SYNERGY OF CLASSICAL METHODS AND DIGITAL TECHNOLOGIES: NEW APPROACHES TO MODELING UNCERTAINTY AND STRUCTURAL SHIFTS IN THE ECONOMY

Анотація

This paper explores the evolution of economic modeling. Models today face two key
challenges: rising economic uncertainty and major structural shifts from digitalization. We argue that
the future of effective modeling is not a choice between classical models (like DSGE or CGE) and
new digital tools (like ML or NLP), but rather their deep synergy.
Modern economic processes are very complex. This creates a challenge for traditional modeling
methods. We see two main problems: First, digitalization is creating new markets and new types of
business (like the "digital economy"), which are difficult to describe with standard models. Second,
high economic uncertainty has become a key factor that influences investment and growth. However,
uncertainty itself is difficult to measure and put into a model [2]. Classical economic tools, especially
DSGE (Dynamic Stochastic General Equilibrium) and CGE (Computable General Equilibrium)
models are still the standard "workhorses" for policy analysis. DSGE models give us a strong
37
theoretical base to analyze monetary and fiscal policy. The model for a small open economy is a
perfect example [3]. CGE models are very useful for analyzing trade and connections between
different industries. Researchers are already adapting CGE models to study the new digital sector.
However, digital technologies now offer tools to help solve the limits of these classical models.
We focus on two key areas. For a long time, "uncertainty" was just an idea. It was hard to use in an
empirical model. But new digital tools like NLP (Natural Language Processing) and Big Data analysis
have changed this. Researchers can now analyze millions of newspaper articles to create objective,
real-time data, like the Economic Policy Uncertainty (EPU) index [2]. This index can be directly
integrated into DSGE models [3] to test how uncertainty shocks affect investment and growth in an
open economy. ML (Machine Learning) is not just for forecasting. As Susan Athey (2019) explains,
ML is also a powerful new method for causal analysis and for finding patterns in very large, complex
datasets [1]. For structural models, this is very important. For example, ML can help us move beyond
the simple idea of a single "representative agent" and build models that are more realistic.
To model the modern economy, we need a hybrid approach. We should not choose between
"theory" (classical models) and "data" (digital tools). Instead, we must focus on their synergy. The
future of economic modeling is in integrated models. These models will use the structure from
classical theory (DSGE/CGE) but will be powered by new data from NLP (like uncertainty indexes)
and analyzed with new methods from ML. This will help us create more realistic models for an
economy that is always changing.