Smiling robot as illustration for AI risks by MIM326 from Pixabay

AI risks: Researchpost #140

AI risks: 16x new research on AI, job risks, migration, climate, food, GHG accounting, biodiversity, broadband, return measures, listed private equity, Ethereum etc. by Lars Hornuf, Marc Elsberg and many more (#: SSRN downloads as of Aug. 24th,2023)

Social and ecological research: AI risks

Firing risks: Does Climate Risk Affect Employment Decisions? International Evidence by Claude Francoeur, Faten Lakhal, Hamza Nizar, Zvi Singer as of Aug. 13th, 2023 (#32): “Using a cross-country sample of 31,200 observations for the period 2011–2019, we find that climate risk due to extreme weather events is positively associated with underinvesting in labor and, in particular, with over-firing employees. … The results also show that the underinvestment behavior is less severe for firms that are more socially responsible” (abstract).

AI risks for jobs: The Short-Term Effects of Generative Artificial Intelligence on Employment: Evidence from an Online Labor Market by Xiang Hui, Oren Reshef, and Luofeng Zhou as of Aug. 21st, 2023 (#203): “This paper studies the short-term effects of generative AI and LLMs (Sö: Large language models) on labor outcomes by estimating the effect of ChatGPT on the employment of workers in a large online labor market (Sö: Upwork). Across the board, we find that freelancers who offer services in occupations most affected by AI experienced reductions in both employment and earnings. The release of ChatGPT leads to a 2% drop in the number of jobs on the platform, and a 5.2% drop in monthly earnings. …. top employees are disproportionately hurt by AI” (p. 13). My comment: I include HR service companies with good E, S and G-Ratings in my SDG-aligned portfolios because they help many people to find temporary and permanent new jobs.

AI risks (2): The Algorithmic Explainability “Bait and Switch” by Boris Babic and I. Glenn Cohen as of Aug. 20th, 2023 (#15): “Explainability in artificial intelligence and machine learning (“AI/ML”) is emerging as a leading area of academic research and a topic of significant regulatory concern. … We argue that for explainability to be a moral requirement — and even more so for it to be a legal requirement — it should satisfy certain desiderata which it currently does not, and possibly cannot. … the currently prevailing approaches to explainable AI/ML are (1) incapable of guiding our action and planning, (2) incapable of making transparent the actual reasons underlying an automated decision, and (3) incapable of underwriting normative (moral/legal) judgments, such as blame and resentment. This stems from the post hoc nature of the explanations offered by prevailing explainability algorithms. As we explain, that these algorithms are “insincere-by design,” so to speak” (abstract).

E-deficits: Is advanced digitalisation the philosopher´s stone or a complex challenge? – Experiences from Austrian and German EA practice by Birthe Uhlhorn, Gesa Geißler, Alexandra Jiricka-Pürre as of June 28th, 2023 (#9): “… While research increasingly discusses digital developments and their influence on procedural steps, the uptake of advanced digital tools remains limited amongst planning professionals in Germany and Austria. Practitioners still share concerns related to data quality, causalities and legal securities among others. … In addition, EA practice (Sö: Environmental assessment) in Germany and Austria lacks strategic discussion on the opportunities and challenges of digitalisation so far … this paper confirms the potential of digital solutions to improve the quality of EA processes and accelerate EA practice … ” (p. 15).

Pro/Con Migration: Attitudes to Migration and the Market for News by Razi Farukh, Matthias Heinz, Anna Kerkhof, and Heiner Schumacher as of Aug. 21st, 2023 (#6): “For Germany, we found that most national news outlets adopt an attitude to migration that is in between the two ideological extremes, but closer to pro- than to anti-migration campaigns. … Only the largest newspaper in Europe – the tabloid newspaper Bild – changed its attitude to migration from very positive to fairly negative within a few months, most likely in order to cater to readers’ political preferences. For Hungary, we found that the attitude to migration is on average more negative than in Germany. … for the US, we found that, the average attitude to migration in the market for news is comparable to that in Germany. However, both the most positive and the most negative news outlet in the our US sample are fairly large, which suggests that the degree of polarization in this market is substantial“ (p. 29).

Food risks: Coping with Climate Shocks: Food Security in a Spatial Framework by Diogo Baptista, John Spray, and D. Filiz Unsal of the International Monetary Fund as of Aug. 23rd, 2023 (#6): “… we show that (i) climate shocks are already having large negative impacts on GDP, nutrition and welfare, (ii) these impacts are disproportionately harming those households which are remote and food insecure, (iii) poverty and food insecurity exacerbates the impact of shocks. We go on to show that policy to lower the cost of trade and migration can lower the impact from climate shocks by allowing households alternative sources of income and affordable food“ (p. 34/35).

Responsible investment research

Green risks: Greening the Economy: How Public-Guaranteed Loans Influence Firm-Level Resource Allocation by Bruno Buchetti, Ixart Miquel-Flores, Salvatore Perdichizzi, and Alessio Reghezza as of July 14th, 2023 (#185): “First, we established that European banks face a ”green-transition-risk,” as less polluting firms have higher probabilities of default (PDs) than their more polluting counterparts (browner firms). … This higher implicit risk, called ”green transition-risk,” leads to a natural preference for lending to more polluting firms (browner firms). Secondly, we discovered that deploying PGLs (Sö: Public-guaranteed loans) during the pandemic resulted in a relative increase in lending to greener firms … PGLs eliminate (or drastically reduce) the ”green-transition-risk“ (p. 25).

GHG data: A rapid review of GHG accounting standards by Jimmy Jia, Kaya Axelsson, Abrar Chaudhury, and Evan Taylor as of July 29th, 2023 (#52): “We did a rapid systematic review of GHG (Sö: Green house gas) accounting standards to find that all are derivative works of the GHG Protocol. Further, commonly used GHG accounting standards are based on three methodologies. We found that the field converges quickly and there are fewer options than expected …“ (abstract).

Biodiversity-hole: Biodiversity Confusion: The impact of ESG biodiversity ratings on asset prices by Wei Xin, Lewis Grant, Ben Groom, and  Chendi Zhang as of Aug. 14th, 2023 (#35): “The biodiversity components of ESG ratings are analysed …. biodiversity ratings are largely uncorrelated to firm characteristics other than via firm size, and do not predict stock returns. … A suite of tests suggests that biodiversity as measured in ESG ratings does not appear to provide useful additional information for financial decision makers“ (abstract).

Rules or fiduciary? EU ‚Rule-based‘ ESG Duties for Investment Funds and their Managers under the European ‚Green Deal‘ by Sebastiaan Niels Hooghiemstra as of Aug. 15th, 2023 (#269): “This contribution focusses on explaining that the recently introduced “ESG duties” for European investment funds and their managers under European financial regulatory laws can be classified as “rule-based ESG duties,” largely substituting traditional corporate law “ESG fiduciary duties” applying to European investment funds and their managers” (abstract). My comment: My fund is compliant with Article 9 SFDR and has a social focus. For investors many of the current reporting requirements are not very helpful.

Other investment research: AI risks

Social broadband: Broadband Internet and the Stock Market Investments of Individual Investors by Hans K. Hvide, Tom G. Meling, Magne Mogstad, and Ola L. Vestad as of Aug. 2nd, 2023 (#422): “We find that broadband use leads to increased stock market participation, to improved portfolio allocation for existing investors, and to increased participation in bonds, bond funds, and unlisted stocks. We do not find adverse effects of internet use; for example, access to high-speed internet does not lead to excessive stock trading among existing investors, except possibly for the very most active investors. … Over the broadband expansion period, we observe a broad trend towards increased internet-based information acquisition and learning. … the effects of broadband on stock market participation are stronger for younger, lower-income, and lower-wealth individuals, who have the lowest stock market participation rates and likely the lowest financial literacy to begin with …“ (p. 34/35). My comment: This is one reason why I include telecom infrastructure providers and servicers in my SDG-aligned portfolios.

Wrong measures? How Should Returns to Long Term Investors be Measured? by Hendrik Bessembinder, Te-Feng Chen, Goeun Choi, and John Wei as of Aug. 2nd, 2023 (#131) “Widely studied databases contain stock returns measured at the monthly horizon. The most common method of aggregating this information across multiple months is to compute arithmetic means of the monthly returns. … However, arithmetic mean returns are potentially very misleading as to investors’ experiences across multiple months. … We use a broad sample of over 71,000 stocks to demonstrate the extent to which conclusions regarding long-term investment performance can differ across measures, with the goal of guiding market observers to the measure that is most relevant for the task at hand”.

Listed PE: Thematic Investing With Big Data: The Case of Private Equity by Ludovic Phalippou as of March 13th, 2023 (#1159): “Using natural language processing, we score companies based on the frequency with which news articles contain both their names and terms Private Equity and Leveraged Buy-Out. An index is then created … with the weights set as a function of a company exposure to this theme. … this listed private equity index is highly correlated to commonly used private equity fund market indices …. In addition, our index has similar returns as non-tradable LBO fund indices” (abstract). My comment: Since many years, I include listed private equity in my alternatives allocations for traditional ETF-portfolios.

Good reporting: The Value of Publicly Available Information on Acquired Firms in Corporate Acquisitions by Dan Givoly, Songyi Han, and Sharon P. Katz as of July 5th, 2023 (#62): “Acquiring privately held firms enables acquirers to benefit from liquidity and information risk discounts extracted from the owners of private firms. The information risk arises from the information asymmetry between the acquirer and the private firm due to the lack of public information. … Our study analyzes the outcomes of acquisitions of three types of target firms: private firms, public firms, and quasi-private firms, i.e., privately-owned firms that are subject to financial reporting obligations. …. the outcomes of acquisitions … are significantly more favorable for the acquisition of quasi-private firms than for acquisitions of both public and private firms. Further, despite all the measures employed by acquirers to mitigate the higher information risk involved in acquiring private firms, including potential price discounts, they do not fully compensate for this added risk“ (p. 26/27).

Fin-MaL = Fin-Good? Financial Machine Learning by Bryan Kelly and Dacheng Xiu as of July 25th, 2023 (#22312): “We survey the nascent literature on machine learning in the study of financial markets. We highlight the best examples of what this line of research has to offer and recommend promising directions for future research. This survey is designed for both financial economists interested in grasping machine learning tools, as well as for statisticians and machine learners seeking interesting financial contexts where advanced methods may be deployed” (abstract). My comment: See my recent publication AI: Wie können nachhaltige AnlegerInnen profitieren? – Responsible Investment Research Blog ( or How can sustainable investors benefit from artificial intelligence? – GITEX Impact – Leading ESG Event 2023

OK crimes? Cybercrime on the Ethereum Blockchain by Lars Hornuf, Paul P. Momtaz, Rachel J. Nam, and Ye Yuan as of Aug. 15th, 2023 (#557): “We identify more than 1.78 million transactions that are externally verified to be linked to cybercrime, corresponding to an aggregate amount of $1.65 billion of funds lost. … we find that victims increase their overall risk-taking … we show that victim and cybercrime addresses differ systematically, leading to variation that can be exploited in predictive models to screen for cybercriminals ex ante“ (p. 39).

Geo-Engineering: oC Celsius (kostenpflichtig) ist der neueste Öko-Polit-Thriller von Marc Elsberg vom März 2023. Dabei geht es um Geo-Engineering und dessen potenziellen ökologischen und politischen Chancen und Risiken. Celsius ist kein wissenschaftliches Buch, aber es sollte zum Nachdenken anregen.


Advert for German investors:

Sponsor my research by investing in and/or recommending my global small/midcap mutual fund (SFDR Art. 9). The fund focuses on social SDGs and uses separate E, S and G best-in-universe minimum ratings and broad shareholder engagement with currently 29 of 30 engaged companiesFutureVest Equity Sustainable Development Goals R – DE000A2P37T6 – A2P37T; also see Active or impact investing? – (