Prof. Bryan Caplan
Week 10: Endogenous Policy Theory, II: Public Choiceand Macro Policy
I. Public Choice and Macro Policy
A. The traditional modeling assumption: market agentsmaximize their own utility functions; political agents maximize a socialwelfare function.
B. Implication: Political processes can fail due tolack of ability, but not lack of desire.
C. The Public Choice approach: model market andpolitical agents symmetrically as own-U fn maximizers. Don't assume that failed policies arewell-intentioned; rather assume that what appear to be policy"failures" are a success given policy-makers' U fns.
II. Seigniorage, I: Microanalytics
A. Traditional approach: outside forces somehow makemoney growth high. Or high money growthis necessary for high employment. Orhigh rate of price growth forces policy-makers to accommodate with high rate ofmoney growth.
B. But note: Governments can raise revenue just byprinting up new base money. This iscalled “seigniorage” or “the inflation tax.”
C. Public Choice theory of high inflation: money growthis high because the government wants it to be high in order to raiserevenue. (Rothbard quote,pp.177-178). Moreover, there is noreason to think that the revenue raised by money creation is allocated tomaximize a SWF; far more likely that revenue is used in self-interest ofpolicy-makers.
D. This suggests a very simple model of money supplydetermination: Governments maximize seigniorage; they pick the moneysupply growth rate to get at the peak of the Laffer curve for inflation taxrevenue.
E. While this is a clean and simple theory, it is hardto see its application in modern, industrial countries - though it explains alot about LDC’s and many historical instances.
D. A model of seigniorage (from Mankiw 1987):
1. Government has to satisfy a present-value budgetconstraint (which is equivalent to ruling out explosive debt).
2. Expenditure is exogenous (so government is not necessarilyat the maximum of the inflation tax Laffer curve).
3. Money demand is just a linear function of output(note: this assumes away the most obvious deadweight cost of inflation stemmingfrom the increase in nominal interest rates).
4. Government has two revenue sources: seigniorage, andan output tax. Both taxes havedeadweight losses; the government picks the mix of taxes.
5. Government picks the tax mix that minimizes sum ofdeadweight costs, s.t. the lifetime balanced-budget constraint. (Note: anon-altruistic government might want to do this. Why?)
E. Implications of the model: government uses bothtaxes; use of both taxes increases as expenditures increase; deadweight burdensof taxation are equalized at the margins. This includes the intertemporal margin - so a government following thismodel will smooth taxes over time.
F. Empirical results work: nominal interest rates andinflation positively related to T/Y.
1. Other explanations of the data? Would a generic "big government"theory explain the same facts?
III. Seigniorage, II: Hyperinflation and Expectations
A. Note: A 1-shot increase in the money supply collectsseigniorage without increasing the inflation tax. It's like a lump-sum tax.
B. The lower people expect the inflation tax to be, thegreater the value of real balances.
C. Thus, the lower people expect the inflation tax tobe, the greater the pay-off to the government of a 1-shot increase in the moneysupply.
D. Problem: The more you try these "1-shot"increases, the more people are likely to expect similar increases in thefuture.
E. Taking advantage of short-run seigniorage Laffercurve may leave you to the right of the maximum of the long-run Laffercurve. This seems to be the case inmany hyper-inflations - eventually, real revenue declines even as the tax rategoes up.
F. Hyperinflation greatly reduces real balancesheld. Disinflation leads people toreturn to normal levels of real balance holdings. This is just the flip side of the "discrete jump" inthe price level that accompanies a change in the rate of monetary growth.
G. Irony: A hyperinflating government intent ondisinflating can take advantage of what would otherwise be a discrete fall inthe price level with one last 1-shot inflation.
IV. Seigniorage, III: The Unpleasant MonetaristArithmetic
A. Models like King and Plosser emphasize theendogeneity of the “money-multiplier.” A very different perspective is to view the monetary base as itselfendogenous - though in a very different sense. Not endogenous from the point of view of the government as a whole, butendogenous from the point of view of a single agency - namely, the monetaryauthority.
B. Sargent and Wallace’s “unpleasant monetaristarithmetic”: if you hold future expenditure and taxation decisions constant,AND ruleout explosive debt growth, then only a narrow set of monetarypolicies are feasible (from that one agency’s perspective) because the centralbank prints money to make up the difference.
C. Suppose the Fed chooses a fixed money growth rule -then its choices are even more tightly constrained.
D. If there is significant sensitivity of money demandto the nominal interest rate (and thereby to expected future inflation), thenS-W point out that it may not even be possible to temporarily reduce inflationwith monetary policy!
V. The Central Bank as Bureaucracy
A. PC view of bureaucracies as constrainedbudget-maximizers can be extended to central banks. Main variations on this theme:
B. Fed favors discretion over rules (or alternativearrangements like free banking) to maintain its power and budget. As White puts it, with Friedman's monetaryrule, the Fed staff could virtually be reduced to one person.
C. Similarly, economic crises may not be bad for theFed. Another way to increase power andbudget.
D. The Fed as a scapegoat/lightning rod to protectCongress from public dissatisfaction.
E. Note: the Fed is self-funding. It just keeps a fraction of seigniorage,giving the remainder to the Treasury. If it gets to keep a constant fraction, the Fed has a strong incentiveto increase total seigniorage and thereby increase its budget.
F. Two empirical studies confirm this:
1. Toma finds positive correlation between annualseigniorage and annual budget.
2. Shugart and Tollison find that Fed employment"Granger-causes" changes in the monetary base.
VI. Political Business Cycles
A. Traditional approach: If monetary policy can affectoutput and employment in the short-run, then policy-makers will use this powerto smooth fluctuations and compensate for shocks unanticipated by the privatesector.
B. Public Choice approach: Just because money mattersdoes not mean that it will be used beneficially. Rather it will be used to help keep incumbent in power by raisingmoney supply growth to stimulate the economy during election year. The recession generated by the subsequentmonetary tightening is no accident, but a deliberate choice to impose short-runlosses when they are least likely to be remembered by the electorate.
C. Original model: Nordhaus.
1. Assume expectations-augmented Phillips curve withadaptive expectations.
2. Myopic voting based on election year's economicconditions.
3. Vote-maximizing politicians.
4. Incumbent controls policies.
5. Implies: Term-length cycle manipulated by incumbentto have low unemployment and low inflation during next election year.
6. Thus: expansionary policy during election year;contractionary right after to get ready for next election.
D. McCallum's complaint: Violates RE. Mixed empirical evidence.
E. Alesina and Sachs alternative PBC.
1. Partisan model: Dems care more about unemploymentand less about inflation that Reps.
2. Voters have RE, but election generates uncertainty.
3. Whoever wins was expected to win with less thanprobability 1.
4. Therefore, if Dems win, there is an unexpectedexpansionary shock; if Reps win, thereis an unexpected contractionary shock.
5. Seems to work empirically: Dems have prosperity infirst two years; Reps have recession. In years 3-4, same unemployment, but lower inflation if Reps are inoffice.
F. PBC debate helps illustrate the difference betweenPC with underlying assumption of voter irrationality with later developments.
A. Why isn't seigniorage at its maximum in allcountries?
B. Which if any of these theories work best?
C. Is there any way to combine time consistency and thePC approach?