Денежно-кредитная нестабильность в странах c формирующимся рынком тема диссертации и автореферата по ВАК РФ 00.00.00, кандидат наук Сикхвал Швета

• Сикхвал Швета
• кандидат наук
• 2025

Оглавление диссертации кандидат наук Сикхвал Швета

Table of Contents

Introduction

Chapter 1 Effects of US Interest Rate Shocks in the Emerging Market Economies: Evidence from Panel Structural VAR

1.1 Introduction

1.2 Literature Review

1.3 Data and Methodology

1.3.1 Data

1.3.2 Methodology

1.3.2.1 Identification of US interest rate shocks

1.3.2.2 The PSVAR model

1.4 Empirical Results and Discussion

1.4.1 Results from the U.S. SVAR

1.4.2 Results from PSVAR

1.5 Extension and Robustness

1.5.1 Extension: Regional Analysis

1.5.2 Robustness

1.6 Conclusion

Chapter 2 Quantifying the Spillover Effects of U.S. Economic Policy Uncertainty on Emerging Market Economies Using GMM-PVAR Model

2.1 Introduction

2.2 Literature Review

2.3 Data and Methodology

2.3.1 Data

2.3.2 Econometric Methodology

2.3.2.1 Preliminary Analysis

2.3.2.2 GMM estimation of PVAR model

2.3.2.3 Impulse Response Functions

2.4 Empirical Analysis

2.4.1 Pre-estimation results

2.4.1.1 Descriptive Statistics

2.4.1.2 Stationarity Tests

2.4.1.3 Lag-selection criterion

2.4.1.4 Stability Test

2.4.2 Evidence from GMM-PVAR estimation

2.4.3 Generalized Impulse Response Function Analysis

2.5 Robustness

2.5.1 Sensitivity to Forward Orthogonal Transformation

2.5.2 Sensitivity to the Alternative US EPU Index

2.5.3 Addressing Potential Endogeneity of the Oil Price Uncertainty Index

2.5.4 Addressing Outliers by Excluding Large Countries like China and India

2.6 Conclusion

Chapter 3 Comparative Analysis of Machine Learning Models for Money Demand Forecasting in the Indian Economy

3.1 Introduction

3.2 Literature Review

3.3 Dataset, Model Specification and Methodology

3.3.1 Dataset

3.3.2 Model Specification and Preliminary Analysis

3.3.3 Methodology

3.3.3.1 Autoregression of Order

3.3.3.2 Random Forest Regression

3.3.3.3 Gradient boosting

3.3.3.4 Xtreme Gradient Boosting

3.3.3.5 Support Vector Regression

3.3.3.6 Least Absolute Shrinkage and Selection Operator

3.3.3.7 Long Short-Term Memory

3.4 Model Validation

3.4.1 Expanding Window Cross-validation with Time-Series Split

3.4.2 Hyperparameter Tuning

3.4.3 Evaluation Metrics

3.4.4 Forecasting Accuracy

3.5 Empirical Analysis

3.6 Conclusion

Conclusion

Professional Significance

References

Appendices

Appendix A1: Supporting Data and Results for Chapter

Appendix A2: Supporting Data and Results for Chapter

Appendix A3: Supporting Data and Results for Chapter

Appendix A4: AI Disclosure

Appendix A5: Russian translation of the dissertation / Перевод диссертации на русский язык

Введение диссертации (часть автореферата) на тему «Денежно-кредитная нестабильность в странах c формирующимся рынком»

Introduction

Motivation1

Monetary stability, defined as low and predictable inflation, stable exchange rates, and well-functioning financial markets, is fundamental to sustained economic growth and development. While advanced economies possess robust financial systems and strong policy instruments to buffer against shocks, emerging market economies (EMEs) face distinct challenges in achieving monetary stability. These economies are often characterized by underdeveloped financial markets, weaker institutional frameworks, limited policy credibility, and a high dependence on foreign capital. Their integration into the global financial system exposes them to external shocks, particularly from monetary and economic policy shifts in advanced economies like the United States.

A key factor driving this external vulnerability is the dominance of the U.S. dollar in global transactions and the widespread influence of U.S. monetary policy. Changes in U.S. interest rates and broader economic policy uncertainty (EPU) have profound consequences for EMEs, frequently triggering financial volatility, inflationary pressures, and policy constraints for their central banks. U.S. monetary tightening, for instance, often results in capital outflows from EMEs as global investors reallocate funds toward higher returns in advanced economies. This capital flight depreciates EME currencies, raises the cost of servicing foreign-denominated debt, and fuels inflation through increased import costs.

The repeated exposure of EMEs to external shocks has been evident across several financial crises. The Latin American debt crisis of the early 1980s exemplifies how rising global interest rates, driven by the U.S. Federal Reserve's efforts to combat inflation, significantly increased external debt servicing costs for countries like Mexico, Brazil, and Argentina. Many of these economies had relied heavily on external borrowing throughout the 1970s, and the sudden surge in repayment costs, coupled with falling commodity prices, led to liquidity crises, currency devaluations, and prolonged economic contractions. The crisis exposed the risks associated with excessive reliance on external financing and the limitations of fixed exchange rate regimes in absorbing external shocks.

1 I would like to express my gratitude to Marek P. Dabrowski for his valuable feedback and insightful suggestions, which contributed significantly to the development of this dissertation.

Similar vulnerabilities surfaced during the 1997 Asian Financial Crisis, when large-scale capital flight and speculative attacks on fixed exchange rate regimes resulted in currency collapses across Thailand, Indonesia, and South Korea. The crisis exposed the structural weaknesses of these economies, particularly their dependence on short-term foreign capital inflows. When investors suddenly withdrew their funds, governments were forced to seek International Monetary Fund (IMF) assistance, leading to deep economic contractions and long-term structural adjustments.

The 2008 Global Financial Crisis (GFC) further demonstrated the external dependence of EMEs. Initially, large-scale asset purchases under U.S. quantitative easing (QE) flooded global markets with liquidity, leading to capital inflows into EMEs and asset price inflation. However, when the U.S. Federal Reserve signaled its intention to scale back QE in 2013 (the so-called "taper tantrum"), EMEs experienced sharp capital outflows, currency depreciations, and rising inflationary pressures. This episode illustrated that even shifts in expectations of U.S. monetary policy can generate significant financial instability in EMEs, reinforcing their exposure to external shocks.

More recently, the COVID-19 pandemic led to a rapid "flight to safety," with investors pulling capital from EMEs and reallocating funds into U.S. assets. This sudden stop in capital inflows triggered currency depreciation, increased borrowing costs, and heightened financial instability in many developing economies. In response, EME central banks implemented aggressive monetary easing to stabilize their economies. However, as inflation surged globally in 2021, the subsequent tightening of U.S. monetary policy once again created financial stress for EMEs, underscoring the persistent challenge they face in shielding their economies from external disruptions.

These crisis episodes illustrate what has been described as the global financial cycle (Rey, 2015), where monetary conditions in advanced economies drive synchronized movements in capital flows, asset prices, and exchange rates across countries. EMEs, due to their structural weaknesses, remain especially susceptible to these fluctuations, often finding that their domestic monetary policies are constrained by external conditions.

While changes in U.S. interest rates remain a central driver of financial instability in EMEs, broader U.S. EPU has also emerged as a significant external risk factor. Unlike direct monetary policy shifts, which have clearer transmission mechanisms, uncertainty surrounding fiscal policy, trade policy, and regulatory changes can create financial instability by affecting investor sentiment,

capital flows, and exchange rate volatility. Periods of heightened EPU often lead to risk aversion, prompting capital outflows from EMEs, currency depreciation, and increased inflationary pressures. However, the precise mechanisms through which U.S. EPU transmits to EMEs remain less understood, with some studies emphasizing financial market reactions and others pointing to indirect channels such as global trade linkages and commodity price fluctuations. A more detailed understanding of these transmission channels is critical for policymakers in EMEs to design effective responses to external uncertainty.

In this context, the ability to accurately forecast key macroeconomic variables becomes increasingly crucial for managing monetary instability in EMEs. One such variable is money demand, which plays a central role in shaping monetary policy responses. While some EMEs have adopted inflation-targeting frameworks that prioritize price stability over monetary aggregates, monitoring money demand remains essential for liquidity management, financial stability, and inflation control, particularly in large, partially open economies like India.

However, traditional econometric models often struggle to capture the non-linearities and regime shifts that characterize EME economies, particularly during periods of heightened uncertainty and external shocks. This limitation can lead to unreliable forecasts, increasing the risk of suboptimal policy decisions. Recent advances in machine learning (ML) offer a promising alternative, providing more flexible and adaptive forecasting tools that can better account for changing economic conditions. By applying ML-based forecasting techniques to money demand in India, this dissertation seeks to provide insights into how emerging economies can develop more resilient and responsive policy tools to navigate monetary instability.

By integrating an analysis of external shocks (U.S. interest rate changes and EPU) with an internal forecasting solution (ML-based money demand prediction), this dissertation offers a comprehensive framework for understanding and addressing monetary instability in EMEs. In doing so, it contributes to ongoing discussions on how EMEs can enhance their policy autonomy, improve forecasting accuracy, and develop more adaptive monetary strategies in an increasingly interconnected global financial system.

Brief Literature Review

Effects of US interest rate shocks in the EMEs

The literature on the impact of U.S. monetary policy and economic performance on global macroeconomic conditions is extensive. However, much of this research has focused on advanced economies, with relatively less attention paid to EMEs. The dominant role of the U.S. dollar and Federal Reserve policies creates significant spillover effects, yet important gaps remain in understanding how these effects vary across a large, diverse set of EMEs, particularly in response to unconventional policy measures.

The effects of U.S. monetary policy on EMEs operate through multiple channels, including exchange rates, capital flows, and financial linkages. Studies such as Dahlhaus and Vasishtha (2020) show that unexpected tightening of U.S. monetary policy leads to significant capital outflows from EMEs, amplifying economic instability during periods of global uncertainty. Similarly, Rey (2015, 2016) highlights the "global financial cycle," where U.S. monetary policy dictates global risk sentiment, constraining the monetary autonomy of EMEs and leading to synchronized capital flow reversals. However, much of the existing literature focuses primarily on conventional policy tools, such as the federal funds rate, and pays less attention to the effects of unconventional measures, which have become increasingly relevant since the GFC.

Unconventional monetary policy, particularly QE, has introduced new complexities. Anaya et al. (2017) demonstrate that QE drives capital inflows into EMEs, reducing borrowing costs and fostering short-term growth, while Lim et al. (2014) emphasize the dual nature of QE-induced inflows, which support growth in the short term but heighten risks of capital flow reversals during policy normalization.

Despite growing evidence of U.S. monetary spillovers, there is limited research on cross-regional heterogeneity in EME responses. Bhattarai and Park (2021) identify significant regional differences, attributing Latin America's higher vulnerability to structural weaknesses, such as greater external debt reliance and institutional fragility. Similarly, Canova (2005) demonstrates that U.S. monetary shocks have pronounced effects on Latin America, influencing output, inflation, and trade balances, with smaller, trade-dependent economies experiencing greater disruptions. In contrast, Asian EMEs exhibit more resilience due to stronger foreign exchange

reserves and proactive policy measures (Eichengreen and Gupta, 2015). Feldkircher and Huber (2016) extend this analysis, showing that financial integration amplifies spillovers, with more financially open economies facing greater vulnerabilities.

Spillover effects of US EPU on EMEs

Economic uncertainty significantly influences economic activity by prompting firms and households to delay investment and consumption decisions, as highlighted in foundational works by Bernanke (1983), Dixit and Pindyck (1994), and Bloom (2014). Empirical studies, such as Bloom (2009), show that volatility shocks can cause short-term declines in industrial production, followed by prolonged recovery phases. In the U.S. context, Baker et al. (2016), Bachmann et al. (2013), and Caggiano et al. (2017) demonstrate that elevated EPU negatively impacts investment, output, and employment. Caggiano et al. (2017) further reveal that these effects are amplified when monetary policy is constrained by the zero lower bound.

Colombo (2013) and Alam (2015) find that U.S. policy uncertainty significantly affects the Euroarea and Canada, emphasizing the interconnectedness of global economies. For EMEs, Bhattarai et al. (2020) show that heightened U.S. uncertainty leads to capital outflows, currency depreciation, and financial stress, exacerbating vulnerabilities. This "flight to safety" behavior, where investors shift capital to safer U.S. assets, highlights the role of global risk sentiment in amplifying the effects of U.S. EPU on EMEs.

The literature identifies several channels of international spillovers through which uncertainty affects economic activity, including the "wait-and-see" effect, precautionary savings, and elevated risk premiums (Arellano et al., 2012; Christiano et al., 2014). While these mechanisms are well-documented, studies often focus on individual countries or small groups of economies, offering only limited insights into broader trends. For instance, Carriere-Swallow and Céspedes (2013) and Gourio et al. (2015) explore the impacts of U.S. uncertainty on specific EMEs, but their findings vary, with some studies suggesting reduced capital flows and others observing increased inflows. This duality emphasizes the complexity of these interactions and the need for a more comprehensive approach.

Forecasting money demand using ML techniques

Under a turbulent global environment, forecasting key macroeconomic variables becomes even more critical for EMEs struggling with external shocks. Among these variables, money demand plays an important role in understanding liquidity conditions, monetary policy transmission, and, more broadly, financial stability, particularly when U.S.-driven volatility disrupts capital flows and exchange rates. Traditional econometric models (e.g., ARIMA, VAR) often struggle to account for non-linearities and regime shifts, reducing their accuracy during periods of external turbulence (Stock & Watson 2003). Systematic comparisons between ML and traditional models are especially needed for monetary aggregates, which can fluctuate sharply under external shocks.

The dynamics of money demand and its relationships with macroeconomic variables such as output and inflation have long been central to economic research. Studies such as Bahmani-Oskooee (1996) and Akinlo (2006) demonstrate stable long-run relationships in economies as diverse as Japan and Nigeria, while Adil et al. (2020) and Aggarwal (2016) find mixed evidence of money demand stability in India when considering financial innovation or temporary interest-rate shocks. More recent work, like Barnett et al. (2022), challenges conventional measures by showing that Divisia monetary aggregates outperform simple-sum approaches, suggesting further refinements in how we model money demand.

ML techniques have emerged as promising alternatives, offering tools to capture complex, nonlinear interactions in macroeconomic data. Goulet Coulombe et al. (2022) highlight the advantages of ML approaches, including regularization and cross-validation, particularly during periods of heightened uncertainty. Studies such as Nguyen et al. (2022) and Pham et al. (2022) demonstrate that advanced models like LSTM and neural networks outperform traditional econometric models in forecasting a wide range of macroeconomic variables, including inflation and monetary aggregates.

Research Gap

Despite extensive research on the impact of U.S. monetary policy on global economic conditions, several critical gaps remain, particularly concerning EMEs:

• Limited focus on unconventional monetary policy spillovers: Most studies emphasize conventional monetary instruments, such as changes in the federal funds rate, while largely

overlooking the effects of unconventional policy measures like QE. Given the increasing reliance on QE, its role in influencing capital flows, exchange rates, and macroeconomic stability in EMEs requires further exploration.

• Heterogeneity in EME responses to external shocks: Existing studies often treat EMEs as a homogenous group, neglecting regional and structural differences that shape their vulnerability to U.S. monetary shocks. Factors such as financial openness, foreign exchange reserves, and debt composition may influence how individual economies respond, but these variations remain underexplored.

• Unclear transmission mechanisms of U.S. EPU: While heightened U.S. EPU is known to influence global financial markets, its precise impact on EMEs remains insufficiently studied. The extent to which uncertainty affects capital flows, investor sentiment, and exchange rate volatility across different EMEs is still not fully understood.

• Forecasting limitations in macroeconomic modeling for EMEs: Traditional econometric models often struggle to capture the non-linearities and structural shifts that characterize EME economies, particularly during periods of external turbulence. The potential of advanced ML techniques to improve forecasting accuracy has not been sufficiently explored in this context.

Addressing these gaps is essential for developing more effective monetary policy strategies tailored to the unique vulnerabilities of EMEs.

Objectives of the Research

To bridge these gaps, this dissertation pursues three key objectives:

1) Examine the effects of U.S. interest rate shocks on EMEs by analyzing how both conventional and unconventional monetary policies influence GDP, inflation, exchange rates, broad money (M3 or M4), and foreign exchange reserves.

2) Investigate the spillover effects of U.S. economic policy uncertainty on EMEs, focusing on how variations in U.S. EPU impact macroeconomic stability indicators, such as economic growth, inflation, exchange rates, and interest rates.

3) Evaluate the potential of advanced machine learning (ML) techniques for forecasting key monetary variables, specifically money demand, in an emerging market context, thereby providing a practical toolkit for policymakers to anticipate and mitigate monetary instability.

Research Questions

Building on the identified research gaps, this dissertation seeks to answer the following key questions:

1) How do U.S. interest rate shocks affect key macroeconomic indicators in EMEs?

• This research examines the effects of U.S. monetary policy shifts, both conventional and unconventional, on inflation, GDP, exchange rates, foreign exchange reserves, and broad money in EMEs.

2) To what extent does U.S. EPU impact macroeconomic stability in EMEs?

• The study explores how heightened U.S. EPU affects economic growth, inflation, and exchange rates, identifying the key transmission mechanisms at play.

3) How can forecasting techniques improve monetary policy decisions in EMEs?

• This research assesses the potential of advanced forecasting models in predicting money demand and explores their role in helping policymakers mitigate the effects of external monetary shocks.

Methods

In order to achieve the objectives of the study, I apply econometric and statistical methods. Particularly,

1) To examine the effects of U.S. interest rate shocks on a broad set of EMEs, I utilize SVAR models incorporating the Wu-Xia shadow rate to derive U.S. interest rate shocks.

• Rationale: The Wu-Xia rate captures both conventional and unconventional (QE) policies, which many standard measures overlook.

• Subsequently, I employ a PSVAR model to analyze how these shocks influence key macroeconomic variables across the panel of EMEs, while controlling for country-specific fixed effects.

• Regional Sensitivities: I then extend the analysis to explore which regions or structural characteristics (e.g., degree of financial openness, reserve adequacy) heighten vulnerability to these spillovers. This approach helps address cross-country heterogeneity by identifying groups of EMEs with distinct responses.

2) To develop a dynamic and comprehensive understanding of how U.S. EPU impacts multiple EMEs, I employ a GMM-based PVAR approach.

• Justification: GMM estimators in a PVAR setting mitigate endogeneity by treating key variables, including U.S. EPU and EME macro indicators, as jointly determined, while also controlling for unobserved heterogeneity.

• Robustness Checks: Because smaller EMEs may not realistically influence U.S. policy, I also examine alternate specifications (e.g., treating U.S. EPU as predetermined) to ensure stability of results.

3) To provide empirical evidence on the suitability of ML methods for improving policy decisionmaking by offering more reliable forecasts, I do a comparative analysis of advanced forecasting techniques such as Random Forest regression, Gradient boosting, extreme gradient boosting, Support Vector Regression, LASSO, and LSTM with AR (1) as a benchmark model.

• Relevance for EMEs: These ML methods offer flexibility in capturing non-linear dynamics and regime shifts, which are common in EMEs facing volatile external environments (e.g., sudden capital flow reversals or policy shocks from the U.S.).

• Application: By forecasting money demand in India over a sample period that includes episodes of global and regional turbulence, I assess whether ML models can adapt more effectively than conventional approaches, thereby informing policymakers about potential instabilities in liquidity conditions.

Data

1) To analyze the impact of U.S. interest rate shocks on a broad set of EMEs, I use a monthly dataset covering the period from 2007 to 2020 for 29 EMEs identified largely through IMF classifications, with some exclusions due to data limitations. This period captures critical global events such as the Global Financial Crisis and the 2013 taper tantrum, during which EMEs experienced intensified external volatility. To represent U.S. monetary conditions in both conventional and unconventional phases, the study uses the Wu-Xia shadow rate (Wu,

Xia 2016). For each EME, three core macroeconomic indicators are examined, including exchange rate against the U.S. dollar, broad money, represented by the M3 or M4 monetary aggregates, and foreign exchange reserves.

2) For examining U.S. EPU spillovers on EMEs, a panel dataset of 39 EMEs (2005-2019) is utilized. The earlier start date offers sufficient coverage of pre- and post-crisis turbulence, while 2019 marks the last year for which the Oil Price Uncertainty index (an additional indicator of global volatility) is consistently available. Alongside the Baker et al. (2016) U.S. EPU index, this dataset includes each EME's real GDP, CPI, short-term interest rates, and nominal effective exchange rates, enabling a comprehensive exploration of dynamic policy uncertainty spillovers.

3) To forecast money demand in India, a monthly dataset (1997-2021) is used, focusing on M1 and M3 aggregates. The study employs the call money rate and government securities yield as interest rate variables, with income approximated by the Index of Industrial Production (IIP), exchange rate dynamics captured through NEER, and financial stability gauged by Bombay Stock Exchange (BSE) market capitalization.

Structure of the Text of Dissertation

The dissertation consists of an introduction, three chapters, a conclusion, a list of references and an appendix. Each chapter examines a key aspect of monetary instability in EMEs, focusing on external vulnerabilities and forecasting approaches.

The first chapter analyzes the effects of U.S. interest rate shocks on EMEs, examining both conventional and unconventional monetary policies. Using PSVAR model, it assesses how changes in U.S. interest rates influence inflation, GDP, exchange rates, foreign exchange reserves, and broad money in EMEs. The chapter also explores regional differences in economic responses.

The second chapter investigates the spillover effects of U.S. EPU on EMEs, focusing on its impact on inflation, exchange rates, and economic growth. Using a GMM-PVAR model, it examines how uncertainty in U.S. policy affects EMEs.

The third chapter conducts a comparative analysis of ML models for forecasting money demand in India, using AR (1) as the benchmark model. It evaluates the predictive performance of these

models in capturing non-linearities. The findings highlight the potential of ML-based forecasting to improve monetary policy formulation in EMEs.