Financial Literacy JEP 2014

8/10/2019 Financial Literacy JEP 2014

1/40

Journal of Economic Literature 2014, 52(1), 544http://dx.doi.org/10.1257/jel.52.1.5

5

1. Introduction

Financial markets around the world havebecome increasingly accessible to thesmall investor, as new products and finan-cial services grow widespread. At the onsetof the recent financial crisis, consumer creditand mortgage borrowing had burgeoned.People who had credit cards or subprimemortgages were in the historically unusualposition of being able to decide how muchthey wanted to borrow. Alternative financial

*Lusardi: George Washington University. Mitchell:University of Pennsylvania. The research reported herein

was performed pursuant to a grant from the TIAACREF

Institute; additional research support was provided bythe Pension Research Council and Boettner Center atthe Wharton School of the University of Pennsylvania.The authors thank Janet Currie, Tabea Bucher-Koenen,Pierre-Carl Michaud, Maarten van Rooij, and StephenUtkus for suggestions and comments, and Carlo de BassaScheresberg, Hugh Kim, Donna St. Louis, and Yong Yu forresearch assistance. Opinions and conclusions expressedherein are solely those of the authors and do not representthe opinions or policy of the funders or any other institu-tions with which the authors are affiliated.

Go to http://dx.doi.org/10.1257/jel.52.1.5 to visit thearticle page and view author disclosure statement(s).

The Economic Importanceof Financial Literacy:Theory and Evidence

A L O S. M*

This paper undertakes an assessment of a rapidly growing body of economic researchon financial literacy. We start with an overview of theoretical research, which casts

financial knowledge as a form of investment in human capital. Endogenizing financialknowledge has important implications for welfare, as well as policies intended toenhance levels of financial knowledge in the larger population. Next, we draw onrecent surveys to establish how much (or how little) people know and identify theleast financially savvy population subgroups. This is followed by an examination of

the impact of financial literacy on economic decision making in the United States andelsewhere. While the literature is still young, conclusions may be drawn about theeffects and consequences of financial illiteracy and what works to remedy these gaps.

A final section offers thoughts on what remains to be learned if researchers are to

better inform theoretical and empirical models as well as public policy. (JEL A20,D14, G11, I20, J26)
http://dx.doi.org/10.1257/jel.52.1.5http://dx.doi.org/10.1257/jel.52.1.5http://dx.doi.org/10.1257/jel.52.1.5http://dx.doi.org/10.1257/jel.52.1.5http://dx.doi.org/10.1257/jel.52.1.5


2/40

Journal of Economic Literature, Vol. LII (March 2014)6

services including payday loans, pawn shops,auto title loans, tax refund loans, and rent-to-own shops have also become widespread.1Atthe same time, changes in the pension land-scape are increasingly thrusting responsibil-ity for saving, investing, and decumulatingwealth onto workers and retirees, whereasin the past, older workers relied mainly onSocial Security and employer-sponsoreddefined benefit (DB) pension plans in retire-ment. Today, by contrast, Baby Boomersmainly have defined contribution (DC) plansand Individual Retirement Accounts (IRAs)during their working years. This trendtoward disintermediation is increasingly

requiring people to decide how much to saveand where to invest and, during retirement,to take on responsibility for careful decumu-lation so as not to outlive their assets whilemeeting their needs.2

Despite the rapid spread of such finan-cially complex products to the retail market-place, including student loans, mortgages,credit cards, pension accounts, and annuities,many of these have proven to be difficult forfinancially unsophisticated investors to mas-ter.3 Therefore, while these developmentshave their advantages, they also impose onhouseholds a much greater responsibility toborrow, save, invest, and decumulate theirassets sensibly by permitting tailored finan-cial contracts and more people to accesscredit. Accordingly, one goal of this paper isto offer an assessment of how well-equippedtodays households are to make these com-plex financial decisions. Specifically we focus

onfinancial literacy, by which we mean peo-

1 See Lusardi (2011) and FINRA Investor EducationFoundation (2009, 2013).

2 In the early 1980s, around 40 percent of U.S.private-sector pension contributions went to DC plans;two decades later, almost 90 percent of such contributions

went to retirement accounts (mostly 401(k) plans; Poterba,Venti, and Wise 2008).

3 See, for instance, Brown, Kapteyn, and Mitchell(forthcoming)

ples ability to process economic informationand make informed decisions about financialplanning, wealth accumulation, debt, andpensions. In what follows, we outline recenttheoretical research modeling how financialknowledge can be cast as a type of investmentin human capital. In this framework, thosewho build financial savvy can earn above-average expected returns on their invest-ments, yet there will still be some optimallevel of financial ignorance. Endogenizingfinancial knowledge has important implica-tions for welfare, and this perspective alsooffers insights into programs intended toenhance levels of financial knowledge in the

larger population.Another of our goals is to assess the effects

of financial literacy on important economicbehaviors. We do so by drawing on evidenceabout what people know and which groupsare the least financially literate. Moreover,the literature allows us to tease out theimpact of financial literacy on economicdecision making in the United States andabroad, along with the costs of financial igno-rance. Because this is a new area of economicresearch, we conclude with thoughts on poli-cies to help fill these gaps; we focus on whatremains to be learned to better inform theo-retical/empirical models and public policy.

2. A Theoretical Frameworkfor Financial Literacy

The conventional microeconomicapproach to saving and consumption

decisions posits that a fully rational andwell-informed individual will consume lessthan his income in times of high earnings,thus saving to support consumption whenincome falls (e.g., after retirement). Startingwith Modigliani and Brumberg (1954) andFriedman (1957), the consumer is positedto arrange his optimal saving and decumula-tion patterns to smooth marginal utility overhis lifetime. Many studies have shown how


3/40

7Lusardi and Mitchell: The Economic Importance of Financial Literacy

such a life cycle optimization process can beshaped by consumer preferences (e.g., riskaversion and discount rates), the economicenvironment (e.g., risky returns on invest-ments and liquidity constraints), and socialsafety net benefits (e.g., the availability andgenerosity of welfare schemes and SocialSecurity benefits), among other features.4

These microeconomic models gener-ally assume that individuals can formulateand execute saving and spend-down plans,which requires them to have the capacity toundertake complex economic calculationsand to have expertise in dealing with finan-cial markets. As we show in detail below,

however, few people seem to have muchfinancial knowledge. Moreover, acquiringsuch knowledge is likely to come at a cost.In the past, when retirement pensions weredesigned and implemented by governments,individual workers devoted very little atten-tion to their plan details. Today, by contrast,since saving, investment, and decumulationfor retirement are occurring in an increas-ingly personalized pension environment, thegaps between modeling and reality are worthexploring, so as to better evaluate wherethe theory can be enriched, and how policyefforts can be better targeted.

Though there is a substantial theoreticaland empirical body of work on the econom-ics of education,5far less attention has beendevoted to the question of how people acquireand deploy financial literacy. In the last fewyears, however, a few papers have begun to

4 For an older review of the saving literature seeBrowning and Lusardi (1996); recent surveys are providedby Skinner (2007) and Attanasio and Weber (2010). A

very partial list of the literature discussing new theoreticaladvances includes Cagetti (2003); Chai et al. (2011); DeNardi, French, and Jones (2010); French (2005); French(2008); Gourinchas and Parker (2002); Aguiar and Hurst(2005, 2007); and Scholz, Seshadri, and Khitatrakun(2006).

5 Glewwe (2002) and Hanushek and Woessmann (2008)review the economic impacts of schooling and cognitivedevelopment.

examine the decision to acquire financialliteracy and to study the links betweenfinancial knowledge, saving, and investmentbehavior (Delavande, Rohwedder, and Willis2008; Jappelli and Padula 2013; Hsu 2011;and Lusardi, Michaud, and Mitchell 2013).6For instance, Delavande, Rohwedder, andWillis (2008) present a simple two-periodmodel of saving and portfolio allocationacross safe bonds and risky stocks, allow-ing for the acquisition of human capital inthe form of financial knowledge ( la Ben-Porath 1967, and Becker 1975). That workposits that individuals will optimally elect toinvest in financial knowledge to gain access

to higher-return assets: this training helpsthem identify better-performing assets and/or hire financial advisers who can reduceinvestment expenses. Hsu (2011) uses asimilar approach in an intrahousehold set-ting where husbands specialize in the acqui-sition of financial knowledge, while wivesincrease their acquisition of financial knowl-edge mostly when it becomes relevant (suchas just prior to the death of their spouses).Jappelli and Padula (2013) also consider atwo-period model but additionally sketch amultiperiod life cycle model with financialliteracy endogenously determined. Theypredict that financial literacy and wealth willbe strongly correlated over the life cycle,with both rising until retirement and fallingthereafter. They also suggest that, in coun-tries with generous Social Security benefits,there will be fewer incentives to save andaccumulate wealth and, in turn, less reason

to invest in financial literacy.

6 Another related study is by Benitez-Silva, Demiralp,and Liu (2009) who use a dynamic life cycle model of opti-mal Social Security benefit claiming against which theycompare outcomes to those generated under a sub-optimalinformation structure where people simply copy thosearound them when deciding when to claim benefits. Theauthors do not, however, allow for endogenous acquisitionof information.


4/40


Each of these studies represents a usefultheoretical advance, yet none incorporateskey features now standard in theoreticalmodels of savingnamely borrowing con-straints, mortality risk, demographic fac-tors, stock market returns, and earnings andhealth shocks. These shortcomings are recti-fied in recent work by Lusardi, Michaud, andMitchell (2011, 2013), which calibrates andsimulates a multiperiod dynamic life cyclemodel where individuals not only selectcapital market investments, but also under-take investments in financial knowledge.This extension is important in that it permitsthe researchers to examine model implica-

tions for wealth inequality and welfare. Twodistinct investment technologies are con-sidered: the first is a simple technology thatpays a fixed low rate of return each period(_

R = 1 +_

r ), similar to a bank account,while the second is a more sophisticatedtechnology providing the consumer accessto a higher stochastic expected return, R(ft),which depends on his accumulated level offinancial knowledge. Each period, the stockof knowledge is related to what the individualhad in the previous period minus a depre-ciation factor: thus f

t+1=ft+it, where

represents knowledge depreciation (due toobsolescence or decay) and gross investmentin knowledge is indicated with it. The sto-chastic return from the sophisticated tech-nology follows the process R(ft+1) =

_

R +r(ft+1) +t+1 (where t is a N(0, 1) iidshock and refers to the standard devia-tion of returns on the sophisticated technol-

ogy). To access this higher expected return,the consumer must pay both a direct cost (c)and a time and money cost () to build upknowledge.7

7 This cost function is assumed to be convex, thoughthe authors also experiment with alternative formulations,

which do not materially alter results. Kzdi and Willis(2011) also model heterogeneity in beliefs about the stockmarket, where people can learn about the statistical pro-cess governing stock market returns, reducing transactions

Prior to retirement, the individual earnsrisky labor income (y) from which he canconsume or invest so as to raise his return (R)on saving (s) by investing in the sophisticatedtechnology. After retirement, the individualreceives Social Security benefits, which area percentage of preretirement income.8Additional sources of uncertainty includestock returns, medical costs, and longevity.

Each period, therefore, the consumers deci-sion variables are how much to invest in thecapital market, how much to consume (),and whether to invest in financial knowledge.

Assuming a discount rate of and o, y,and , which refer, respectively, to shocks in

medical expenditures, labor earnings, andrate of return, the problem takes the formof a series of Bellman equations with the fol-lowing value function Vd(st) at each age aslong as the individual is alive (pe,t >0):

Vd(st) = maxct,it,t

ne,tu (ct/ne,t)

+ pe,t

y

oV (st+1) dFe

(o) dF

e(

y) dF ().

The utility function is assumed to be strictlyconcave in consumption and scaled using thefunctionu(ct/nt), wherentis an equivalencescale capturing family size which changespredictably over the life cycle; and by educa-tion, subscripted by e. End-of-period assets(a

t+1) are equal to labor earnings plus the

returns on the previous periods saving plustransfer income (tr), minus consumption

and costs of investment in knowledge (aslong as investments are positive; i.e., >0).

costs for investments. Here, however, the investment costwas cast as a simplified flat fixed fee per person, whereasLusardi, Michaud, and Mitchell (2013) evaluate morecomplex functions of time and money costs for investmentsin knowledge.

8 There is also a minimum consumption floor; seeLusardi, Michaud, and Mitchell (2011, 2013).


5/40


Accordingly,

at+1

=R(ft+1)(at+ye,t+trtct

(it) cdI (t>0)).9

After calibrating the model using plausibleparameter values, the authors then solvethe value functions for consumers with low/medium/ high educational levels by back-ward recursion.10 Given paths of optimalconsumption, knowledge investment, andparticipation in the stock market, they thensimulate 5,000 life cycles allowing for return,income, and medical expense shocks.11

Several key predictions emerge from thisstudy. First, endogenously determined opti-mal paths for financial knowledge are humpshaped over the life cycle. Second, consum-ers invest in financial knowledge to the pointwhere their marginal time and money costsof doing so are equated to their marginalbenefits; of course, this optimum will dependon the cost function for financial knowledgeacquisition. Third, knowledge profiles differacross educational groups because of peo-ples different life cycle income profiles.

Importantly, this model also predicts thatinequality in wealth and financial knowledgewill arise endogenously without having torely on assumed cross-sectional differencesin preferences or other major changes tothe theoretical setup.12 Moreover, differ-ences in wealth across education groups alsoarise endogenously; that is, some population

9 Assets must be non-negative each period and thereis a nonzero mortality probability as well as a finite lengthof life.

10 Additional detail on calibration and solution meth-ods can be found in Lusardi, Michaud, and Mitchell (2011,2013).

11 Initial conditions for education, earnings, and assetsare derived from Panel Study of Income Dynamics (PSID)respondents age 2530.

12 This approach could account for otherwise unex-plained wealth inequality discussed by Venti and Wise(1998, 2001).

subgroups optimally have low financial lit-eracy, particularly those anticipating sub-stantial safety net income in old age. Finally,the model implies that financial educationprograms should not be expected to pro-duce large behavioral changes for the leasteducated, since it may not be worthwhilefor the least educated to incur knowledgeinvestment costs given that their consump-tion needs are better insured by transferprograms.13This prediction is consistent withJappelli and Padulas (2013) suggestion thatless financially informed individuals will befound in countries with more generous SocialSecurity benefits (see also Jappelli 2010).

Despite the fact that some people willrationally choose to invest little or nothing infinancial knowledge, the model predicts thatit can still be socially optimal to raise finan-cial knowledge for everyone early in life,for instance by mandating financial educa-tion in high school. This is because even ifthe least educated never invest again and lettheir knowledge endowment depreciate, theywill still earn higher returns on their saving,

which generates a substantial welfare boost.For instance, providing pre-labor marketfinancial knowledge to the least educatedgroup improves their wellbeing by an amountequivalent to 82 percent of their initial wealth(Lusardi, Michaud, and Mitchell 2011). Thewealth equivalent value for college graduatesis also estimated to be substantial, at 56 per-cent. These estimates are, of course, specific tothe calibration, but the approach underscoresthat consumers would benefit from acquiring

financial knowledge early in life even if theymade no new investments thereafter.

In sum, a small but growing theoretical lit-erature on financial literacy has made strides

13 These predictions directly contradict at least onelawyers surmise that, [i]n an idealized first-best world,

where all people are far above average, education wouldtrain every consumer to be financially literate and wouldmotivate every consumer to use that literacy to make goodchoices (Willis 2008).


6/40


in recent years by endogenizing the processof financial knowledge acquisition, generat-ing predictions that can be tested empiri-cally, and offering a coherent way to evaluatepolicy options. Moreover, these models offerinsights into how policymakers might enhancewelfare by enhancing young workers endow-ment of financial knowledge. In the next sec-tion, we turn to a review of empirical evidenceon financial literacy and how to measure it inpractice. Subsequently, we analyze existingstudies on how financial knowledge mattersfor economic behavior in the empirical realm.

3. Measuring Financial Literacy

Several fundamental concepts lie at theroot of saving and investment decisions asmodeled in the life cycle setting describedin the previous section. Three such conceptsare: (i)numeracy and capacity to do calcula-

tions related to interest rates, such as com-pound interest; (ii)understanding of inflation;and (iii)understanding of risk diversification.Translating these into easily measured finan-cial literacy metrics is difficult, but Lusardiand Mitchell (2008, 2011a, 2011c) havedesigned a standard set of questions aroundthese ideas and implemented them in numer-ous surveys in the United States and abroad.

Four principles informed the design ofthese questions. The first is Simplicity: thequestions should measure knowledge of thebuilding blocks fundamental to decisionmaking in an intertemporal setting. The sec-

ond is Relevance: the questions should relateto concepts pertinent to peoples day-to-dayfinancial decisions over the life cycle; more-over, they must capture general, rather thancontext-specific, ideas. Third is Brevity: thenumber of questions must be kept short tosecure widespread adoption; and fourth isCapacity to differentiate, meaning that ques-tions should differentiate financial knowl-edge to permit comparisons across people.

These criteria are met by the three financialliteracy questions designed by Lusardi andMitchell (2008, 2011a), worded as follows:

Suppose you had $100 in a savingsaccount and the interest rate was 2 per-cent per year. After 5 years, how much doyou think you would have in the accountif you left the money to grow: [morethan $102; exactly $102; less than $102;do not know; refuse to answer.]

Imagine that the interest rate on yoursavings account was 1 percent peryear and inflation was 2 percent peryear. After 1 year, would you be able to

buy: [more than, exactly the same as,or less than todaywith the money inthis account; do not know; refuse toanswer.]

Do you think that the following state-ment is true or false? Buying a singlecompany stock usually provides a saferreturn than a stock mutual fund. [true;false; do not know; refuse to answer.]

The first question measures numeracy,or the capacity to do a simple calculationrelated to compounding of interest rates.The second question measures understand-ing of inflation, again in the context of asimple financial decision. The third questionis a joint test of knowledge about stocksand stock mutual funds and of risk diver-sification, since the answer to this questiondepends on knowing what a stock is andthat a mutual fund is composed of many

stocks. As is clear from the theoretical mod-els described earlier, many decisions aboutretirement savings must deal with finan-cial markets. Accordingly, it is important tounderstand knowledge of the stock market,as well as differentiate between levels offinancial knowledge.

Naturally, any given set of financial liter-acy measures can only proxy for what indi-viduals need to know to optimize behavior


7/40


in intertemporal models of financial decisionmaking.14 Moreover, measurement erroris a concern, as well as the possibility thatanswers might not measure true financialknowledge. These issues have implicationsfor empirical work on financial literacy, to bediscussed below.

3.1 Empirical Evidence of FinancialLiteracy in the Adult Population

The three questions above were firstadministered to a representative sampleof U.S. respondents aged 50 and older, ina special module of the 2004 Health and

Retirement Study (HRS).15Results, summa-rized intable 1, indicate that this older U.S.population is quite financially illiterate: onlyabout half could answer the simple 2 per-cent calculation and knew about inflation,

14 See Huston (2010) for a review of financial literacymeasures.

15 For information about the HRS, see http://hrsonline.isr.umich.edu/.

and only one third could answer all threequestions correctly (Lusardi and Mitchell2011a). This poor showing is notwithstand-ing the fact that people in this age group hadmade many financial decisions and engagedin numerous financial transactions over theirlifetimes. Moreover, these respondents hadexperienced two or three periods of highinflation (depending on their age) and wit-nessed numerous economic and stock mar-ket shocks (including the demise of Enron),which should have provided them with infor-mation about investment risk. In fact, thequestion about risk is the one where respon-

dents answered disproportionately with Donot know.

These same questions were added to sev-eral other U.S. surveys thereafter, includ-ing the 20072008 National LongitudinalSurvey of Youth (NLSY) for young respon-dents (ages 2328) (Lusardi, Mitchell, andCurto 2010); the RAND American LifePanel (ALP) covering all ages (Lusardi andMitchell 2009); and the 2009 and 2012

TABLE 1F L P U S

Panel A. Distribution of responses to financial literacy questions

Responses

Correct Incorrect DK Refuse

Compound interest 67.1% 22.2% 9.4% 1.3%

Inflation 75.2% 13.4% 9.9% 1.5%

Stock risk 52.3% 13.2% 33.7% 0.9%

Panel B. Joint probabilities of answering financial literacy questions correctly

All 3 responsescorrect

Only 2 responsescorrect

Only 1 responsecorrect

No responsescorrect

Proportion 34.3% 35.8% 16.3% 9.9%

Note:DK = respondent indicated dont know.

Source: Authors computations from 2004 HRS Planning Module
http://hrsonline.isr.umich.edu/http://hrsonline.isr.umich.edu/http://hrsonline.isr.umich.edu/http://hrsonline.isr.umich.edu/


8/40


National Financial Capability Study (Lusardiand Mitchell 2011d).16In each case, the find-ings underscore and extend the HRS resultsin that, for all groups, the level of financialliteracy in the U.S. was found to be quite low.

Additional and more sophisticated con-cepts were then added to the financial lit-eracy measures. For instance, the 2009 and2012 National Financial Capability Surveyincluded two items measuring sophisti-cated concepts such as asset pricing andunderstanding of mortgages/mortgage pay-ments. Results revealed additional gapsin knowledge: for example, data from the2009 wave show that only a small percent-

age of Americans (21 percent) knew aboutthe inverse relationship between bond pricesand interest rates (Lusardi 2011).17 A pass/fail set of 28 questions by Hilgert, Hogarth,and Beverly (2003) covered knowledge ofcredit, saving patterns, mortgages, and gen-eral financial management, and the authorsconcluded that most people earned a failingscore on these questions as well.18Lusardi,Mitchell, and Curto (forthcoming) alsoexamine a set of questions measuring finan-cial sophistication, in addition to basic finan-cial literacy, and found that a large majorityof older respondents are not financiallysophisticated. Additional surveys have alsoexamined financial knowledge in the contextof debt. For example, Lusardi and Tufano(2009a, 2009b) examined debt literacyregarding interest compounding and found

16 Information on the 2009 and 2012 National

Financial Capability Study can be found here:http://www.usfinancialcapability.org/.

17 Other financial knowledge measures include Kimballand Shumway (2006), Lusardi and Mitchell (2009), Yoong(2011), Hung, Parker, and Yoong (2009), and the review inHuston (2010). Related surveys in other countries exam-ined similar financial literacy concepts (see, the DutchCentral Bank Household Survey, which has investigatedand tested measures of financial literacy and financialsophistication, Alessie, van Rooij, and Lusardi 2011).

18 Similar findings are reported for smaller samples orspecific population subgroups (see Agnew and Szykman2011; Utkus and Young 2011).

that only one-third of respondents knew howlong it would take for debt to double if onewere to borrow at a 20 percent interest rate.This lack of knowledge confirms conclusionsfrom Moores (2003) survey of Washingtonstate residents, where she found that peo-ple frequently failed to understand interestcompounding along with the terms and con-ditions of consumer loans and mortgages.Studies have also looked at different mea-sures of risk literacy (Lusardi, Schneider,and Tufano 2011). Knowledge of risk andrisk diversification remains low even whenthe questions are formulated in alterna-tive ways (see, Kimball and Shumway 2006;

Yoong 2011; and Lusardi, Schneider, andTufano 2011). In other words, all of thesesurveys confirm that most U.S. respondentsare not financially literate.

3.2 Empirical Evidence of FinancialLiteracy among the Young

As noted above, it would be useful to knowhow well-informed people are at the start oftheir working lives. Several authors have mea-sured high school students financial literacyusing data from the Jump$tart Coalition forPersonal Financial Literacy and the NationalCouncil on Economic Education. Becausethose studies included a long list of ques-tions, they provide a rather nuanced evalua-tion of what young people know when theyenter the workforce. As we saw for their adultcounterparts, most high school students inthe U.S. receive a failing financial literacygrade (Mandell 2008; National Council on

Economic Education 2005). Similar findingsare reported for college students (Chen andVolpe 1998; and Shim et al. 2010).

3.3 International Evidenceon Financial Literacy

The three questions mentioned earlierand that have been used in several surveysin the United States have also been used inseveral national surveys in other countries.
http://www.usfinancialcapability.org/http://www.usfinancialcapability.org/http://www.usfinancialcapability.org/http://www.usfinancialcapability.org/


9/40


Table 2 reports the findings from the twelvecountries that have used these questionsand where comparisons can be made forthe total population.19For brevity, we onlyreport the proportion of correct and do notknow answers to each question and for allquestions.

The table highlights a few key findings.First, few people across countries can cor-rectly answer three basic financial literacyquestions. In the United States, only 30 per-cent can do so, with similar low percentagesin countries having well-developed financialmarkets (Germany, the Netherlands, Japan,Australia, and others), as well as in nations

where financial markets are changing rap-idly (Russia and Romania). In other words,low levels of financial literacy found in theUnited States are also prevalent elsewhere,rather than being specific to any givencountry or stage of economic development.Second, some of what adult respondentsknow is related to national historical experi-ence. For example, Germans and Dutch aremore likely to know the answer to the infla-tion question, whereas many fewer peopledo in Japan, a country that has experienceddeflation. Countries that were plannedeconomies in the past (such as Romania andRussia) displayed the lowest knowledge ofinflation. Third, of the questions examined,risk diversification appears to be the conceptthat people have the most difficulty grasping.

19 The Central Bank of Austria has used these questionsto measure financial literacy in ten countries in Eastern

Europe and we report the findings for Romania, wherefinancial literacy has been studied in detail (Beckmann2013). These questions have also been fielded in Mexicoand Chile (Hastings and Tejeda-Ashton 2008; Hastings andMitchell 2011; Behrman et al. 2012), India, and Indonesia(Cole, Sampson, and Zia 2011). They have also been usedto measure financial literacy among Sri Lankan entrepre-neurs (de Mel, McKenzie, and Woodruff 2011) and a sam-ple of U.S.-based migrants from El Salvador (Ashraf et al.2011). We do not report the estimates for these countriesbecause they do not always work with representative sam-ples of the population or use samples that can be compared

with the statistics reported in table 2.

Virtually everywhere, a high share of peoplerespond that they do not know the answerto the risk diversification question. Forinstance, in the United States, 34 percentof respondents state they do not know theanswer to the risk diversification question;in Germany 32 percent and the Netherlands33 percent do so; and even in the most risk-savvy countries of Sweden and Switzerland,18 percent and 13 percent respectively,report that they do not know the answer tothe risk diversification question.

The Organization for EconomicCo-operation and Development (OECD)has been a pioneer in highlighting the lack of

financial literacy across countries. For exam-ple, an OECD report in 2005 documentedextensive financial illiteracy in Europe,Australia, and Japan, among others.20Morerecently, Atkinson and Messy (2011, 2012)confirmed the patterns of financial illiteracymentioned earlier in the text across 14 coun-tries at different stages of development infour continents, using a harmonized set offinancial literacy as in the three questionsthat were used in many countries.21

The goal of evaluating student financialknowledge around the world among theyoung (high school students) has recentlybeen taken up by the OECDs Programmefor International Student Assessment(PISA),22 which in 2012 added a moduleon financial literacy to its review of profi-ciency in mathematics, science, and reading.Accordingly, 15-year-olds around the worldwill be able to be compared with regard to

20 Researchers have also examined answers to questionson mathematical numeracy in the England LongitudinalSurvey of Ageing (ELSA; Banks and Oldfield 2007), and inthe Survey of Health, Ageing, and Retirement in Europe(SHARE; Christelis, Jappelli, and Padula 2010).

21 Their survey uses eight financial literacy questionsand focuses on fundamental concepts including the threemain concepts discussed earlier.

22 For more information on the Financial LiteracyFramework in PISA, see: http://www.oecd.org/pisa/pisaproducts/46962580.pdf
http://www.oecd.org/pisa/pisaproducts/46962580.pdfhttp://www.oecd.org/pisa/pisaproducts/46962580.pdfhttp://www.oecd.org/pisa/pisaproducts/46962580.pdfhttp://www.oecd.org/pisa/pisaproducts/46962580.pdf


10/40


TABLE2

CSRFLQW

Intere

strate

Inflation

R

iskDiversification

Authors

Country

Yearof

data

Correct

DK

Correct

DK

C

orrect

DK

All3

correct

Atleast1

dontknow

Numberof

Observations

LusardiandMitchell(2011d)

USA

2009

64.9

%

13.5

%

64.3

%

14.2

%

51.8

%

33.7

%

30.2

%

42.4

%

1,4

88

Alessie,

VanRooij,andLusardi(2011

)

Netherlands

2010

84.8

%

8.9

%

76.9

%

13.5

%

51.9

%

33.2

%

44.8

%

37.6

%

1,6

65

Bucher-KoenenandLusardi(2011)

Germany

2009

82.4

%

11.0

%

78.4

%

17.0

%

61.8

%

32.3

%

53.2

%

37.0

%

1,0

59

Sekita(2011)

Japan

2010

70.5

%

12.5%

58.8

%

28.6

%

39.5

%

56.1

%

27.0

%

61.5

%

5,2

68

Agnew,

Bateman,andThorp(2013)

Australia

2012

83.1

%

6.4

%

69.3

%

13.0

%

54.7

%

37.6

%

42.7

%

41.3

%

1,0

24

Crossan,

Feslier,andHurnard(2011)

N.

Zealand

2009

86.0

%

4.0

%

81.0

%

5.0

%

27.0

%

2.0

%*

24.0

%*

7.0

%

850

BrownandGraf(2013)

Switzerland

2011

79.3

%

2.8

%*

78.4

%

4.2

%*

73.5

%*

13.0

%*

50.1

%*

16.9

%*

1,5

00

ForneroandMonticone(2011)

Italy

2007

40.0

%*

28.2

%*

59.3

%*

30.7

%*

52.2%*

33.7

%*

24.9

%*

44.9

%*

3,9

92

AlmenbergandSve-Sderbergh(201

1)

Sweden

2010

35.2

%*

15.6

%*

59.5

%

16.5

%

68.4

%

18.4%

21.4

%*

34.7

%*

1,3

02

Arrondel,Debbich,andSavignac(201

3)

France

2011

48.0

%*

11.5

%*

61.2

%

21.3

%

66.8

%*

14.6

*

30.9

%*

33.4

%*

3,6

16

KlapperandPanos(2011)

Russia

2009

36.3

%*

32.9

%*

50.8

%*

26.1

%*

12.8

%*

35.4

%*

3.7

%*

53.7

%*

1,3

66

Beckmann(2013)

Romania

2011

41.3

%

34.4

%

31.8

%*

40.4

%*

14.7

%

63.5

%

3.8

%*

75.5

%*

1,0

30

Note:*

indicatesquestionsthathaveslightlydifferentwordingthanthebaselinefinancialliteracyquestionsenumeratedinthetext.


11/40


their financial knowledge. In so doing, PISAhas taken the position that financial literacyshould be recognized as a skill essential forparticipation in todays economy.

3.4 Objective versus Subjective Measures ofFinancial Literacy

Another interesting finding on financialliteracy is that there is often a substantialmismatch between peoples self-assessedknowledge versus their actual knowledge,where the latter is measured by correctanswers to the financial literacy questionsposed. As one example, several surveys includequestions asking people to indicate their self-

assessed knowledge, as indicated by the fol-lowing question used in the United States andalso in the Netherlands and Germany:

On a scale from 1 to 7, where 1 meansvery low and 7 means very high, howwould you assess your overall financialknowledge?

Even though actual financial literacy levelsare low, respondents are generally ratherconfident of their financial knowledge and,overall, they tend to overestimate how muchthey know (table 3). For instance, in the 2009U.S. Financial Capability Study, 70 percentof respondents gave themselves score of 4or higher (out of 7), but only 30 percent ofthe sample could answer the factual ques-tions correctly (Lusardi 2011). Similarfindings were reported in other U.S. sur-veys and in Germany and the Netherlands

(Bucher-Koenen et al. 2012). One exceptionis Japan, where respondents gave themselveslow grades in financial knowledge. In otherwords, though actual financial literacy islow, most people are unaware of their ownshortcomings.

3.5 Financial Literacy and Framing

Peoples responses to survey questionscannot always be taken at face value, a

point well-known to psychometricians andeconomic statisticians. One reason, as notedabove, is that financial literacy may be mea-sured with error, depending on the wayquestions are worded. To test this possibility,Lusardi and Mitchell (2009) and van Rooij,Lusardi, and Alessie (2011) randomly askedtwo groups of respondents the same risk ques-tion, but randomized their order of presenta-tion. Thus half the group received format (a)and the other half format (b), as follows:

(a) Buying a company stock usually pro-vides a safer return than a stock mutualfund. True or false?

OR

(b) Buying a stock mutual fund usuallyprovides a safer return than a companystock. True or false?

They found that peoples responses were,indeed, sensitive to how the question wasworded in both the U.S. American Life Panel(Lusardi and Mitchell 2009) and the DutchCentral Bank Household Survey (DHS; vanRooij, Lusardi, and Alessie 2011). For exam-ple, fewer DHS respondents respondedcorrectly when the wording was buying a

stock mutual fund usually provides a saferreturn than a company stock; conversely,the fraction of correct responses doubledwhen shown the alternative wording: buy-ing a company stock usually provides a saferreturn than a stock mutual fund. This was

not simply due to people using a crude ruleof thumb (such as always picking the first asthe correct answer), since that would gener-ate a lower rather than a higher percentageof correct answers for version (a). Instead,it appeared that some respondents did notunderstand the question, perhaps becausethey were unfamiliar with stocks, bonds, andmutual funds. What this means is that someanswers judged to be correct may instead


12/40


TABLE3

CSRS-F

L

Authors

Country

Dataset

12

3

4

5

6

7

Average

score

Authorscalculations

USA

NFCS2012

3.9%

5.2%

14.9%

33.2%

26.1%

13.6%

5.1

Lusardi(2011)

USA

NFCS2009

7.5%

6.0%

16.2%

32.3%

20.2%

17.5%

5

LusardiandTufano(2009a)

USA

TNSGlobal2007

4.9%

7.7%

19.5%

31.9%

18.9%

10.7%

4.9

Authorscalculationsondatafrom

USA*

ALP2009*

5.3%

11.6%

27.2%

34.7%

16.7%

4.4%

4.6

LusardiandMitchell(2009)*

Bucher-Koenen,Lusardi,Alessie

Netherlands

DHS2010

7.3%

10.9%

23.0%

32.0%

23.4%

3.5%

4.6

andvanRooij(2012)

Bucher-Koenen,Lusardi,Alessie

Germany

SAVE2009

8.3%

14.2%

23.0%

32.2%

15.6%

6.8%

4.5

andvanRooij(2012)

Sekita(2011)*

Japan*

SLPS2010*

71%*

23.3%*

5.6%*

Note:Thistablereportsrespondentsanswerstothequestion:Ona

scalefrom1

to7,where1meansve

rylowand7meansveryhigh,how

wouldyouassess

youroverallfinancialknowledge?

*NotethatthequestionposedinLusardiandMitchell(2009)isdif

ferentandasksthefollowing:How

wouldyouassessyourunderstandingofeconomics

(ona7-pointscale;1meansvery

lowand7meansveryhigh)?InJapan,respondentswereaskedwheth

ertheythinkthattheyknowalotaboutfinanceona

15pointscale(Sekita2011).


13/40


be attributable to guessing. In other words,analysis of the financial literacy questionsshould take into account the possibility thatthese measures may be noisy proxies of truefinancial knowledge levels. 23

4. Disaggregating Financial Literacy

To draw out lessons about which peoplemost lack financial knowledge, we turn nextto a disaggregated assessment of the data.In what follows, we briefly review evidenceby age and sex, race/ethnicity, income andemployment status, and other factors ofinterest to researchers.

4.1 Financial Literacy Patterns by Age

The theoretical framework sketched aboveimplies that the life cycle profile of financialliteracy will be hump-shaped, and survey dataconfirm that financial literacy is, in fact, lowestamong the young and the old.24This is a find-ing which is robust across countries and wereport a selected set of countries infigure 1.

Of course with cross-sectional data, onecannot cleanly disentangle age from cohorteffects, so further analysis is required toidentify these clearly, and below we comment

23 In the 2008 HRS, the financial literacy questionswere modified to assess the sensitivity of peoples answersto the way in which the questions were worded. Resultsconfirmed sensitivity to question wording, especiallyfor the more sophisticated financial concepts (Lusardi,Mitchell, and Curto forthcoming). Behrman et al. (2012)developed a financial literacy index employing a two-step

weighting approach, whereby the first step weighted each

question by difficulty and the second step applied princi-pal components analysis to take into account correlationsacross questions. Resulting scores indicated how finan-cially literate each individual was in relation to the averageand to specific questions asked. The results confirmed thatthe basic financial literacy questions designed by Lusardiand Mitchell (2011a) receive the largest weights.

24 Earlier we made mention of the widespread lack offinancial and economic knowledge among high school andcollege students. At the other end of the work life, financialliteracy also declines with age, as found in the 2004 HRSmodule on financial literacy on people age 50+ and inmany other countries (Lusardi and Mitchell 2011a, 2011c).

further on this point (figure 1a). Nevertheless,it is of interest that older people give them-selves very high scores regarding their ownfinancial literacy, despite scoring poorly onthe basic financial literacy questions (Lusardiand Mitchell 2011a; Lusardi and Tufano2009a) and not just in the United States, butother countries as well (Lusardi and Mitchell2011c). Similarly, Finke, Howe, and Huston(2011) develop a multidimensional measureof financial literacy for the old and confirmthat, though actual financial literacy falls withage, peoples confidence in their own financialdecision-making abilities actually increaseswith age. The mismatch between actual and

perceived knowledge might explain whyfinancial scams are often perpetrated againstthe elderly (Deevy, Lucich, and Beals 2012).

4.2 Financial Literacy Differences by Sex

One striking feature of the empirical dataon financial literacy is the large and persis-tent gender difference described in figure 1b.Not only are older men generally more finan-cially knowledgeable than older women, butsimilar patterns also show up among youngerrespondents as well (Lusardi, Mitchell, andCurto 2010; Lusardi and Mitchell 2009;Lusardi and Tufano 2009a, 2009b). Moreover,these gaps persist across both the basic andthe more sophisticated literacy questions(Lusardi, Mitchell, and Curto forthcoming;Hung, Parker, and Yoong 2009).

One twist on the differences by sex, how-ever, is that while women are less likely toanswer financial literacy questions correctly

than men, they are also far more likely to saythey do not know an answer to a question, aresult that is strikingly consistent across coun-tries (figure 1b).25This awareness of their own

25While statistics are only reported for four countriesin figure 1b, the prevalence of do not know responsesby women is found in all of the twelve countries listed intable 2.


14/40


Panel 1B. By sex(percent providing correct answers to all three financial literacy questions)

Panel 1A. By age group(percent providing correct answers to all three financial literacy questions)

(percent responding do not know at least once to any of the three financialliteracy questions)

70%

60%

50%

40%

30%

20%

10%

0%

70%

60%

50%

40%

30%

20%

10%

0%

60%

50%

40%

30%

20%

10%

0%

USA Germany Netherlands Switzerland



38.3

22.5

59.6

47.5

55.1

35.0

39.3

62.0

34.3

50.0

29.9

43.345.8

29.0

11.8

21.6

MaleFemale

65

MaleFemale

Figure 1. Financial Literacy across Demographic Groups (Age, Sex, and Education)


15/40


lack of knowledge may make women idealtargets for financial education programs.

Because these sex differences in financialliteracy are so persistent and widespreadacross surveys and countries, several research-ers have sought to explain them. Consistentwith the theoretical framework described ear-lier, Hsu (2011) proposed that some sex dif-

ferences may be rational, with specializationof labor within the household leading marriedwomen to build up financial knowledge onlylate in life (close to widowhood). Nonetheless,that study did not explain why financial lit-eracy is also lower among single women incharge of their own finances. Studies of finan-cial literacy in high school and college alsorevealed sex differences in financial literacyearly in life (Chen and Volpe 2002; Mandell

2008).26Other researchers seeking to explainobserved sex differences concluded that tra-ditional explanations cannot fully accountfor the observed male/female knowledgegap (Fonseca et al. 2012; Bucher-Koenenet al. 2012). Fonseca et al. (2012) suggestedthat women may acquire or produce finan-cial literacy differently from men, while

Bucher-Koenen et al. (2012) pointed to apotentially important role for self-confidencethat differs by sex. Brown and Graf (2013)also showed that sex differences are not due

26 It may be possible but untested so far that women,for example young ones, expect they would have someonelater in life (a husband or companion) to take care of theirfinances.

USANetherlands

Switzerland Germany

706050403020100

Panel 1C. Financial literacy by education group(percent providing correct answers to all three financial literacy questions)

8070

6050403020100

8070

6050403020100

80

60

40

20

0

12.619.2

31.3

44.3

63.8

2835.1

41.7

54.4 55.4

69.8

7270.1

5552.451.6

21.7

68.9

44.9

26.6

43.1

Less than HS High-school Some college College Postgraduate Primary Lowersecondary

Middlesecondary

Uppersecondary

Highervocational

University

Lowersecondary

Uppersecondary

non-GDR GDR Post-secondary TertiaryTertiaryUppersecondary

VocationalPrimary or lowersecondary

Figure 1. Financial Literacy across Demographic Groups (Age, Sex, and Education) Continued


16/40


to differential interest in finance and financialmatters between women and men.

To shed more light on womens financial lit-eracy, Mahdavi and Horton (2012) examinedalumnae from a highly selective U.S. womensliberal arts college. Even in this talented andwell-educated group, womens financial liter-acy was found to be very low. In other words,even very well educated women are not par-ticularly financially literate, which could implythat women may acquire financial literacy dif-ferently from men. Nevertheless this debateis far from closed, and additional researchwill be required to better understand theseobserved sex differences in financial literacy.

4.3 Literacy Differences by Educationand Ability

As illustrated in figure 1c, there are sub-stantial differences in financial knowledgeby education: specifically, those without acollege education are much less likely to beknowledgeable about basic financial literacyconcepts, as reported in several U.S. surveysand across countries (Lusardi and Mitchell2007a, 2011c). Moreover, numeracy is espe-cially poor for those with low educationalattainment (Christelis, Jappelli, and Padula2010; Lusardi 2012).

How to interpret the finding of a posi-tive link between education and financialsavvy has been subject to some debate in theeconomics literature. One possibility is thatthe positive correlation might be driven bycognitive ability (McArdle, Smith, and Willis2009), implying that one must control on

measures of ability when seeking to parseout the separate impact of financial literacy.Fortunately, the NLSY has included bothmeasures of financial literacy and of cognitiveability (i.e., the Armed Services VocationalAptitude Battery). Lusardi, Mitchell, andCurto (2010) did find a positive correlationbetween financial literacy and cognitive abil-ity among young NLSY respondents, butthey also showed that cognitive factors did

not fully account for the variance in finan-cial literacy. In other words, substantial het-erogeneity in financial literacy remains evenafter controlling on cognitive factors.

4.4 Other Literacy PatternsThere are numerous other empirical reg-

ularities in the financial literacy literaturethat are, again, persistent across countries.Financial savvy varies by income and employ-ment type, with lower-paid individuals doingless well and employees and the self-employeddoing better than the unemployed (Lusardiand Tufano 2009a; Lusardi and Mitchell2011c). Several studies have also reported

marked differences by race and ethnicity, withAfrican Americans and Hispanics displayingthe lowest level of financial knowledge in theU.S. context (Lusardi and Mitchell 2007a,2007b, 2011b). These findings hold across agegroups and many different financial literacymeasures (Lusardi and Mitchell 2009). Thoseliving in rural areas generally score worse thantheir city counterparts (Klapper and Panos2011). These findings might suggest thatfinancial literacy is more easily acquired viainteractions with others, in the workplace orin the community.27Relatedly, there are alsoimportant geographic differences in financialliteracy; for example, Fornero and Monticone(2011) report substantial financial literacydispersion across regions in Italy and so doesBeckmann (2013) for Romania. Bumcrot,Lin, and Lusardi (2013) report similar differ-ences across U.S. states.

The literature also points to differences

in financial literacy by family background.For instance, Lusardi, Mitchell, and Curto(2010) linked financial literacy of 2328-year-old NLSY respondents to characteristicsof the households in which they grew up,

27 This might also help account for the sex differencesmentioned above, since in many cultures, men are morelikely than women to interact daily with financially knowl-edgeable individuals.


17/40


controlling for a set of demographic and eco-nomic characteristics. Respondents finan-cial literacy was also significantly positivelycorrelated with parental education (in par-ticular, that of their mothers), and whethertheir parents held stocks or retirementaccounts when the respondents were teen-agers. Mahdavi and Horton (2012) reporteda connection between financial literacy andparental background; in this case, fatherseducation was positively associated withtheir female childrens financial literacy.28Inother words, financial literacy may well getits start in the family, perhaps when childrenobserve their parents saving and investing

habits, or more directly by receiving financialeducation from parents (Chiteji and Stafford1999; Li 2009; Shim et al. 2009).

Other studies have noted a nationalitygap in financial literacy, with foreign citizensreporting lower financial literacy than thenative born (Brown and Graf 2013). Othershave found differences in financial literacyaccording to religion (Alessie, van Rooij,and Lusardi 2011) and political opinions(Arrondel, Debbich, and Savignac 2013).These findings may also shed light on howfinancial literacy is acquired.

To summarize, while financial illiteracy iswidespread, it is also concentrated amongspecific population subgroups in most coun-tries studied to date. Such heterogeneityin financial literacy suggests that differentmechanisms may be appropriate for trackingthe causes and possible consequences of theshortfalls. In the United States, those facing

most challenges are the young and the old,women, African-Americans, Hispanics, theleast educated, and those living in rural areas.To date, these differences have not beenfully accounted for, though the theoretical

28 Other studies discussing financial socialization of theyoung include Hira, Sabri, and Loibl (2013) and the refer-ences cited therein.

framework outlined above provides guide-lines for explaining some of these.

5. How Does Financial Literacy Matter?

We turn next to a discussion of whetherand how financial literacy matters foreconomic decision making.29 Inasmuch asindividuals are increasingly being asked totake on additional responsibility for theirown financial well-being, there remainsmuch to learn about these facts. And as wehave argued above, when financial literacyitself is a choice variable, it is important todisentangle cause from effect. For instance,

those with high net worth who invest infinancial markets may also be more likely tocare about improving their financial knowl-edge, since they have more at stake. In whatfollows, we discuss research linking financialliteracy with economic outcomes, taking intoaccount the endogeneity issues as well.

5.1 Financial Literacy and EconomicDecisions

The early economics literature in this areabegan by documenting the link betweenfinancial literacy and several economicbehaviors. For example Bernheim (1995,1998) was among the first to emphasize thatmost U.S. households lacked basic financialknowledge and that they also used cruderules of thumb when engaging in savingbehavior. More recently, Calvet, Campbell,and Sodini (2007, 2009) evaluated Swedishinvestors actions that they classified as

mistakes. While that analysis included nodirect financial literacy measure, the authorsdid report that poorer, less educated, andimmigrant households (attributes associ-ated with low financial literacy, as notedearlier) were more likely to make financialerrors. Agarwal et al. (2009) also focused on

29 For a review of the role of financial literacy in theconsumer behavior literature, see Hira (2010).


18/40


financial mistakes, showing that these weremost prevalent among the young and the old,groups which normally display the lowestfinancial knowledge.

In the wake of the financial crisis of200809, the U.S. federal government hasalso begun to express substantial concernabout another and more extreme case ofmistakes, namely where people fall prey tofinancial scams. As often noted, scams tendto be perpetrated against the elderly, sincethey are among those with the least finan-cial savvy and often have accumulated someassets.30A survey of older financial decision-makers (age 60+) indicated that more than

half of them reported having made a badinvestment, and one in five of those respon-dents felt they had been misled or defraudedbut failed to report the situation (FINRAInvestor Education Foundation 2006). AsBaby Boomers age, this problem is expectedto grow (Blanton 2012), since this cohort is apotentially lucrative target.

Several researchers have examined the rela-tionships between financial literacy and eco-nomic behavior. It is much harder to establisha causal link between the two, and we will dis-cuss the issue of endogeneity and other prob-lems in more detail below. Hilgert, Hogarth,and Beverly (2003) uncovered a strong cor-relation between financial literacy and day-to-day financial management skills. Severalother studies, both in the United States andother countries, have found that the morenumerate and financially literate are alsomore likely to participate in financial markets

and invest in stocks (Kimball and Shumway

30 In 2011 Americans submitted over 1.5 million com-plaints about financial and other fraud, up 62 percent in

just three years; these counts are also likely understate-ments (FTC 2012). Financial losses per capita due to fraudhave also increased over time: the median loss per victimrose from $218 in 2002 to $537 in 2011. Similarly the U.S.SEC (2012) warns about scams and fraud and other poten-tial consequences of very low financial literacy, particularlyamong the most vulnerable groups.

2006; Christelis, Jappelli, and Padula 2010;van Rooij, Lusardi, and Alessie 2011; Yoong2011; Almenberg and Dreber 2011; Arrondel,Debbich, and Savignac 2012). Financial liter-acy can also be linked to holding precaution-ary savings (de Bassa Scheresberg 2013).

The more financially savvy are also morelikely to undertake retirement planning, andthose who plan also accumulate more wealth(Lusardi and Mitchell 2007a, 2007b, 2011a,2011d). Some of the first studies on theeffects of financial literacy were linked to itseffects on retirement planning in the UnitedStates. These studies have been replicatedin most of the countries covered in table 2,

showing that the correlation between finan-cial literacy and different measures of retire-ment planning is quite robust.31 Studiesbreaking out specific components of financialliteracy tend to conclude that what mattersmost is advanced financial knowledge (forexample, risk diversification) and the capac-ity to do calculations (Lusardi and Mitchell2011b; Alessie, van Rooij, and Lusardi 2011;Fornero and Monticone 2011; Klapper andPanos 2011; Sekita 2011).

Turning to the liability side of the house-hold balance sheet, Moore (2003) reportedthat the least financially literate are also morelikely to have costly mortgages. Campbell(2006) pointed out that those with lowerincome and less education (characteristicsstrongly related to financial illiteracy) wereless likely to refinance their mortgages duringa period of falling interest rates. Stango andZinman (2009) concluded that those unable

to correctly calculate interest rates out of astream of payments ended up borrowing

31 The link between financial literacy and retirementplanning also robust to the measure of financial literacyused (basic versus sophisticated financial knowledge;Lusardi and Mitchell 2009, 2011b), how planning is mea-sured (Lusardi and Mitchell 2007a, 2009, 2011a; Alessie,

van Rooij, and Lusardi 2011), and which controls areincluded in the empirical estimation (van Rooij, Lusardi,and Alessie 2011).


19/40


more and accumulating less wealth. Lusardiand Tufano (2009a) confirmed that the leastfinancially savvy incurred high transactioncosts, paying higher fees and using high-costborrowing. In their study, the less knowl-edgeable also reported that their debt loadswere excessive, or that they were unable tojudge their debt positions. Similarly, Mottola(2013) found that those with low financialliteracy were more likely to engage in costlycredit card behavior, and Utkus and Young(2011) concluded that the least literate werealso more likely to borrow against their401(k) and pension accounts.

Moreover, both self-assessed and actual

literacy is found to have an effect on creditcard behavior over the life cycle (Allgood andWalstad 2013). A particularly well-executedstudy by Gerardi, Goette, and Meier (2013)matched individual measures of numericalability to administrative records that provideinformation on subprime mortgage holderspayments. Three important findings flowedfrom this analysis. First, numerical abilitywas a strong predictor of mortgage defaults.Second, the result persisted even after con-trolling for cognitive ability and generalknowledge. Third, the estimates were quan-titatively important, as will be discussed inmore detail below, an important finding forboth regulators and policymakers.

Many high-cost methods of borrowing haveproliferated over time, with negative effectsfor less savvy consumers.32 For instance,Lusardi and de Bassa Scheresberg (2013)examined high-cost borrowing in the United

States, including payday loans, pawn shops,auto title loans, refund anticipation loans,and rent-to-own shops. They concluded thatthe less financially literate were substantially

32 The alternative financial services (AFS) industry hasexperienced tremendous growth in the United States: in2009, the Federal Deposit Insurance Corporation esti-mated the industry to be worth at least $320 billion interms of transactional services (FDIC 2009).

more likely to use high-cost methods of bor-rowing, a finding that is particularly strongamong young adults (age 2534) (de BassaScheresberg 2013). While most attention hasbeen devoted to the supply side, these stud-ies suggest it may also be important to lookat the demand side and the financial literacyof borrowers. The large number of mortgagedefaults during the financial crisis has like-wise suggested to some that debt and debtmanagement is a fertile area for mistakes;for instance, many borrowers do not knowwhat interest rates were charged on theircredit card or mortgage balances (Moore2003; Lusardi 2011; Disney and Gathergood

2012).33It is true that education can be quite influ-

ential in many of these arenas. For instance,research has shown that the college edu-cated are more likely to own stocks and lessprone to use high-cost borrowing (Haliassosand Bertaut 1995; Campbell 2006; Lusardiand de Bassa Scheresberg 2013). Likewise,there is a very strong positive correlationbetween education and wealth holding(Bernheim and Scholz 1993). But for ourpurposes, including controls for educationalattainment in empirical models of stockholding, wealth accumulation, and high-costmethods of borrowing does not diminish thestatistical significance of financial literacyand, in fact, it often enhances it (Lusardiand Mitchell 2011a; Behrman et al. 2012;van Rooij, Lusardi, and Alessie 2011, 2012;Lusardi and de Bassa Scheresberg 2013).Evidently, general knowledge (education)

and more specialized knowledge (financialliteracy) both contribute to more informedfinancial decision making. In other words,investment in financial knowledge appearsto be a specific form of human capital, rather

33 Disney and Gathergood (2012) reported that UKconsumer credit customers systematically underestimatedthe cost of borrowing, while the least financially literatehad higher average debt-to-income ratios.


20/40


than being simply associated with more yearsof schooling. Financial literacy is also linkedto the demand for on-the-job training (Clark,Matsukura, and Ogawa 2013) and being ableto cope with financial emergencies (Lusardi,Schneider, and Tufano 2011).

5.2 Costs of Financial IgnorancePre-retirement

In the wake of the financial crisis, manyhave become interested in the costs of finan-cial illiteracy as well as its distributionalimpacts. For instance, in the Netherlands,van Rooij, Lusardi, and Alessie (2012) esti-

mate that being in the 75th versus the 25thpercentile of the financial literacy indexequals around 80,000 in terms of differen-tial net worth (i.e., roughly 3.5 times the netdisposable income of a median Dutch house-hold). They also point out that an increase infinancial literacy from the 25th to the 75thpercentile for an otherwise average indi-vidual is associated with a 1730 percentagepoint higher probability of stock market par-ticipation and retirement planning, respec-tively. In the United States, simulations froma life-cycle model that incorporates financialliteracy shows that financial literacy alonecan explain more than half the observedwealth inequality (Lusardi, Michaud, andMitchell 2013). This result is obtained bycomparing wealth-to-income ratios acrosseducation groups in models with and withoutfinancial literacy, which allows individuals toearn higher returns on their savings. For this

reason, if the effects of financial literacy onfinancial behavior can be taken as causal, thecosts of financial ignorance are substantial.

In the United States, investors are esti-mated to have foregone substantial equityreturns due to fees, expenses, and activeinvestment trading costs, in an attempt tobeat the market. French (2008) calculatesthat this amounts to an annual total cost ofaround $100 billion, which could be avoided

by passive indexing. Since the least finan-cially literate are unlikely to be sensitive tofees, they are most likely to bear such costs.Additionally, many of the financially illiteratehave been shown to shun the stock market,which Cocco, Gomes, and Maenhout (2005)suggested imposed welfare losses amountingto four percent of wealth. The economic costof underdiversification computed by Calvet,Campbell, and Sodini (2007) is also substan-tial: they concluded that a median investor inSweden experienced an annual return loss of2.9 percent on a risky portfolio, or 0.5 per-cent of household disposable income. Butfor one in ten investors, these annual costs

were much higher, 4.5 percent of disposableincome.

Costs of financial ignorance arise not onlyin the saving and investment arena, but alsoinfluence how consumers manage their lia-bilities. Campbell (2006) reported that sub-optimal refinancing among U.S. homeownersresulted in 0.51 percent per year highermortgage interest rates, or in aggregate,$50100 billion annually. And as noted above,the least financially savvy are least likely torefinance their mortgages. Gerardi, Goette,and Meier (2013) showed that numericalability may have contributed substantially tothe massive defaults on subprime mortgagesin the recent financial crisis. According totheir estimates, those in the highest numeri-cal ability grouping had about a 20 percent-age point lower probability of defaulting ontheir subprime mortgages than those in thelowest financial numeracy group.

One can also link debt literacy regardingcredit card behaviors that generate fees andinterest charges to paying bills late, goingover the credit limit, using cash advances,and paying only the minimum amount due.Lusardi and Tufano (2009a) calculated thecost of ignorance, or transaction costsincurred by less-informed Americans andthe component of these costs related to lackof financial knowledge. Their calculation of


21/40


expected costs had two componentsthelikelihood and the costs of various credit cardbehaviors. These likelihoods were deriveddirectly from empirical estimates using thedata on credit card behavior, debt literacy,and a host of demographic controls thatinclude income. They showed that, whileless knowledgeable individuals constituteonly 29 percent of the cardholder popula-tion, they accounted for 42 percent of thesecharges. Accordingly, the least financiallysavvy bear a disproportionate share of thecosts associated with fee-inducing behaviors.Indeed, the average fees paid by those withlow knowledge were 50 percent higher than

those paid by the average cardholder. And ofthese four types of charges incurred by less-knowledgeable cardholders, one-third wereincremental charges linked to low financialliteracy.

Another way that the financially illiter-ate spend dearly for financial services is viahigh-cost forms of borrowing, including pay-day loans.34While the amount borrowed isoften low ($300 on average), such loans aremade to individuals who have five or moresuch transactions per year (Ernst, Farris,and King 2004). It turns out that these bor-rowers also frequently fail to take advantageof other, cheaper opportunities to borrow.Agarwal, Skiba, and Tobacman (2009) stud-ied payday borrowers who also have accessto credit cards, and found that two-thirdsof their sample had at least $1,000 in creditcard liquidity on the day they took out theirfirst payday loan. This points to a pecuniary

mistake: given average charges for paydayloans and credit cards and considering a two-

week payday loan of $300, the use of creditcards would have saved these borrowers

34 Americans paid about $8 billion in nance charges toborrow more than $50 billion from payday lenders in 2007;the annual interest rates on such loans are often very high,over 400 percent. See Bertrand and Morse (2011) and thereferences therein.

substantial amountsaround $200 per year(and more if they took out repeated paydayloans). While there may be good economicreasons why some people may want to keepbelow their credit card limits, includingunexpected shocks, Bertrand and Morse(2011) determined that payday borrowersoften labored under cognitive biases, similarto those with low financial literacy (Lusardiand de Bassa Scheresberg 2013).

5.3 Costs of Financial Ignorancein Retirement

Financial knowledge impacts key out-comes, including borrowing, saving, and

investing decisions, not only during theworklife, but afterwards, in retirement, aswell. In view of the fact that people over theage of sixty-ve hold more than $18 trillionin wealth,35this is an important issue.

Above we noted that financial literacy isassociated with greater retirement plan-ning and greater retirement wealth accu-mulation.36 Hence, it stands to reasonthat the more financially savvy will likelybe better financially endowed when theydo retire. A related point is that the morefinancially knowledgeable are also betterinformed about pension system rules, paylower investment fees in their retirementaccounts, and diversify their pension assets

35See for instance Laibson (2011).36 See for instance Ameriks, Caplin, and Leahy (2003);

van Rooij, Lusardi, and Alessie (2012); and Lusardi andMitchell (2007a, 2007b; 2009). It is worth noting that edu-cation also plays a role, as pointed out by Poterba, Venti,

and Wise (2013) who find a substantial association betweeneducation and the postretirement evolution of assets.For example, for two-person households, assets growthbetween 1998 and 2008 was greater for college graduatesthan for those with less than a high school degree, pro-ducing over $600,000 in assets for the richest quintile, to$82,000 for the lowest asset quintile. As in the theoreti-cal model described previously, households with differentlevels of education will invest in different assets, allowingthem to earn different rates of return. It remains to be seen

whether this is because of differential financial literacyinvestments, or simply due to general knowledge gleanedthrough education.


22/40


better (Arenas de Mesa et al. 2008; Chanand Stevens 2008; Hastings, Mitchell, andChyn 2011).37 To date, however, relativelylittle has been learned about whether morefinancially knowledgeable older adults arealso more successful at managing theirresources in retirement, though the pres-ence of scams among the elderly suggeststhat this topic is highly policy-relevant.

This is a particularly difficult set of deci-sions requiring retirees to look ahead to anuncertain future when making irrevocablechoices with far-reaching consequences.For instance, people must forecast their(and their partners) survival probabilities,

investment returns, pension income, andmedical and other expenditures. Moreover,many of these financial decisions areonce-in-a-lifetime events, including whento retire and claim ones pension and SocialSecurity benefits. Accordingly, it would notbe surprising if financial literacy enhancedpeoples ability to make these important andconsequential decisions.

This question is especially relevant whenit comes to the decision of whether retir-ees purchase lifetime income streams withtheir assets, since by so doing, they insurethemselves against running out of incomein old age.38 Nevertheless, despite the factthat this form of longevity protection is veryvaluable in theory, relatively few payoutannuities are purchased in practice in vir-tually every country (Mitchell, Piggott, andTakayama 2011). New research points to theimportance of framing and default effects in

this decision process (Agnew and Szkyman

37 Gustman, Steinmeier, and Tabatabai (2010) notethat financial knowledge is not the same thing as cognitivefunctioning, since the latter is not associated with greaterknowledge of retirement plan rules.

38Several authors have also linked financial literacyand knowledge about retirement saving. For instance,Agnew et al. (2007) show that employees who were theleast financially knowledgeable were 34 percent less likelyto participate voluntarily, and 11 percent less likely to beautomatically enrolled, in their companys 401(k) plan.

2011; Brown, Kapteyn, and Mitchell forth-coming). This conclusion was corroboratedby Brown et al. (2011), who demonstratedexperimentally that people valued annuitiesless when they were offered the opportunityto buy additional income streams, and theyvalued annuities more if offered a chanceto exchange their annuity flows for a lumpsum.39Importantly for the present purpose,the financially savvy provided more consis-tent responses across alternative ways ofeliciting preferences. By contrast, the leastfinancially literate gave inconsistent resultsand respond to irrelevant cues when pre-sented with the same set of choices. In other

words, financial literacy appears to be highlyinfluential in helping older households equipthemselves with longevity risk protection inretirement.

Much more must be learned about howpeoples financial decision-making abilitieschange with age, and how these are relatedto financial literacy. For instance, Agarwal etal. (2009) reported that the elderly pay muchmore than the middle-aged for 10 finan-cial products;40 the 75-year-olds in theirsample paid about $265 more per year forhome equity lines of credit than did the50-year-olds. How the patterns might varyby financial literacy is not yet known, butit might be that those with greater baselinefinancial knowledge are better able to dealwith financial decisions as they move into thesecond half of their lifetimes.41

39 These findings are not attributable to differences inindividuals subjective life expectancies, discount rates, riskaversion, borrowing constraints, political risk, or other con-

ventional explanations (Brown et al. 2011).40 These include credit card balance transfers; home

equity loans and lines of credit; auto loans; credit cardinterest rates; mortgages; small business credit cards;credit card late-payment fees; credit card over-limit fees;and credit card cash-advance fees.

41 This could be particularly important inasmuch asKorniotis and Kumar (2011) find that cognitive decline isfastest with age for the less educated, lower earners, andminority racial/ethnic groups.


23/40


5.4 Coping with Endogeneityand Measurement Error

Despite an important assembly of factson financial literacy, relatively few empiri-cal analysts have accounted for the potentialendogeneity of financial literacy and theproblem of measurement error in financialliteracy alluded to above. In the last fiveyears or so, however, several authors haveimplemented instrumental variables (IV)estimation to assess the impact of financialliteracy on financial behavior, and the resultstend to be quite convincing. To illustrate theingenuity of the instruments used, table 4

lists several studies along with the instru-ments used in their empirical analysis. Someof the descriptive evidence on financial liter-acy discussed earlier may explain why theseinstruments may be anticipated to predictfinancial literacy.

It is useful to offer a handful of commentson some of the papers with particularlystrong instruments. Christiansen, Joensen,and Rangvid (2008) used the opening ofa new university in a local areaarguablyone of the most exogenous variables onecan findas instrument for knowledge, andthey concluded that economics educationis an important determinant of investmentin stocks. Following this lead, Klapper,Lusardi, and Panos (2012) used the num-ber of public and private universities inthe Russian regions and the total numberof newspapers in circulation as instrumentsfor financial literacy. They found that finan-

cial literacy affected a variety of economicindicators including having bank accounts,using bank credit, using informal credit,having spending capacity, and the availabil-ity of unspent income. Lusardi and Mitchell(2009) instrumented financial literacy usingthe fact that different U.S. states mandatedfinancial education in high school at dif-ferent points in time and they interactedthese mandates with state expenditures on

education. Behrman et al. (2012) employedseveral instruments, including exposure toa new educational voucher system in Chile,to isolate the causal effects of financial lit-eracy and schooling attainment on wealth.Their IV results showed that both finan-cial literacy and schooling attainment werepositively and significantly associated withwealth levels.

Van Rooij, Lusardi, and Alessie (2011)instrumented financial literacy with thefinancial experiences of siblings and parents,since these were arguably not under respon-dents control, to rigorously evaluate the rela-tionship between financial literacy and stock

market participation. The authors reportedthat instrumenting greatly enhanced themeasured positive impact of financial lit-eracy on stock market participation. Theseinstruments were also recently used byAgnew, Bateman, and Thorp (2013) to assessthe effect of financial literacy on retirementplanning in Australia. Bucher-Koenen andLusardi (2011) used political attitudes at theregional level in Germany as an instrument,arguing that free-market oriented support-ers are more likely to be financially literate,and the assumption is that individuals canlearn from others around them. The study byArrondel, Debbich, and Savignac (2013) alsoshows some differences in financial literacyacross political affiliation.

Interestingly, in all these cases, the IVfinancial literacy estimates always prove tobe larger than the ordinary least squaresestimates (table 4). It might be that peo-

ple affected by the instruments have largeresponses, or there is severe measurementerror, but on the other hand, it seems clearthat the noninstrumented estimates of finan-cial literacy may underestimate the trueeffect.

Despite these advances, one might worrythat other omitted variables could still influ-ence financial decisions in ways that couldbias results. For example, unobservables such


24/40


25/40


as discount rates (Meier and Sprenger 2013),IQ (Grinblatt, Keloharju, and Linnainmaa2011), or cognitive abilities could influ-ence saving decisions and portfolio choice(Delavande, Rohwedder, and Willis 2008;Korniotis and Kumar 2011). If these can-not be controlled for, estimated financialliteracy impacts could be biased. However,Alessie, van Rooij, and Lusardis (2011) workusing panel data and fixed-effects regres-sion as well as IV estimation confirmed thepositive effect of financial literacy on retire-ment planning, and several studies, as men-tioned earlier (Gerardi, Goette, and Meier2013), account explicitly for cognitive ability.

Nevertheless, they show that numeracy hasan effect above and beyond cognitive ability.

A different way to parse out the effects offinancial literacy on economic outcomes is touse a field experiment in which one group ofindividuals (the treatment group) is exposedto a financial education program and theirbehavior is then compared to that of a secondgroup not thus exposed (the control group).Yet even in countries with less developedfinancial markets and pension systems, finan-cial literacy impacts are similar to those foundwhen examining the effect of financial literacyon retirement planning and pension participa-tion (Lusardi and Mitchell 2011c). For exam-ple, Song (2011) showed that learning aboutinterest compounding produces a sizeableincrease in pension contributions in China.Randomized experimental studies in Mexicoand Chile demonstrated that more finan-cially literate individuals were more likely to

choose pension accounts with lower admin-istrative fees (Hastings and Tejeda-Ashton2008; Hastings and Mitchell 2011; Hastings,Mitchell, and Chyn 2011). More financiallysophisticated individuals in Brazil were alsoless affected by their peers choices in theirfinancial decisions (Bursztyn et al. 2012).

The financial crisis has also provided alaboratory to study the effects of financial lit-eracy against a backdrop of economic shocks.

For example, when stock markets droppedsharply around the world, investors wereexposed to large losses in their portfolios.This, combined with much higher unem-ployment, has made it even more importantto be savvy in managing limited resources.Bucher-Koenen and Ziegelmeyer (2011)examined the financial losses experienced byGerman households during the financial cri-sis and confirmed that the least financially lit-erate were more likely to sell assets that hadlost value, thus locking in losses.42In Russia,Klapper, Lusardi, and Panos (2012) foundthat the most financially literate were signifi-cantly less likely to report having experienced

diminished spending capacity and had moreavailable saving. Additionally, estimates fromdifferent time periods implied that financialliteracy better equips individuals to deal withmacroeconomic shocks.

Given this evidence on the negative out-comes and costs of financial illiteracy, weturn next to financial education programs toremedy these shortfalls.

6. Assessing the Effects of FinancialLiteracy Programs

Another way to assess the effects of finan-cial literacy is to look at the evidence onfinancial education programs whose aimsand objectives are to improve financialknowledge. Financial education programs inthe United States and elsewhere have beenimplemented over the years in several dif-ferent settings: in schools, workplaces, and

libraries, and sometimes population sub-groups have been targeted. As one example,several U.S. states mandated financial educa-tion in high school at different points in time,generating natural experiments utilized byBernheim, Garrett, and Maki (2001), one ofthe earliest studies in this literature. Similarly,

42 Part of this behavior could also be due to liquidityconstraints.


26/40


financial education in high schools hasrecently been examined in Brazil and Italy(Bruhn, Legovini, and Zia 2012; Romagnoliand Trifilidis 2013). In some instances, largeU.S. firms have launched financial educa-tion programs (Bernheim and Garrett 2003,Clark and DAmbrosio 2008, and Clark,Morrill, and Allen 2012a, 2012b). Often theemployers intention is to boost defined ben-efit pension plan saving and participation(Duflo and Saez 2003, 2004; Lusardi, Keller,and Keller 2008; Goda, Manchester, andSojourner 2012). Programs have also beenadopted for especially vulnerable groups,such as those in financial distress (Collins

and ORourke 2010).Despite the popularity of the programs,

only a few authors have undertaken care-ful evaluations of the impact of financialeducation programs. Rather than detailingor reviewing the existing literature,43 herewe instead draw attention to the key issueswhich future researchers must take intoaccount when evaluating the effectivenessof financial education programs.44 We alsohighlight key recent research not reviewedin prior surveys.

A concern emphasized above in section 2is that evaluation studies have sometimesbeen conducted without a clear understand-ing of how financial knowledge is developed.That is, if we define financial literacy as aform of human capital investment, it standsto reason that some will find it optimal toinvest in financial literacy while others willnot. Accordingly, if a program were to be

judged based on specific behavioral changes

43 See for instance Collins and ORourke (2010);Gale, Harris, and Levine (2012); Hastings, Madrian, andSkimmyhorn (2012); Hathaway and Khatiwada (2008);Lusardi and Mitchell (2007b); Lyons et al. (2006); andMartin (2007). Hira (2010) provides a broad overview ofresearch on financial education over a long time span.

44 Two good discussions by Fox, Ba

Documents

Financial Literacy JEP 2014