Financial data science: the birth of a new financial research paradigm complementing econometrics?

Abstract

Financial data science and econometrics are highly complementary. They share an equivalent research process with the former's intellectual point of departure being statistical inference and the latter's being the data sets themselves. Two challenges arise, however, from digitalisation. First, the ever-increasing computational power allows researchers to experiment with an extremely large number of generated test subjects (i.e. p-hacking). We argue that p-hacking can be mitigated through adjustments for multiple hypothesis testing where appropriate. However, it can only truly be addressed via a strong focus on integrity (e.g. pre-registration, actual out-of-sample periods). Second, the extremely large number of observations available in big data set provides magnitudes of statistical power at which common statistical significance levels are barely relevant. This challenge can be addressed twofold. First, researchers can use more stringent statistical significance levels such as 0.1% and 0.5% instead of 1% and 5%, respectively. Second, and more importantly, researchers can use criteria such as economic significance, economic relevance and statistical relevance to assess the robustness of statistically significant coefficients. Especially statistical relevance seems crucial, as it appears far from impossible for an individual coefficient to be considered statistically significant when its actual statistical relevance (i.e. incremental explanatory power) is extremely small. ARTICLE HISTORY

Perspectives on the Economics of the Environment in the Shadow of Coronavirus

“…When we do so (Model 4), it emerges that considering the entire universe of bonds issued by European energy companies, the ECB's portfolio is tilted not only to those energy companies that are more GHG intensive, but also to companies which are less transparent on their GHG Performance as well as those companies who are more likely to oppose progressive climate action. Having established statistical significance, we investigate the economic and statistical relevance (Brooks et al 2019). Inspecting the economic relevance of the GHG intensity variable in Model 3, we find GHG intensity to have the largest marginal effects.…”

Significance, relevance and explainability in the machine learning age: an econometrics and financial data science perspective

“…Notes: This table extends Table 2 of Brooks et al (2019) by the concept of explainability and further refines some of its content to reflect advances in econometrics and financial data science. Since machine learning applications represent economic as well as broader societal use cases, we amended the name of the column presented on the right.…”

“…Such inability for spot checking is unlikely to build trust towards users in credence services such as Significance Conventional statistical significance levels of 1, 5% and 10% may need to strengthen to 0.1%, 0.5% and 1% given the vastly increasing statistical power of big data. (see Brooks et al 2019, Table 1)…”

“…Hence, in introducing this special issue, we extend the work of Brooks et al (2019) and discuss the concepts of significance, relevance and, in particular, explanainability from a statistical and societal (e.g. economic) perspective.…”

“…2 Statistical relevance thereby indicates how commonly the respective coefficient occurs as explanation of the dependent variable's behaviour. In other words, while coefficients compete for statistical relevance, their statistical significance is, in principle, indifferent of each other 3 and largely dependending on the available degrees of freedrom (Brooks et al 2019).…”