The first chapter of this thesis deals with households’ consumption-savings trade off in monetary policy models. The model’s fit with the data can be markedly improved first by the use of households specific interest rates, second by modelling households as inattentive agents. The second chapter proves analytically that the Taylor principle is valid in a model with both staggered prices and wages. According to the Taylor principle, when the central banker overreacts to inflationary shocks, he rules out sun spot equilibria and thus reduces economic fluctuations. This property is well known in models with staggered prices only, it had to be established analytically when wages are staggered. The third chapter quantifies domestic production (house-chores, cooking, handy work…) taking place out of the market. Valuing these activities induces an upward revision of economic activity but a downward revision of its growth. Methodological questions linked to the valuation of domestic work are discussed in details. The fourth chapter deals with the evaluation of structural reforms in institutional models (IMF, European Commission, ECB). The analysis of the core mechanisms of these models and their sensitivity to the calibration bring about two recommendations, first a systematic use of sensitivity analysis and more fundamentally a rethink of their long term mechanisms (in particular for labour supply).