Paper: arXiv 2503.01893

Authors: Maya Vilenko

Abstract

Inflation prediction guides decisions on interest rates, investments, and wages, playing a key role in economic stability. Yet accurate forecasting is challenging due to dynamic factors and the layered structure of the Consumer Price Index, which organizes goods and services into multiple categories. We propose the Bi-directional Hierarchical Recurrent Neural Network (BiHRNN) model to address these challenges by leveraging the hierarchical structure to enable bidirectional information flow between levels. Informative constraints on the RNN parameters enhance predictive accuracy at all levels without the inefficiencies of a unified model. We validated BiHRNN on inflation datasets from the United States, Canada, and Norway by training, tuning hyperparameters, and experimenting with various loss functions. Our results demonstrate that BiHRNN significantly outperforms traditional RNN models, with its bidirectional architecture playing a pivotal role in achieving improved forecasting accuracy.

Complexity vs Empirical Score

  • Math Complexity: 7.5/10
  • Empirical Rigor: 8.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper introduces a novel neural network architecture with mathematical modifications like bidirectional hierarchical information flow and parameter constraints, showing advanced math. It is highly empirically rigorous, evidenced by training, tuning hyperparameters, experimenting with loss functions, backtesting on three real-world inflation datasets (US, Canada, Norway), and releasing code on GitHub.

Research Flowchart

  flowchart TD
  A["Research Goal<br>Inflation Prediction using CPI Hierarchy"] --> B["Data Input<br>US, Canada, Norway CPI Datasets"]
  B --> C["Methodology<br>Bi-directional Hierarchical RNN BiHRNN"]
  C --> D["Computational Process<br>Hyperparameter Tuning & Loss Function Experiments"]
  D --> E["Key Outcome<br>Significantly Outperforms Traditional RNNs"]

Evidence Card

What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card

Proposes

The paper proposes a Bi-directional Hierarchical Recurrent Neural Network (BiHRNN) model for inflation forecasting.

Source sentence
We propose the Bi-directional Hierarchical Recurrent Neural Network (BiHRNN) model to address these challenges by leveraging the hierarchical structure to enable bidirectional information flow between levels.

Assumes 2

  • The time series errors are Gaussian. statistical
    Source sentence
    Assuming Gaussian errors, the likelihood of the time series is modeled as a product of normal distributions
  • Hierarchical informative prior connects each node’s parameters to its parent’s parameters. model
    Source sentence
    A hierarchical informative prior connects each node’s parameters to its parent’s, with the prior 𝑝(𝜃𝑛|𝜃𝜋𝑛,𝜏𝜃𝑛) using 𝜏𝜃𝑛 as a precision parameter.

Data

  • US Consumer Price Index (CPI-U) · U.S. Bureau of Labor and Statistics (BLS) macro public
    January 2012 to September 2023
    Source sentence
    The US dataset spans from January 2012 to September 2023
  • Canada Consumer Price Index (CPI) · Statistics Canada macro public
    January 2013 to February 2023
    Source sentence
    The Canada dataset from January 2013 to February 2023
  • Norway Consumer Price Index (CPI) · Statistics Norway macro public
    January 2009 to May 2023
    Source sentence
    The Norway dataset from January 2009 to May 2023

Costs

Not applicable The paper reports no trading or execution result.

Finds 1

  • The results show that the BiHRNN consistently outperforms other models in terms of predictive accuracy and stability. performance
    Source sentence
    The results in Tables 5.2, 5.3, and 5.1 show that the BiHRNN consistently outperforms other models in terms of predictive accuracy and stability.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 28 related papers). Coverage is partial, so this is neither replication nor refutation.

Code

https://github.com/mayavilenko/Maya-Thesis — the authors' own repository

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.

Cited References (89)

  1. An improved temporal convolutional network with attention mechanism for photovoltaic generation forecasting (2023) — link
  2. Are Transformers Effective for Time Series Forecasting? (2022) — link
  3. Forecasting CPI inflation components with Hierarchical Recurrent Neural Networks (2023) — link
  4. Improving inflation forecasts using robust measures (2024) — link
  5. Forecasting CPI inflation under economic policy and geopolitical uncertainties (2024) — link
  6. Deep hierarchical LSTM networks with attention for video summarization (2022) — link
  7. The model confidence set (2011)
  8. Hierarchical Attention Networks for Document Classification (2016)
and 81 more
  1. Explainable Multivariate Time Series Classification: A Deep Neural Network Which Learns to Attend to Important Variables As Well As Time Intervals (2021)
  2. Self-attention with relative position representations (2018) — link
  3. Time Series Analysis: Forecasting and Control (1970)
  4. Long Short-Term Memory (1997)
  5. Deep Learning (2016)
  6. Macroeconomics and Reality (1980)
  7. Forecasting Inflation: Econometric Models Versus Machine Learning Models (2019)
  8. A Hybrid Model for Inflation Forecasting with Machine Learning and Econometric Models (2021)
  9. Forecasting: principles and practice (2011)
  10. Optimal forecast reconciliation for hierarchical and grouped time series through trace minimization (2019)
  11. A Contribution to the Mathematical Theory of Epidemics (1927)
  12. Prediction and Analysis of COVID-19 Positive Cases Using Deep Learning Models: A Descriptive Case Study of India (2020)
  13. Time Series Forecasting of COVID-19 Transmission in Canada Using LSTM Networks (2020)
  14. Using online prices for measuring real consumption across countries (2018)
  15. The billion prices project: Using online prices for measurement and research (2016)
  16. Measuring and testing dependence by correlation of distances (2007)
  17. A Brief Introduction to Boosting (1999)
  18. XGBoost: A Scalable Tree Boosting System (2016) — link
  19. Decision tree methods: applications for classification and prediction (2015)
  20. Random forests (2001)
  21. Measuring nonlinear dependence in time-series, a distance correlation approach (2012)
  22. Support Vector Regression Machines (1996) — link
  23. Empirical evaluation of gated recurrent neural networks on sequence modeling (2014) — link
  24. Recurrent neural networks for prediction: learning algorithms, architectures and stability (2001)
  25. Consumer price index prediction using Long Short Term Memory (LSTM) based cloud computing (2020)
  26. Inflation targeting: lessons from the international experience (2018)
  27. Inflation dynamics during the financial crisis (2017)
  28. Inflation Targeting and Financial Stability (2012)
  29. Searching for Activation Functions (2017)
  30. Nonlinear inflation forecasting with recurrent neural networks (2019)
  31. A review of recurrent neural networks: LSTM cells and network architectures (2019)
  32. Long short-term memory (1997)
  33. A critical review of recurrent neural networks for sequence learning (2015)
  34. Gate-variants of gated recurrent unit (GRU) neural networks (2017)
  35. The relation between unemployment and the Rate of change of money wage rates in the United Kingdom, 1861–1957 (1958)
  36. Are Phillips curves useful for forecasting inflation? (2001)
  37. Smooth Transition Autoregressive Models — A Survey Of Recent Developments (2002)
  38. Testing linearity against smooth transition autoregressive models (1988)
  39. Forecasting Inflation (2013)
  40. Semiparametric ARX neural-network models with an application to forecasting inflation (2001)
  41. Inflation forecasting using a neural network (2005)
  42. Neural network models for inflation forecasting: an appraisal (2012)
  43. Forecasting inflation with thick models and neural networks (2005)
  44. The impact of machine learning on economics (2018)
  45. Machine learning at central banks (2017)
  46. Forecasting Inflation in a data-rich environment: the benefits of machine learning methods (2021)
  47. Machine learning: an applied econometric approach (2017)
  48. Approximation and learning theory (1992)
  49. A comparison of linear and nonlinear univariate models for forecasting macroeconomic time series (1998)
  50. Why has US inflation become harder to forecast? (2007)
  51. Modeling inflation after the crisis (2010)
  52. Inflation Forecasting Using Machine Learning Methods (2018)
  53. How is Machine Learning Useful for Macroeconomic Forecasting? (2022)
  54. The M4 Competition: 100,000 time series and 61 forecasting methods (2020)
  55. Objectivity, reproducibility and replicability in forecasting research (2018)
  56. Bagged neural networks for forecasting Polish (low) inflation (2019)
  57. Are Phillips curves useful for forecasting inflation? (2001)
  58. Linear models, smooth transition autoregressions, and neural networks for forecasting macroeconomic time series: A re-examination (2005)
  59. Forecasting Euro area inflation: Does aggregating forecasts by HICP component improve forecast accuracy? (2005)
  60. Understanding and forecasting aggregate and disaggregate price dynamics (2014)
  61. Core inflation and trend inflation (2016)
  62. The macroeconomic and fiscal implications of inflation forecast errors (2018)
  63. Forecasting aggregated vector ARMA processes (1987)
  64. Multi-horizon inflation forecasts using disaggregated data (2010)
  65. Hierarchical forecasting (2020)
  66. Essays in hierarchical time series forecasting and forecast combination (2018)
  67. Forecast combinations (2006)
  68. Inflation persistence, the NAIRU, and the great recession (2014)
  69. Optimal inflation and the identification of the Phillips curve (2019)
  70. A comparison of direct and iterated multistep AR methods for forecasting macroeconomic time series (2006)
  71. The Lag in Effect of Monetary Policy (1961)
  72. The Phillips Curve: Back to the 60s? (2016)
  73. What’s up with the Phillips Curve? (2020)
  74. Do disaggregated CPI data improve the accuracy of inflation forecasts? (2012)
  75. Sectoral inflation persistence and optimal monetary policy (2020)
  76. Forecasting inflation with online prices (2020)
  77. Comparing Predictive Accuracy (1995)
  78. Trend, Seasonal, and Sectorial Inflation in the Euro Area (2020)
  79. Sectoral inflation and the Phillips curve: What has changed since the Great Recession? (2018)
  80. Prospects for inflation in a high pressure economy: Is the Phillips curve dead or is it just hibernating? (2020)
  81. Simulating historical inflation-linked bond returns (2018)

0 of 89 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.