This dissertation contains three essays in international macroeconomics and financial time series forecasting. In the first essay, I show, numerically, that a two-country New-Keynesian Sticky Prices model, driven by monetary and productivity shocks, is capable of explaining the highly positive correlation across the industrialized countries' inflation even though their cross-country correlation in money growth rate is negligible. The structure of this model generates cross-country correlations of inflation, output and consumption that appear to closely correspond to the data. Additionally, this model can explain the internal correlation between inflation and output observed in the data. The second essay presents two important results. First, gains from monetary policy cooperation are different from zero when the elasticity of substitution between domestic and imported goods consumption is different from one. Second, when monetary policy is endogenous in a two-country model, the only Nash equilibria supported by this model are those that are symmetrical. That is, all exporting firms in both countries choose to price in their own currency, or all exporting firms in both countries choose to price in the importer's currency. The last essay provides both conditional and unconditional predictive ability evaluations of the aluminum futures contracts prices, by using five different econometric models, in forecasting the aluminum spot price monthly return 3, 15, and 27-months ahead for the sample period 1989.01-2010.10. From these evaluations, the best model in forecasting the aluminum spot price monthly return 3 and 15 months ahead is followed by a (VAR) model whose variables are aluminum futures contracts price, aluminum spot price and risk free interest rate, whereas for the aluminum spot price monthly return 27 months ahead is a single equation model in which the aluminum spot price today is explained by the aluminum futures price 27 months earlier. Finally, it shows that iterated multiperiod-ahead time series forecasts have a better conditional out-of-sample forecasting performance of the aluminum spot price monthly return when an estimated (VAR) model is used as a forecasting tool.

L'objectif de cette thèse est d'évaluer l'interdépendance macroéconomique entre pays développés sur les récentes décennies et, en particulier, à la suite de la crise financière de 2007-09 aux États-Unis. Pour cela, on utilise différentes hypothèses de modélisation dans les trois chapitres principaux que constituent la thèse permettant de capturer la dimension internationale des cycles économiques : modèle VAR en panel permettant de modéliser l'interdépendance entre les pays directement, modèle VAR simple en utilisant des variables domestiques et étrangères, et modèle DSGE à 2 pays permettant de modéliser directement les mécanismes réels et financiers qui lient les pays entre eux. Notre résultat principal est que la dimension internationale est importante pour expliquer la dynamique macroéconomique des pays développés sur les trois dernières décennies, que les variables soient réelles, nominales ou financières. Néanmoins, le rôle des facteurs étrangers ne croit pas dans le temps comme on pourrait le penser avec l'accentuation de la mondialisation de ces dernières décennies. Aussi, en regardant les crises économiques récentes aux États-Unis et de la zone euro, nous confirmons que la crise financière américaine de 2007-09 présente un choc plus important comparé aux standards historiques, qui s'est propagé à la zone euro à travers les liens financiers internationaux. Au contraire, la crise des dettes publiques de la zone euro de 2011 est un choc relativement standard, similaire aux chocs observés pendant la crise du Système Monétaire Européen (SME) de 1992-93, et affectant principalement les économies européennes. / The aim of this thesis is to evaluate macroeconomic interdependence between developed economies over the recent decades and, in particular, following the 2007-09 US financial crisis. For that purpose, we use several modeling assumptions across the three main chapters of the thesis to capture the international dimension of business cycles across countries: panel VAR model to model countries interdependence directly, simple VAR model with both domestic and foreign variables, and two-country DSGE model to model the real and financial mechanisms that link countries together. Our main result is that international dimension is important to explain the macroeconomic dynamics of developed economies over the last three decades and for either real, nominal and financial variables. Nevertheless, the role of foreign factors does not grow over time as would be expected with the increase in globalization of the recent decades. Also, looking at the recent economic crises in the US and the euro area, we confirm that the 2007-09 US financial crisis features a bigger shock relative to historical standards, which propagated to euro area economies through international financial linkages. In contrast, the 2011 euro area sovereign debt crisis features a standard shock, comparable to those observed in previous European crises like the 1992-1993 ERM crisis, and affecting mostly European economies.