We examine the effect of media freedom on public expenditure efficiency using a panel of 134 advanced and developing countries from 1994 to 2016. Our findings provide robust evidence that greater media freedom significantly enhances government spending efficiency. This effect is more pronounced in countries with higher levels of democracy, per capita income, human capital, internet access, and strong fiscal rules. Furthermore, we show that media freedom primarily influences efficiency through reduced corruption, increased transparency and accountability in the public sector, and stronger electoral competition. These results underscore the vital role of media freedom in fostering good governance and efficient public resource allocation, particularly in a context where press freedom is declining and fiscal constraints are tightening.

This thesis examines the crypto-energy-macrofinance-environment nexus as an integrated systemlinking digital assets, energy markets, macrofinancial conditions, and environmental constraints. It starts from the premise that cryptocurrencies cannot be treated as isolated speculative assets,since their valuation, volatility, and systemic role are increasingly shaped by energy costs, financial uncertainty, and sustainability-related forces. The thesis is structured progressively : from conceptual synthesis to empirical mapping, and from structural identification to forecasting. The empirical analysis first documents strong and persistent multiscale connectedness among cryptocurrencies, energy variables, macrofinancial indicators, and environmental assets. Using awavelet TVP-VAR framework on U.S. data from January 2014 to June 2025, it shows that theTotal Connectedness Index generally ranges between 70% and 90%, peaking at 93% over long horizons. It also identifies successive contagion regimes: energy-crypto dominance before 2020,financial synchronization during COVID-19, and a macro-energy-environment configuration that gradually evolves into a digital-sustainability regime by 2025. Short-run contagion is mainly speculative and energy-driven, whereas long-run connectedness becomes increasingly anchored ininflation, industrial production, and electricity prices. The thesis then refines spillover measurement through the Clustered Network Connectedness framework, which separates within-cluster comovement from between-cluster contagion. Using daily data from January 2022 toOctober 2024, it shows that standard generalized connectedness measures overestimate total spillovers by about 8-12%. The energy cluster appears as a persistent net transmitter, with netconnectedness peaking at around +0.96, while cryptocurrencies shift from net receivers to nettransmitters after the FTX episode. Strong energy-to-environment transmission is also documented, with about 60% of environmental return variance explained by energy shocks. MonteCarlo results confirm that this gap reflects a structural identification bias rather than a finite-sample artifact. Finally, the thesis turns connectedness into a predictive signal. Clustered Forecasting Error Variance Decomposition spillover matrices are interpreted as dynamic directed graphs and embedded in a temporal graph-learning architecture combining graph attention and gated recurrent dynamics to forecast Bitcoin returns. The results show that predictive gains are horizon-dependent :limited at the shortest horizon, more pronounced at intermediate horizons, and most pronounced at long horizons. Within-cluster dependence matters most at medium horizons, whereas cross-cluster spillovers dominate at long horizons. These gains further translate into economically significant gains in volatility-targeted trading strategies. Overall, the thesis shows that the crypto-energy-macro-environment nexus is a structured transmission system whose architecture matters for interpretation, systemic-risk measurement, and forecasting.

This paper examines the impact of low-carbon transition risks on sovereign borrowing costs. Using two unbalanced panel datasets covering 125 countries from 1995 to 2019, we estimate extended models of the macroeconomic determinants of short- and long-term sovereign debt costs. We include key indicators capturing exposure to transition risks, such as fossil resource abundance, the carbon intensity of GDP, and the share of renewable energy in total energy consumption. Results show that fossil resource abundance and a higher renewable energy share are associated with lower borrowing costs, whereas greater carbon intensity raises sovereign debt costs. Financial markets therefore appear to reward fossil resource endowments while penalizing carbon-intensive uses of these resources. This reveals a contradictory signal: fossil resource wealth lowers the cost of public borrowing, as it is perceived as a form of implicit collateral, while the actual use of these resources increases borrowing costs, reflecting a carbon risk premium.