AMF Scientific Advisory Board meets to discuss the shareholdings of French households
Research presented at the Scientific Advisory Board meeting of 10 October 2017 focused on the core issue of retail investors' shareholdings.
Luc Arrondel (Paris School of Economics, member of AMF Scientific Advisory Board) and one of his co-authors (Hector Calvo-Pardo, University of Southampton) presented the results of a model that aims to measure how social interactions affect stock market participation. The discussion was led by Gunther Capelle-Blancard.
Marie-Hélène Broihanne (University of Strasbourg) gave an overview of the results of three academic papers (two of which she co-authored) that use responses to MiFID ‘know your customer’ (KYC) questionnaires to assess how accurately they can predict investment in risky assets. The discussion was led by Patrice Poncet.
Presentation by Luc Arrondel (PSE) and Hector Calvo-Pardo (University of Southampton): “Informative social interactions”
In their opening statement, the authors presented a few stylised facts on investment in risky assets, more particularly direct and indirect shareholdings, which have been in decline for a decade, even within more wealthy households. This trend is not unique to France; it has also been observed in the other major euro area economies. In fact, the PATER survey, which serves as the basis for the research presented to the Scientific Advisory Board, reveals a number of positive trends since the previous survey in 2014, including higher expected returns, a greater willingness to invest in risky assets and more confidence in financial intermediaries.
The paper presented to the Board assesses how social and information networks influence the financial investment behaviour of households. The authors have developed a theoretical model, according to which the demand for risky assets depends on the private information at investors’ disposal and the information they receive through social interactions. The model predicts that higher social connectedness implies higher demand for risky assets. Such an effect is magnified as the precision of the signals pooled from these social interactions improves.
The model was tested on data from surveys carried out by TNS Sofres in December 2014 and May 2015 on a representative sample of the French population. These data reveal the nature of a respondent’s social circle, and more specifically the members of this social circle with whom they specifically interact on financial matters, i.e. their financial circle.
The authors conclude that:
When an individual has a financial circle, they are more likely to participate on the equity markets and invest in shares because their social interactions adjust their expected returns. The more precise the information they receive, the greater the effect.
These interactions are destined to become even more significant owing to the growth of social networking technology.
Social interactions can therefore represent a very powerful channel through which financial education and information can influence stock market participation.
Social interactions may be considered a good substitute for financial advice, particularly if such advice is of poor quality.
Social interactions are not, however, exempt from risk, particularly the mindless imitation of behaviours observed within the broader social circle.
The members of the Scientific Advisory Board thanked the authors for their interesting work. Discussions on the decline in stock market participation focused on aspects of demand (particularly the role of real estate) and supply (less privatisation and the role of prudential regulation), as well as the effects of the growth in high-frequency trading. On the other hand, it would appear that taxation, which remained fairly constant during the research period, has little effect. The Board felt it was a shame that since employee stock ownership plans were not open to everyone, they had not been factored into the analysis. Discussions on how social interactions affect stock market participation focused mainly on investors’ behavioural bias and forecasting errors (one of the main conclusions of the paper presented to the Board was that informative social interactions would help to correct these errors).
Presentation by Marie-Hélène Broihanne (University of Strasbourg): “MiFID questionnaire answers: stock market participation, appetite for information and investor sentiment”
Marie-Hélène Broihanne, Professor of Finance at the University of Strasbourg, presented the results of three separate studies (two of which she co-authored) that use responses to MiFID ‘know your customer’ (KYC) questionnaires. The first paper, co-authored with Hava Orküt and titled MiFID questionnaire answers and stock market participation, combines clients’ responses to MiFID questionnaires with actual shareholding data from the bank's records. The overall dataset covers more than 70,000 retail clients between 2007 and 2015. The other two papers use responses to questionnaires distributed by a Belgian investment bank, which in theory has a significantly different client base owing to its positioning in a more specialised market segment. The dataset includes information and questionnaire responses from 45,000 individual investors between 2008 and 2012. One of these papers, co-authored with A. Bellofatto and titled Appetite for information in mandatory profiling of individual investors, attempts to measure how a client’s willingness to respond to an additional, optional questionnaire affects their investment behaviour. The other paper, written by C. D’Hondt and P. Roger, focuses on the extent to which investor sentiment can predict future returns, depending on whether they use financial advice or information. Paradoxically, it would appear that the sentiment of more ignorant investors is a better predictor of returns. The paper is titled Investor sentiment and stock return predictability: The power of ignorance.
Paper 1: The authors use a binary logistic model to show that the information provided in the questionnaires relating to risk perception and tolerance of losses is a good predictor of equity investment decisions. In addition to these behavioural factors, other individual parameters are included in the model as control variables: socio-demographic factors (age, gender, location, marital status, socio-professional category) on the one hand, and information relating to finances and assets (income, outstanding loans, unit-linked life insurance policies or retirement savings plans, length of banking relationship) on the other.
Paper 2: In the second paper, the authors divide retail investors into two groups based on whether, in addition to the mandatory MiFID appropriateness test, they agreed to fill in an additional, optional suitability questionnaire entitling them to automated financial advice. Those who agreed (Group S) are deemed to have shown a greater appetite for information than those who refused (Group A). Meanwhile, the investors’ portfolio data is reconstructed using the Eurofidai database. After a matching procedure that reconciles the observable characteristics of the two groups, univariate and multivariate analysis is performed on various elements of trading behaviour. The investors who completed the optional questionnaire have more diversified portfolios, including a wider range of financial products, and receive better returns than those who did not.
Paper 3: In the third paper, C. D’Hondt and P. Roger used data from the Belgian investment bank to construct a unique, quantitative measuring tool that provides an intuitive indicator of investors’ optimism (this measure having been introduced by P. Roger in a 2014 paper). The authors show that the sentiment of less well informed investors (A) is a more powerful predictor of the returns from a long-short portfolio based on size than the sentiment of the other investors (S).
The Scientific Advisory Board members praised the quality of the papers presented to them, and in particular the innovative way in which they compared clients’ own declarations in questionnaires with their actual investment behaviour. This type of research is more developed in the US, however the structural difference between the markets prevents us from simply transposing the results from American studies into continental Europe. The Board discussed the main findings from the papers, namely: the predictive power of disclosed preferences regarding the ownership of risky assets; the fact that an apparent appetite for information does indeed translate into different behaviour; and the power of the sentiment of less well-informed investors to predict returns.
These discussions focused on the biases created by the questionnaires, which are caused by the very nature of the questions themselves and can be exacerbated when responses are given face to face with an advisor. Moreover, investors may decide to restrict their financial knowledge as much as possible in order to enjoy greater levels of protection. The limited time period covered by the sample, and the omission of certain variables, may also be a source of bias. Several suggestions were put forward with a view to removing some of these biases: introducing technical questions to the MiFID questionnaires in order to check that clients’ declarations about their financial knowledge are reflected in their actual understanding of financial theory; distinguishing between direct and indirect investment in shares (investors’ behaviour can differ greatly depending on whether they hold physical securities or fund units); and introducing information about real estate assets, which are playing an ever greater role in portfolios.
The Board acknowledged the considerable differences between the client questionnaires, in terms of structure, content and quality, as well as the remedial work needed to achieve the initial objective of these questionnaires, namely to make clients more aware of the risks involved in investing.
Head of publications: The Executive Director of AMF Communication Directorate. Contact: Communication Directorate – Autorité des marches financiers 17 place de la Bourse – 75082 Paris cedex 02