Stress Testing as a Tool for Monitoring and Modelling the Dynamics of Business Activity of Manufacturing Enterprises in Russia in the Face of Market Shocks: Short-term Scenarios of Industry Tendencies
34 Pages Posted: 1 Jun 2020
Date Written: May 27, 2020
The article proposes a methodology for using macro-level stress testing based on the results of business tendency surveys to study possible scenarios for the development of crisis dynamics triggered by external unforeseen supply and demand shocks, as in the case of the COVID-19 pandemic, as well as a review of existing approaches in the field of stress testing and building stress indices with an emphasis on methods based on vector autoregressive models and their various modifications.
The basis for empirical calculations is data from business tendency surveys of the leaders of Russian manufacturing enterprises, reflecting their combined estimates of the current state of business activity. Based on the results of business tendency surveys, four composite indices were formed reflecting various aspects of business activity of enterprises: demand index, production index, finance index and employment index. Index values calculated monthly from 2008 to March 2020 were used to build the Bayesian vector autoregressive model (BVAR). This model was used to predict the dynamics of indices under the condition of four possible shock scenarios: short-term shock, V-shaped shock, W-shaped shock and U-shaped shock. Moreover, for each of the scenarios, cases of a shock of demand, a shock of production, and a simultaneous shock of demand and production were separately considered.
The results indicated the key role of demand in the dynamics of all the indices under consideration, the W-shaped shock, as the worst of the considered scenarios, as well as the relatively greater sensitivity of the employment index to the demand index and the finance index to the production index.
Keywords: stress testing, business tendency studies, scenario analysis, manufacturing, vector autoregressive model, Bayesian methods, digital Indicators, COVID-19.
JEL Classification: C32, E27, E37, G17
Suggested Citation: Suggested Citation