NIMS - Dissertações de Mestrado em Estatística e Gestão da Informação (Statistics and Information Management)
URI permanente para esta coleção:
Navegar
Entradas recentes
- Integrating Climate Covariates into Mortality Forecasting Models: A Feasibility Study for Longevity Risk PricingPublication . Matteucci, Alessandro; Bravo, Jorge Miguel VenturaThis thesis investigates whether climate-related determinants—specifically extreme heat intensity—can be meaningfully incorporated into actuarial mortality modelling and whether such integration is empirically justified within the context of Southern European populations. Motivated by the growing intersection between environmental risk, demographic change, and longevity risk management, the study examines how exogenous climate variables may enter mortality-intensity frameworks traditionally structured around age, period, and cohort effects. At a conceptual level, the thesis formalizes the integration of a standardized heatanomaly indicator into a mortality modelling architecture consistent with the Generalized Age–Period–Cohort (GAPC) paradigm. At an empirical level, rather than estimating a fully stochastic climate-augmented GAPC model, the study implements a reduced-form Poisson Generalized Linear Model (GLM) panel specification. This approach allows the direct identification of the marginal association between regional annual heat anomalies and mortality rates at ages 65–85, while controlling for flexible age effects, region-specific heterogeneity, and common temporal shocks. Using harmonized mortality, exposure, and climate data for Southern European regions, the analysis finds that the estimated heat coefficients are generally small in magnitude and not consistently statistically significant under cluster-robust inference. Even where statistical significance is detected, the implied proportional change in annual mortality associated with a one-standard-deviation increase in heat intensity remains modest. Within the annual fixed-effects framework adopted, extreme heat anomalies do not emerge as dominant drivers of mortality dynamics. The contribution of the thesis lies in establishing a disciplined methodological bridge between climate indicators and actuarial mortality modelling. By linking demographic forecasting, environmental risk, and actuarial finance, the study offers an early step toward climate-aware longevity modelling systems.
- Forecasting Commodity Volatility under Geopolitical Risk: Evidence from the Russia–Ukraine WarPublication . Reis, Vasco Montez Barata Coelho Dos; Damásio, Bruno Miguel PintoThis study examines the impact of geopolitical risk on the volatility of Brent crude oil, gold and wheat in the period from 2020 and 2025, with particular focus on the Russia–Ukraine war. Understanding how geopolitical tensions influence commodity markets is essential for investors, policymakers and risk managers as these assets play a central role in global energy, financial stability and food security. The analysis focuses on identifying the extent to which geopolitical shocks influence commodity market volatility and on comparing the forecasting performance of econometric models with that of machine learning techniques . Daily price data were collected from Yahoo Finance and combined with the Geopolitical Risk Index, and the empirical analysis relied on GARCH-type volatility models (GARCH, EGARCH and GJR-GARCH), a GARCH–MIDAS specification incorporating the Geopolitical Risk Index as a low-frequency driver, and two non-linear machine learning methods, Random Forest and Gradient Boosting. The results show that increases in geopolitical risk significantly amplify volatility, with econometric models being more effective in capturing short term clustering and machine learning methods providing stronger predictive accuracy out of sample. These findings highlight the value of incorporating geopolitical indicators into risk management frameworks and clarify the channels through which geopolitical tensions shape commodity market dynamics , offering evidence that supports more reliable forecasting in periods of heightened uncertainty
- Measuring the impact of climate-induced hydrological and meteorological risks on the non-life insurance sector until 2050/2100 in AustriaPublication . Wagenbauer, Ina Katharina; Bravo, Jorge Miguel VenturaClimate change is expected to intensify weather-related hazards in Austria, raising concerns about future insurance losses and the resilience of traditional risk analysis strategies. This thesis analyses the correlation between climatic variability and non-life insurance payouts in Austria. This was accomplished by developing a regression model in order to forecast prospective losses under two climate scenarios, which are a mild baseline scenario (RCP4.5) and a high-risk scenario (RCP8.5). Due to the lack of detailed non-life insurance payout data, an Elastic Net model was created, as it works well with small datasets. The study utilises a seasonal dataset of historical payouts from the Austrian non-life insurance market and climate indicators from 27 weather stations to develop an Elastic Net regression model that quantifies the statistical correlation between climate variables and insurance losses, accounting for temporal trends and seasonal variations. Due to the limited availability of detailed non-life insurance claims data and the confidential nature of insurance records, a simple modelling technique has been employed that focuses on mean temperature, summer days, and rainy days as key predictors. The results reveal that payouts have been going up steadily over time, and the scenario forecasts indicate that future losses will be much larger under both climate trajectories. Although the RCP8.5 scenario results in slightly higher payouts than RCP4.5 until 2100, the gap between scenarios is minor relative to the overall increase. Despite its linear framework and simplified representation of climate-loss correlations, the model offers a clear and cautious projection of possible future scenarios. Additionally, the results highlight the necessity for non-life insurers in Austria to include climate risk in long-term planning, adjust pricing, and allocate resources towards adaptation and loss-prevention initiatives. Therefore, it is crucial to include scenario analysis with more detailed data and focus on a specific region or hazards to receive enhanced results. So, the thesis concludes by identifying key directions for future research, particularly regarding data availability, regional risk differentiation, and the development of more advanced modelling approaches, if the available data is sufficient.
- Tarifação em Seguros Não Vida: Um estudo do ramo MarítimoPublication . Figueiredo, Carlos Filipe Mendes Dutra; Damásio, Bruno Miguel PintoA tarifação em seguros Não-Vida é um dos factores mais importantes para uma constante adaptação ao mercado concorrencial, permitindo o cálculo de prémios mais justos para os segurados sem prejuízo da rentabilidade dos produtos de seguros. Constatando-se a necessidade de modelos robustos e ajustados ao nível de risco do objecto segurado, pretende-se com este projecto investigar como desenvolver um modelo para cálculo do Prémio Puro das coberturas de risco danos próprios no ramo marítimo de seguros. Foi feito o estudo separado dos modelos de frequência e severidade, identificando os factores tarifários mais relevantes para cada uma destas componentes. Com recurso a R, foram desenvolvidos modelos baseados na metodologia clássica dos Modelos Lineares Generalizados e, por outro lado, foram explorados métodos de machine learning, mais especificamente o Gradient Boosting. As duas metodologias foram avaliadas comparativamente concluindo-se que os MLG foram a metodologia com melhor desempenho na modelação da frequência e o Gradient Boosting na modelação da severidade. Ambas as abordagens apresentam ser opções válidas para desenvolver modelos de tarifação das coberturas de risco em estudo, confirmando os métodos de machine learning como possível metodologia auxiliar ou alternativa à metodologia clássica.
- Optimizing Hospital Billing Processes: A Process Mining approach to Performance AnalysisPublication . Eneh, Kaodichinma Chisom; Neves, Joana Paisana Pires Costa dasHospitals’ billing processes are often slow, fragmented and difficult to control. To address this, the study applied process mining within a regional hospital’s ERP system, analysing three years of event logs (over 100,000 billing cases) through the CRISP DM and DSRM frameworks. First, the extracted and transformed event logs were loaded onto Celonis; next, fuzzy and heuristic mining in Celonis uncovered 18 core billing activities and 950 process variants. Conformance checking against an ideal six step “happy path” workflow showed only 34 % of cases strictly followed the prescribed sequence. Surprisingly, 66 % of cases that deviated ran on average 5.4 days faster than fully conforming cases. Overall performance analysis revealed wide variability: the median end to end cycle time was 122 days, with some cases stretching out to 1,032 days. The transition from “FIN” to “RELEASE” emerged as the main bottleneck, delaying 67 % of all cases. Building on these insights, this study proposed various targeted interventions, including rule-based automation of key hand‐offs, parallel validation streams to reduce idle waiting, exception‐handling frameworks to streamline deviations, and real-time monitoring dashboards for operational transparency. These recommendations aim to speed up billing throughput while preserving compliance and auditability. In demonstrating how process mining can surface hidden inefficiencies across multiple departments, the study offers a replicable methodology for continuous improvement. It also calls for future work to extend the approach across institutions, incorporate qualitative staff feedback, and pilot the proposed automations to quantify cost savings and operational impact.
- Regulatory Equivalence and Liquidity Risk Management: Impact on Banking Efficiency in AngolaPublication . Dala, Hedosanjos Clinton João; Ashofteh, AfshinLiquidity risk management is fundamental to the stability and operational resilience of banking institutions, particularly in emerging economies such as Angola, where the Banco Nacional de Angola (BNA) has pursued regulatory equivalence with European Union standards since 2020. Given the limited availability and granularity of banking data, this study adopts an illustrative empirical approach using GARCH family models such as GARCH(1,1), EGARCH(1,1), and TGARCH(1,1), to explore return volatility dynamics in Angolan commercial banks over the period 2014–2024. The models serve as practical demonstrations of how conditional heteroskedasticity and asymmetric responses to shocks may be analysed in data-constrained contexts rather than as definitive empirical evidence of strategic effectiveness. The findings show persistent volatility, leverage effects, and significant asymmetries, particularly under the EGARCH specification, consistent with patterns observed in emerging markets. Overall, the study demonstrates the applicability of volatility modelling to the Angolan banking sector and provides qualitative insights into liquidity risk behaviour, regulatory convergence, and operational efficiency.
- Efficiency in Public Procurement of Healthcare Services: A Quantitative Analysis of Portuguese ContractsPublication . Novo, Mariana Pires; Damásio, Bruno Miguel Pinto; Pinheiro, Flávio Luís PortasThis research explores the management of public procurement within Portugal’s healthcare sector, a domain of strategic national relevance that represents more than 10% of GDP and plays a critical role in both population well-being and fiscal responsibility. Despite its centrality, this area has received limited empirical attention regarding the technical efficiency of contract allocation and execution. The study aims to identify efficiency gaps and propose data-driven strategies for performance improvement, drawing on a data set of completed healthcare procurement contracts retrieved from the BASE.Gov platform. A methodological framework that combines stochastic frontier analysis (SFA), benchmarking techniques, and machine learning (random forest) was employed to measure efficiency levels, uncover key performance drivers, and simulate alternative policy scenarios. The findings reveal widespread inefficiencies, with a large share of contracts performing significantly below optimal levels, and marked regional disparities, particularly in Lisbon and Madeira. The simulation results suggest that aligning underperforming contracts with high-efficiency benchmarks, or leveraging predictive models for early risk identification, could yield considerable fiscal savings. These insights underscore the value of integrating benchmarking and predictive analytics into procurement systems, offering policy-relevant contributions to institutional modernization, enhanced transparency, and sustainable management of public resources.
- Bond Immunization Strategies Based on Duration Model: A Study of Angola Government BondPublication . Ndala, Jasmim Pedro Pinto; Ashofteh, AfshinThe goal of this research project is to illustrate the bond immunization strategies based on the duration Model, precisely the Mauculay Doration, from the perspective of controlling interest rate risk. Our empirical analysis focuses on the Angola Government debt bond over a specified period, considering the calendar of titles issued on the Angolan securities market. The yields are recorded on February and July of each year from 2025 to 2029. The performance of this strategy is tested over 1-, 2-, 3-, and 4-year horizons. The results confirmed that duration-matching portfolios provide good immunization performance. This research project concludes with implications for investors or portfolio managers.
- Integração do mercado de ações na SADC: Determinantes de integração do mercado de ações na SADCPublication . Pinto, José Eduardo Nganga; Bravo, Jorge Miguel VenturaContexto e objetivo: A integração dos mercados de ações é um desafio central para o desenvolvimento económico regional. Este estudo avalia o grau de integração do mercado de ações na SADC no período 2014–2019 para 5 dos 16 Estados-Membros. Métodos: Estima-se um modelo DCC-GARCH com rendibilidades diárias, assumindo distribuição tStudent para acomodar caudas pesadas e eventos extremos. Avaliam-se volatilidades condicionais dinâmicas e correlações condicionais dinâmicas como métricas de integração. Dados: São usadas séries diárias de rendibilidades de mercados de 5 dos 16 países na SADC no período 2014–2019. Resultados principais: São identificadas volatilidades condicionais dinâmicas elevadas e persistentes na generalidade dos mercados; As correlações condicionais dinâmicas são baixas e, por vezes, negativas, sugerindo ausência de integração ao nível regional; O par África do Sul–Namíbia destaca-se com β > 0,95 (forte persistência da volatilidade) e correlação média ≈ 0,90, indicando maior integração bilateral face aos restantes mercados. Conclusões: No período analisado, o mercado de ações dos cincos países analisados pertencentes à SADC não se encontram integrados como bloco, embora existam núcleos de maior integração (p.ex., África do Sul–Namíbia) que ajudam a identificar determinantes relevantes. Implicações de política: A integração efetiva requer compromisso político, institucional e económico entre Estados-Membros, incluindo: (i) concretização de uma zona de livre comércio; (ii) harmonização fiscal, cambial e regulatória; (iii) reforço de um enquadramento jurídico independente; (iv) cooperação técnica; e (v) investimento em infraestrutura tecnológica para mercados financeiros. Implicações para investidores: Identificam-se oportunidades tanto em mercados com elevada correlação e estabilidade (integração maior) como em mercados com baixa correlação, que podem oferecer diversificação e arbitragem de risco. Limitações e investigação futura: Resultados circunscritos a 2014–2019 para apenas 5 EstadosMembros a um único enquadramento (DCC-GARCH com t-Student). Investigações futuras poderão alargar o horizonte temporal, testar especificações alternativas, incorporar mais Estados-Membros e choques macroeconómicos e institucionais.
- Hybrid Pension Plans and Risk Sharing: A Simulation-Based Comparison of Defined Benefit and Collective Defined Contribution Models in the Dutch Pension SystemPublication . Martins, Maria Isabel Guerreiro; Bravo, Jorge Miguel VenturaConcerns about demographic ageing and financial sustainability have led to increased attention to reform of the pension system in recent years. Traditional Defined Benefit (DB) plans are giving way to more flexible and risk-sharing Collective Defined Contribution (CDC) models in the Netherlands, a country renowned for its robust financed system. This thesis investigates the transition of the Dutch pension system from a traditional DB structure to a CDC model, a reform aimed at improving long-term sustainability and intergenerational fairness. Using a dynamic and empirically calibrated simulation over a 60-year horizon, the study assesses the financial and demographic evolution of pension funds under both the DB and the CDC schemes. The model incorporates stochastic investment returns, salary growth, and cohort-based mortality from the AG2024 tables, as well as institutional rules that govern contributions, benefits, and solvency thresholds. A dual-phase structure enables the comparison between DB and CDC regimes, with calibration based on empirical data from the Dutch Central Bank. It is to applied also scenario testing, including baseline, longevity shocks, low-return environments, and policy reforms, to examine how various conditions influence the behaviour of pension funds. The methodology emphasises empirical relevance and policy alignment, offering a structured framework to assess the effects of systemic pension reform in a rapidly ageing society.