Personal Finance and Artificial Intelligence (AI): Don't be Fooled by the Stork!
When we use artificial intelligence for managing our personal finance, we often see it as a barrier protecting our decisions from the impact of our emotions. Yet, it could also be misleading, particularly in correlation and causation analysis. Edouard Camblain, our behavioural finance expert and Head of Strategic Projects & Development at Societe Generale Private Banking, gives us the low down.
Storks and high heels: blurring the line between causation and correlation
Causation error which is broadly defined as:
the erroneous assumption that because one event follows another, the first event is the cause of the second event (known as post hoc ergo propter hoc(1));
the inversion of meaning between cause and effect;
or the erroneous assumption that one thing is the cause of another when in fact they are two effects of one underlying cause.
There are many examples of causation error. The most prominent is the “stork effect” — sometimes used as a synonym of causation error — according to which villages where storks nest have higher birth rates, proving that babies are in fact delivered by storks. But you don’t need to be rocket scientist to see that the most logical explanation is that storks prefer nesting in small, rural villages that in sprawling big cities where the relative birth rate is lower.
In the finance arena, three correlation theories for forecasting economic cycles have been hyped up:
The “hemline index”, introduced in 1926 by the economist George Taylor, which suggests that the length of women’s skirts is an indicator of a country’s economic health;
The “lipstick index”, coined by businessman Leonard Lauder in 2001, according to which lipstick sales are an inverse indicator of economic health;
The “high heel index”, created by IBM in 2011(2) based on a study over a 100-year period, according to which the height of high-heels tend to increase during times of economic recession.
Alas, none of these have proved perfectly accurate. Unrealistic, they fall into the first category of causation errors by assuming that two successive events make one the cause/consequence of the other.
Some even go to absurd lengths in their quest financial market forecasting models(3). David Leinweber(4) is known for taking the satire to the point of irony, demonstrating that between 1981 and 1993 the US S&P index (the index of the 500 largest US listed companies) was 75% correlated with butter production in Bangladesh, 95% including American butter... and 99% when including the number of sheep of both countries!
AI: the champion of correlation let down by causal links
While technological progress such as processing capacity, big data, and artificial intelligence effortlessly identify correlations, they have a much harder time with causal links. Human error — and causation error in particular — could well persist.
Artificial intelligence is often presented as a "miracle” cure for the influence our emotions have on financial decision-making. But the confusion between correlation and causation is one of the effects that can result from our emotions; and it presents a real disadvantage in the area of personal finance, for the search for advanced indicators on investment trends (corresponding to causes) may lead to entirely ineffectual strategies.
And the conviction of having found the winning formula thanks to a solid correlation (bringing an overconfidence bias into play…) runs a high risk of dismissing the rules of caution. And bear in mind that while the above “correlations” did work at certain points in time, they did not hold over the long term.
As a final point, I cannot emphasise enough that even strong historical corrections are in no way a prediction of the future — they are far from being a reliable indication of correlations and/or future performance!
And since I cannot be sure shorter articles are more read than longer articles (causation), or whether it’s all the same (correlation)... I’ll stop here!
(1) Latin for “after this, therefore because of this”.
(3) Take a look at the website of Tyler Vigen who has built a collection of erroneous correlations:
(4) Author of the book Nerds on Wall Street: Math, Machines and Wired Markets.
Would you like to discuss this subject further with us?
Grille Disclaimer
GENERAL WARNING:
Societe Generale Private Banking is the Societe Generale Group business line operating through its registered office within Societe Generale S.A. and departments, branches or subsidiaries, located in the territories mentioned below, acting under the brand name “Société Générale Private Banking” and distributors of this document.
This document, of an advertising nature, has no contractual value. Its content is not intended to provide an investment service, it does not constitute investment advice or a personalised recommendation on a financial product, or a personalised advice or recommendation on insurance, or a solicitation of any kind, legal, accounting or tax advice from any entity reporting to Société Générale Private Banking.
The information contained is for information purposes only, may be modified without prior notice, and is intended to communicate information that may be useful for decision-making. Any information on past performance reproduced does not guarantee future performance.
The private bankers of the Société Générale Private Banking entities are available to potential investors to provide them with more information on the variations, within the Société Générale Private Banking entity concerned, the theme presented in this document.
This document is confidential, intended exclusively for the person who consults it, and may not be communicated or disclosed to third parties, nor reproduced in whole or in part, without the prior written consent of the Société Générale Private Banking entity concerned.
No entity reporting to Société Générale Private Banking may under any circumstances be held liable for any decision taken by an investor based solely on the information contained in this document.
The Societe Generale Group maintains an effective administrative organisation taking all necessary measures to identify, control and manage conflicts of interest. To this end, Societe Generale Private Banking entities have put in place a conflict of interest management policy to manage and prevent conflicts of interest. For more details, Societe Generale Private Banking clients can refer to the conflict of interest policy available on request from their private banker.
Societe Generale Private Banking has also put in place a policy for handling complaints made by its clients, available upon request from their private banker or on the Societe Generale Private Banking website (www.privatebanking.societegenerale.com).
SPECIFIC WARNINGS BY JURISDICTION
France: Unless expressly stated otherwise, this document is published and distributed by Société Générale, a French bank authorised and supervised by the Autorité de Contrôle Prudentiel et de Résolution, 4, place de Budapest, CS 92459, 75436 Paris Cedex 09, under the prudential supervision of the European Central Bank (“ECB”) and registered with ORIAS as an insurance intermediary under number 07 022 493 orias.fr. Societe Generale is a French limited company with capital of €1,010,261,206.25 at 1 February 2023, with its registered office located at 29 boulevard Haussmann, 75009 Paris, and with a unique identification number of 552 120 222 R.C.S. Paris. More details are available on request or at www.privatebanking.societegenerale.com.
Luxembourg: This document is distributed in Luxembourg by Société Générale Luxembourg, a public limited company registered with the Luxembourg Trade and Companies Register under number B 6061 and an authorised credit institution governed by the Commission de Surveillance du Secteur Financier (“CSSF”), under the prudential supervision of the European Central Bank (“ECB”), whose registered office is located at 11 avenue Émile Reuter – L 2420 Luxembourg. More details are available on request or at www.societegenerale.lu. No investment decision of any kind could result from reading this document alone. Société Générale Luxembourg accepts no responsibility for the accuracy or other characteristics of the information contained in this document. Societe Generale Luxembourg accepts no responsibility for the actions taken by the addressee of this document solely on the basis of this document, and Societe Generale Luxembourg does not present itself as providing advice, in particular as regards investment services. The opinions, views and forecasts expressed in this document (including its annexes) reflect the personal opinions of the author/authors and do not reflect the opinions of other persons or of Société Générale Luxembourg, unless otherwise stated. This document was prepared by Société Générale. The CSSF has not carried out any analysis, verification or control over the content of this document.
Monaco: This document is distributed in Monaco by Société Générale Private Banking (Monaco) S.A.M., located at 11 avenue de Grande Bretagne, 98000 Monaco, Principality of Monaco, governed by the Autorité de Contrôle Prudentiel et de Résolution and the Commission de Contrôle des Activités Financiers. Financial products marketed in Monaco may be reserved to qualified investors in accordance with the provisions of Law No. 1.339 of 07/09/2007 and Sovereign Order No. 1.285 of 10/09/2007. More details are available on request or at www.privatebanking.societegenerale.com.
Switzerland: This document may constitute advertising within the meaning of the Financial Services Act (“FinSA”). It is distributed in Switzerland by SOCIETE GENERALE Private Banking (Switzerland) SA (“SGPBS” or the “Bank”), whose head office is located at rue du Rhône 8, CH-1204 Geneva. SGPBS is a bank authorised by the Swiss Financial Market Supervisory Authority (FINMA). This document cannot be considered as investment advice or recommendation by SGPBS. The Bank recommends obtaining professional advice before acting or not acting on the basis of this document and accepts no responsibility for the content of this document. Financial instruments, including in particular units of collective investment schemes and structured products, may only be offered in accordance with FinSA. Further information is available on request from SGPBS or at www.privatebanking.societegenerale.com.
This document is not distributed by SG Kleinwort Hambros Bank Limited in the United Kingdom or by its branches in Jersey, Guernsey and Gibraltar acting together under the brand name “Kleinwort Hambros”. Consequently, the information provided and any offers, activities and financial and wealth information presented do not concern these entities and may not be authorised by these entities or adapted in these territories. Further information on the activities of Societe Generale’s private banking entities located in the territories of the United Kingdom, the Channel Islands and Gibraltar, including additional legal and regulatory information, are available at www.kleinworthambros.com.
Decryption of the projection bias
Behavioural finance is defined by the application of psychology to finance: it thus differs from classical financial...
Insights by our Wealth Engineering expert, Stéphane Maljevac.
Hospitality is a vibrant sector that’s always on the move and influenced by a multitude of economic, social and...