Client Involvement in Expert Advice Antibiotics in Finance?

Transcription

1 Client Involvement in Expert Advice Antibiotics in Finance? Andreas Hackethal, Christine Laudenbach, Steffen Meyer, and Annika Weber May 19, 2017 ABSTRACT We use minutes from 19,000 financial advisory sessions at a large German bank to measure how client-advisor interaction, and especially the degree of active client involvement, affects portfolio outcomes. Nearly forty percent of sample clients trade not only through their advisor but also independently, and it is especially them who introduce own investment ideas to their advisor and also otherwise intervene in the standardized advisory process. We find that client involvement increases the likelihood that bank advisors recommend single stocks instead of equity mutual funds and that advisor recommendations carry a home bias. We show that both effects reduce portfolio diversification without improving after-cost portfolio performance. Our findings parallel the phenomenon of doctors prescribing antibiotics catering to client requests, even if inappropriate. We thank Roman Inderst, Thomas Mosk, Jenny Pirschel, Matthias Rumpf, Nate Vellekoop as well as seminar participants at the University of Mannheim, the SAFE Internal Research Conference, Bad Homburg, the SAFE Household Finance Workshop, Ruedesheim, and the SMYE 2017, Halle, for valuable comments. All errors are our own. Goethe University Frankfurt. Corresponding author: Annika Weber, annika.weber@hof.uni-frankfurt.de Leibniz University Hanover

2 Numerous medical studies find a significant share of antibiotics prescriptions to be inadequate, exposing patients to adverse side effects and ultimately contributing to the development of resistances. Doctors cite patient requests as one of the reasons for inadequate prescriptions, for example in case of viral illnesses, for which antibiotics provide no benefit. 1 Rather than educating patients on their erroneous prior, medical experts may choose to cater to patient demand in an attempt to safeguard the patient relationship. Hence, laymen advisees may exert negative impact on the quality of professional advice. In this paper, we ask whether these general phenomena carry over to individual investors and their interactions with professional financial advisors. Our main contribution is to demonstrate that part of the variation in advised clients portfolio composition can be attributed to specific client requests and advisors respective responses. Neither regulators nor existing studies on the quality of financial advice have so far zoomed in on the role of client involvement in the advisory process, mostly because available data sets do not allow to disentangle client and advisor effects. Rather, any observed difference between advised and self-directed accounts is typically fully attributed to the supply side. 2 Yet, first evidence suggests that clients may take important influence on the process and ultimately on the outcomes of financial advice. Bhattacharya, Hackethal, Kaesler, Loos, and Meyer (2012) show that many individual investors are unwilling to follow even unbiased advice. Campbell (2016) warns that this reluctance of individual investors to follow advice may reduce the quality of advice they receive as, in order to close a deal, advisors may decide to go along even with unsound client requests. In fact, experimental studies on general advice taking show that advisees tend to overvalue their own opinion relative to advisor input and discount the advice received the more, the more distant it results from their own priors (Yaniv, 2004; Yaniv and Milyavsky, 2007; Sniezek and Buckley, 1995). Relatedly, Guiso and Jappelli (2006) find that investors who are overconfident with regard to their private information are less willing to rely on financial advisors and are more likely to collect information on their own. Borgsen, Glaser, and Weber (2012) report that clients approaching their advisor with the motivation to obtain reassurance for an own investment idea are less likely to follow along with the recommendations received. As a consequence, expert advisors may choose not to argue against the beliefs of clients, regardless of their merit. In line with this hypothesis, Agnew, Bateman, 1 Dekker, Verheij, and van der Velden (2015) for instance find that overprescribing was highest in cases where patients explicitly expected to obtain antibiotics to alleviate their suffering. See also Roberts, Albert, Johnson, and Hicks (2015), Shapiro, Hicks, Pavia, and Hersh (2014), and references therein. 2 Empirical evidence from individual investor data (Hoechle, Ruenzi, Schaub, and Schmid, 2016a,b; Chalmers and Reuter, 2015; Hackethal, Haliassos, and Jappelli, 2012a; Karabulut, 2013) and fund-level data (Christoffersen, Evans, and Musto, 2013; Christoffersen et al., 2013; Bergstresser, Chalmers, and Tufano, 2009) suggests that brokers and financial advisors take negative impact on their clients portfolio performance. Inderst and Ottaviani (2009, 2012a,b) theoretically model the principal-agent problem inherent to the advisor-advisee relationship that is aggravated by financial inducements paid to advisors by product providers. Some studies find a soothing effect of financial advice on most common behavioral biases such as underdiversification (Hoechle et al., 2016b; Gaudecker, 2015; Hackethal et al., 2012a), home bias (Kramer, 2012; Bluethgen, Gintschel, Hackethal, and Mueller, 2008) and the disposition effect (Hoechle et al., 2016b; Shapira and Venezia, 2001). Other studies find that advice consumption helps to achieve tax-efficient allocations (Cici, Kempf, and Sorhage, 2017), improves general financial planning (Finke, 2013; Lusardi and Mitchell, 2011), or encourages risk-taking (Gennaioli, Shleifer, and Vishny, 2015). 1

3 Eckert, Iskhakov, Louviere, and Thorp (2017) show in an online video experiment that a catering strategy, in which advisors provide advice on easy topics in line with anticipated client expectations, positively influences perceived advisor trustworthiness and competence, irrespective of the quality of subsequent recommendations. Mullainathan, Noeth, and Schoar (2012) and Anagol, Cole, and Sarkar (2017) provide evidence from audit studies that financial advisors cater to the beliefs and existing financial strategies of their clients in order to safeguard current and future business. In the model by Gennaioli et al. (2015) financial advisors have strong incentives to pander to their clients irrational wish to invest in overvalued assets as this generates higher fees. We exploit a unique dataset containing the minutes on 18,787 financial advisory sessions between bank advisors and their 7,505 clients at a large German branch-based bank from 2010 through Written minutes of financial advisory sessions are mandatory for all financial advisors in Germany since January These minutes document, besides other client-related information, the financial products discussed during the session and the final investment recommendations of the advisor along with his corresponding justifications. The internal guidelines of the sample bank ask advisors to match their clients with actively managed mutual funds preselected by the central investment committee of the bank. However, the advisor is free to discuss any other investment and product ideas brought forward by the client on own initiative. At the end of the session, the two parties jointly decide which products best respond to the client s needs. Hence, both input by the advisor and the client may impact final recommendations. We link the information from the advisory minutes to administrative data on the complete history of client portfolio holdings, security transactions, and the degree to which clients make use of financial advice. This comprehensive data allows us to assess the degree of client involvement in the advisory process and its implications for advised portfolio outcomes. We use the term involvement to refer to an advised client s own dedication to his personal finances. Involvement subsumes the collection of own information and the development of own research ideas and strategies, which a client may present to his advisor for validation or, despite of using advice on other transactions, execute independently. 3 Clients more involved in their financial decision-making are expected to discuss more actively with their financial advisor, and challenge any advice received more against own priors regarding criteria for successful investments. We also hypothesize that advisors may feel a stronger need to pander to more involved clients so that such clients do not turn away and place their trades on their own. We use two different measures for client involvement. First, we proxy client independence in financial decision-making by the share of self-directed purchases in all purchases conducted during the sample period. Clients with a higher share of independent transactions are assumed to be more involved. Our second measure of client involvement, labeled client ideas is more precise in nature. Using textual analysis of the justifications advisors provide for each single recommendation, we identify advisor recommendations that actually trace back to a client idea or specific request (e.g. The client seeks to invest in funds investing in stocks from South Korea or Indonesia or The client brought up this specific fund 3 Note that all clients in our sample make use of financial advice, however to a varying degree. 2

4 idea himself ). We assume that clients with own ideas are generally more financially involved. The two measures are positively but far from perfectly correlated. We start out from the observation that there is a large heterogeneity in the structure of advised (equity) portfolios and investigate whether client involvement can explain this heterogeneity beyond traditional factors such as client risk attitude and portfolio volume, and factors that have been recently identified as relevant, such as unobserved advisor heterogeneity (Foerster, Linnainmaa, Melzer, and Previtero, 2016). We find that client involvement hardly contributes to the explanation of the equity share in client portfolios, but that both our measures of client involvement have a large effect on the composition of equity portfolios. Equity portfolios of more involved clients exhibit a greater home bias and also a larger share of single stocks as opposed to equity funds. A ten percentage point increase in the share of trades that are conducted without an advisor goes along with a domestic equity share (single stock share) that is 2.7 (4.0) percentage points higher compared to average base rates of 25 (21) percent. In a second step, we run regressions at the individual recommendation-level in order to investigate whether advisors pander to existing portfolio structures of clients. We find that after controlling for numerous client characteristics and advisor fixed effects, client consented advisor recommendations differ substantially depending on ex ante equity portfolio composition and also on client involvement: advisors are more likely to recommend single stocks and domestic instruments to clients whose portfolios already exhibit a higher share in these products before the meeting. For instance, a ten percentage point increase in the ex ante portfolio share for single stocks raises the probability of single stock recommendations by one percentage point - a large effect compared to the share of single stocks in all equity recommendations of just six percent. Interaction terms indicate that this kind of pandering is even more pronounced for more involved clients. Hence, the observed differences in portfolio structures across clients of different involvement do not only result from independent trades but also from advisor suggestions responding to clients existing investment priors, just as in the antibiotics example above. In a third step we asses the impact of clients active involvement in the process of advised financial decision-making on the effectiveness of financial advice. We find that both advised clients self-directed transactions as well as direct client involvement in the process of financial advice reduce the diversifying effect of advice. However, more independent clients and clients introducing own ideas save significantly on annual portfolio fees. More, if the more involved clients exhibit superior skill in security picking or market timing, higher returns might compensate for the observed higher degree of underdiversification that comes with client involvement. However, we find no evidence that more involved clients achieve better investment results compared to their less involved peers, who rely more on advice. Client involvement is not associated with higher after-fee portfolio Sharpe ratios and therefore does not improve portfolio efficiency in terms of return realized per unit of risk taken. Our findings contribute to recent research providing empirical and experimental evidence in support of catering strategies applied by (financial) advisors. Of course, we do not doubt that 3

5 financial incentives play an important role in shaping advisor recommendations. Also, our results may coexist with the influence of personal advisor preferences in shaping advice within less standardized advisory models. However, we raise attention to important demand side effects in financial advice. Our research furthermore adds to the discussion of the multiple roles played by financial advisors. A number of recent studies alludes to the fact that improved portfolio performance might be only part of what clients expect advisors to deliver. Use of financial advisors in addition may have psychological value for different client types (Gennaioli et al., 2015; Del Guercio and Reuter, 2014; Pagel, 2016). Rather than merely assisting in eliciting a client s risk preferences and in building a portfolio, advisors may also serve as scape goats or sounding boards to validate opinions and ideas of their more independent and active clients (Chang, Solomon, and Westerfield, 2016; Borgsen et al., 2012). Less independent clients may above all appreciate the peace of mind they obtain from their advisor. Such peace of mind might spur their market participation and improve their investment outcomes compared to a counterfactual portfolio without any advice even if the advice those passive clients obtain is highly standardized and occasionally self-serving (Gennaioli et al., 2015; Chalmers and Reuter, 2015). A. Institutional Framework I. Institutional Framework and Data Using data from one of Germany s largest retail branch banks, we shed light on the process of bank financial advice as it is common also in many other European economies such as Sweden, Italy or France. Other than in the US where financial advice to retail clients primarily is provided by independent financial advisors, banks dominate the market for retail financial advice in these countries. 4 In fact, the great majority of German households takes financial decisions with the input of their house bank financial advisor. 5 At the sample bank, an advisor is assigned to each client who opens a bank account. Bank customers may conduct their financial transactions independently or optionally can consult their advisor first. The median retail client in our sample consults with his financial advisor not more often than once a year, which is in line with national industry averages (Eurogroup Consulting, 2013). Complete delegation of financial decisions is rare in Germany and mostly restricted to the upper end of customers in terms of wealth (Hackethal, Inderst, and Meyer, 2012b). Within a client segment, advisors at a branch are randomly assigned to clients, but then are meant to be their long-term designated contact person at the bank. Advisor changes are possible yet according to 4 With a high importance of independent financial advisors, the United Kingdom marks an exception among European countries. 5 Figures for German households relying on financial advice range from 60 to 80 percent across studies (Burke and Hung, 2015; Bundesverband Deutscher Banken, 2009; Chater, Huck, and Inderst, 2010), depending on the definition of the sources of advice. In addition, 66% of individual stocks, 72% of mutual funds and 72% of bonds sold to individual investors are distributed by banks (Chater et al., 2010). Participation in financial assets outside retirement accounts is low in Germany. Only 12% of German respondents report to hold stocks or bonds, 10% report to hold investment funds in the Eurobarometer Survey in

6 the bank mostly result from an advisor leaving the branch or resigning. As a result of the random allocation procedure, a single advisor usually has a quite heterogeneous client portfolio, which is exemplarily illustrated by the wide dispersion of the portfolio value in an advisor s client portfolio (see Panel A of Figure 1). [Insert Figure 1 about here] Sample advisors are full-time employees of the bank who have completed a three-year vocational training; they are paid a fixed wage in accordance with the collective wage agreement of the banking industry. Variable components must not amount to more than ten percent of total salary and typically base on team or branch performance, acquisition of new client assets, as well as surveyed client satisfaction. 6 However, although advisors are not directly compensated for their individual performance, career concerns may provide indirect additional incentives (Hoechle et al., 2016a). Advised bank customers pay for financial advice mostly indirectly through product fees (Hackethal et al., 2012b). This is in spite of efforts in other countries such as the United Kingdom and the United States to abolish inducements. The events of the global financial crisis have put consumer financial protection to the top of policymakers agenda. In consequence, retail financial advisors in Germany had to adapt to a series of new regulatory requirements. Already before the crisis, the German Securities Trading Act required financial advisors to elicit information on clients personal and financial situation, risk preferences, and investment goals, as well as on their level of experience with different financial instruments prior to making a recommendation. Ultimately, advisors must only recommend financial products that suit the client profile. Furthermore, they are required to disclose any commission payments by third parties resulting from a client s transactions. Effective January 1st, 2010, the German legislator in addition requires written records of investment advisory meetings. By law, the minutes have to inform on the cause or motivation and the duration of the advisory session, the personal situation and specific needs of the client, the financial products discussed, and the advisors final recommendations along with their justification. 7 advisor has to sign the protocol and hand a copy to the client prior to the execution of any trade advised. 8 The minutes are supervised and audited by the German Federal Supervisory Authority (BAFIN). 9 Providing rich detail on the client-advisor interview, the records form a valuable basis for the analyses in this paper. 6 In addition, banks are allowed to pay their employees annual bonuses based on the success of the branch, the team and/or the entire bank. This bonus is not allowed to exceed 120% of a monthly gross salary. 7 34(2a) WpHG (Securities Trading Act), 14(6) WpDVerOV (Regulation on Investment Services, Conduct, and Organization). In the circular (WA 31-Wp /0010) to all banks the German Federal Supervisory Authority (BAFIN) strongly demands free text entries to record client situation, particular client concerns and justifications for final recommendations. 8 In the case of on-the-phone advice, the client may agree to initiate a transaction prior to receiving a copy of the protocol. In this case, the client has the right to cancel the transaction in case the protocol is incomplete or incorrect. 9 Audits rely on customer complaints as well as on-site visits. In 2013 the BAFIN audited advisory minutes in Germany. In cases the advisory minutes did not fully meet the standards. These cases did not cluster in single institutions. The 5

7 As a result of regulatory tightening and ensuing compliance concerns, financial advice at large bank institutions has become much more standardized, especially in the retail client segment. Our sample bank entertains a central investment committee that continuously screens the market for actively managed mutual funds to create and update short lists of financial products to be recommended to clients of different risk categories. In particular, the standard product lists feature a limited number of actively managed funds of different asset compositions and risk classes. Single stocks do not appear on the standard recommendation lists. Advisors at the bank are expected to pick their product recommendations to clients from these short lists. While contributing to client diversification, this fund-based approach is consistent with the bank responding to financial incentives to sell mutual funds. The bank s standardized advisory model thus concedes limited discretion to the individual advisor. However, advisors have some leeway to deviate from the standard recommendation list in order to respond to specific client requests. This is reflected in the observed distribution of product recommendations across the product universe. When considering all product types, 1,261 different ISINs make up for the 26,765 purchase recommendations observed. Overall, the ten products recommended most often as measured by their value share in total purchase recommendations already account for about half of all purchases recommended. These popular products all are mutual funds from the short list defined by the bank s investment committee, seven are in-house products. Panel B of Figure 1 illustrates this dominance of a small number of financial products. With fund recommendations (72 percent of purchase recommendations), standardization is even more pronounced: 177 different mutual funds make up for 18,965 fund purchase recommendations in our sample and the ten most popular funds make up 68 percent of fund purchase recommendations. As a consequence of the high degree of standardization, advisor-specific effects that have been found to explain a large share of the cross-sectional heterogeneity of portfolios advised by independent advisors (Foerster et al., 2016; Linnainmaa, Melzer, and Previtero, 2016) are likely to play a much smaller role in our setting. B. Sample Investors and Use of Finanical Advice Our final data set includes information on 18,787 advisory sessions of 7,505 clients and their 487 advisors distributed over the period January 2010 to November We restrict our analysis to the sample bank s retail client segment featuring the traditional bank client valuing long-standing relationships with his main bank. Business clients and the very wealthy in the wealth management segment are excluded from the sample, as they are likely to be treated differently by the bank. 10 Sample investors are split roughly equally by gender and are relatively old, with the median investor being 65 and older, a feature shared by comparable data sets on bank clients in Germany (Hackethal et al., 2012a). Sample investors are rather risk averse, with the median investor being willing to take on moderate financial risk, while 25 percent of sample clients select in the two highest risk categories (Panel A of Table I). 10 In practical terms, we exclude all clients with portfolio holdings of EUR 500,000 and beyond. 6

8 [Insert Table I about here] Client portfolio information is reported in Panel B of Table I. The distribution of the portfolio size over the sample period is right-skewed. While average conditional portfolio holdings amount to EUR 56,000, the median investor holds a portfolio of around EUR 32,000. These figures compare to the EUR 54,200 (EUR 16,600) average (median) financial assets held by German households conditional on participation in financial assets estimated by the German Central Bank for 2014 (Deutsche Bundesbank, 2016). With the median investor conducting two trades (one purchase) per year, the majority of sample clients are rather inactive investors (Panel C of Table I). Given the, on average, relatively low risk-tolerance, low investor activity, but at the same time representative portfolio size, the portfolios held at the sample bank are unlikely to constitute play money accounts. With, on average, over 60 percent of total value, fund holdings by far dominate client holdings; only eight percent of clients do not participate in funds during the sample period. Yet, low-fee Exchange Traded Funds are widely missing in the sample, amounting to only one percent of total client holdings. Single stocks make up about nine percent of the average portfolio. Besides funds and single stocks, corporate and government bonds account for a considerable 23 percent of total portfolio holdings. About 80 percent of investors participate in equity over the sample period, with an average portfolio share in equity of 35 percent. 11 Single stocks on average account for one fifth of client equity holdings. Yet, only 30 percent of sample clients participate in single stocks at least once during the sample period. Conditional on investing in single stocks, these make up on average 54 percent of equity holdings. In line with the literature, the conditional average number of different stocks held is 3.3 (90th percentile is 8). In addition, clients at our sample bank exhibit a strong home bias; German equity accounts for 25 percent of the average equity portfolio, which compared to Germany s share in the global asset market capitalization of under three percent 12 is considerable. The upper 10 percent of equity investors invest more than 85 percent domestically. The remaining equity investments largely come in the form of internationally diversified equity funds. The average investor conducts 80 percent of security purchases with the input of an advisor (Panel C of Table I). The remaining security purchases are self-directed trades. From the advisory minutes we extract additional information about the meetings (Panel D of Table I). Details on the textual analysis of the minutes will be given later in the paper. For the average investor, the data includes records of 2.5 advisory meetings over the four-year sample period. Over 80 percent of consultations are initiated by the advisor. Usually, meetings take place face-to-face in one of the bank s branch offices, only in 20 percent of cases advice is received over the phone. 63 percent of advisory interviews last longer than 30 minutes. The most common motivations for an advisor interview are the availability of excess liquidity for new investments, a general portfolio check-up or the current market situation. 11 Following (Foerster et al., 2016), in calculating client holdings we assume that balanced funds invest half in equity, half in fixed income securities. 12 See Worldbank (2016). 7

9 The advisory minutes provide detailed information on the products recommended, referenced by their International Security Identification Number (ISIN). Panel E of Table I provides descriptive statistics on the content of advice. We follow Hoechle et al. (2016b) by concentrating on purchases, which make up 65 percent of all advised trades. The reason is that sales transactions may be driven by various forces that are hard to control for (e.g. client liquidity need or other personal events). On average, an advisory session results in 1.4 purchase recommendations (median is one). Accounting for over 70 percent of all recommendations, funds dominate bank financial advice. Especially active balanced funds are frequently recommended (40 percent of fund recommendations). Single stocks are hardly recommended. 12,303 purchase recommendations involve an equity product (including balanced funds). Comparing the structure of recommended transactions to that of client holdings in Panel B, investment in single stocks and purely domestic equity is recommended far less often. This is in line with existing studies documenting that financial advice contributes to better portfolio diversification (Gaudecker, 2015; Hoechle et al., 2016b; Hackethal et al., 2012a; Bluethgen et al., 2008; Shapira and Venezia, 2001) and lower home bias (Kramer, 2012; Bluethgen et al., 2008). 13 However, the general recommendation pattern is also consistent with advisors responding to incentives for distributing expensive actively managed mutual funds. C. Measures of Client Involvement in the Advice Process By client involvement we generally understand an advised client s own dedication to his personal finances. This subsumes the collection of own information or the development of own research ideas and strategies, which may also result in security transactions that are conducted without the input of an advisor. More, clients more involved in their financial decision making are also likely to discuss more actively with their financial advisor or even introduce own ideas in the client-advisor interview. We use two different measures in order to identify client input into the advisory process - client independence and client ideas. Independence is measured as the share of self-directed purchases in all purchases conducted over the sample period. Therefore, the measure is not based on a specific advisory meeting, but on the clients overall advised trading behavior. While most contributions treat being an advised client as a zero-one classification, Hoechle et al. (2016b) point to the importance of considering the degree to which clients make use of advice when evaluating advised against non-advised portfolios. 14 particular, our interpretation of this first measure is that more independent clients who besides with their advisor also conduct security purchases on their own are more likely to collect own information and develop own investment ideas and strategies. They may also discuss more actively with their advisor and evaluate any advice given against a stronger prior. 15 Figure 2 unveils great 13 Yet, Hoechle et al. (2016b) find no reduction in home bias through advice. 14 Other than in Hoechle et al. (2016b) we are not confined to a rule of thumb for identifying advised trades. The authors define a trade as advised if it takes place within four days of a meeting. This definition is not suitable for our sample. Due to frictions, it may take longer until an advised trade gets executed. More, clients also execute a non-negligible number of trades around the advisory meeting in securities that have not been discussed during the meeting (see Figure AI in Appendix A). 15 For example, among investors purchasing funds outside their retirement accounts through an advisor, only one In 8

10 heterogeneity in investors reliance on financial advice: while 58 percent of clients in our sample rely on their advisor in all their security purchase decisions, 24 percent of otherwise advised clients conduct less than half of their purchases with advisor input. Another 18 percent of clients use advice in more than half of their purchases, yet still buy financial securities independently. [Insert Figure 2 about here] Our second measure of client involvement, client idea, is obtained directly from the advisory minutes. The image files of the original advisory minutes were converted into machine readable text using optical character recognition (OCR) software and analyzed using textual analysis. An important asset of this data source is the information on single securities discussed during the interview along with their justification given by the advisor. The single products discussed are referenced by their ISIN, allowing to identify the exact nature of the product. Justifications provided by the advisor emphasize the diversification potential of products discussed but also the active management of frequently recommended funds, promising the client peace of mind by not having to take care of his finances himself. Yet, these reasons also provide definite evidence on different sorts of client input in the advisory process, e.g. specific client ideas or requests. We search for signals of client involvement in the minutes and define a dummy being one if an advised transaction traces back to a client idea or specific request. Given the great degree of individuality in formulations, typos and abbreviations, evidence on clients own investment ideas had to be hand-collected from the minutes. The following two statements found in the minutes are examples for such client ideas: You wish to invest in this stock and we provided you with our assessment of the firm or The XYZ investor s Club recommended this asset. This is why you wish to purchase it. We list a series of most common example phrases pointing to clients own ideas in Table AI in Appendix A. [Insert Table II about here] Table II provides descriptive statistics on clients by their degree of involvement in the process of financial advice according to our two measures. In Panel A, client independence is used to measure client involvement. For illustrative purposes, we divide clients into three brackets of independence levels. Clients are defined to exhibit high independence, if they conduct more than half of their purchases without advisor input. Clients of medium independence conduct less than half but still a positive number of purchases independently. Low independence clients consult with their advisor on all security purchases. Descriptive results reveal that less independent clients are relatively riskaverse, passive and older investors holding rather small portfolios, pointing to a relatively lower level of experience in financial matters. They may lack the confidence, interest, or time to take financial decisions on their own and therefore are willing to hand over control of their personal finances to their trusted advisor. Confidence in the own (financial) ability has been identified a key driver of third responded to fully convey the lead in decision making to the advisor in a survey administered by the Investment Company Institute. The majority of investors described themselves as involved in the management of their portfolios. 20 percent even indicated to take the lead in advised financial decision making themselves (Leonard-Chambers and West, 2006). 9

11 advice seeking and usage. Individuals lacking confidence in their own decision-making abilities tend to be more likely to seek advice from others and to rely on it (Gino, Brooks, and Schweitzer, 2012; Bonaccio and Dalal, 2006). Thereby, trust in the advisor or the advice received acts as an important mediator (see Bonaccio and Dalal (2006) for a review). In the models by Georgarakos and Inderst (2014) and Gennaioli et al. (2015), trust in financial advice (the advisor) acts as a substitute for own financial capability and confidence. Descriptive results at hand on clients personal characteristics associated with the degree of independence are in line with this literature. A very similar picture emerges from Panel B of Table II using client ideas as measure of client involvement. Here, we define a client as having own ideas if at least one purchase recommendation he receives over the sample period traces back to an own investment idea or specific client request. 381 clients (5 percent) of clients introduce own investment ideas in the advisory process according to this definition. Taking a closer look at portfolio holdings, descriptive statistics suggest that both higher client independence as well as own client ideas are associated with portfolio structures that more strongly deviate from the bank s standard advice concept: clients who are less reliant on the advisor also hold a lower share of funds and concentrate their equity portfolio more strongly in single stocks and domestic equity. Overall, descriptive results provide first evidence that client involvement indeed has an influence on the composition of advised portfolios. Defining a client as advised based on a simple zero-one distinction, as it was predominantly done in the literature so far, may hence not be sufficient when studying the impact of financial advice on client portfolio outcomes. Supportive evidence for our measure of investor independence as a measure of client involvement in the advisory process comes from the finding that more independent investors are significantly more likely to initiate a client meeting on their own. This effect is especially pronounced for clients approaching their advisor with own ideas. More, we find clients approaching their advisor with own ideas to be generally more independent in their purchases. II. Client Involvement and Advice Outcomes A. Does involvement help to explain heterogeneity in portfolio compositions? The above descriptive evidence suggests that the degree of a client s dedication to his finances and active involvement in the process of financial advice may help explain part of the heterogeneity in advised portfolio outcomes. Yet, given the differences in client characteristics across groups, it is necessary to evaluate these findings in a multivariate setting. We begin by analyzing the cross-sectional variation in clients portfolio structures at the end of the month following an advisory interview. 16 More specifically, we investigate whether client involvement helps to explain the variation in advised clients portfolios beyond traditional factors, in particular client risk tolerance and unobserved advisor heterogeneity. Using the approach of Foerster et al. (2016), who demonstrate the explanatory power of advisor heterogeneity on clients 16 For example, if the advisory interview has taken place on July 18, 2012, we look at portfolio holdings at the end of August

12 portfolio composition, we run pooled cross-sectional regressions of the form y ij = βx i + θz i,t 1 + µ t + µ a + ɛ ij, (1) where the dependent variable is one of three measures of portfolio composition of client i observed at the end of the month following on advisory meeting j. The first measure of portfolio composition is the share of equity in the total portfolio. In addition, we use diversification properties of the equity part of an advised client s portfolio as dependent variables, namely the domestic equity share (home bias) and the share in the equity portfolio that is held in the form of single stocks. x i is one of our two measures of client involvement. Z i,t 1 is a vector containing client characteristics: risk tolerance, age, gender, investment horizon, income, and financial wealth held with the bank (including cash). While most client characteristics are constant over time, time-variant portfolio characteristics are measured as of the end of the month preceding the advisory meeting. µ t controls for year and month fixed effects. Following Foerster et al. (2016), we include advisor fixed effects µ a to control for unobserved advisor heterogeneity. We estimate the model using OLS. Standard errors are clustered at the advisor level. [Insert Table III about here] Table III reports the regression estimates. According to neoclassical portfolio theory as well as regulatory requirements, we should expect that differences in risk tolerance explain a large part of the heterogeneity in equity shares. Columns (1) and (2) demonstrate that the advisory process the bank has established results in portfolio allocations in line with the theory. Client risk tolerance is the key predictor of an investor s equity share and explains most of the variance in risk taking. The equity share increases with income, portfolio value and investment horizon and decreases with age, in line with many life-cycle models. Adding advisor fixed effects in Column (3) does not alter signs or magnitudes of effects. Given the highly standardized process sample advisors operate in, it is not surprising that unobserved advisor heterogeneity explains much less of the observed heterogeneity in advised clients portfolio structures compared to the sample of independent financial advisors in Foerster et al. (2016). Being interested in the demand side of financial advice, we add our measures of client involvement in Columns (4) and (5). We find a positive and significant effect of client independence as well as client ideas on clients portfolio equity share. The magnitude of the coefficients are however rather low, indicating that client involvement, after controlling for client risk tolerance, has limited impact on clients portfolio equity share. Turning to the composition of clients equity portfolios, columns (6) to (8) use a clients domestic equity share after an advisory meeting as a dependent variable. Here, coefficients of both involvement measures are of much larger relative magnitude. While a ten percentage point increase in the share of trades that are conducted without an advisor increases the equity share by only 1 percentage point (base rate being 26 percent), the same 10 percentage point increase increases domestic share and stock share by 2.7 and 4 percentage points, respectively, compared to base rates of on average 25 (10 percent) and 21 (20 percent), respectively. Coefficients on the client idea 11

13 dummy being one if at least one of the clients purchase recommendations over the sample period can be traced back to an own idea or specific request are even larger in magnitude. Thus rather than on overall risk taking, clients degree of involvement seems to have an effect on the composition of equity portfolios, with more independent clients holding equity portfolios concentrating on few single stocks and domestic equity in line with the literature on independent individual investors. From the above results, however, we are not able to isolate the effect of financial advice on client portfolio outcomes. Rather than advisors catering to the (perceived) priors of more involved clients as in the antibiotics example, the differences in portfolio structures may simply result from more independent clients shaping their portfolios mainly through self-directed trades. Table IV provides descriptive statistics on the trading behavior of clients by degree of involvement, splitting both advised and non-advised trades by content. [Insert Table IV about here] Panel A of Table IV reveals that both self-directed trades and advisor purchase recommendations differ significantly by degree of client independence: while the share of fund recommendations is lowest to clients who rely on advice the least, these clients also trade less in funds when trading independently compared to their peers relying on advice in at least half of their purchases. Also, while the most independent clients trade most strongly in single stocks and domestic equity when buying securities on their own, the recommendations they receive from advisors show the same tendencies. With lower levels of independence, on the other hand, the pattern of products recommended moves closer to the bank s standard advisory approach, emphasizing mutual funds and hardly recommending single stocks. Panel B of table IV shows similar results for clients with and without own ideas. A preliminary conclusion from the figures in Table IV thus is that both channels are at work: while more independent clients shape their (equity) portfolios according to their diversification preferences and own investment strategies by trading without advice, the figures also point to the possibility that advisors cater to these (perceived) preferences and priors of their clients. Rather than educating their clients on the benefits of diversification, advisors may choose to go along with both implicit and explicit client requests, recommending stocks and domestic equity to those already holding more concentrated portfolios. To assess this hypothesis in more detail, we investigate in the following whether advisor suggestions, as in the antibiotics example, tilt toward clients existing investment priors. B. Do advisors tilt toward involved clients existing investment priors? In a second set of regressions, we estimate how clients existing portfolio strategies and degree of involvement affect the outcome of the meeting documented in the minutes. Note that we regard the recommendations documented in the minutes as a consensus between the advisor and the client, which may be driven by both advisor and client input. 12

14 We regress an indicator of the type of equity recommendation on our measures of client involvement and the given set of control variables in an OLS regression model of the form y i = βx i + γs i,t 1 + δx i s i,t 1 + θz i,t 1 + µ t + µ a + ɛ ij, (2) where the unit of observation is a single recommendation provided to client i in any of the advisory meetings over the sample period. The dependent variable is an indicator of the type of equity recommendation (single stock, domestic equity product, domestic stock). x i is one of two measures of client involvement. 17 To estimate the degree to which advisors cater to their clients existing strategies, we include previous month-end equity portfolio shares dedicated to single stocks and domestic equity (s i,t 1 ), respectively. This forces us to restrict the analysis to clients who enter the advisory meeting with positive holdings, as, except for the case where clients explicitly state their investment preferences, advisors may not infer any information on the client s current strategy for his equity portfolio. To investigate whether advisors tendency to cater to client priors is stronger with more independent investors, we interact the first two terms in some of the specifications. Z i,t 1 again contains client characteristics that are either time-constant or measured as of the end of the month preceding the interview. µ t controls for year and month fixed effects. Advisor fixed effects µ a control for unobserved advisor heterogeneity. We estimate the model using OLS. Standard errors are clustered at the advisor level. In columns (1) to (5) of Table V the dependent variable is an indicator equal to one if a single stock rather than a diversified equity fund is recommended for purchase. Our main variable of interest is the share of the equity portfolio dedicated to single stocks when entering the meeting. We find a significantly positive effect, pointing to advised transactions catering to clients existing strategies. A 10 percentage point increase in the previous stock share, raises the probability of being advised a single stock by one percentage point, which, compared to the base rate of 6 percent among all equity recommendations, is considerable. In line with the above reflections that more involved clients may enhance a catering tendency of advice, we add the measures of client involvement to the regression. For the sake of an easier interpretation, we use dummies for the above independence brackets. Being highly independent adds an additional 2 percentage points to the probability of an advised stock purchase. More, the positive and significant coefficient on the interaction term in column (3) reveals that advisors more strongly cater involved clients. Turning to our second involvement measure, we are able to shed additional light on whether the tilt in recommendation indeed results from client-side effects. The results show that a client s own idea considerably increases the probability of a single stock recommendation (Column 4). Having already larger share in single stocks again reinforces this phenomenon (Column 5). We rerun the same specifications using domestic equity recommendations (Columns 6-8) and domestic stock transactions (Columns 9-11) respectively as dependent variables. We are able to replicate 17 For the ease of interpretation of interaction effects we make use of dummies for the three client independence categories introduced in Tables II and IV. 13

15 previous results - recommendations for single and domestic stocks are more frequent for clients with a higher previous share in these products and even more likely for those among them who are more independent or come up with own investment ideas. An alternative explanation to the above results is that clients self-select certain advisors who provide them with the advice they want to hear (Mullainathan et al., 2012). Figure 3 provides graphical evidence that independent clients in our sample do not concentrate on a small number of advisors who may provide advice in line with client priors. Rather, the distribution of clients across advisors appears to be quite balanced with respect to client independence (Panel A), equity share (Panel B), and equity portfolio shares of single stocks and home equity (Panel C). Overall, we find that clients who are more involved receive different recommendations than their less involved peers, presumably on the expense of portfolio diversification. While our analysis so far has focused on single purchase recommendations and their diversification potential, we will now investigate in more detail whether client involvement compromises the effectiveness of financial advice in building efficient portfolios. III. Client Involvement and Effectiveness of Financial Advice Mullainathan et al. (2012) define sound financial advice as advice that moves a client towards a low-cost, diversified portfolio. Along this definition, we investigate the effects of bank financial advice on advised clients portfolio diversification and costs by degree of involvement in the advisory process. A. Diversification Properties of Advised Portfolios As a first indication of the effectiveness of financial advice, we track the change in portfolio diversification properties over the sample period by degree of client involvement (see Table VI). In particular, we observe changes in advised clients portfolio composition from the end of the month preceding the client s first advisory meeting and the end of the month following on the last meeting recorded during the sample period. Our data allows us to track these changes for 5,778 clients (77 percent). 18 [Insert Table VI about here] Results on client independence in Panel A of Table VI show that the portfolio structure of most independent clients does not significantly change over the sample period. Their portfolio shares in funds and equity, as well as their stock share in equity and domestic equity shares remain unchanged. In managing their equity portfolio, independent clients seem to stick to their strategy of investing in single stocks and predominantly domestic equity. Moving along the client spectrum 18 For the remaining clients, this is inhibited by missing entries, client entries or exits to the sample that do not make us observe client portfolio holdings at each point in time during the sample period. For example, 1,557 of meetings are first-time meetings with the client having had no positive security holdings with the bank before. 14