given educational intervention (e.g., Lührmann et al. 2015). The effectiveness of the programs is usually measured by tracking self-reported financial behavior of the participants which is elicited by means of a questionnaire (e.g., Lusardi and Mitchell 2007; Bell et al. 2009; Lührmann et al. 2015).

Fernandes et al. (2014) have recently conducted a meta-analysis in which they include the entire universe of published and unpublished studies on financial education interventions and standardize effect sizes reported in the original studies when aggregating the respective findings. Thus, the conclusions derived from this meta-analysis are arguably less prone to criticism on methodological grounds. The authors evaluate as much as 90 studies in which a financial education intervention has been examined and conclude that, on aggregate, financial education interventions explain no more than 0.1% of the variance in financial behavior. Moreover, they confirm prior findings showing that the already small effects of financial education initiatives tend to decline over time. Specifically, they show that even large interventions have only little impact when financial decisions are made 20 months or later after the subject has received a financial education unit. Taken together, Fernandes et al. (2014) confirm—on more solid empirical grounds—the conclusions drawn in earlier studies which survey the literature on financial education (Hastings et al. 2013; Hung et al. 2009), i.e. a lack of conclusive evidence as to whether a positive impact of financial education can be observed for consumers’ financial behavior. Recently, Lührmann et al. (2015) extend the literature by an assessment of the effectiveness of a financial education program carried out among German high school students (My Finance Coach) and generally find a positive impact of that intervention on students’ financial knowledge. With respect to financial behavior, the evidence depends on the different financial domains examined. While students participating in the program on average exhibit a higher propensity to suppress impulsive purchases, the authors do not find evidence of a significant increase with regard to savings.

Fernandes et al. (2014) highlight that their findings should not be interpreted as evidence against the usefulness of financial education since a large number of financial education programs has never been evaluated and non-assessed initiatives obviously did not enter the sample. A number of reasons for the rather surprising finding of little effectiveness of educational interventions have been discussed in the literature. First, Fernandes et al. (2014) as well as Lusardi and Mitchell (2014), question the quality and motivation of teachers with respect to personal finance issues. As such, Way and Holden (2009), for example, find that less than 20% of high school teachers felt well prepared to teach personal finance topics. Moreover, individuals are heterogeneous in various dimensions. Lusardi and Mitchell (2013) show in a theoretical model that due to this heterogeneity, not everyone should change its behavior after receiving standardized financial training.

Moreover, while the average impact of financial education may be low, the literature has identified several circumstances in which an intervention might be promising. Lusardi and Mitchell (2014), in their review of related research, claim that financial education programs have to be targeted to specific groups in order to incorporate the heterogeneity of individuals. For example, the authors argue that females are ideal targets for specific financial literacy programs, since—other than the average male subject—they are aware of their low financial literacy levels. In their meta-analysis, Fernandes et al. (2014) have identified future directions for designing more successful programs. Specifically, they suggest that improving individuals’ soft skills—e.g. their confidence to be proactive and their willingness to take investment risks—is likely more promising than teaching financial knowledge about, e.g., compound interest. In addition, since the authors find that the effect of financial education declines over time, promising programs should be designed as just-in-time interventions tied to a particular decision. In a recent study, Goedde-Menke et al. (2017) even suggest that one potential explanation for why financial literacy programs are mostly ineffective is the very fact that being financially literate is typically equated with having specific financial knowledge rather than having a basic understanding of fundamental economic concepts. In their study among German adolescents, the authors document that basic economic skills beneficially relate to both individual debt attitudes and behaviors while financial literacy levels turn out to be insignificant. Accordingly, they conclude that a stronger consideration of fundamental economic concepts in financial literacy programs might be a fruitful way to increase their effectiveness.

While financial education programs have enjoyed strong political support as a means to address poor financial decision-making, Willis (2011) stresses that, besides the fact that they are unable to turn everyone into a financial expert, this should not be the path to take for reasons of efficient division of labor alone (see Sect. 6.1). Thus, instead of trying to educate inexperienced individuals, an alternative way to enhance the quality of their decisions on a market for financial services and products characterized by a growing complexity might be to delegate the job by relying on the services offered by professional financial advisors.³⁷ Indeed, a large proportion of households seek expert advice when making financial decisions. Bluethgen et al. (2008) indicate that roughly 80% of individual investors in Germany turn to financial advisors for their investment decisions. In the U.S., 81% of the households investing in mutual funds, outside a retirement plan, rely on financial advice (ICI 2007), and 75% of them seek advice before conducting stock market or mutual fund transactions (Hung and Yoong 2010).

However, the potential benefits of financial advice hinge on two important conditions. First, the advice itself must be accurate, suitable, and consistent with the client’s goals. Whether financial expert intervention indeed benefits consumers remains controversial, though. While some studies concerned with household finance suggest that financial counseling can help individuals develop better financial practices and reduce their debt levels and delinquency rates (Collins and O’Rouke 2010; Agarwal et al. 2011), the evidence as to whether individuals’ investment decisions benefit from expert financial advice is rather mixed. It could be shown that professionally-managed portfolios are better diversified (Bluethgen et al. 2008; Gerhardt and Hackethal 2009) and exhibit weaker disposition effects (Shapira and Venezia 2001) than portfolios of self-directed retail investors. Yet, a number of contributions in the field find that advised accounts are on average associated with higher costs, lower returns and inferior risk-return tradeoffs (Bergstresser et al. 2009; Hackethal et al. 2012; Kramer 2012) and conclude that advisors do not add value through their investment recommendations when judged relative to passive investment benchmarks (Foerster et al. 2014). Also, while there is some consensus that advice can improve retail investor portfolio decisions if conflicts of interest are mitigated (Bhattacharya et al. 2012; Hung and Yoong 2010), a typical advisor’s incentive structure does in fact create a conflict of interest, leading advisors to reinforce biases of their clients instead of correcting them (Mullainathan et al. 2012) and tilt their recommendations towards costly transactions (Hoechle et al. 2015).

Notably, however, while the German government has done little to improve consumers’ financial literacy by means of financial education initiatives, addressing the supply-side issues of retail financial markets has been the top priority of German regulatory authorities in recent years. With respect to financial advice, for instance, regulators now require banks to ask their clients for their prior investment experience before advising them on risky financial products. Moreover, financial advisors are required to assess the risk propensity of their clients before they are allowed to provide them with recommendations. Additionally, banks are obliged to provide advisees with product information sheets disclosing arguably decision-relevant characteristics of financial products. Finally, regulators require advisors to prepare a detailed transcript of each client meeting which has to be authorized by the advisee. The rationale behind these regulations is that the reasons which prevent people from benefiting from financial advice are essentially rooted in the supply side and increasing access to neutral advice should solve the problem of poor financial decision-making. Similarly, regulatory authorities in the Netherlands and in the U.K. have recently enforced a new legislation prohibiting commissions for brokers and advisors altogether. Taken together, we state that regulators in Germany and elsewhere in the world have implemented a number of different measures banning incentives for biased financial advice. In what follows, we thus do not focus on potential conflicts of interest with respect to financial advice.

A second condition which must hold for professional financial advice to be able to substitute for financial literacy, is that low literate individuals must of course seek the support of professional advisors in the first place. Otherwise, measures imposed by regulators to protect consumers arguably will not benefit those who need them most. Thus, knowledge about how financial literacy relates to the demand for financial advice has recently received increasing attention among academics and policymakers.

The notion that financial advice can substitute for low levels of financial literacy rests on the assumption that less knowledgeable individuals face higher hurdles with regards to the collection and processing of information and thus save more on information and search costs when turning to an advisor. Moreover, low literate households may be less aware of potential conflicts of interest arising from advisors’ typical incentive structures and hence more willing to assign an advisor with planning their personal finances (Inderst and Ottaviani 2009). Georgarakos and Inderst (2011) sketch an analytical framework of individual behavior in the context of financial advice using a “cheap-talk”-game in which uninformed investors must decide whether or not to trust the advice they receive whereas informed advisors can opt to ignore the advice. Thus, in their model, consumer information and financial advice are substitutes. Using a large-scale survey among households in 15 EU countries, the authors empirically confirm that trust only matters for the less literate consumers. Similarly, Disney et al. (2015) recently analyze the decision of indebted consumers in the U.K. to ask for financial advice in the form of credit counseling and conclude that professional counseling substitutes for financial literacy: answering an additional financial literacy question correctly reduces the likelihood of an individual seeking assistance by roughly 60%. As can be seen from Table 5, which provides a summary of the empirical evidence regarding the link between financial literacy and advice-seeking, other studies also provide evidence pointing to a negative relationship between financial literacy and the demand for expert financial advice. In the U.S., Hung and Yoong (2010) conduct an experiment among defined-contribution plan holders in the RAND American Life Panel and show that the least sophisticated were most likely to take advice, and Chalmers and Reuter (2012), applying a demographics-based financial literacy proxy, find that younger, less highly educated, and less well-paid (i.e. on average less financially sophisticated) university employees are more likely to demand financial advice on defined-contribution retirement planning.

However, a growing number of studies in the field challenge the negative relationship between peoples’ knowledge in financial matters and their propensity to seek expert assistance and instead point to a complementarity between financial literacy and financial advice. Bucher-Koenen and Koenen (2015) present a model in which advisors have an incentive to provide better advice to consumers who they perceive to be better informed, thus pointing to the fact that financial literacy and the quality of financial advice are complements rather than substitutes. In their analytical framework, it is assumed that advisees with better financial knowledge more likely understand the advice they obtain. This in turn provides the advisor with more incentives to develop sound recommendations for the financially sophisticated investors. Drawing on the 2008 and 2009 waves of the German SAVE household survey, the authors find that smarter investors indeed receive better advice, thus confirming their model predictions. In a related study, Bhattacharya et al. (2012) find that those customers of a German online broker who opted to obtain financial advice in a field study were among the most financially literate clients, and Hackethal et al. (2012) show that advisors of a large German retail bank are matched with wealthier and older investors (proxied to be more financially sophisticated), which also points to a complementarity of financial literacy and financial advice. The authors interpret their findings with respect to both the demand-side and the supply-side of advice. On the one hand, higher opportunity costs of time may induce wealthier clients to make use of financial advice, although they are relatively better prepared to perform the task themselves. On the other hand, financial advisors with commission-based compensation systems should have an incentive to prefer clients with substantial amounts of assets who are more likely to generate significant trading fees.

Empirical evidence indicating that professional financial advice serves as a complement rather than a substitute for financial literacy is not limited to the German market, though. In the US, Collins (2012) uses data from the 2009 FINRA Financial Capability Survey and finds that individuals with higher incomes, higher educational attainment, and higher levels of financial literacy are most likely to receive financial advice. Similarly, Finke (2013) draws on the 2008 wave of the National Longitudinal Survey and documents that financial sophistication increases the demand for financial advice: those individuals most likely to seek advice are not those who are most prone to make financial mistakes. Corroborating this empirical evidence, van Rooij et al. (2011b) exploit the Dutch Household Survey (DHS) and show that people who are less financially literate rely more on informal sources of financial advice, such as friends and family.

Finally, the finding that low literate individuals rely less on expert advice also ties in with psychological evidence, which challenges the notion that people are sophisticated enough to turn to advice in order to overcome their own lack of financial capability. By contrast, this literature suggests that individuals who do not know much about a subject tend not to recognize their ignorance and therefore typically fail to seek better information (Kruger and Dunning 1999).

In a recent contribution, Kramer (2014) suggests that the ambiguous results as to the role of financial knowledge for advice-seeking may at least partly be explained by the different approaches that have been employed to elicit financial literacy. He concludes that studies using self-assessed financial literacy typically find a negative relationship with demanding advice, while those that rely on test-based financial literacy report a positive or insignificant relationship. This discrepancy between the role of self-assessed versus test-based financial literacy implies a role for overconfidence as proposed in the model of Guiso and Jappelli (2006) in which overconfident investors are less willing to rely on information provided by financial advisors, banks or brokers and more likely to collect information directly because they perceive their self-collected information to be of better quality than it actually is. Using the DHS survey data, Kramer (2014) explicitly differentiates between self-reported and objectively elicited financial literacy and shows that confidence in one’s own knowledge is negatively related to asking for help, while actual expertise is not significantly associated with seeking professional financial advice. This finding is confirmed in another recent study based on the DHS data, which analyzes the role of financial literacy and financial advice for households’ portfolio diversification and does not find a significant association between financial literacy and the likelihood of turning to financial advice (von Gaudecker 2015). However, most losses from insufficient diversification are spotted among overconfident investors, which neither are financially literate nor consult with financial advisors.

While seeking financial advice is an important step for the low literate in order to arrive at more informed decisions, one would also like to know if they choose to follow the advice they receive in order to properly assess the potential of professional advice as a substitute for financial literacy. Clearly, financial advice does not translate into sound financial decisions if individuals do not act on the recommendations of their advisors. Surprisingly, however, the question of whether advisees in fact implement the advice they receive is still largely unanswered and very few contributions have considered the role of financial literacy when it comes to following expert advice. The rightmost column of Table 5 summarizes the empirical evidence pertaining to this link.

Calcagno and Monticone (2015) present a stylized model of strategic interaction between advisees and better informed advisors with conflicts of interest. Unlike previous research, however, the authors allow for different degrees of interaction intensity ranging from consulting the advisor in order to enhance one’s information set to completely delegating all decisions to her. Consistent with the framework of Bucher-Koenen and Koenen (2015), the model predicts that the more financially sophisticated are more likely to consult with financial advisors because they anticipate that they will receive valuable information from their advisors. However, once low literate individuals decide to seek the help of financial experts, they are more likely to delegate the decision-making entirely to the advisor. Using data from a survey among the retail customers of Italy’s largest bank, the authors empirically confirm the model predictions showing that financial literacy indeed appears to play a role when it comes to how people implement the advice which they receive from their advisors.

Several studies drawing on German data corroborate the negative relationship between financial literacy and the propensity to follow financial advice. Hackethal et al. (2011), who study the trading behavior of advised retail clients using data from German brokerage accounts, find that they are less likely to implement the advice given to them when their financial sophistication is higher. Bucher-Koenen and Koenen (2015) also document a negative relationship between financial knowledge and advisees’ self-reported propensity to follow advisors’ recommendations. Finally, Stolper (2016) directly matches the recommendations of financial advisors at a German advisory firm with their clients’ post-advice account activity and find that the degree of following the advice is highest for those exhibiting the lowest levels of financial literacy.

To rationalize the ambiguous impact of financial literacy on the use of financial advice, it is argued in the literature that financial sophistication carries two dimensions, i.e. the ability to understand advice on the one hand, and the literacy to question it as well as to process information privately (a possibility which Bucher-Koenen and Koenen 2015, refer to as outside option), on the other hand. While the skills to understand the advice increase the propensity of demanding it, the competence to challenge the advice, along with the ability to collect and handle private information, reduces the likelihood of following it. This is because the financially sophisticated advisee understands the advice and only opts to follow it if she prefers the recommendations to searching on her own, while she ignores it otherwise.

To conclude, the evidence on whether financial advice can be considered a substitute for financial literacy is inconsistent. While some studies document the required negative relationship between the two potential channels to improve financial decision-making, several studies challenge this view by showing that those who would benefit the most from advice are least likely to seek it. Once the advice has been obtained, however, recent contributions to the literature suggest that it is indeed the low literate who are most likely to follow it.