Abstract
A common trends model for grow national income, private consumption, government consumption and net taxes is estimated on US data. The system has two cointegrating vectors and thus two common stochastic trends, interpreted as a technology trend and a public sector trend. The two temporary shocks are interpreted as a private demand and government financing shock, respectively. Theoretical models suggest that the two cointegrating vectors could be due to the private and public sectors' intertemporal budget constraints. We find two co-integrating vectors, as predicted by no-Ponzi game constraints on the sectors. However, a stronger version of the no-Ponzi game constraint is a solvency condition, which implies particular co-integrating vectors. These cointegration vectors are both rejected for the sample period, indicating that the public sector will not be able to repay its debt if the current policy is maintained. However, the private sector is at the same time accumulating wealth, which is consistent with predictions from a Ricardian model. Further, the equivalence theorem pridicts that private consumption should be unaffected by financing shocks. Data, however, indicate that there is a significant short run effect on both income and private consumption from the financing shock, but the effect indicates that increasing taxes is accompanied by increasing private consumption, contrary to both standard Ricardian and Keynesian models. In the theoretical world, this type of pattern could be generated in models with risk averse individuals and uncertainty about future taxes