Let us temporarily make the simplifying assumption that the marginal propensity to spend out of income is unity. Although, as will be shown below, this assumption is not necessary for the quantity theory, it is a classic quantity theory case. Armed with this assumption, consider a case where the government has the same income velocity as the private economy. In this case the balanced budget multiplier is zero: the government is merely substituting itself for private firms or households in the income-expenditure chain. On the other hand, assume that the government's marginal Marshallian k is zero, i.e., that the government holds no additional cash balances when tax receipts and expenditures rise by the same amount. In this case, we have the classical balanced budget multiplier of unity. This is because the government's expenditure raises income by an equal amount without reducing private expenditures at all. Third, the government's k may be greater than zero, but less than that of the private economy. In this case the balanced budget multiplier is greater than zero, but less than unity. This is the case recently considered by Selden.1 Finally, the government's k may be greater than the private k, and if so, the balanced budget multiplier is negative.2 How do these quantity theory balanced budget multipliers look from the viewpoint of Keynesian theory? It turns out that Keynesian theory is not able to handle these cases, for if the marginal propensity to consume (the Keynesian analogue of the marginal propensity to spend) is unity, there is no equilibrium income level to be computed by multiplier theory. To apply this specifically to the balanced budget theorem, consider what happens to both of its proofs if the marginal propensity to consume is unity. The first proof, which is to compare the tax and expenditure chains, then looks as follows:

This paper criticizes Temin's hypothesis that the Great Depression was caused by an exogenous decline in consumption in 1930. Using Temin's own consumption function, as well as two other ones on the levels of the data, there is little support for Temin's hypothesis. First difference regressions support Temin's hypothesis if one uses a dummy variable for 1930. But if one looks instead at the residuals from the regressions, then the data provide only very limited support for it.

I. Introduction, 454. — II. Minimizing the number of inequities, 458. — III. Effect on total revenue of giving favorable or unfavorable treatment to new taxpayers, 459.

The question of whether large banks should be allowed to fail brings us face to face with a conflict between two social goals. On the one hand, the goal of optimal resource allocation suggests that even very large banks, like other firms, should be allowed to fail. On the other hand, the stabilization goal suggests that, given the present institutional structure, failures of large banks should be prevented lest they lead to runs on other banks and to a significant reduction in the money stock. The solution suggested here for this conflict is small changes in the institutional structure.

Of the twelve characteristics of modern monetarism, five are explicit in Hume's writings: the quantity theory, the Chicago transmission process, private sector stability, the vertical Phillips curve, which Hume originated, and preference for free markets. Two others, irrelevance of allocative detail and focus on the price level as a unit, are implicit. Preference for reduced-form models fits Hume's theory of causation. Preference for stable money growth fits the whole tenor of Hume's discussion. Two propositions on targets and indicators were irrelevant in Hume's day, but Hume rejected the monetarists' strong opposition to inflation.

I. Introduction, 275. — II. Consumption, 280. — III. Investment, 282. — IV. Removing the assumptions, 284. — V. Conclusion, 289.