Neural network models for assessing the impact of intangible assets on microeconomic indicators
Relevance
ESG factors are becoming more and more important for the development of companies every year. Such factors include environmental indicators, social factors and corporate governance factors. These factors affect both the company's operations and its perception by investors and the external environment, which makes the study of ESG factors and their dynamics an important task from the point of view of each company's stakeholders.
Project goal
is to develop and train a neural network NLP model with a well-defined distance function on the corpus of ESG report texts to identify ESG factors and their impact on ESG ratings and financial performance of major companies.
Project task
is to adapt existing language models, in particular the BERT model (modifying the final layers of a neural network to explicitly control the behaviour of the distance function in the embedding space of individual text sentences).
-
Data source:
Reports and other related documents classified by public companies as ESG-disclosure. The largest 500 companies traded on the US exchanges NYSE and NASDAQ from 2010 to 2020 were selected for analysis.
