In this post Dr Ekaterina Svetlova, Associate Professor in Finance and Accounting in ULSB, discusses her new book assessing the influence of financial models on markets and society. The book questions the assumption that financial markets have become a purely analytical and quantitative place and argues that human influence has not disappeared, but it unfolds in the multifaceted interplay between users and models in the practice of markets.
The recent stock market correction has again raised the question of who is in control of markets: humans or technology. Nasdaq CEO Adena Friedman said on CNBC that “humans are definitely in charge of the decisions in the market” and that “the algorithms are written basically on the back of a human decision.” At the same time, CNBC quoted influential banking analyst Dick Bove who claimed that “the United States equity markets has been captured by out-of-control technological investment systems”. My new book broadly addresses this controversy.
It challenges the accusations made towards financial models (not just algorithms) in the recent years, especially in the aftermath of the 2008 crisis. The arguments behind these accusations are familiar: financial models are abstract and unworldly constructs so that their users are predestined to be misguided. Thus, the argument goes, as insufficient models became widespread tools for decision-making in financial markets, the vast majority of market participants were seduced by their mathematical sophistication and followed them towards alleged safety.
The book argues, however, that models cannot be condemned indiscriminately. Generally, it claims that the discussion about dangers induced by financial models is based on the principal misunderstanding of models’ roles in markets. The aforementioned accusations are rooted in a conceptualization of models as “calculative tools” which directly guide investment decisions. In other words, there is an implied assumption that models tell people what to do and that the latter blindly follow models’ advice. However, this book suggests – and demonstrates using various empirical examples – that financial models do not ultimately determine investors’ decisions and actions. The role of models is subtler: they do not dictate financial decisions but make them possible in different ways. This insight clearly relativizes the performativity thesis that claims that models make markets.
Financial decision-making is characterized by radical uncertainty which is not calculable. Thus, calculations provided by models are unable to grasp the ever-changing, uncertain reality of markets. In other words, there is always a gap between models and (market) reality. However, financial market participants cannot allow themselves to be detached from the markets. They are a part of the markets themselves and, thus, their decisions cannot be based on model calculations only. That is, on calculations that leave out expectations, emotions, stories, judgments and – importantly – the modelling efforts of other market participants.
Models are constantly connected with markets in action-like decision-making which implies more than calculation. This connecting involves engaging with the world by simultaneously observing, deciding and taking actions and happens in various ways and styles.
The book uses qualitative research methods to investigate and systematize distinct practices of model use in financial decision-making. This analysis and systematization is an important novelty of the book. The case studies clearly highlight that there is no separation between calculation, judgment, decision-makers and markets. Rather, there are constant shifts of attention from models to markets and back as well as the constant formation of judgments and their application, of which models and markets are a subject but also a part. It is about judgment with models and about models, with markets (what do others think?) and about markets (what is my view?).
The book empirically identifies three general modes of “bringing together” models, markets and users in the process of action-like decision-making: “qualitative overlay”, “backing out”/”implied modelling” and “models as opinion proclaimers.”
In the case of “qualitative overlay”, the model results are compared with a pre-formed judgment and serve as an anchor but not as an ultimate guide for decisions. For example, fund managers often use the Discounted Cash Flow model as a point of departure for decision-making but frequently overlay it by their judgment call rendering the model just a supportive tool.
In the case of “backing out” and “implied modelling”, models are used to observe the markets and to figure out the mistakes in model users’ estimates in order to correct the mistakes or to identify new investment opportunities. Good examples here are reverse engineering in portfolio management and “reflexive modelling” in the field of merger arbitrage. In the latter case, arbitrageurs “back out” the implied probability of a merger, the probability namely that is assigned to the merger’s success by the market. Then, the “implied” probability is compared to the traders’ subjective views. If there is a disagreement (“dissonance”), traders start to ask: “What am I missing?” or “What do my models not see?” Hence, merger arbitrageurs use models to determine the behaviour of other investors, to compare their own views with the views of others and to establish a position based on this socially informed calculation.
Finally, models can be applied to express market participants’ pre-formed opinions about the market or a security. The book shows how security analysts form their views about a company and then identify the numerical parameters that should be inserted into a model in order to support their pre-formed views.
Furthermore, the book makes a general distinction between frontstage use of models (decision-selling and justification) and back-stage use (decision-making). In other words, it demonstrates that there are not only different styles of financial models’ involvement in decision-making but also that models play various roles in decision-selling. They provide legitimacy for decisions, perform impression management, convince others to invest and help to reach a consensus; those secondary functions can overlay the primary goals of model use (i.e. calculation) and render models less important for immediate decision-making.
The analysis of cultures of model use in the book suggests that model users are by no means “model dopes”; they do not “blind out” model deficiencies but rather consciously make “insufficient” models work in a not uniform way. The central argument is that it is exactly because the styles of model use vary that there is no way that different users derive the same results from their models and make the same decisions.
Hence, models’ influence on markets and society is not straightforward; their power is far more fragile than widespread criticism would indicate. The general claim that financial markets have become a purely analytical and quantitative place might be exaggerated. Human influence has not disappeared; rather, it unfolds in the multifaceted interplay between users and models in the practice of markets. The book demonstrates that we find large “pockets” where human judgment and stories are as important as the complicated formulas and algorithms. Hence, the general designation of financial models as the ultimate villains is questionable because their influence on markets is clearly mediated by the practices of their use.