Access the presentation of the AMF
Access our missions: regulate, supervise, inform and protect
On its December 5 meeting, the Scientific Advisory Board discussed a paper on the role of markets in financing the economy and another one on the behavioral biases of financial analysts.
The Scientific Advisory Board met on 5 December 2017. Tristan Roger, winner of the 2017 AMF Young Researcher Award, presented the results of his research on financial analysts’ behavioural biases. Patrick Artus presented an article on the role of financial markets in the funding of companies, written in collaboration with Laurence Boone.
Patrick Artus (Natixis, AMF Scientific Advisory Board member) and Laurence Boone (AXA IM, Scientific Advisory Board member) note that financial markets are less and less used to finance private companies. Indeed, in the US, a large withdrawal of financial assets has taken place over the past 15 years. The number of listed companies, as well as that of initial public offerings is clearly decreasing. What is more, securities issuance, most of which are bonds is essentially used to finance shares buybacks. In Europe, net issues on financial markets remain positive, but at low levels. Additionally, this trend is not offset by a rebound in bank credit.
According to the authors, this has to be linked to the increase in the rate of self-financing, which contributes to limiting the need to resort to financial markets. They highlight two potential explanatory factors: on the one hand, the lowering of the relative price of investment, combined with the increase in companies’ savings, and, on the other hand, greater labour market flexibility (which translates into a share of the value added that is unfavourable to employees). Regarding this general observation, France is a somewhat peculiar case. Indeed, France is the only OECD country where companies’ self-funding rates lies below 100%, which could explain a greater use of external financing to fund investments.
Other explanatory factors are mentioned, such as the development of alternative funding (boom in private equity, and on the unlisted securities markets more broadly). The authors also consider that financial regulations probably contributed to drive investors (and especially institutional investors) away from holding listed securities. Holdings of corporate bonds and shares are said to have decreased among banks and pension funds. For insurers, it remained broadly constant.
The authors point to a radical paradigm shift, in which excess savings are no longer used to finance companies, but rather governments, just as is the case in Japan. Along those lines, financial markets are only used to value companies and risks.
This model shift could potentially have significant consequences, most notably a worsening in the transmission of monetary policies to the real economy through the interest rate channel. Furthermore the authors raise the issue of the consequences on market liquidity, but reserve this matter for further investigation.
During the discussion, Board members qualified some of the conclusions drawn, mentioning in particular the differences between European and American markets. A closer study of the stockpiling of assets rather than of issue flows also enables such matters to be put in perspective. A separate analysis on securities’ issuance and redemption, as well as on net and gross flows could prove most enlightening. Additionally, a disaggregated analysis by industry sector would be useful.
Tristan Roger (Université Paris-Dauphine, Winner of the 2017 AMF young researcher award) presented the results of a joint work with Patrick Roger and Alain Schatt. The authors analyse the extent to which some cognitive biases that have been identified by neuro-science investigations, number representation in the present case, can be observed among financial analysts.
Neuro-science demonstrates that the human brain visualizes numbers in a straight line. In civilizations where people write from left to right, small numbers tend to be located on the left while large numbers are on the right, and conversely. Furthermore, among remote tribes, as well as among pre-school age children, it seems that numbers are processed on a logarithmic scale (i.e. numbers are perceived in relative rather than absolute terms) whereas the teaching of mathematics tends to create a linear representation of numbers for small quantities.
The authors combined the historical prices of 6 000 shares listed on the American market with a database that gathered price forecasts made by more than 9000 analysts between 2000 and 2013 (I/B/E/S). They show that financial analysts have different expectations on shares depending on whether their unit value is large or small: there exists a monotonic decreasing relationship between the unit price of a stock and the yield forecast provided by analysts. Hence when expressed in different terms, anticipations on “small price stocks” (whether optimistic or pessimistic) have a larger relative magnitude than for “large unit-price stocks”, a phenomenon that the authors call « small price bias ». This bias is confirmed by an analysis of recommendations (buy, sell …) as well as by a study focused on stock split operations (perceived optimism concerning price expectation increases simply because of the effective reduction in the unit face-value of the security).
To sum up, analysts are indeed affected by cognitive biases in number processing. For small price stocks, price variations are visualized along a linear scale (absolute price) while for large price stocks, analysts tend to process numbers in terms of yields (logarithmic scale). This bias could probably also be found among other populations of financial markets players, and could explain some irrational phenomena, such as abnormal returns and high volatility of small price stocks, or anomalous yields observed after stock split operations. A lab experiment is being considered to complement this analysis.
The members of the Scientific Advisory Board discussed the magnitude of the bias in question in relation to the other behavioural biases identified by the academic literature (excess optimism, excess self-confidence, herding behaviour …). They also asked whether data existed in order to replicate the analysis for other financial markets participants, and/or in other jurisdictions. They hypothesize that the tick size could play a significant role in the phenomenon observed by the authors.