An Overview of Dynamic Model Averaging Techniques in Time‐series Econometrics

Nonejad, Nima

doi:10.1111/joes.12410

Cited by 26 publications

(13 citation statements)

References 173 publications

(447 reference statements)

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Values of this parameter close to zero imply (extremely) large observational volatility. In turn, the parameter generates time variation in the full model set (Catania and Nonejad (2018); Nonejad (2021)). Values of this parameter close to zero induce (extremely) fast switches among models.…”

Section: Dynamic Model Averagingmentioning

confidence: 99%

See 1 more Smart Citation

Forecasting Inflation in Open Economies: What Can a NOEM Model DO?

Duncan¹,

Martínez‐García²

2022

gwp

View full text Add to dashboard Cite

This paper evaluates the forecasting ability when inflation is viewed as an inherently global phenomenon through the lens of the workhorse New Open Economy Macro (NOEM) model. The NOEM model emphasizes the importance of cross-country spillovers arising through trade and its reduced form solution can be represented by a finite-order VAR which provides a tractable model of inflation forecasting. We use Bayesian techniques to estimate this VAR specification-we name it NOEM-BVAR-and pseudo-out-of-sample forecasts to assess its forecasting performance at different horizons in a diverse set of 18 countries. On average, the NOEM-BVAR specification produces a similar or even lower root mean square prediction error (RMSPE) than its standard competitors, which include both purely statistical models and theoretically-based forecasting models (e.g., Phillipscurve-type alternatives and others with global inflation measures). In a number of cases, the gains in smaller RMSPEs are statistically significant, especially at short horizons. The NOEM-BVAR model is also accurate in predicting the direction of change for inflation and is often better than its competitors along this dimension as well. Even though purely statistical models can be useful prediction tools, the NOEM-BVAR is attractive among those forecasting models motivated by economic theory.

show abstract

Section: Dynamic Model Averagingmentioning

confidence: 99%

“…3 For additional technical details and the multiple applications of DMA in macroeconomics forecasting, see alsoCatania and Nonejad (2018) andNonejad (2021).…”

mentioning

confidence: 99%

Forecasting Inflation in Open Economies: What Can a NOEM Model DO?

Duncan¹,

Martínez‐García²

2022

gwp

View full text Add to dashboard Cite

show abstract

“…Muitas aplicac ¸ões do mundo real estão inseridas no contexto de séries temporais. Uma aplicac ¸ão bastante realizada e de grande interesse para a comunidade acadêmica e em geral é a predic ¸ão de séries temporais, que é utilizada em diversas áreas do conhecimento como em sistemas biológicos [2], em econometria [3], na teoria de sistemas de controle [4], meteorologia [5], entre outras. A natureza caótica de algumas séries temporais é uma complexidade que motiva o desenvolvimento de modelos matemáticos que sejam aptos a representar a dinâmica não linear [6].…”

Section: Introduc ¸ãOunclassified

Predição de Séries Temporais Caóticas baseada em Modelo ARMA via Algoritmo Adaptativo FVSS-NLMS

Noronha¹

2021

Anais Do 15. Congresso Brasileiro De Inteligência Computacional

View full text Add to dashboard Cite

Neste trabalho é utilizado o modelo ARMA (do inglês, AutoRegressive Moving Average) para a predição de séries temporais caóticas. A estimativa do vetor de pesos do modelo ARMA pode ser obtida através de um algoritmo de otimização baseado em gradiente descendente estocástico, de tal forma que o desempenho de predição de uma série temporal caótica realizada pelo modelo ARMA é influenciado pelo valor do tamanho de passo. Para isto, neste trabalho é proposta uma nova versão do algoritmo NLMS (do inglês, Normalized Least Mean Square) com o tamanho de passo adaptado por um MFIS (do inglês, Mamdani Fuzzy Inference System) para atualização da estimativa do vetor de pesos do modelo ARMA. O desempenho do modelo ARMA para a predição de séries temporais caóticas foi comparado com os resultados obtidos pelas versões tradicionais dos algoritmos LMS (do inglês, Least Mean Square) e NLMS.

show abstract

“…(Catania & Nonejad, 2018;Nonejad, 2021). Values of this parameter close to zero induce (extremely) fast switches among models.…”

mentioning

confidence: 99%

“…33 33 For additional technical details and the multiple applications of DMA in macroeconomics forecasting, see also Catania and Nonejad (2018) and Nonejad (2021). …”

mentioning

confidence: 99%

Forecasting inflation in open economies: What can a NOEM model do?

Duncan

Martínez‐García

2023

Journal of Forecasting

View full text Add to dashboard Cite

This paper evaluates the forecasting ability when inflation is viewed as an inherently global phenomenon through the lens of the workhorse New Open Economy Macro (NOEM) model. The NOEM model emphasizes the importance of cross‐country spillovers arising through trade, and its reduced form solution can be represented by a finite‐order VAR that provides a tractable model of inflation forecasting. We use Bayesian techniques to estimate this VAR specification—we name it NOEM‐BVAR—and pseudo‐out‐of‐sample forecasts to assess its forecasting performance at different horizons in a diverse set of 18 countries. On average, the NOEM‐BVAR specification produces a similar or even lower root mean square prediction error (RMSPE) than its standard competitors, which include both purely statistical models and theoretically‐based forecasting models (e.g., Phillips‐curve‐type alternatives and others with global inflation measures). In a number of cases, the gains in smaller RMSPEs are statistically significant, especially at short horizons. The NOEM‐BVAR model is also accurate in predicting the direction of change for inflation and is often better than its competitors along this dimension as well. Even though purely statistical models can be useful as prediction tools, the NOEM‐BVAR is an attractive tool among those forecasting models motivated by economic theory.

show abstract

An Overview of Dynamic Model Averaging Techniques in Time‐series Econometrics

Cited by 26 publications

References 173 publications

Forecasting Inflation in Open Economies: What Can a NOEM Model DO?

Forecasting Inflation in Open Economies: What Can a NOEM Model DO?

Predição de Séries Temporais Caóticas baseada em Modelo ARMA via Algoritmo Adaptativo FVSS-NLMS

Forecasting inflation in open economies: What can a NOEM model do?

Contact Info

Product

Resources

About