• A
  • A
  • A
  • ABC
  • ABC
  • ABC
  • А
  • А
  • А
  • А
  • А
Regular version of the site
  • HSE University
  • Publications of HSE
  • Articles
  • Практика идентификации ненаблюдаемых компонент в траектории ВВП: потенциальный уровень и краткосрочные разрывы

Article

Практика идентификации ненаблюдаемых компонент в траектории ВВП: потенциальный уровень и краткосрочные разрывы

Вопросы статистики. 2015. № 10. С. 14-25.
The authors systematize the most well-known concepts and definitions of potential output and gaps according to the various economic schools approaches. The article also provides a typology of basic econometric methods for estimating potential level and output gap in the national gross product dynamics. The authors give a brief description of various econometric methods of statistical decomposition of the Gross Value Added dynamics for long-term potential level and short-term deviation – from simple statistical filtering to complex dynamic stochastic general equilibrium models. In particular, they describe properties of one-dimensional filters, which based on statistical evaluation of interrelations in the output dynamics only, with decomposing long-term level by probabilistic methods without taking into account any settings of economic theory. Also, authors examine advantages and disadvantages of multi-dimensional and structural methods for estimating potential levels and output gaps, which allow consider-ing the economic interrelations through inclusion of additional relevant variables. These include multidi-mensional Hodrick-Prescott filter, multivariate models with unobserved components, structural vector autoregression models, and stochastic dynamic general equilibrium model. The analysis of decomposition methods and numerous studies devoted to this problem lead to conclusion that decomposition of time series for long-term and cyclical components is not a trivial task, which may have a unique solution. Many of decomposition methods are sensitive to input data, including presence of additional parameters in the model, therefore more complex algorithms may show poorer re-sults than algorithms based on statistical analysis, without taking into account an economic nature of in-vestigated series. Finding the best decomposition method in various countries is possible only through empirical studies.