CONTRIBUTIONS Blanchett

Financially Sound Households Use Financial Planners, Not Transactional Advisers by David M. Blanchett, Ph.D., CFA, CFP®

David M. Blanchett, Ph.D., CFA, CFP®, is head of Executive Summary research at Morningstar Management. He is an adjunct professor of wealth • Financial advisers can add decisions, followed by those using management at The American College. He is a recipi- significant value for clients, but the Internet, while those work- ent of the Journal’s 2007 Financial Frontiers Award empirical evidence documenting ing with a transactional adviser and its 2014, 2015, and 2019 Montgomery-Warschauer this effect is mixed. were making the worst financial Award. • This paper explores how house- decisions. hold financial decision-making • Selection bias is a potential issue Households are becoming increas- varies by four sources of informa- with the results, since the decision ingly responsible for myriad financial tion: financial planners; transac- to work with a financial planner decisions, such as determining how tional financial advisers; friends; is a positive indicator of financial much to save for retirement, how to or the Internet. decision-making and potentially invest those savings, when to retire, etc. • Five aspects of decision-making endogenous to variables con- Given the complexity of these decisions were explored: portfolio risk sidered. However, these findings levels; savings habits; life insur- do suggest that the potential and the general lack of financial literacy ance coverage; revolving value of financial advice may vary among U.S. households (Lusardi and card balances; and emergency significantly based on the nature Mitchell 2014), financial advisers savings using the six most recent of the financial engagement and should seemingly be well-positioned waves of the Survey of Consumer that households are likely better to help improve household financial (2001 to 2016). off working with an adviser that decision-making. Indeed, a growing body • Households working with a is more comprehensive (e.g., a of theoretical research has noted the financial planner were found to be financial planner) than transac- potential value of financial advisers in a making the best overall financial tional in nature. variety of domains; however empirical evidence on the topic is mixed and capturing value created in other domains coverage, revolving balances, generally suggests households with (e.g., savings rates or life and emergency savings) varied across financial advisers do no better (or even coverage). Another could be that certain four information sources: - worse) than those without, especially in types of advisers are providing valuable ners, transactional financial advisers, investment-related domains. services (e.g., financial planners) that friends, or the Internet. By decomposing The general lack of empirical evidence are not consistently captured in the financial advisers into two types, it is on the improved outcomes or decision- relatively broad “” possible to better understand if any making for households working with description. differences exist by the type of the advice financial advisers is not positive for the This paper used the six most recent engagement. financial advice profession. Empirical waves of the Survey of Consumer The analysis focused on the soundness evidence on this topic is lacking for a Finances (2001 to 2016) to explore how of various household financial decisions variety of possible reasons. One could household decision-making across five (e.g., does the household have any be that the empirical research, which financial planning domains (portfolio revolving credit card ?) versus more is largely investment-focused, is not risk level, savings habits, life insurance outcome-oriented variables (e.g., wealth

30 Journal of Financial Planning | April 2019 FPAJournal.org Blanchett CONTRIBUTIONS or level of savings). Focusing on decisions uniform, either, so research that does not terjee (2014) also explored the potential better captured differences in multiple attempt to control for advice type may value of financial advisers. domains, reduced issues associated likely produce biased results. Households that work with a financial with reverse causality (because clients Overall, the basic question “Do adviser tend to have higher incomes, with more wealth become increasingly financial advisers add value?” is not be wealthier, more educated, older, and attractive to financial advisers and it necessarily well-defined in the empirical more financially literate (Burke and may be difficult to determine the role of literature, given the significant differ- Hung 2015). These individuals also tend the financial adviser with respect to the ences in the scope of services provided to have higher risk tolerance (Hanna wealth creation), and controlled for the by financial advisers. It is likely that 2011). Research on consumer financial fact that higher wealth (or more savings) potential and realized benefits of decisions increasingly points to the doesn’t necessarily imply the household financial advice vary by adviser type. importance of financial sophistication is behaving optimally (e.g., adequate life This paper will explore this specific topic as a determinant of sound financial insurance may reduce available savings, in greater detail. decision-making (Campbell 2006), but it is a vital component of a sound therefore controlling for household financial plan for most households). Literature Review demographics is an important aspect of Households working with a financial The lack of financial literacy of U.S. any type of empirical analysis. planner were found to be making the households (Lusardi and Mitchell 2014) One problem with identifying any “best” financial decisions, in the aggre- would suggest financial advisers have the type of empirical benefit associated with gate as well as in four of the five domains potential to add significant value, both working with a financial adviser is that considered, while households working in investing and non-investing domains. the decision to hire a financial adviser is with a transactional adviser were For example, from an investment not random and is potentially endog- making the “worst” financial decisions. perspective, Odean (1998) found that enous to whatever outcome variable is Selection bias is a potential issue with tend to underperform by sell- considered. For example, it may be that the results, since the decision to work ing winners too soon and holding losers wise and financially prudent decision- with a financial planner is a positive too , a tendency labeled the “disposi- makers are more likely to hire financial indicator of financial decision-making tion effect” (Shefrin and Statman 1985). advisers. Similarly, an who was and potentially endogenous to variables A financial adviser who is aware of this already making sound financial decisions considered; however, these findings do effect can either make clients aware of it may hire a financial adviser with the goal at least suggest financial planners are to help mitigate it, or take discretion of of helping him or her make even better adding the most value among the infor- the account (assuming the adviser is not financial decisions. For these investors, mation sources considered, especially disposed to the same effect). it would be difficult to disentangle the compared to transactional advisers. Exploring the potential value of actual impact of the adviser on decisions, Households using the Internet scored financial advice is a growing field of had the investor not hired the adviser second to financial planners on overall research. Theoretical research on the (i.e., correlation does not necessarily financial soundness. This is noteworthy value of financial advice has focused on imply causation). given the growing use of the Internet the potential value of making optimal Early empirical evidence on the value as the primary information source for financial decisions compared to some of a financial adviser focused largely on households included in the analysis, type of naïve benchmark (i.e., what investment-related domains and noted increasing from 3 percent in 2001, to the household would be assumed to do mixed findings. For example, research 40 percent in 2016, as well as given its without the adviser). For example, Hanna has noted positive (Grinblatt and Kelo- relatively low cost (especially compared and Lindamood (2010) and Blanchett harju 2000; Shapira and Venezia 2001; to many financial advisers). However, and Kaplan (2013) both used utility-based and Barber, Lee, Liu, and Odean 2008), the better outcomes associated with the models to explore the potential value of and negative (Bergstresser, Chalmers, Internet have been declining over time financial advisers. Both found that the and Tufano 2009; Mullainathan, Noeth, (from 2001 to 2016), so it is not clear to value of financial advice can be significant and Schoar 2012; Hackethal, Haliassos, what extent this relation will persist in and potentially exceed common financial and Jappelli 2012; and Chalmers, John- the future. adviser fees, although the true expected son, and Reuter 2014) effects of advisers All financial advice is not the same; value will vary by client. Additional on investment outcomes. However, the nor are adviser types. Thus, one shouldn’t research by Kinniry, Jaconetti, DiJoseph, majority of research has suggested inves- expect the potential value of advice to be and Zilbering (2014) and Grable and Chat- tors using financial advisers are no better

FPAJournal.org April 2019 | Journal of Financial Planning 31 CONTRIBUTIONS Blanchett off (or potentially worse off, especially value of working with a financial adviser types of financial advisers. Martin and after fees) than those without. is weak for a variety of reasons, such as Finke (2014) is one example. They noted The possible benefits of a financial misaligned incentives, lack of general households using more comprehensive adviser extend beyond investment ability, and segmentation/identification. financial advisers generated more wealth domains, and some research has With respect to incentives, depending than those without any help, as well explored these areas. For example, on the domain explored, it may not as versus those advisers providing less Warschauer and Sciglimpaglia (2012) actually be in the financial adviser’s holistic services. noted how advisers can assist with best interest to help the client make the emergency fund management, debt optimal decision, if that decision does Sources of Financial Information management, insurable risk reduction, not align with the adviser’s method of for Households investment risk control, goal assess- compensation. For example, Del Guercio Robust data on household financial ment, and and estate assessment. and Reuter (2014) noted how brokers information sources may be found in the Engelmann, Capra, Noussair, and Berns face a weaker incentive to generate Survey of Consumer Finances (SCF). (2009) suggested financial planners alpha, and Christoffersen, Evans, and The SCF is a triennial cross-sectional may help clients focus on long-term Musto (2013) suggested fee sharing survey of U.S. families conducted by goals by reducing -term anxiety alters broker incentives and can be the Federal Reserve Board that includes from market volatility. Burke and Hung particularly harmful to investors when information on families’ balance sheets, (2015) suggested that working with a brokers’ incentives are not aligned with , income, and demographic financial adviser helps improve financial their clients’ interests. characteristics. Heckman, Saey, Kim, and savings habits. Financial advisers may also not be as and Letkiewicz (2016) evaluated the Research by Martin and Finke (2014), capable as they should be. For example, validity of the measures of financial Finke, Huston, and Waller (2009), Linnainmaa, Melzer, Previtero, and planner use in publicly available datasets Cho, Gutter, Kim, and Mauldin (2012), Foerster (2018) found financial advisers and suggested the SCF was one of the among others, has noted a positive make the same poor investment deci- two most promising datasets, as there relationship between the use of financial sions as their clients (such as frequent are a variety available.4 advisers and savings. Additional research trading, return chasing, use of active This specific question in the SCF on decisions surrounding life insurance funds, and under-diversification). asks the respondent about the source of (Finke, Huston, and Waller, 2009), No state or federal law requires financial information: emergency savings (Bhargava and Lown financial advisers to hold designations.1 2006), and disability insurance (Scott Of the one million What sources of information do you and Finke 2013) have also noted better professionals in the U.S. today, only (and your family) use to make decisions outcomes for households working with approximately 80,000 financial advisers about saving and ? (Do financial advisers. Most of these studies, hold the Certified Financial Planner you call around, read newspapers, though, did not control for selection (CFP®) designation,2 the most popular magazines, material you get in the mail, bias. Marsden, Zick, and Mayer (2011) financial advising designation, followed use information from television, radio, attempted to control for simultaneity by the Chartered Financial Consultant the Internet, or advertisements? Do you bias and reverse causation and found (ChFC) designation, with 55,0003 get advice from a friend, relative, , no statistically significant difference in designees (Raskie, Martin, Lemoine, , banker, broker, or financial self-reported retirement savings or short- and Cummings 2018). Job titles also planner? Or do you do something else?) term growth in retirement account asset often provide little insight into the scope values for those using a financial adviser. of services provided by the adviser, at The response to this question was However, they did note that meeting least partially due to lack of regulatory used to determine a household’s source with a financial adviser was associated requirements. of financial information.5 If multiple with setting long-term goals, calculating Identifying the scope of the advice sources were provided by the respondent, retirement needs, retirement-account engagement (i.e., the type of financial the first response provided was assumed diversification, use of supplemental adviser) can be difficult, especially when to be the primary information source. retirement accounts, retirement using well-known publicly available data- Financial advisers were classified into confidence, and higher levels of savings sets (Heckman, Saey, Kim, and Letkie- two types: financial planners and trans- in emergency funds. wicz 2016). Limited research documents actional advisers. If “financial planner” Recall that empirical evidence on the how households fare using different was the response, the financial adviser

32 Journal of Financial Planning | April 2019 FPAJournal.org Blanchett CONTRIBUTIONS was deemed to be a financial planner. If “banker” or “broker” was selected, Figure 1: Sources of Investment and Savings Information for Households Ages 25 to 55 the financial adviser was deemed to be transactional. Lawyer or accountant 100% !"#$#%"$&'(&$##)* ,#-)*#)- 3-4)* 6% responses were not included in either 768 15% ;8 7<8 13% 12% 22%7:8 =8 7=8 24% 16% financial adviser group because these are 7:8 7>8 558 not professions typically associated with 80% 7:8 5:8 598 18% 7>8 22%;78 7<8 24% financial planning. 16%7<8 968 7=8 18% The term “transactional adviser” 60% 6% 3% 40% was used, versus the actual response of 7:8 17% 7?8 31% 25% banker or broker, to reflect the likely 40% ;98 scope of services associated with the 41% 45% advice. Being a broker (or banker) and Respondents % Total 30% 23% 22% 19% a financial planner is not mutually 20% exclusive; many advisers work for a 17% 18% broker-dealer (who are technically 10% 15% 15% 15% 0% brokers) that provides comprehensive 2001 2004 2007 2010 2013 2016 financial planning services. Therefore, Survey Year the response to the question was Financial Planner Friend Internet Transactional Adviser Other assumed to be based on the nature of Source: Survey of Consumer Finances services being provided, where the advis- ers providing more holistic services were sentative of the entire U.S. population, tional analysis (not included in Table 1) referred to as “financial planners,” and yet likely better reflected the cohort was conducted where households were advisers who are less holistic in nature, of investors who would potentially be split based on respondent age—those and likely more transaction-oriented interested in working with a financial above and below the age of 40. The (e.g., broker or banker) were “transac- adviser (e.g., it is unlikely a household results were very similar for both tional advisers.” The “Internet” was also with no income and no savings would groups. One reason for the relatively considered an information source (for seek the services of a financial adviser). large growth in the use of the Internet those who selected the Internet) and a Figure 1 includes information about for this analysis was that only relatively “friends” information source was created the distribution of the use of these four young households were included (all as a combination of the “call-around” advice sources for the six waves of the are age 55 or younger). This relation and “friend/relative” responses. SCF included in the analysis. Household may not hold at all ages (e.g., respon- Instead of including all available weights were included when estimating dents over the age of 80). households, the test group was limited the percentages. The percentage of households using a to households that were assumed to be As shown in Figure 1, the Internet financial planner increased over the study potentially interested in considering appears to be displacing the “Friends” period, from 10 percent in 2001 to 18 per- financial advice, as well as those that and “Other” sources of financial infor- cent in 2016, while the percentage using would consider guidance among the five mation since 2001. For example, Friends transactional advisers remained relatively domains considered. To be included, and Internet were 45 percent and 3 unchanged. On average, approximately the respondent must have been between percent of information sources in 2001, 34 percent households were using either ages 25 and 55;6 the household must respectively, but changed to 19 percent type of financial adviser—a financial have had at least $5,000 in financial and 40 percent, respectively, by 2016. planner or a transactional adviser—over assets and retirement assets (note these This suggests that households who may the entire period. assets are not mutually exclusive); and have asked a (relatively unsophisticated) Collins (2012) noted financial advice the household had to have income friend a financial question historically usage in the U.S. was 20 percent to 33 and normal wage income above $25,000 are increasingly going online to find an percent based on different sources, while annually (again, these definitions are answer instead. Hanna (2011), using data from SCFs not mutually exclusive). All values were The growth in the use of the Internet from 1998 to 2007, noted advice usage converted to 2016 dollars. These filters has been relatively similar across age from 21 percent to 25 percent. The likely created a dataset that was not repre- groups within this dataset. An addi- reason this estimate of financial adviser

FPAJournal.org April 2019 | Journal of Financial Planning 33 CONTRIBUTIONS Blanchett

Table 1: Logistic Regression Where the Dependent Variable Is the Information Source

Financial Planner Friend Internet Transactional Adviser

Source Odds Odds Odds Odds Value Ratio Value Ratio Value Ratio Value Ratio Intercept –2.040** –2.192** 1.037 0.188 Age 0.002 1.002 -0.006 0.994 0.005 1.005 0.002 1.002 ln(Income) –0.011 0.989 0.164** 1.178 –0.341** 0.711 –0.126 0.882 ln(Financial Assets) 0.029 1.029 –0.009 0.991 0.029 1.030 0.018 1.018 Years of 0.002 1.002 –0.039** 0.961 0.107** 1.112 –0.042** 0.959 Female? 0.362** 1.436 0.068 1.070 –0.348** 0.706 0.244* 1.277 Single? –0.320** 0.726 0.118 1.125 0.072 1.074 –0.228* 0.796 Non-white/Hispanic? –0.012 0.988 –0.004 0.996 0.152* 1.164 –0.026 0.974 Notes: * Signi cant at the 5% level; ** signi cant at 1% level. use (34 percent) is higher than other data (Montalto and Sung 1996) when subjective. This analysis focused more research is because households that do running regressions using the SCF, the on household decisions (the process), not meet the previously noted income or logistic regressions in this study were versus more outcome-oriented variables asset requirements were excluded from based on a single aggregated value for (wealth or savings levels). Focusing the analysis (recall that the dataset is not each household. (Additional information on decisions reduced potential issues representative of all U.S. households, about the household-level aggregation associated with reverse causality, because rather households that are more likely to approach is provided in the analysis clients with more wealth become be investors). section.) Using a single value for each increasingly attractive to financial advis- household decreased the standard errors ers and it may be difficult to determine Who Uses Each Information Source? for the regression. the role of the financial adviser with To better understand which household The logistic regression results in Table respect to the wealth creation, as well as attributes were associated with the 1 are somewhat inconsistent with past the fact more wealth doesn’t necessarily selection of each of the four potential research exploring who uses a financial imply the household has made (or is information sources, a series of logistic adviser. For example, Burke and Hung making) optimal financial planning regressions were performed. The depen- (2015) noted that households with a decisions. dent variable for the logistic regressions financial adviser tended to be wealthier, As noted previously, the analysis used was the information source selected. The have higher incomes, be more educated, data from the 2001, 2004, 2007, 2010, independent variables were respondent older, and more financially literate 2013, and 2016 waves of the SCF. To be age, total household income, total house- (through a meta-analysis). The logistic included in the analysis, the respondent hold financial assets, respondent years of regressions shown in this analysis (Table had to be between the ages of 25 and 55, education, whether the respondent was 1) suggest that households headed by a have children, have a minimum of wage female (this is a dummy variable that female and those who are married are income and normal income of $25,000, equals one if the respondent is female), more likely to use a financial planner and at least $5,000 in financial assets whether the household was single (this or transactional adviser, but there is no and retirement assets. Five financial is a dummy variable that equals one statistically significant relation between decision-making domains were consid- if the household is not married), and age, income, or financial assets. With ered for the analysis: (1) portfolio risk whether the household was non-white respect to the use of friends as an infor- appropriateness; (2) savings habits; (3) or Hispanic (also a dummy variable). All mation source, these households tended life insurance coverage; (4) revolving values were translated in 2016 dollars. to have higher income levels but lower credit card debt; and (5) emergency sav- Only households that met the previously levels of education. For the Internet, ings. These tests are introduced below: noted criteria were included in these these households had lower levels of Portfolio risk appropriateness. regressions. The results of the logistic income, more education, and were more This test determined if the household’s regressions are included in Table 1. likely to be male and not white. retirement assets were invested in a Although it is common to use the portfolio that had a risk level that would repeated-imputation inference (RII) Analysis generally be considered prudent, given method to correct for underestimation of Determining the soundness of financial the respondent’s age. For the analysis, variances due to imputation of missing decision-making for a household is the equity level of retirement assets (e.g.,

34 Journal of Financial Planning | April 2019 FPAJournal.org Blanchett CONTRIBUTIONS

401(k)s, IRAs, etc.) was determined Figure 1:2: TkTarget Retirement Assets Equity Allocation and compared with the Morningstar® (Morningstar® Moderate Lifetime IndexSM Glide Path) Moderate Lifetime IndexSM based on the respondent’s age, assuming a retirement 100% !"# $%&'#()*+, -..#/(0123'4516#/(0123'4 age of 65. 78 9:;88< =88;88< >?;88< To be considered prudently invested, 7= 9:;88< =88;88< >?;88< the equity level must be within 25 80% 77 9:;88< =88;88< >?;88< 7: 9:;88< =88;88< >?;88< percentage points of the Morningstar 7@ 9:;88< =88;88< >?;88< Moderate Lifetime Index (25 points 60% 7A 97;99< =88;88< >B;99< above or below the glide path, bounded 7> 97;9B< =88;88< >B;9B< 40% 7B 97;9@< =88;88< >B;9@< by 100 percent and 0 percent, respec- 7? 97;?9< =88;88< >B;?9< 79 97;?:< =88;88< >B;?:< tively). The glide path, or equity target, Equity Allocation 20% :8 97;BB< =88;88< >B;BB< for the Morningstar Moderate Lifetime := 97;>B< =88;88< >B;>B< Index and the respective upper and :7 97;AA< =88;88< >B;AA< 0% lower bounds targets are included in :: 97;@B< =88;88< >B;@B< 20 :@ 25 97;:?30< =8835;88< 40>B;:?<45 50 55 60 65 70 Figure 2. :A 97;=?< =88;88< >B;=?Age< :> 9=;97< =88;88< >>;97< This was effectively a test that the Glide Path Upper Bounds Lower Bounds :B 9=;>B< =88;88< >>;>B< portfolio was diversified and reasonably consistent with a general target risk level given the investor’s age. Only retirement habits were the focus, versus the amount is highly unlikely to achieve through assets were considered because these are of savings, to simplify the analysis and investing in the markets (especially on a typically savings directed toward a single because of SCF data limitations related risk-adjusted and after-tax basis). There- goal (retirement) with a relatively similar to savings variables. fore, it was assumed the households begin date (approximately age 65). Life insurance coverage. This should not maintain any revolving credit There will of course be situations where domain focused on whether the balances. If the household did maintain the allocations should deviate from the household had face value life insurance any revolving credit, it was assumed the target; therefore, this was viewed more at least equal to the total wage income household was making a poor decision in as a general test to ensure the household of the household. All households in this this domain. had their retirement assets invested in a analysis had children; therefore, it is Emergency savings. The final test reasonable manner. The 25-point band reasonable to assume that some level explored whether the household had created a relatively wide range that would of life insurance would be desirable adequate emergency savings. This include virtually every target-date mutual for most households. The ideal level of was calculated by dividing total liquid fund family series in the U.S. market. coverage was not estimated. The vast savings—which included balances in Savings habits. This test focused on majority of households in this dataset checking accounts, savings accounts, whether the household had a savings should have more than the relatively low mutual funds, and money plan in place. The specific text of the threshold of just one times wage income. market demand accounts—by average SCF question was: “Which of the This test showed whether the household normal income. The goal was to have at following statements on this page comes had thought about life insurance enough least three months income set aside in closest to describing your (and your to make even a relatively de minimis emergency savings. husband/wife/partner’s) saving habits?” purchase. It is likely that some house- Each household in the SCF had five There were six potential responses such holds may have no need life insurance; implicates, or observations. Each of as not saving at all, saving whatever is however, the analysis controlled for basic the five tests were conducted for each left over at the end of the month, or demographic data and the target was a implicate, resulting in 25 total tests for some type of savings plan (e.g., saving relatively low threshold. a household. The results of the test for the income of one family member, Revolving credit card debt. This each implicate were combined, based on saving non-regular income, and a regular question focused on whether the implicate weights, to get a “pass rate” for savings program). For this analysis, so household had any revolving credit card the respective domain. Pass rates ranged long as the household had some type of debt at the end of the month. Interest from 0 percent, where none of the savings plan in place, it was considered rates on credit cards typically exceed implicates passed, to 100 percent, where to have good savings habits. Savings 15 percent7—a “return” the household all the implicates passed. The scores at

FPAJournal.org April 2019 | Journal of Financial Planning 35 CONTRIBUTIONS Blanchett

addition, the four sources of financial Figure 1:3: TkPercentage of Households Passing the Respective information were included as dummy Financial Planning Test by SCF Year variables, which is whether the house- hold financial information source was a 80 !"#$ %&$'(&)*&+/#,*0-*. 12-+3#45**'(-"+61-7$#91$8"":*'+9#$:+<;=""4$'2"18>+/#0*12- ?@@A BCDECFAG GEDEAH?F FGDE@IEC FIDCFFIF FCD?AFE financial planner, a friend, the Internet, ?@@B F@D@?F??GFDFIBB? FED?BEHC FBDB?IFE FFDB@?@B or a transactional adviser. For each 70 ?@@H BIDFF?A? GHDIHC?? GADEFFIE F@DICIHB FGDH@BFF information source, the coefficient was ?@A@ BIDIHACA FGDC@@IBGCD@GIEI FBDFCCEG F@DBEC@C set to equal 1 if the household used that 60 ?@AC F?DA@AEG?DHCEGA GIDA?BA FHDABHB FHDHHFB ?@AG BIDAI?EF GFDCIFI? GBDBCB@HBEDFAF@FFIDIEHHH information source, otherwise it was 50 zero. Weights for each household were included in logistic regressions. The 40 results are included in Table 2. The sign and statistical significance 30 of the coefficients varied by test. The coefficients for years of education and 20

Percentage of Households Meeting Target Meeting of Households Percentage Portfolio Savings Life Credit Card Emergency financial assets were always positive Risk Habits Insurance Debt Savings and significant for four of the five tests 2001 2004 2007 2010 2013 2016 (all but portfolio risk). This suggests households with more education and the individual domain level were then largely binary. For example, the percent- more financial assets tend to make better averaged to get the aggregate financial age of households where all implicates financial decisions. soundness score for the household. either passed or failed the individual The coefficient for age was negative Demographic control variables (e.g., metric ranged from 76.8 percent (for (and statistically significant) for those total financial assets) are also created portfolio risk) to 99.9 percent (for same four domains (all but portfolio using the implicate weights (i.e., the the savings test). Therefore, given the risk), which suggests older households weighted average of the implicate relatively binary nature of the individual are making worse decisions; however, values for that household) so that each results, the values were transformed and the odds ratio was not that different from household had a single set of values. a logistic regression was performed. 1, which implies the economic impact Figure 3 provides insight into the For the logistic regression, for each of age is relatively low. There was quite percentage of households that passed the domain, a value of 1 was assigned a bit of inconsistency across some of the respective tests for each of the six SCF for that test if the pass rate for the other variables. For example, the sign datasets included in the analysis. household was 50 percent or greater, and statistical significance of the income In Figure 3, most of the tests (except otherwise, it was assigned a value of variable varied across domains. for savings habits) have approximately a zero. Note, this transformation was only The financial planner coefficients 50 percent pass rate. This was somewhat performed for individual domain tests. were the most positive for all but the intentional to ensure there was disper- The aggregate values were much more credit card metric (where it was second), sion in each domain across households varied; only 9 percent of households but only statistically significant for (i.e., all households were not passing or had a score of zero or 1. Therefore, this three of the five domains. The Internet failing for a given domain). The fact that transformation was not necessary when coefficients were the second best for all only approximately half of households reviewing the aggregate results. but the credit card metric (where it was passed each test would suggest there Similar to the logistic regressions first), while the transactional adviser and is a large potential benefit for financial exploring information source usage in friend coefficients were generally the advisers to help the households make Table 1, a number of independent vari- worst or second worst coefficients. better financial decisions. ables were included in these next logistic For the portfolio risk appropriateness regressions, including age, household domain logistic regression, only the Results income, total household financial assets, financial planner coefficient was positive Regarding results for the five individual respondent years of education, whether and statistically significant. This suggests planning domains, while there was the respondent is female, whether the the probability of having a portfolio that significant variation in the aggregate household is single, and whether the is even generally consistent with age was results, the individual metrics were household is non-white or Hispanic. In higher if the household used a financial

36 Journal of Financial Planning | April 2019 FPAJournal.org Blanchett CONTRIBUTIONS

Table 2: Individual Metric Logistic Regression Results

Portfolio Risk Savings Habits Life Insurance Credit Card Debt Emergency Savings Odds Odds Odds Odds Odds Value Ratio Value Ratio Value Ratio Value Ratio Value Ratio Intercept –1.725 ** –8.605 ** 1.601 ** –3.043** –1.148* Age 0.003 1.003 –0.032 ** 0.969 –0.007 * 0.993 –0.019** 0.981 –0.038** 0.963 ln(Income) 0.059 1.061 0.448 ** 1.565 –0.254 ** 0.776 –0.086 0.918 –0.749** 0.473 ln(Financial Assets) 0.041 1.042 0.496 ** 1.642 0.220 ** 1.246 0.356** 1.427 0.917** 2.502 Years of Education 0.017 1.017 0.071 ** 1.074 0.056 ** 1.058 0.067** 1.070 0.103** 1.108 Female? 0.094 1.099 –0.021 0.979 0.119 1.126 –0.364** 0.695 –0.044 0.957 Single? –0.018 0.982 0.318 ** 1.374 –0.989 ** 0.372 0.399** 1.491 –0.029 0.972 Non-white/Hispanic? 0.005 1.005 0.141 1.152 0.017 1.017 –0.172** 0.842 –0.159** 0.853 Financial Planner? 0.203 * 1.225 0.082 1.085 0.261 ** 1.298 –0.031 0.970 0.202* 1.224 Friend? –0.054 0.947 –0.074 0.928 0.093 1.097 –0.146* 0.865 0.007 1.007 Internet? 0.087 1.091 –0.068 0.935 0.209 ** 1.233 –0.006 0.994 0.144 1.155 Transactional Adviser? –0.003 0.997 –0.310 ** 0.733 0.118 1.125 –0.102 0.903 –0.051 0.951 Notes: * signi cant at the 5% level; ** signi cant at 1% level.

Table 3: Average Financial Planning Score OLS Regression Results

Model 1 Model 2 Model 3 Model 4 Coe t stat Coe t stat Coe t stat Coe t stat Intercept 21.445** 4.340 58.024 ** 90.244 55.044** 37.922 20.273** 4.068 Age 0.319** –10.545 0.071* 2.290 –0.322** –10.643 ln(Income) –3.720** –7.569 –3.611** –7.344 ln(Financial Assets) 7.424** 31.774 7.400** 31.723 Years of Education 1.250** 11.125 1.210** 10.755 Female? –0.781 –0.898 –0.776 –0.893 Single? –2.613** –3.588 –2.536** –3.486 Non-white/Hispanic? –1.022 –1.889 –1.072* –1.983 Financial Planner? 3.405 ** 3.715 3.383** 3.691 3.087** 3.754 Friend? –0.636 –0.805 –0.625 –0.791 –0.341 –0.481 Internet? 2.283 ** 2.663 2.281** 2.660 2.014** 2.613 Transactional Adviser? –1.266 –1.462 –1.280 –1.480 –0.549 –0.707 Observations 8,078 8,078 8,078 8,078 R² 19.91% 0.56% 0.63% 20.28% Adjusted R² 19.84% 0.51% 0.57% 20.17% Notes: * signi cant at the 5% level; ** signi cant at 1% level. planner, but effectively random for the performed, each including different sets income potentially warrant greater study, other information sources. of available independent variables. The given that both are typically positively Individual test results may be interest- results of the OLS regressions are shown associated with financial sophistication. ing; however, the aggregate financial in Table 3. The base demographic variables (e.g., soundness metric was the primary focus Households that make better financial Model 1) explained a significant degree of this analysis. For this, an ordinary decisions tended to be younger, have more of household financial soundness least squares (OLS) regression was per- lower incomes, more financial assets, than the financial information source formed where the dependent variable higher levels of education, have a male (e.g., Model 2), as evidenced by the R² was the average pass rate across the five respondent, are married, and are white. values (see Table 3). This suggests while domains for each household. The same The most significant variables were the source of financial information is independent variables as past regres- financial assets and years of education, important, other household attributes sions were included in these regressions, both of which had positive coefficients were materially more so. and the regressions include household (which is consistent with past research). Households that used financial weights. Four separate regressions were The negative coefficients for age and planners as their financial information

FPAJournal.org April 2019 | Journal of Financial Planning 37 CONTRIBUTIONS Blanchett

scored almost 5 percent higher than the Figure 1:4: TkRelative Aggregate Financial Planning Score Across SCF average in 2001, while households using the Internet in 2016 scored 2 percent 6 below average (which was the worst among the four sources considered). 4 There are a variety of potential reasons for this. One may be that the benefits 2 associated with the Internet were due largely to “early adopters,” and as usage 0 increased, the caliber and intentions of SCF Year –2 2001 2004 2007 2010 2013 2016 Average Internet users have declined. This gets to Score vs. Average vs. Score 0.29 3.16 2.23 3.24 1.43 1.04 1.90 the fundamental issue around selection Di erence in Aggregate Di erence –4 -1.94 -2.42 0.51 -2.77 -1.39 -0.10 -1.35 bias that is difficult to control for in this 2001 5.120040 -0.44 2007-0.17 1.86 20100.34 -1.201334 0.89 2016 type of analysis. This topic is also likely -3.45 -0.29 -2.58 -2.34 -0.38 0.40 -1.44 Financial Planner Friend Internet Transactional Adviser worth exploring in future research.

Implications for Financial Advisers source were making the best decisions a transactional adviser were doing one The results of this analysis are consistent of the groups studied, followed by those thing really well and everything else with the growing body of research that using the Internet. Households that were poorly; they were doing everything working with a financial adviser can using a transactional adviser were mak- poorly. It’s possible there are aspects of result in better outcomes, as well as ing the worst decisions, and households households that selected transactional empirical research suggesting that finan- using friends were the second worst. advisers that were not controlled for in cial advisers can actually make some It cannot be concluded that working this analysis or other areas where they households worse off. How it is possible with a financial planner is the reason improve outcomes affecting these results. that financial advisers can both help and those households were making better Future research may provide clarity here. hurt their clients? This is largely due financial choices due to potential A secondary analysis was performed to the relatively ambiguous nature of selection bias; the outcome could be to see how the aggregate scores have the term “financial adviser.” Financial endogenous to the selection of the changed across the four information advisers can provide significantly differ- information source. However, these sources. This analysis was similar to ent scopes of services and advisers can findings at least imply that working with the information in Figure 3; however, be compensated in myriad ways. This a financial planner can help households instead of looking at the individual heterogeneity creates significant issues make better financial decisions, while results, this analysis compared the aver- when attempting to empirically assess working with a transactional adviser may age aggregate score for each household, the “value” of financial advice. actually result in worse decisions. based on household information source, These findings strongly suggest that It is not known why households work- and then controlled for the SCF year. financial advisers who focus on financial ing with a transactional adviser were This approach ensured the average score planning are having a positive impact making the worst decisions of the groups among the four sources for each SCF was on households, especially compared to studied; although one can speculate. zero. The results are shown in Figure 4. financial advisers that are more trans- One possibility is that these households Overall, the time varying results in actional in nature. These results should may have a false sense of confidence Figure 4 are relatively similar to the not be misconstrued to suggest financial about their financial soundness because regression results shown in Table 3, advisers cannot provide value if they are they get advice in a few domains and where the financial planner values paid primarily through commissions, or think that are covered in all domains were consistently the highest, and the that certain types of adviser registration when they are, in fact, not. One problem transactional adviser and friend were methods are worse than others. What with this hypothesis is that households typically and consistently the lowest. matters are the services being provided working with a transactional adviser Note, however, the reduction the aver- to the client and consequently how were doing the worst in effectively every age score among households that used the client perceives the nature of the domain considered. In other words, the Internet. Households using the Inter- relationship. Helping clients accomplish it’s not that households working with net as the primary information source goals typically requires more than just

38 Journal of Financial Planning | April 2019 FPAJournal.org Blanchett CONTRIBUTIONS selling a product, such as a mutual fund made the worst decisions. does appear to be declining over time. or annuity—it requires a financial plan It cannot be concluded that using a The potential value of the Internet as with ongoing management. Financial financial planner entirely explains better a source of financial information and advisers that provide these services are decision-making of those households advice is notable given the significant not likely to be described as transac- due to implications around selection increase in usage over the last 15 years or tional in nature; rather they are likely be bias. However, these findings do suggest so, especially if its role as an information described as financial planners. that the potential value associated with source continues to increase into the working with a financial adviser could future. Conclusions differ significantly by adviser type. This paper explored the quality of five These findings also have important Endnotes household financial planning decisions implications for future research exploring 1. See the SEC’s “Investor Bulletin: Top Tips for (portfolio risk level, savings habits, life the value of financial advice, especially in Selecting a Financial Professional,” posted August insurance coverage, revolving credit card an empirical setting. Any kind of analysis 25, 2016 at sec.gov/investor/pubs/invadvisers.htm. balances, and emergency savings) across that focuses primarily on transactional 2. See CFP Board professional demographics data four information sources (financial advisers may significantly different at cfp.net/news-events/research-facts-figures/ planners, transactional financial advisers, conclusions on the value of financial cfp-professional-demographics. friends, or the Internet). The quality of advice than one focused on advisers that 3. See The American College of Financial Services household decisions was found to vary are comprehensive. data at theamericancollege.edu/designations- across information sources. Households Additionally, there is significant evi- degrees/ChFC-CFP. using a financial planner made the best dence that households using the Internet 4. The other is the National Longitudinal Study of decisions, followed by the Internet. are making better-than-average financial Youth (bls.gov/nls/home.htm). Households using a transactional adviser planning decisions, although the benefit 5. A similar question in the SCF asks about

Financial Planning for Cancer Program

Connecting Families Battling Cancer to Free, Quality Financial Advice Funded through a grant from the Foundation for Financial Planning (FFP), FPA is excited to collaborate with FFP and their grantees, Family Reach and Consumer Education and Training Services (CENTS), to engage FPA members in helping families facing cancer manage the financial challenges associated with treatment. We’re proud of our volunteer members who demonstrate such heart, compassion and a commitment to transform lives through the power of financial planning, and hope you will consider joining them. To sign up to volunteer or to learn more, visit OneFPA.org/Cancer or contact Kurt Kaczor at KKaczor@ OneFPA.org.

Financial Planning for Cancer Partners

Photo Courtesy of Family Reach

FPAJournal.org April 2019 | Journal of Financial Planning 39 CONTRIBUTIONS Blanchett

sources of information for borrowing money or Maximize? Evidence from their Brokers’ Incen- paper/k7f7zFKd. obtaining credit. This analysis only considered tives.” Journal of Finance 68 (1): 201–235. Lusardi, Annamaria, and Olivia S. Mitchell. 2014. “The the savings and investment question because the Collins, J. Michael. 2012. “Financial Advice: A Economic Importance of Financial Literacy: Theory majority of tests cover only these domains. Addi- Substitute for Financial Literacy?” Financial and Evidence.” Journal of Economic Literature 52 (1): tionally, the household sources vary across the Services Review 21 (4): 307–323. 5–44. two questions, which would create additional Del Guercio, Diane, and Jonathan Reuter. 2014. Marsden, Mitchell, Cathleen D. Zick, and Robert classification groups that would complicate the “Mutual Fund Performance and the Incentive to N. Mayer. 2011. “The Value of Seeking Financial analysis. Invest in Active Management.” Journal of Finance Advice.” Journal of Family and Economic Issues 32 (4): 6. This specific filter is important later in the 69 (4): 1,673–1,704. 625–643. analysis when determining life insurance coverage Engelmann, Jan, C. Monica Capra, Charles Noussair, Martin, Terrance, and Michael Finke. 2014. “A adequacy as well as savings habits. and Gregory Berns. 2009. “Expert Financial Comparison of Retirement Strategies and Financial 7. See current credit card data at Advice Neurobiologically ‘Offloads’ Financial Planner Value.” Journal of Financial Planning 27 (11): bankrate.com/finance/credit-cards/current- Decision-Making Under Risk.” PLoS one 4 (3): 46–53. interest-rates.aspx. e4957. Montalto, Catherine Phillips,­ and Jaimie Sung. 1996. Finke, Michael, Sandra Huston, and William Waller. “Multiple Imputation in The 1992 Survey Of References 2009. “Do Contracts Impact Comprehensive Consumer Finances.” Journal of Financial Counseling Barber, Brad M., Yi-Tsung Lee, Yu-Jane Liu, and Financial Advice?” Financial Services Review 18 and Planning 7: 133–147 Terrance Odean. 2008. “Just How Much Do (2): 177–194. Mullainathan, Sendhil, Markus Noeth, and Antoinette Individual Investors Lose by Trading?” Review of Grable, John E., and Swarn Chatterjee. 2014. “Reducing Schoar. 2012. “The Market for Financial Advice: Financial Studies 22 (2): 609–632. Wealth Volatility: The Value of Financial Advice as An Study.” National Bureau of Economic Bergstresser, Daniel, John Chalmers, and Peter Measured by Zeta.” Journal of Financial Planning 27 Research Working Paper No. w17929. Available at Tufano. 2009. “Assessing the Costs and Benefits (8): 45–51. nber.org/papers/w17929. of Brokers in the Mutual Fund Industry.” Review of Grinblatt, Mark, and Matti Keloharju. 2000. “The Odean, Terrance. 1998. “Are Investors Reluctant to Financial Studies 22 (10): 4,129–4,156. Investment Behavior and Performance of Various Realize their Losses?” The Journal of Finance 53 (5): Bhargava, Vibha, and Jean M. Lown. 2006. “Pre- Investor Types: A Study of Finland’s Unique Data 1,775–1,798. paredness for Financial Emergencies: Evidence Set.” Journal of Financial Economics 55 (1): 43–67. Raskie, Sterling, Jason Martin, Craig Lemoine, and from the Survey of Consumer Finances.” Financial Hackethal, Andreas, Michael Haliassos, and Tullio Benjamin Cummings. 2018. “The Value of Financial Counseling and Planning 17 (2): 17–26. Jappelli. 2012. “Financial Advisers: A Case of Designations: A Consumer Perspective.” Journal of Blanchett, David, and Paul Kaplan. 2013. “Alpha, Babysitters?” Journal of Banking and Finance 36 Financial Planning 31 (6): 44–54. Beta, and Now… Gamma.” Journal of Retirement 1 (2): 509–524. Scott, Janine K., and Michael S. Finke. 2013. “The (2): 29–45. Hanna, Sherman. 2011. “The Demand for Financial Demand for Disability Insurance.” Financial Services Burke, Jeremy, and Angela Hung. 2015. Do Financial Planning Services.” Journal of 10 Review 22 (1): 1–12. Advisers Influence Savings Behavior? RAND (1): 36–62. Shapira, Zur, and Itzhak Venezia. 2001. “Patterns of Corporation e-book. Available at rand.org/pubs/ Hanna, Sherman and Suzanne Lindamood. 2010. Behavior of Professionally Managed and Indepen- research_reports/RR1289.html. “Quantifying the Economic Benefits of Personal dent Investors.” Journal of Banking and Finance 25 Campbell, John. “Household Finance.” 2006. Journal Financial Planning.” Financial Services Review 19 (2): (8): 1,573–1,587. of Finance 61 (4): 1,553–1,604. 111–127. Shefrin, Hersh, and Meir Statman. 1985. “The Disposi- Chalmers, John, Woodrow Johnson, and Jonathan Heckman, Stuart, Martin Seay, Kyoung Tae Kim, and tion to Sell Winners too Early and Ride Losers too Reuter. 2014. “The Effect of Design Jodi Letkiewicz. 2016. “Household Use of Financial Long: Theory and Evidence.” Journal of Finance 40 on Employer Costs and Employee Retirement Planners: Measurement Considerations for (3): 777–790. Choices: Evidence from Oregon.” Journal of Public Researchers.” Financial Services Review 25: 427–446. Warschauer, Thomas, and Donald Sciglimpaglia. 2012. Economics (116): 17–34. Kinniry, Francis M., Colleen M. Jaconetti, Michael “The Economic Benefits of Personal Financial Cho, Soo Hyun, Michael Gutter, Jinhee Kim, and A. DiJoseph, and Yan Zilbering. 2014. “Putting Planning: An Empirical Analysis.” Financial Services Teresa Mauldin. 2012. “The Effect of Socialization a Value on Your Value: Quantifying Vanguard Review 21 (3): 195–208. and Information Source on Financial Manage- Adviser’s Alpha.” Vanguard White Paper. Available at ment Behaviors Among Low-and Moderate- vanguard.com/pdf/ISGQVAA.pdf. Citation Income Adults.” Family and Consumer Sciences Linnainmaa, Juhani, Brian Melzer, Alessandro Blanchett, David M. 2019. “Financially Sound Research Journal 40 (4): 417–430. Previtero, and Stephen Foerster. 2018. “Financial Households Use Financial Planners, Not Christoffersen, Susan, Richard Evans, and David K. Advisers and Risk-Taking.” Working paper. Avaial- Transactional Advisers.” Journal of Financial Musto. 2013. “What Do Consumers’ Fund Flows able at aeaweb.org/conference/2018/preliminary/ Planning 32 (4): 30–40.

40 Journal of Financial Planning | April 2019 FPAJournal.org