| Abstract: |
Forecasting a key macroeconomic variable, consumer price index (CPI)
inflation, for BRIC countries using economic policy uncertainty and
geopolitical risk is a difficult proposition for policymakers at the central
banks. This study proposes a novel filtered ensemble wavelet neural network
(FEWNet) that can produce reliable long-term forecasts for CPI inflation. The
proposal applies a maximum overlapping discrete wavelet transform to the CPI
inflation series to obtain high-frequency and low-frequency signals. All the
wavelet-transformed series and filtered exogenous variables are fed into
downstream autoregressive neural networks to make the final ensemble forecast.
Theoretically, we show that FEWNet reduces the empirical risk compared to
single, fully connected neural networks. We also demonstrate that the
rolling-window real-time forecasts obtained from the proposed algorithm are
significantly more accurate than benchmark forecasting methods. Additionally,
we use conformal prediction intervals to quantify the uncertainty associated
with the forecasts generated by the proposed approach. The excellent
performance of FEWNet can be attributed to its capacity to effectively capture
non-linearities and long-range dependencies in the data through its adaptable
architecture. |