Prepayment risk creates cash flow uncertainties that complicate valuation, hedging, and risk management in mortgage lending. Prediction is indeed difficult: borrowers respond nonlinearly to economic drivers, prepayment events are rare, and shifting portfolio composition contaminates comparisons over time. We model prepayment on French residential securitized mortgages observed quarterly from August 2020 through February 2022 and benchmark a single-hidden-layer neural network against logistic regression, deeper network architectures, Random Forest, and XGBoost. Machine learning models consistently outperform the logistic baseline. The shallow network improves discriminative ability by roughly 20%, and XGBoost attains a mean out-of-sample AUC of 0.82, which evidences the interaction and threshold effects that traditional models like logistic regression cannot represent. SHAP and LIME attributions agree on the driver hierarchy. Under near-zero rates, refinancing incentives lose much of their predictive power, and prepayment is shaped by a broader set of conditions, real estate dynamics above all, followed by local unemployment and finally also by the financial incentive to refinance.
This study examines how fairness perceptions influence cryptocurrency investors and market behavior, using the Ultimatum Game as the conceptual anchor. We specify a framework that links distributive, procedural, and informational justice to buy and sell intentions, holding horizons, and community participation, with emotions such as anger and trust acting as mediators. To address the distinction between investor-level perceptions and market-level volatility, the revised design combines online survey responses with token-day trading data, event indicators, on-chain concentration measures, disclosure records, and community-activity proxies. Survey data are used to test individual behavioral and emotional channels, whereas market data are used to estimate abnormal returns, realized volatility, and difference-in-differences effects around fairness shocks. The revised analysis therefore avoids inferring market volatility directly from respondent opinions. Findings indicate that perceived unfairness raises selling intention and shortens holding horizons, while objectively observed fairness shocks are associated with higher realized volatility. Heterogeneity checks show stronger effects for assets with high decentralization and active communities. The study offers design levers for transparent rules, governance safeguards, and clear disclosure cadence.
Portfolio credit risk models typically incorporate the default dependence between obligors that arises from common risk factors. However, they often ignore obligor-to-obligor default contagion effects, which can propagate through economic networks. We present a structural framework that can be layered onto existing factor models to capture higher-order network-based default contagion. Our approach yields efficient and tractable computations that are applicable to heterogeneous singly connected networks, which can represent empirically observed dependence structures such as supply chains, corporate groups and hierarchical governments. As a key contribution, we develop an expectation-maximization algorithm that estimates default contagion parameters from historical creditworthiness information. Our simulation-based numerical results show that the impact of default contagion on risk measures is well estimated by our framework.
This study investigates whether cryptocurrency adoption moderates the impact of fiat currency depreciation and instability on economic development across 118 countries from 2021 to 2024. Using unbalanced panel data and fixed-effects estimations with clustered, bootstrapped, and Driscoll-Kraay standard errors, as well as the Generalized Method of Moments, the results reveal that both depreciation and instability significantly hinder development. While cryptocurrency adoption itself is associated with a negative impact on development, it mitigates the adverse effects of depreciation and instability. Notably, this moderating effect is evident only in emerging and developing economies. However, despite these benefits through the channels of government expenditures, trade openness, foreign direct investment, and financial depth, the broader harms of cryptocurrency adoption outweigh the gains. Consequently, aggregate economic development declines more on net when cryptocurrency adoption is introduced to counteract depreciation and instability than when these conditions occur without adoption. Our paradox recommends that while cryptocurrency adoption can serve as a tool to address monetary challenges, its drawbacks should be managed before it can contribute positively to development.
The paper proposes a nonlinear optimal control method for treating the control and stabilization problem of Keen's macroeconomic model which describes the debt-labor interaction and the associated economic growth cycles. The primary state-space model of the debt-labor dynamics is considered to have as state variables the wages share, the employment share and the debt-share, while its control inputs are the rate of growth of the labor productivity, the rate of growth of the labor force and the debt's interest rate. It is proven that the dynamic model of the debt-labor interaction is differentially flat and a nonlinear optimal control method is developed for it. To apply this control method, approximate linearization is performed with the use of Taylor-series expansion while an algebraic Riccati equation has to be solved at each sampling instance. The proposed control method avoids complicated changes of state variables and state-space model transformations while the control inputs it computes are used directly on the initial nonlinear model of the controlled system. It achieves optimality because of enabling convergence of the state variables of the debt-labor dynamics to the targeted setpoints under minimal variations of the control inputs. The paper's nonlinear optimal control method is compared to flatness-based control implemented in successive loops.
Stochastic resonance (SR), the counterintuitive enhancement of weak signals by noise, is well-established in physics but rarely demonstrated empirically in finance. We analyze 47 years of S&P 500 data using wavelet decomposition and Hidden Markov Models (HMMs) to reveal a quarterly cyclical signal masked by market trends. At optimal volatility (sigma(& lowast;)=0.22), signal-to-noise ratio (SNR) increases by 47.66dB (95% CI: [46.11, 48.58]dB, Cohen's d=74.2, p<0.001). We propose a novel mechanism: volatility breaks trend dominance through residence-time asymmetry collapse, decoupling trend from cycle. Kramers rate theory provides excellent validation (R-2=0.981), confirming regime transitions follow thermally-activated escape dynamics. Out-of-sample validation on Gold futures (XAU/USD, 2008-2020) replicates the phenomenon (sigma(& lowast;)=0.09, gain =46.21 dB), establishing cross-asset universality. Results reframe market noise as information rather than risk.
Purpose: This study aims to develop a dynamic, risk-sensitive asset allocation framework for pension funds in Ghana. It is specifically motivated by the 2022-2023 domestic debt exchange program, which exposed critical vulnerabilities in pension portfolios heavily concentrated in government securities.Design/methodology/approach: The research employs a continuous-time stochastic model that incorporates equities, risk-free treasury bills, and defaultable sovereign bonds. Government bond default probabilities are modeled using the Cox-Ingersoll-Ross process, while pension liabilities evolve stochastically. The framework is evaluated using a 10-year Monte Carlo simulation with 10,000 iterations based on empirical Ghanaian data from 2013 to 2023.Findings: Results indicate that sovereign default intensities are mean-reverting but highly cyclical, spiking during fiscal stress. Portfolios concentrated in domestic bonds suffer significant wealth erosion under default conditions, while aggressive portfolios are highly sensitive to credit shocks. In contrast, balanced portfolios consistently deliver superior risk-adjusted outcomes.Research limitations/implications: The study's findings are based on the specific macroeconomic and institutional context of Ghana. Future research could test the model in other emerging markets with similar fiscal volatility.Practical implications: The findings advocate for a paradigm shift among Ghanaian pension funds toward liability-driven investment (LDI) strategies and greater diversification into equities, inflation-linked, and alternative assets. The study strongly recommends regulatory reform by the National Pensions Regulatory Authority (NPRA) to enable this necessary diversification and enhance long-term portfolio resilience.Social implications: The research highlights the direct link between pension fund solvency and societal well-being. Effective implementation of the proposed framework is critical for protecting the retirement savings of millions of Ghanaians, thereby reducing old-age poverty, maintaining social stability, and upholding public trust in the national pension system.Originality/value: This paper provides a novel, integrated framework that explicitly models stochastic sovereign default risk within pension fund asset allocation - a critical factor often overlooked in emerging market contexts. It offers evidence-based strategic and regulatory prescriptions to preserve solvency and ensure sustainable retirement outcomes in high-volatility, fiscally constrained environments.
Using the financial parameters of firms listed on the Borsa Istanbul BIST 30 index from 2004 to 2024, this study aims to identify anomalies in various time periods and calculate dynamic portfolio weights based on these findings. Beyond typical portfolio techniques based on fixed assumptions, a strategy responsive to market conditions has been created in light of the turbulent character of financial markets. In addition to fundamental financial measurements, anomaly scores were developed annually for each business in the research and utilized in portfolio development. The Random Forest technique was utilized to ascertain the variable significance levels in this case, whereas the Isolation Forest approach was employed for anomaly identification. The dynamic portfolio outperformed the traditional equal-weighted approach in the majority of years, according to comparative studies utilizing criteria like return, Sharpe ratio, and maximum value loss. Specifically, it was discovered that anomaly scores contributed to the attainment of risk-return equilibrium. The weights assigned to debt and market valuation ratios in the model have influenced portfolio performance, and their effects on returns are now clearly visible. The application findings suggest that portfolio management can benefit from judgments based on risk indicators that fluctuate over time in addition to historical data.
This paper develops a novel option-theoretic framework for valuing carry-forward tax losses in investment portfolios. We demonstrate that the tax shield generated by a realized loss is mathematically equivalent to a capped long call spread (bull call spread), with payoff bounded by the statutory tax rate multiplied by the loss amount. Using risk-neutral valuation techniques, we derive closed-form expressions for the present value of tax shields and establish clear bounds on their economic benefit. The framework reveals that prospective tax liability on future gains replicates an uncapped short call option struck at the original portfolio value plus the realized loss. Our analysis provides transparent decision criteria for tax-loss harvesting strategies and unifies derivative pricing with tax-aware portfolio management. Empirical analysis using Indian market parameters shows that the present value of tax shields typically falls well below their nominal ceiling, providing important guidance when harvesting strategies create genuine economic benefit.
The global credit outlook is worsening for various industries as international trade tensions intensify. As AI becomes more embedded in business processes, it is crucial to understand its impact on a company’s creditworthiness to ensure financial stability. This research empirically investigates how enterprise-level AI adoption affects credit risk, using data from Chinese A-share-listed firms from 2012 to 2024. Fixed-effects regression analysis indicates that adopting AI significantly reduces the firm’s credit risk. Further examination of underlying mechanisms identifies three primary channels through which AI reduces credit risk: asset volatility, internal control quality, and financial leverage. Heterogeneity tests indicate that the credit risk reduction effect is stronger in firms within highly digitalized, high-tech industries, and with lower financing constraints. The AI adoption is significantly correlated with reduced credit risk exposure, providing valuable insights for existing research on organizational digitalization and risk mitigation strategies. Furthermore, these findings hold key implications for stakeholders, suggesting that corporate managers can align AI adoption with risk management, especially internal control, and governments can provide more support for the AI adoption of innovative, digitally strong firms.
The stock market is an important investment channel and asset allocation tool for investors. How to use portfolio strategy to combine different stocks together to maximize investment return and risk control is the most important concern for investors and a very important issue in the field of finance. In this paper, we propose a deep reinforcement learning (DRL)-based machine learning approach (DSR_RL) for stock portfolio selection and prediction of investment returns. An innovative financial model that integrates the differential Sharpe ratio of multiple stocks as the agent reward function is proposed. In addition, we propose an integrated learning framework with an alternate training scheme (ILAT) that ensembles the decisions of multiple reinforcement learning algorithms to approximate a theoretically optimal portfolio. Compared with classical portfolio strategies and existing DRL baselines, DSR_RL achieves annualized returns of around 80% on our selected stock portfolios, while ILAT further improves the best-case annualized return to 94.23% (on the rising-stock portfolio); meanwhile, the average Sharpe ratio is improved by about 0.15. The effectiveness of the proposed method is demonstrated on U.S. equity portfolios under different market regimes and financial cycles.
This paper examines the relationship between oil prices and commodity futures across energy-linked assets, industrial metals, precious metals, and agro-food commodities in the pre-COVID period, during the COVID-19 pandemic, the post-COVID and the Russia-Ukraine war period. Using NARDL and QNARDL models, we capture asymmetric short- and long-run effects of oil returns, trading volume, and open interest, complemented by wavelet coherence analysis to assess time-frequency co-movements. In the pre-COVID period, energy assets (gasoil, natural gas) showed strong coherence with oil, industrial metals showed moderate mid-frequency co-movements, agro-food commodities were largely decoupled, and gold minimally correlated. During COVID-19, long-term positive co-movements appeared for oil-natural gas, oil-cocoa, oil-gold, and oil-palladium, while oil-copper and oil-white maize showed negative coherence, reflecting crisis-induced contagion. Energy and precious metals experienced intensified speculation, agricultural commodities acted mainly as hedges, and natural gas exhibited regional asymmetries and lead-lag shifts. Gold functioned as a safe-haven, though occasionally constrained by forced liquidations. In the post-COVID and Russia-Ukraine war period, short-term coherence surged for most commodities, white maize, zinc, palladium, cocoa, and gold, highlighting rapid contagion under geopolitical shocks, whereas oil-natural gas and oil-copper correlations were lower, suggesting hedging potential. Overall, co-movement and hedging properties are crisis-specific, emphasizing the value of sub-period analysis for investment and risk-management strategies.
This study examines the influence of capital structure on firm value, considering the moderating role of market perception. Market perception is reflective of information perceived by outsiders in the context of signaling theory. We have applied ordinary least squares (OLS) and feasible generalized least squares (FGLS) regressions to a panel data set of 281 nonfinancial firms listed at the Pakistan Stock Exchange (PSX) from 2017 to 2022. Our analysis indicates a negative influence of capital structure, measured through total, short-term, and long-term debt ratios on the firm value. Additionally, the moderating influence of market perception is found to be positive. It suggests that market perception can alleviate the adverse effects of higher debt ratios on firm value. These findings emphasize the importance of a favorable market perception along with efficient financial management for firm value. Moreover, the findings have significant implications for firms' managers and policymakers. They must make appropriate decisions regarding capital structure and cultivate strong market perceptions that yield optimum firm value.
In an era focused on the evolving perspectives of gender equality, this study investigates the effect of boardroom gender diversity on key corporate decisions and their outcomes. Tobit and logit regression and generalized method of moments (GMM) are used to study the impact of boardroom gender diversity on firm efficiency, dividend payout, leverage, profitability and insolvency. The underlying study estimates the efficiency scores of nonfinancial firms using data envelopment analysis and insolvency using the Altman Z-score emerging market model. The data from 261 listed firms in the nonfinancial sector of the Pakistan Stock Exchange over five years (from 2017 to 2021) are employed for hypothesis testing. The findings that a gender diverse board with female representation improves the firm's efficiency are of interest to policymakers and shareholders. The study highlights that regulators in the developing economies should devise policies to increase the female representations in the corporate boards. The findings suggest that the presence of female in the board minimizes agency issues, as the strategic choices of female members are inclined primarily toward the increase in shareholder value. Female board members are more risk-averse, better decision makers and focus on improving the firm's value. This study is a novel attempt to estimate the role of gender in the efficiency of listed firms.
This study examines the predictive relationship between macroeconomic indicators and the performance of utility and equity ETFs using a hybrid machine learning (ML) framework. Six variables, FDI, GDP, GDP growth rate, industrial production, net exports, and population, were analyzed. A hybrid approach combining Grey Relational Analysis (GRA) and Random Forest (RF)based feature engineering was used to enhance forecasting accuracy. Five (ML) models (RF, DT, KNN, NB, and SVM) were tested under two specifications: baseline variables versus RF-selected features. Results show that the hybrid models consistently outperformed baseline models across R-2, RMSE, and MAE, with the support vector machine delivering the best performance. Robustness was confirmed across multiple data splits and validated using Diebold-Mariano tests. The findings highlight key macroeconomic drivers of ETF returns and demonstrate the benefits of integrating feature selection with ensemble learning.
In the literature, on optimal portfolio selection, closed-form solutions are rare, specially when market friction is present. This becomes even more challenging if market friction comes in the form of liquidity risk. This paper bridges this gap by formulating a portfolio selection problem with liquidity risk as a stochastic control problem with Constant Relative Risk Aversion (CRRA) utility function. Assuming that liquidity costs are generated by rebalancing the portfolio over finite intervals, rather than by instantaneous position changes, we derive a closed-form expression for the optimal portfolio weights. This formulation offers two notable advantages. First, it explicitly takes liquidity costs into the wealth dynamics and provides an exact analytical solution to the problem. Second, this formulation has the feature to trace back to the classical [Merton, RC (1969). Lifetime portfolio selection under uncertainty: The continuous-time case, The Review of Economics and Statistics, 51(3), 247-257] model in the case of perfect asset liquidity. In high-liquidity regimes, our numerical results show that new optimal portfolio weights initially exhibit variation but ultimately converge to the frictionless solution proposed by Merton [(1969). Lifetime portfolio selection under uncertainty: The continuous-time case, The Review of Economics and Statistics, 51(3), 247-257] as the terminal time T approaches. Finally, the quantitative impact of this new solution is thoroughly analyzed, providing valuable insights into portfolio optimization under liquidity constraints.
This study explores how firms adjust their green innovation (GI) activities as a business strategy in response to economic policy uncertainty (EPU). Using a data set of listed firms in China over the period 2007-2023, we find that firms increase their GI engagement during periods of increased EPU. We also find that the positive association between EPU and GI activities is more pronounced in firms with lower investor attention, when investor sentiment is optimistic, and in firms with higher reputation capital. Our analysis suggests that GI activities may help firms to alleviate the impact of increasing economic uncertainties. We also argue that the strategic role of GI deserves attention from managers, investors and policymakers alike.
Purpose: This study explores the moderating role of inflation targeting (IT) regime in shaping stock market volatility drivers, leveraging a machine learning (ML) approach to elucidate the complex interplay between monetary policy regimes and macro-financial channels. By analyzing key macro-financial interactions, we aim to provide actionable insights for policymakers and investors in emerging markets.Methodology: This study analyzes 16 IT and 16 non-IT emerging markets (1995:Q1-2024:Q1) using macroeconomic data, policy signals, country-specific factors, and 11 ML techniques (regularization, trees, neural networks) alongside various GARCH-type models. Furthermore, both mean SHAP importance and SHAP interaction techniques are used to rank features and uncover macro-financial channel interactions.Findings: Findings reveal distinct volatility drivers across regimes. Non-IT markets react sharply to immediate market signals and external vulnerabilities, amplified by macro-financial channels, notably institutional quality interacting with global risk. Conversely, IT regimes exhibit greater stability through structured macroeconomic fundamentals and disciplined policy frameworks that buffer short-term shocks. The IT regime's predictability transforms volatility dynamics, replacing sentiment-driven fluctuations with lagged fundamentals and external risk factors, fostering more resilient financial environments compared to non-IT contexts.Practical implications: Non-IT policymakers should strengthen institutions and monitor external balances to mitigate volatility, while considering IT adoption. IT countries must maintain credible regimes with flexible exchange rates to stabilize markets and boost confidence.Originality/Value: This study pioneers optimal volatility measures and ML methods for emerging markets, using SHAP to analyze macro-financial channels and IT's moderating role, advancing theory and guiding policy and investment strategies.
This paper discusses the optimal constant rebalanced portfolio choice problem of an insurer under the safety-first criterion. The surplus process of the insurer is assumed to follow the Cramer-Lundberg model. The insurer invests its surplus in a financial market, which consists of one risk-free bond and n risky assets, whose prices follow an n-dimensional jump-diffusion process. Using the safety-first criterion as a measure of the overall risk of the insurer, we explore the problem of selecting the optimal investment strategy for an insurer under the constraints of acceptable disaster levels. The martingale integral property is used to transform the dynamic problem into a static one, and explicit expressions for the optimal investment strategy are obtained. Based on this, the impact of factors such as disaster level, claim intensity, premium rate and claim size on the optimal investment strategy is analyzed, and some economic explanations are given. Finally, actual data from financial markets are used to simulate how insurance companies allocate investment funds.
