Trading activity, financial intermediation, and price formation in foreign exchange markets

Wang, Tongtong (2026) Trading activity, financial intermediation, and price formation in foreign exchange markets. PhD thesis, University of Glasgow.

Abstract

This thesis investigates how trading activity, financial intermediation, and dealer balance-sheet frictions jointly shape price formation in foreign exchange (FX) markets. This thesis uses proprietary bilateral over-the-counter (OTC) datasets from two of the world’s largest foreign exchange dealer banks to construct a unified framework, which links micro-level trading dynamics with intermediate asset pricing theory. The analysis is divided into three interconnected empirical sections. Each section builds on the previous one, progressively discussing how trading volume, order flow, and dealer constraints interact to determine exchange rate dynamics.

Chapter 2 examines the relationship between trading volume, order flow, and volatility in FX markets by using weekly data on twelve currencies from 2001 to 2012. The analysis employs both realised volatility proxy and implied volatility. The results confirm a robust positive volume–volatility relationship across most currency pairs, which is consistent with information-based trading models. However, a portfolio-sorting approach reveals that this relationship is state-dependent. When trading volume increases, its marginal impact on volatility decreases. This result suggests that high-volume periods are dominated by non-informational activity such as liquidity provision and inventory adjustment. Introducing directional order flow adds substantial explanatory power. Portfolios characterised by strong buying pressure exhibit significantly positive volatility coefficients, while those with selling pressure show negative coefficients. This indicates an asymmetric information channel. Disaggregating order flow by customer type shows heterogeneity: retail (private client) order flow generates a volatility impact five to seven times larger than institutional flow, despite accounting for only 10–2% of total volume. These findings suggest that, compared to total trading volume, trading direction and trader identity are also major drivers of foreign exchange volatility.

Chapter 3 shifts the focus to the predictive content of FX trading volume for exchange rate returns and introduces dealer balance-sheet constraints as the key conditioning variable. This chapter uses unique data from two dealers to demonstrate that trading volume can predict subsequent USD appreciation. However, this predictive ability is limited to the period after the crisis, and only under balance sheet constraint. Before the 2008 crisis, the same volume carried no predictive content. This structural break is documented using three proxies for balance-sheet stress: the CDS spreads, intermediary leverage, and the LIBOR-OIS spread. We sort the sample by the median of each proxy yield and get a consistent result: volume predicts returns exclusively during high-constraint periods. These findings challenge the prevailing view that FX volume predictability arises primarily from asymmetric information.

Chapter 4 identifies and explains the supply-demand mechanism through which dealer balance-sheet frictions affect price formation. The chapter directly estimates the slope of the US dollar supply curve by interacting normalised volume with dealer CDS spreads. The results show that the supply curve steepens sharply during periods of balance-sheet stress. This implies that, with limited intermediary capacity, the same level of trading pressure will produce a larger exchange rate response. Event-time analyses around CDS shocks confirm that FX prices respond to trading pressure when both balance-sheet constraints and trading volume are jointly elevated. On the demand side, the chapter documents that surges in dollar demand during crises originate primarily from leveraged financial clients. Their trading was highly sensitive to the dealers’ capital situation, while the cash flows of non-financial clients were unrelated to the dealers’ capital. This interaction between capital-dependent demand and a steep supply curve lead the positive correlation between foreign exchange trading volume and subsequent dollar appreciation after periods of crisis.

In summary, this thesis makes three contributions. First, it provides comprehensive evidence that the volume–volatility relationship in FX markets is positive. And it is driven by trade direction and trader heterogeneity rather than aggregate volume alone. Secondly, it confirms that foreign exchange trading volume contains predictive information about prices, and this is a mechanistic consequence of the post-crisis balance sheet constraints. Third, it points out that debt overhang makes the dollar supply curve steeper and the demand curve rise, thereby amplifying the impact of trading activity on prices. These findings have implications for exchange rate modelling, risk management, and financial stability policy. This thesis suggests that monitoring dealer balance-sheet health is essential for understanding FX market dynamics.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Keywords: Financial, financial crises, asset pricing, trading volume, bond interest rates, international financial markets, foreign exchange, information and market efficiency, event studies.
• Subjects: H Social Sciences > HG Finance
• Colleges/Schools: College of Social Sciences > Adam Smith Business School
• Supervisor's Name: Cerrato, Professor Mario and Ramian, Dr. Hormoz
• Date of Award: 2026
• Depositing User: Theses Team
• Unique ID: glathesis:2026-86180
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 14 Aug 2026 09:15
• Last Modified: 14 Aug 2026 09:16
• Thesis DOI: 10.5525/gla.thesis.86180
• URI: https://theses.gla.ac.uk/id/eprint/86180