Deep learning models are increasingly being applied to macroeconomic forecasting. However, most studies report aggregate accuracy without isolating the contribution of individual architectural components. This paper proposes a multi-scale attention-augmented bidirectional gated recurrent unit with a direct linear residual branch and introduces a controlled walk-forward ablation protocol to quantify the marginal value of each component. Experiments on eleven monthly United States (U.S.) indicators from the Federal Reserve Economic Data – monthly database (FRED-MD), spanning March 1959 to September 2019, show that the linear residual branch accounts for most of the predictive gain. In contrast, multi-scale convolution and temporal attention each contribute less than a 2.0% additional reduction in root mean square error (RMSE). These results suggest that the deep architectural components offer only limited marginal benefit on this dataset. This diagnostic finding clarifies the practical role of deep learning components in low-signal macroeconomic data and establishes a reproducible evaluation standard for future architectural studies.
References
[1] Makridakis, S., Spiliotis, E., Hollyman, R., Petropoulos, F., Swanson, N., Gaba, A. (2025) The M6 forecasting competition: Bridging the gap between forecasting and investment decisions. International Journal of Forecasting, 41(4), 1315-1354.
[2] Kant, D., Pick, A., Winter, J. D. (2025) Nowcasting GDP using machine learning methods: D. Kant et al. AStA Advances in Statistical Analysis, 109(1), 1-24.
[3] Chapman, J. T., Desai, A. (2023) Macroeconomic predictions using payments data and machine learning. Forecasting, 5(4), 652-683.
[4] Chen, L., Pelger, M., Zhu, J. (2024) Deep learning in asset pricing. Management Science, 70(2), 714-750.
[5] Schlieper, P., Dombrowski, M., Nguyen, A., Zanca, D., Eskofier, B. (2024) Data-centric benchmarking of neural network architectures for the univariate time series forecasting task. Forecasting, 6(3), 718-747.
[6] Catlin, C. (2025) Adaptive forecasting in dynamic markets: an evaluation of AutoTS within the M6 competition. International Journal of Forecasting, 41(4), 1485-1493.
[7] Stoneman, D., Duca, J. V. (2024) Using deep (machine) learning to forecast US inflation in the COVID‐19 era. Journal of Forecasting, 43(4), 894-902.
[8] Vancsura, L., Tatay, T., Bareith, T. (2025) Enhancing policy insights: machine learning-based forecasting of euro area inflation HICP and subcomponents. Forecasting, 7(4), 63.
[9] Semenoglou, A. A., Spiliotis, E., Assimakopoulos, V. (2023) Image-based time series forecasting: a deep convolutional neural network approach. Neural Networks, 157, 39-53.
[10] Xiao, Z., Liu, J., Cao, X., Wang, K., Li, D., Liu, Q. (2025) Time-series forecasting method based on hierarchical spatio-temporal attention mechanism. Sensors, 25(13), 4001.
[11] Yang, Z., Gonsalves, T. (2025) Segmented-Crossformer – a novel and enhanced cross-time and cross-dimensional Transformer for multivariate time series forecasting. Forecasting, 7(3), 41.
[12] Jiang, H., Pan, S., Dong, Y., Wang, J. (2024) Probabilistic electricity price forecasting based on penalized temporal fusion transformer. Journal of Forecasting, 43(5), 1465-1491.
[13] de Vilmarest, J., Werge, N. (2025) An adaptive volatility method for probabilistic forecasting and its application to the M6 financial forecasting competition. International Journal of Forecasting, 41(4), 1514-1520.
[14] Lin, Y. (2024) Progressive neural network for multi-horizon time series forecasting. Information Sciences, 661, 120112.
[15] Dowlut, N., Gobin-Rahimbux, B. (2023) Forecasting resort hotel tourism demand using deep learning techniques – a systematic literature review. Heliyon, 9(7), e18385.
[16] Singh, V., Sahana, S. K., Bhattacharjee, V. (2024) Integrated spatio-temporal graph neural network for traffic forecasting. Applied Sciences, 14(24), 11477.
[17] Arsenault, P. D., Wang, S., Patenaude, J. M. (2025) A survey of explainable artificial intelligence (XAI) in financial time series forecasting. ACM Computing Surveys, 57(10), 1-37.
[18] Elsharkawy, I., Hooberman, B., Kahn, Y. (2025) Uncertainty quantification from ensemble variance scaling laws in deep neural networks. Machine Learning: Science and Technology, 6(3), 035040.
[19] Ma, Y., Wang, W., Cheng, Y. (2026) Forecasting crude oil futures prices: a stacking model with mixed – frequency predictors, machine learning, and forecast combination. Journal of Futures Markets, 46(11), 1867-1888.
[20] Çınarer, G. (2025) Hybrid deep learning and stacking ensemble model for time series-based global temperature forecasting. Electronics, 14(16), 3213.
Share and Cite
Geng, Y., Yang, T. (2026) Research on Deep Learning-based Economic Forecasting Algorithms. Scientific Research Bulletin, 3(5), 28-35. https://doi.org/10.71052/srb2024/NUMZ6301