Executive summary

This paper investigates the subjective understanding and perception of financial information and their impact on investment decisions. A consumer testing approach is applied in order to explore: i) how different representation formats (or Templates) are appraised in terms of complexity, usefulness and information content, ii) how different Templates influence risk perception, iii) how different Templates affect willingness to invest. A sample of 254 Italian investors were submitted different Templates, each delivering in different modes the same information on risk, return and costs of four financial instruments (two structured bonds - one outstanding and the other newly issued - negotiated on the Italian retail bond market and two Italian listed stocks). Risk is alternatively disclosed through four approaches. The first relies on a synthetic risk indicator, aggre-gating information on market, liquidity and credit risks. The second discloses unbundled quantitative measures of the market risk (volatility and value at risk), the liquidity risk (turn-over ratio) and the credit risk (Moody’s official rating and expected default probability). Both the synthetic and unbundled formats compare the risk/return characteristics of the product with the risk/return attributes of a benchmark portfolio. The third mode is based on what-if scenarios. The fourth resorts to probabilistic modelling of expected returns. Costs are disclosed according to three options. The first shows the impact of costs on the internal rate of return. The second highlights the impact of costs on principal and interest. The third unbundles the product fair value into its bond and derivative components, with specific indication about costs. First, investors were asked to rate the complexity and the usefulness of the Templates and to assess the riskiness of the presented products. In order to control for familiarity bias, in the first stage of the test neither the issuer’s name nor the type of the assets were disclosed. Perceived complexity turns out to rise moving from the syn-thetic representation to the unbundled one and reaches its highest for the performance scenarios (both what-if and probabilistic). As for usefulness, both what-if and probabilistic modelling are perceived to be less useful than the synthetic and unbundled representations. Perceived complexity and perceived usefulness of financial information are generally inversely related: in other words, the higher the complexity of the information, the lower the perceived usefulness. Second, in order to assess the relation between information disclosure and risk perception investors were asked to rank products by their riskiness. In general, risk perception results to be positively affected by perceived complexity of the information disclosure. The percentage of respondents correctly ranking the risk of products is higher when unbundled formats are used, whereas performance scenario representations are associated with a higher percentage of wrong answers in ranking products’ riskiness. In details, risk tends to be more frequently over-estimated when participants inspect the what-if scenario representation and to be more frequently under-estimated when probabilistic modelling is taken into account. Finally, respondents were asked how much they would invest in each product, given an initial endowment, a time horizon and an investment objective. This allowed observing propensity towards investment driven exclusively by the representation of the financial information. As expected, perceived complexity results to be the main driver of the willingness to invest, since it always contributes to reduce propensity to invest. To this respect, perceived complexity seems to trigger a standard adverse selection problem: it is as if difficulty of understanding cast individuals into uncertainty, leading them to abstain from entering into the market. Financial knowledge, personal traits and investment habits do play a role in the perception of complexity and risk as well as in the attitude towards investment, although with a certain degree of heterogeneity across different representation modes. Higher levels of financial knowledge are generally negatively associated with perceived complexity and with indecision individuals may experience in the assessment of products’ risk. However, being less hesitant is generally associated with the wrong risk ranking. Another interesting consideration is that, in line with the insights of the behavioural literature, in our sample high financial ‘literate’ individuals are not necessarily free of inclination towards behavioural biases. This evidence, coupled with a positive correlation between risk propensity (as measured through the Grable & Lytton test) and the inclination towards behavioural biases, would point to a latent variable, i.e. the overconfidence fed by a good level of financial knowledge, driving the positive relation between high knowledge and inclination towards behavioural biases. This point claims for financial education initiatives attuned also as debiasing programs, in order to be an effective investor protection tool. Finally, making frequently investment decisions, delegating investment choices to an expert, trusting fi-nancial advisors are all associated with an easier understanding of financial information and a higher propensity to invest. This evidence indirectly confirms that financial experts and advisors may actually make the difference, by playing an educational role and, by this way, changing individuals’ attitude towards financial choices. Overall, the present paper shows that risk perception is context-dependent and mainly determined by the way financial information is disclosed. It adds to the existing literature by providing new evidence on the impact of framing of different representation modes, partially overlapping with the formats mandated by regulators and super-visors and/or used by the industry. Relying on the actual appraisal elicited from a sample of Italian investors, the study provides insights on how people actually read and understand financial information, which may turn useful in the design of financial disclosure and investor education programmes. For instance, it highlights that simplifying financial disclosure is not sufficient to ensure correct risk perception and unbiased investment choices. Moreover, evidence about investors’ heterogeneity and behavioural biases affecting risk perception supports the idea that the ‘optimal’ disclosure may not exist and the ‘one-size-fits-all’ approach cannot be effective in ensuring a suitable level of investors protection. This paper is in line with the approach adopted by some regulators increasingly engaged in the definition of evidence-based rules and may offer useful insights for the design of effective investor education programmes.

We wish to thank Giuseppe D’Agostino and Giovanni Siciliano, for very helpful comments. A special thanks goes to Prof. Francesco Chelli of Università Politecnica delle Marche for valuable advice in framing the sampling procedure. We also thank Chiara Cavalletti and Laura Conti for their excellent research assistantship and the students of the internship program of Università Politecnica delle Marche, Serena Cappelletti, Francesco Postacchini and Thi Anh Tuyet Valentina Tran, that performed scrupulous data entry and data clearing. We would also like to thank Abi, Assoreti and Federcasse, all the Financial Institutions (Intesa San Paolo, Unicredit, Monte Paschi Siena, Banca Nazionale del Lavoro, Banca Sella, Banca Fideuram, and Banca di Credito Cooperativo di Fano) and all the investors that agreed to participate in the research project. The responsibility for any mistakes and for the opinions expressed remains our own. The ideas and positions in the paper are personal views of the authors and cannot be attributable to Consob.