Abstract
This paper offers cross-sectional and data-intensive insights into Robo-advisory portfolio structures. For this purpose, we scrape portfolio recommendations for 16 German Robo-advisors. Our sample accounts for about 78% of assets in the German Robo-advisory market. We analyze about 243.000 pairs of recommended portfolios and their corresponding client characteristics. Our results show that current Robo-advice offers limited individualization. Variables that matter in modern portfolio choice like the amount and nature (beta) of human capital or shadow assets are largely ignored. Instead, portfolio recommendations are designed to meet investor preconceptions or the regulator’s understanding of portfolio choice. While ensuring consumer trust and regulatory approval makes business sense, it also limits the economic benefits of Robo-advisors.\(^1\)
Similar content being viewed by others
Notes
See https://www.Robo-advisor.de from June 2021.
CRRA utility is still the mainstream utility function in finance for very good reasons, apart from analytical tractability. It is compatible with stable risk premia over the last 200 years, even though individuals became many times wealthier.
We group income into the wealth bucket as the present value of future savings reflects an investor’s human capital on her balance sheet. Higher levels of human capital are for most employees very bond-like (only one Robo asks for the profession as an input and the variable is not significant) and should hence increase the optimal allocation to risky assets.
Kritzman and Rich (1998) provide a taxonomy for alternative utility functions.
Lusardi et al. (2017) show that financial knowledge is a key determinant of equity market participation and Foltyn (2020) shows a positive relationship between experience and average shares in risky assets. Ampudia and Ehrmann (2014) show that the impact of experience can go either way (increase or decrease participation).
Superficially, we can label learning from past returns via Bayesian updating as experience. However, in Berk and Green (2004) investors simply learn about the ability of managers to generate positive or negative alpha from most recent realized returns. Depending on the sign of past returns, they decide to invest or not as investors need to chase promising funds before other investors do. Each additional flow dilutes alphas down towards zero. In our view, it would be highly irrational to base long-term asset allocation recommendations on personal investment biographies (across different time horizons). The right approach is to use economic state variables instead.
Suppose goal-based portfolio 1 is optimally short asset A, while portfolio 2 is long asset A. Joint optimization will lead to a partial or complete offset of these positions. Separate long-only optimizations will not.
See ESMA (2018), pp. 14–15.
For financial practitioners, this is not new. The asset allocation model of Black and Litterman (1992) probably owes most of its success to a solution that is strongly anchored in a prior portfolio familiar to all investors (market portfolio).
See Venables and Ripley (2002), page 146 for a description of our methodology.
Our tree only ends up with seven final nodes (despite the existence of 11 portfolios). This is due to the use of (10-fold) cross validation to find the optimal complexity parameter (limiting the tree size). While more complex and better fitting trees can be found with changes in complexity parameters, the initial splits remain unchanged. Multi-class classification trees arrive at the same hierarchy.
References
Ampudia M. and M. Ehrmann (2014), Macroeconomic Experiences and Risk Taking of Euro Area Households, ECB Working Paper Series #1652
Bach, L., L. Calvet, and P. Sodini. 2020. Rich Pickings? Risk, Returns and Skill in the Portfolio in the Portfolios of the Wealthy, American Economic Review 110 (9): 2703–47.
Beketov, M., K. Lehmann, and M. Wittke. 2018. Robo Advisors: Quantitative Methods Inside the Robots. Journal of Asset Management 19: 363–370.
Bhattacharya, U., A. Hackethal, S. Kaesler, B. Loos, and S. Meyer. 2012. Is Unbiased Financial Advice to Retail Investors Sucient? Answers from a Large Field Study. Review of Financial Studies 25 (4): 975–1032.
Black, F., and R. Litterman. 1992. Global Portfolio Optimization, 28–43. September/October: Financial Analysts Journal.
Campbell, J.Y., and L.M. Viceira. 2002. Strategic Asset Allocation. Oxford University Press.
Campbell, J. 2006. Household Finance. The Journal of Finance 61 (4): 1553–160.
Campbell, J. 2016. Restoring Rational Choice: The Challenge of Consumer Financial Regulation. American Economic Review 106: 1–30.
Cochrane J. 1999. Portfolio Advice for a Multifactor World, NBER Working Paper No. w7170, Available at SSRN: https://ssrn.com/abstract=217489
Cochrane J. 2021. Portfolios for long term investors, NBER working paper
Das, S.R., H. Markowitz, J. Scheid, and M. Statman. 2010. Portfolio Optimization with Mental Accounts. Journal of Financial and Quantitative Analysis 45 (2): 311–334.
Deloitte .2016. Cost Income Ratios and Robo-Advisory. https://www2.deloitte.com/de/de/pages/financial-services/articles/robo-advisory-in-wealth-management.html
ESMA. 2018. Guidelines on certain aspects of the MiFID II suitability requirements, ESMA35-43-1163.
Faloon, M., and B. Scherer. 2017. Individualization of Robo-advice. Journal of Wealth Management 20: 30–36.
Fisher, A., C. Rudin, and F. Dominici. 2018. Model Class Reliance: Variable Importance Measures for Any Machine Learning Model Class, from the ”Rashomon” perspective. arXiv preprint arXiv:1801.01489
Foerster, S., J.T. Linnainmaa, B.T. Melzer, and A. Previtero. 2017. Retail Financial Advice: Does One Size Fit All? Journal of Finance 72 (4): 1441–1482.
Foltyn, R. 2020. Experience Based Learning, Stock Market Participation and Portfolio Choice, SSRN working paper
Gennaioli, N., A. Shleifer, and R. Vishny. 2015. Money doctors. Journal of Finance 70 (1): 91–114.
Grealish, A. and P. Kolm. 2021. Robo Advisory: From Investing Principles and Algorithms to Future Developments. In Machine Learning in Financial Markets: A Guide to Contemporary Practice, eds. Capponi A and C. Lehalle. Cambridge: Cambridge University Press, forthcoming
Grinblatt, M., M. Keloharju, and J. Linaaunmaa. 2011. IQ and Stock Market Participation. Journal of Finance 66 (6): 2121–2164.
Hildebrand, C., and A. Bergner. 2020. Conversational Robo advisors as surrogates of trust: onboarding experience, firm perception, and consumer financial decision making. Journal of the Academy of Marketing Sciencehttps://doi.org/10.1007/s11747-020-00753-z.
Hoechle, D., S. Ruenzi, N. Schaub, and M.M. Schmid. 2017. The Impact of Financial Advice on Trade Performance and Behavioral Biases. Review of Finance 21: 871–910.
Hsu, C. 2012. What Drives Equity Market Non-Participation? North American Journal of Economics and Finance 23: 86–114.
Jacovi, A., Marasović, A., Miller, T. and Goldberg, Y. 2020. Formalizing Trust in Artificial Intelligence: Prerequisites, Causes and Goals of Human Trust in AI. arXiv preprint arXiv:2010.0748.
Kritzman, M.P., and D. Rich. 1998. Beware of Dogma: The Truth About Time Diversification. Journal of Portfolio Management 24: 66–77.
Lusardi, A., P.C. Michaud, and O.S. Mitchell. 2017. Optimal Financial Knowledge and Wealth Inequality. Journal of Political Economy 125: 431–477.
Merton, Robert C. 1971. Optimum Consumption and Portfolio Rules in a Continuous-Time Model. Journal of Economic Theory 3: 373–413.
Merton R. .2017. The future of Robo advisors, CNBC video https://www.cnbc.com/2017/11/05/mit-expert-robert-merton-on-the-future-of-Robo-advisors.html.
Mullainathan, S., Noeth, M. and A. Schoar (2012), ’The market for financial advice: An audit study’, NBER Working Paper (17929).
Puhle, M. 2019. The Performance and Asset Allocation of German Robo-Advisors. Society and Economy, 41(3).
Sapienza, P., A. Toldra-Simats, and L. Zingales. 2013. Understanding Trust. The Economic Journal 123 (573): 1313–1332.
Scherer, Bernd and Lehner, Sebastian, What Drives Robo-Advice? .2021. Available at SSRN: https://ssrn.com/abstract=3807921 or doi: https://doi.org/10.2139/ssrn.3807921
Soehnke M., J. Branke, and M. Motahari. 2020. Artificial Intelligence in Asset Management, working paper, Judge Business School.
Tertilt M. and P. Scholz. 2018. To Advise, or Not to Advice - How Robo-Advisors Evaluate the Risk Preferences of Private Investors, Journal Of Wealth Management, Fall 2018, v21, n2, 70–84.
Torno A. and S. Schildmann. 2020. What Do Robo-Advisors Recommend? An Analysis of Portfolio Structure, Performance and Risk. In Enterprise Applications, Markets and Services in the Finance Industry, eds. B. Clapham, J. and A. Koch. FinanceCom 2020. Lecture Notes in Business Information Processing, vol. 401. Springer, Cham. doi: https://doi.org/10.1007/978-3-030-64466-6_6.
Torno, A., D. R. Metzler, and V. Torno. 2021. Robo-What?, Robo-Why?, Robo-How? A Systematic Literature Review of Robo-Advice. Proceedings of the 25th Pacific Asia Conference on Information Systems (PACIS), Dubai.
Van Rooij, M., A. Lusardi, and R. Alessie. 2011. Financial literacy and stock market participation. Journal of Financial Economics 101 (2): 449–472.
Venables, W.N., and B.D. Ripley. 2002. Modern Applied Statistics with S. New York: Springer.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The material presented is for informational purposes only. The views expressed in this material are the views of the authors. The authors claim no conflicts of interest.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix: Statistical model
Appendix: Statistical model
This Appendix illustrates our statistical model for a specific Robo-advisor (Scalable) with 7290 input permutations. We run an OLS regression with (ordered, if applicable) factors as independent variables. The dependent variable is the recommended equity allocation. Our results are shown in table 6. The intercept of 53.483% represents the base case allocation to equities. All other regression coefficients describe the marginal effects of answering questions on the robo-advisor homepage across all 7290 choice sets. As we deal with mostly ordered factors, we can not use one hot encoding or Helmert contrasts but rather use orthogonal polynomial contrasts.Footnote 12 The extensions .L, .Q, .C denote coefficients from linear, quadratic and cubic regression terms.
For example, the contrast coding for experience is as ordered factor with three levels (0,1,2) where we add the number of “yes” answers with respect to product and financial service experience. Table 5 shows the corresponding contrast. A regression coefficient of 1.699 for the linear contrast on knowledge means that an investor twice ticking the box “none” receives a \(-0.7\cdot 1.699\%=-1.18\%\) (percentage points) lower equity recommendation than the base line allocation, while an investor with extensive knowledge will receive a recommendation to add \(1.18\%\) to the baseline allocation. For the full effect, we need to add the quadratic contrast or alternatively look at partial dependence plots. Responses to investment goals are easier to interpret as they are modeled as unordered factors using one-hot dummy encoding. A value of -12.349 for wealth preservation means a (ceteris paribus) decrease of 12.349% in the recommenced equity allocation for all investors ticking this box.
Our regression model explains 97% of the variance of equity allocations. The standard error of the regression (standard deviation of fitted versus actual portfolio recommendations) is 4.96, i.e 2/3 of all predictions are within +/− 4.96% difference to the true value, even though our model did not use any interaction term. However, almost the same performance can be achieved by only including risk aversion and investment goals. The explanatory power fall slightly to 95%. All other variables only account for an additional 2% in explanatory power.
Next, we want to more formally interrogate our regression model to find the most influential variable(s). For this purpose we borrow from the literature on interpretable machine learning and employ the following model agnostic algorithm suggested by Fisher et al. (2018). For each variable we randomly permute the values of that particular feature and recompute the chosen performance metric, in our case \(R_{\rm perm}^{2}\). We then record the difference between the baseline metric and the permutated metric \(R_{\rm base}^{2}-R_{\rm perm}^{2}\) as our importance score. In order to understand the added value of a given variable we look at model results when the observations for the variable under investigation are reshuffled. Shuffling an important variable will lead to a much larger drop in explanatory power than shuffling an unimportant variable. We repeat this procedure 100 times and estimate the average importance score. The results are shown in figure 1. This confirms our earlier results. Risk aversion and investment goals are the most important variables. Creating noise in these variable leads to the most severe reduction in explanatory power across all variables.
Finally, we check the direction of influence for each question in the Robo-advisor’s questionnaire. In order to account for nonlinear contrasts we need to compute the cumulative effect of all polynomial terms. For this purpose, we use partial dependence plot as shown in Fig. 2. The idea of partial dependence plots is to estimate a statistical model using the original data and then use this fitted model to make predictions from a modified data set. The modified data set is a complete copy of the original data set, except for the variable of interest where all realizations are replaced by a particular value. The average across all predictions is then used as best estimate for the partial variation of interest. After repeating this process for all level of the variable of interest we can plot this variable against the average responses in a scatterplot. This plot is called partial dependence plot and shows both direction and magnitude of influence.
We can confirm our results by employing a regression tree in Fig. 3. . Regression trees use explanatory variables to consecutively split the data (using only one variable at each node) into pure clusters with as little intra-cluster variation as possible. Clusters do not need to have the same size (do not need to contain the same number of variations). Instead of making continuous predictions all combinations of explanatory variables that lead into a given terminal node carry the same prediction. In our context (recommended equity allocations from questionnaire inputs, regression trees offer some advantages over linear regressions. The first split selects the most important variable, while the sequence of splits is able to model non-linearities. This allows us to find otherwise hidden nonlinear interactions. While linear regressions can also uncover nonlinear interactions by including all possible cross terms, this requires as many right-hand-side variables as data points and thus results in a loss of all degrees of freedom. Our fitted regression tree identifies the same set of variables as most important in explaining the cross section of equity recommendations. Interestingly its standard error of 1.98 is less than half of a linear regression model, which we take as evidence for nonlinear interactions not covered by a linear regression. High equity recommendations are reserved for investors with low risk aversion, agrressive goals, large levels of wealth and sufficient experience.
Our fitted regression tree regards time horizon as the first variable to split the data on. Suppose we would be only allowed to split the data once into clusters with as little internal dispersion of equity weights as possible. Our regression tree would then recommend to use the variable time horizon. In this sense time horizon is the most important variable. Short and medium term horizon investors receive recommended allocations between 21.14% (node 4) and 49.73% (node 8) equities while long horizon investors obtain portfolios between 53.6% (node 10) and 73.95% (node 13)%. All percentages are predictions from the regression. tree. The standard error of the regression tree (standard deviation of fitted versus actual portfolio recommendations) is 5.51. Five from six nodes use the variables investment goal and time horizon. This again confirms our previous analysis. Predicted equity recommendations as a function of questionnaire replies rise from left to right. The most aggressive allocations are reserved to long term investors with retirement objectives that react to losses by increasing their equity allocations. Investors with short time horizon looking to fund an upcoming expense receive small equity allocations.Footnote 13
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Scherer, B., Lehner, S. Trust me, I am a Robo-advisor. J Asset Manag 24, 85–96 (2023). https://doi.org/10.1057/s41260-022-00284-y
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1057/s41260-022-00284-y