Ricardian equivalence, rational expectations, and the permanent income hypothesis.

Ricardian Equivalence, Rational Expectations, and the Permanent Income Hypothesis

IN CONVENTIONAL KEYNESIAN MACROECONOMIC THEORIES, government bonds were commonly treated as net wealth to the private sector. (1) Assuming that government spending is constant, an increase in government

debt would therefore increase consumption. Barro (1974) argues instead that the way government spending is financed has no effect on private sector behavior. According to Barro's Ricardian equivalence proposition, a tax cut today implies increased taxes in the future, as long as the stream of government spending is not reduced. The household sector treats the bonds issued to finance such a tax cut as equivalent to the present value of the implied increase in future tax liabilities. A tax cut does not lead to an increase in consumption or interest rates in the Ricardian case. Instead, the tax cut increases savings in order to pay for the required increase in future taxes. Barro demonstrates that Richardian equivalence holds for a maximizing household in an overlapping generations model that assumes an operative bequest motive, nondistortionary taxes, and perfect capital markets. Barro (1989) surveys Ricardian equivalence.

In theory, wealth effects of government bonds are possible under modified assumptions of the Barro model. Chan (1983), Blanchard (1985), Barsky, Mankiw, and Zeldes (1986), and Feldstein (1988) show how Ricardian equivalence might not hold. However, Barro (1974, 1989) argues that negative as well as positive wealth effects may occur. Judd (1987) demonstrates this in a theoretical model with distortionary taxes, finite lives, and adjustment costs. Wealth effects could cancel each other or be altogether negligible. The issue cannot be settled on theoretical grounds. An answer to whether Ricardian equivalence is a good approximation to reality has to come from empirical studies.

The empirical tests of Ricardian equivalence have been carried out in basically two ways. The first way is to determine whether larger deficits are associated with higher interest rates. (2) The second way is to assess whether increases in government bonds lead to increases in household wealth and consequently in consumer spending. In my paper, I will choose the second approach. Seater (1985), Aschauer (1988), and Barro (1989) present surveys of empirical consumption studies. The results of these studies are mixed with respect to Ricardian equivalence. Results are sensitive to the sample period and variables included, and the tests may have low power. (3) Typically, three major problems can be identified in most empirical specifications of Ricardian equivalence tests in a consumption model. First, theoretical equations, which are cast in terms of expected future values, are often approximated in the empirical equations by a distributed lag on realized past values. (4) Second, Evans (1988) points out that the consumption function is ordinarily not derived from a model that nests the Ricardian equivalence hypothesis and an alternative of Ricardian nonequivalence. (5) Third, it is not usually established whether the underlying life cycle or permament income model is supported by the data. Ricardian equivalence has been extensively tested within consumption models, however. (6)