Consistent Forecasting vs. Anchoring of Market Stories: Two Cultures of Modeling and Model Use in a Bank

Abstract

ArgumentIt seems theoretically convenient to construe knowledge practices in financial markets and organizations as "applied economics." Alternatively or additionally, one might argue that practitioners draw on economic knowledge in order to systematically orient their actions towards profit-maximization; models, then, are understood as devices that make calculative rationality possible. However, empirical studies do not entirely confirm these theoretical positions: Practitioners' actual calculations are often not "framed" by models; organizations and institutions influence the choice and adoption of models; and different professional groups in financial markets have diverging attitudes towards model calculations. In order to account for this diversity, this article proposes the concept of cultures of economic and financial expertise; the concept focuses on the patterns of knowledge practices and knowledge-related self-definitions of groups within financial organizations and markets; it also facilitates an analysis of the relations between and, more specifically, the hierarchies among different practices and identities. The article then goes on to explore the process of foreign exchange forecasting in a particular bank. The description immediately reveals that two groups are involved in this process: economists and analysts. These groups maintain quite different practices and self-descriptions in relation to models: While the economists in the bank use the models as organizational resources for consistent forecasting procedures and observe data with the help of simple model structures, analysts approach model forecasts from the perspective of critics: They see limits in the variablecentered, as opposed to a "thematic" approach and they disregard a model's imposed temporality. Nevertheless, analysts use model forecasts as anchors for developing their own "paths" and stories about possible future expectation changes in the markets. The specific division of labor between economists and analysts, and specifically the dominant role of analysts in the forecasting process, fit into a larger picture: The rise of institutional investors in the foreign exchange markets and their demand for genuine market knowledge for speculative investment projects has contributed to the rise and dominance of analysts' culture. of expertise include accounting (Power 2010), risk management (Kalthoff 2005;

The virtue of simplicity: On machine learning models in algorithmic trading

“…Various SSF studies have explored the reciprocal influence of models, equations, and algorithms on one hand, and developers and users on the other. MacKenzie's work on the performativity of the Black-Scholes options pricing model is an example of a tone-setting contribution to the field (MacKenzie, 2003(MacKenzie, , 2008MacKenzie and Millo, 2003), but several other thorough empirical studies have deepen the understanding of human-model interactions and entanglements in the financial markets (such as MacKenzie, 2011;Spears, 2014a, 2014b;Millo and MacKenzie, 2009;Svetlova, 2012Svetlova, , 2013Wansleben, 2014). One thing that differentiates the machine learning models examined in this study from, for example, the discounted cash flow valuation models examined by Svetlova (2012Svetlova ( , 2013 is the capacity of the former class of models' to learn and optimise without interference from model developers and users.…”

The Politics of Uncertainty

“…A number of the book's chapters reflect on a wide range of models. These include the economic forecasting models used by banks (Wansleben 2014); the actuarial and parametric models central to the (re)insurance industry (Johnson 2013); the infectious disease models that predict patterns of spread and impact and the many models that aim to predict the impacts of natural hazards -from floods to volcanoes to earthquakes (Hough 2002). Here as elsewhere, modelling struggles to make sense of uncertain, complex systems, often aiming to predict future patterns.…”

Model Talk: Calculative Cultures in Quantitative Finance

“…Calculative idealists have great faith in their models and no worries about basing decisions on model output. Calculative pragmatists, on the other hand, do not expect formal models to produce accurate representations of reality (Svetlova 2018, 69-70; Svetlova and Dirksen 2014; Wansleben 2014, 608). Although not opposed to model use, pragmatists or skeptics “treat model output as the starting point for…the exercise of judgment,” and thus not as the primary basis for decision-making (Mikes 2011, 240).…”

“… 3. There are numerous examples of social studies of finance (SSF) scholarship exploring dynamic entanglements of humans and models in various areas of the financial industry. Studies have explored model use and modeling cultures in investment banking (Beunza and Stark 2012; Lépinay 2011; MacKenzie 2011; MacKenzie and Spears 2014a, 2014b; Wansleben 2013, 2014) and asset management firms (Sevtlova 2012, 2013, 2018). More recent studies have looked at the use of algorithms in high-frequency trading (Borch and Lange 2016; Lange 2016; MacKenzie 2017, 2018; Seyfert 2016) and credit rating agencies (Besedovsky 2018).…”