Integrated perspectives on macroeconomic dynamics: time series disaggregation, financial shock transmission, output clusters and monetary policy communication complexity

Chisha, Keegan (2025) Integrated perspectives on macroeconomic dynamics: time series disaggregation, financial shock transmission, output clusters and monetary policy communication complexity. PhD thesis, University of Glasgow.

Abstract

The absence of high frequency economic data and granular information has historically tended to limit the depth of evidence-based research inquiries in the literature aimed at understanding macroeconomic relationships. This thesis takes an empirical approach to address these gaps by, first, advancing an alternative simple data-driven method for generating high frequency and “flash” estimates for macroeconomic analysis. Second, it departs from the literature by determining linguistic complexity per-topic and showing how overall document complexity can obscure the specific topical elements whose linguistic complexity influence financial market movements. The thesis consists of three main parts:

The first chapter addresses the challenge of converting low frequency time series into high frequency time series. Traditionally, when high frequency data is needed but unavailable, researchers have heavily relied on simple regression-based temporal disaggregation methods to transform low frequency data into higher-frequency estimates. However, these methods face significant limitations - they typically utilise a small number of covariates due to multicollinearity issues and are more suited for explaining statistical relationships than predicting economic outcomes. To address these shortcomings, this study introduces a high dimensional time series disaggregation approach based on Partial Least Squares (PLS) regression. Using U.S. data from the FRED database, comparative analyses of temporal disaggregation and nowcasting demonstrate that the proposed data-driven approach offers an advantage over traditional methods for generating desired high frequency time series and prediction of flash estimates.

The second chapter constructs missing state-level historical quarterly output growth data for the U.S. by leveraging existing annual output growth data by state, state-specific economic indicators, and the PLS method proposed in the preceding chapter. The estimated historical dataset is used in two subsequent analyses: first, the data is used to analyse the differential impact of an aggregate financial shock on state-level real economic activity; and second, it is used to investigate the interconnectedness of economic activity across U.S. states. The study results confirm differentiated response of state-specific out-put growth to an aggregate financial shock and that the U.S economy forms three main economic clusters with two small ones. These applications offer valuable insights into the differentiated impact of an aggregate financial shock and the dynamics of economic activity connectedness among the individual states. The results highlight weaknesses of one-size-fits-all monetary or fiscal policy in favour of state-specific policy responses.

In the final chapter, I provide a novel approach that determines complexity of specific contents of the monetary policy statements. I examine the response of the 10-year government bond liquidity to the linguistic complexity of the overall central bank communication, and that of the individual topics. In order to generate evidence, I study the market liquidity outcomes when the level of linguistic complexity of monetary policy statements is low and when it is high for the Reserve Bank of South Africa (SARB). Descriptive results show that, first, liquidity is generally low on announcement days relative to the 10-day control day. Density distributional shifts to the right reveal even lower liquidity in the bond market, particularly, when linguistic complexity is high. The sensitivity of liquidity is more when the complexity of forecasting is high. Volatility of liquidity is also found to respond to linguistic complexity of the SARB communication. Main regression results indicate that liquidity is particularly responsive to complexity of overall communication and some specific topics - for example, Forecasting. High communication complexity of forecasting is found to lead to higher quoted spreads suggesting disrupted trading. These findings suggest that market participants respond distinctly to different levels of specific topic communication clarity, providing central banks practical insights to target improvements that promote stable market and economic outcomes.

Therefore, this thesis contributes to the existing literature in three distinctive ways. First, it addresses the limitations of traditional methods of temporal disaggregation of time series by proposing a simple novel, data-driven approach. Second, adds to the existing literature by producing fresh evidence confirming the differential impact of an aggregate financial shock on state-specific output growth. Finally, contributes a novel approach to grade central bank communication complexity by topic, improving financial market understanding and presenting opportunities for pinpointed improvements in central bank communication.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Subjects: H Social Sciences > HG Finance
• Colleges/Schools: College of Social Sciences > Adam Smith Business School > Economics
• Supervisor's Name: Korobilis, Professor Dimitris and Nareklishvili, Dr. Maria
• Date of Award: 2025
• Depositing User: Theses Team
• Unique ID: glathesis:2025-85516
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 14 Oct 2025 13:26
• Last Modified: 14 Oct 2025 13:26
• Thesis DOI: 10.5525/gla.thesis.85516
• URI: https://theses.gla.ac.uk/id/eprint/85516