Kossi Agbanda*, Luca Cerasoli**

*We study debt sustainability and pollution mitigation in a two-country model with perfect capital mobility. Pollution abatement is financed through taxes or public bonds, while households can invest in productive capital or public bonds. We show that when the interest rate is low, the economy may converge to an equilibrium characterized by high capital accumulation, sustainable public debt, and improved environmental quality. However, depending on the initial levels of debt and capital, the economy may be relegated to an Environmental - Poverty Trap. We then analyze the role of fiscal policy and find that taxes on production and income can promote capital accumulation and environmental quality by expanding the tax base that finances mitigation. In contrast, higher public spending on abatement may crowd out productive investment and weaken debt sustainability. Fiscal instruments also affect the international distribution of public debt. Higher income taxes in the home country tend to increase foreign public debt, whereas higher production taxes can reduce it under certain conditions and generate positive welfare effects. Overall, the effectiveness of fiscal policy depends on the level of interest rates and on the extent to which households internalize pollution externalities.

**Since the beginning of the twenty first century, copulas have been introduced in f inance, becoming a powerful tool to disentangle complex and non linear depen dence structures between assets and other variables. Yet, dependencies do not remain static over time, rendering Static Copulas unreliable. This article explores a data-driven rolling window estimation (RWE) approach to model dynamic dependence between weather conditions in the USA, and the wheat Future prices, a setting where meteorological shocks have direct financial implications. Using principal component analysis to compress high dimensional weather data, we are able to remain in the bivariate case for our Copulas, which will later prove to have a negative impact in terms of information loss. The RWE method is benchmarked against the gold standard in terms of dynamic copulas, the [Patton(2006)] model for Elliptical and Archimedean copulas. Then, the accuracy of the forecasts of the [Patton(2006)] models is tested against a simple ARMA(1,1) model. Then from this simplistic base, we will derive some more sophisticated dependence benchmarks and forecasting methods, namely DCCGARCH and GAS as in [Engle(2002)] and [Creal et al.(2013)Creal, Koopman, and Lucas] on the depen dence side, and EGARCH and Stochastic Volatility models by [Nelson(1991)], [Jacquier et al.(1994)Jacquier, Polson, and Rossi], and [Taylor(1986)].