This paper estimates a two-country model with a global bank, using US and Euro Area (EA) data. Empirically, a model version with a bank capital requirement outperforms a structure without such a constraint. A loan loss originating in one country triggers a global output reduction. Banking shocks matter more for EA macro variables than for US real activity. Banking shocks account for about 2%-5 % of the unconditional variance of US GDP and for 3%-14 % of the variance of EA GDP. During the 2007-09 recession, banking shocks accounted for about 15 % of the fall in US and EA GDP, and for more than a third of the fall in EA investment and employment.

This paper shows empirical evidence and theory consistent with the US government using debt to adjust public consumption to news about long-term growth. First, using historical forecasts from the Congres-sional Budget Office (CBO) since 1984, I find that government pur-chases and deficits are positively correlated with expectations about long-term tax revenue and GDP growth. I also document that these facts are robust to controlling for current growth and to using à-la-Kalman estimated forecast values for a longer time span. Second, I present a simple open economy RBC model with stochastic produc-tivity trend and endogenous public purchases decided by a forward-looking government. Calibrated to the US economy, the model pro-duces moments similar to those observed in the data. Finally, the role of imperfect information is found negligible to explain these facts. JEL codes: H63, H68, E21 I am grateful to my adviser Robert Kollmann and to Université Libre de Bruxelles (ULB) for granting mini-Arc funding to my research project. I am also grateful to Antonio Moreno-Ibañez, Ugo Panizza and Roberto Pancrazi for useful comments. I thank seminar attendants at ECARES-ULB, U. de Navarra (UN), U. Pública de Navarra (UPNA) and U. Autónoma de Barcelona (UAB), as well as participants of the 2014 T2M and U. of