Prof. Bryan Caplan
Econ 816
http://www3.gmu.edu/departments/economics/bcaplan
Spring, 2000
I. Friedman's critique of "Dinosaur Keynesianism" (Friedman 1968)
A. The primary determinant of nominal GDP is monetary policy. Naive fiscalism is wrong, but so are sophisticated hybrid fiscal/monetary theories. In ISLM terms: vertical LM curve.
1. Alternate version: Fiscal policy is too difficult to use for short-term stabilization in the U.S. political environment.
B. Changes in nominal GDP affect real GDP if they differ from what agents "expected." Expectations and/or their consequences are fixed in the short-run but adjust to reality in the long-run.
1. Friedman talks about inflationary expectations, but in principle people could form expectations about all of the derivatives of the money supply (note that the textbook Keynesian case begins with price levels, but then shifts to inflation).
C. These two observations imply the expectations-augmented Phillips curve. In the short-run, monetary policy can be used to pick any desired point on the short-run Phillips curve. But in the long-run, the position of the Phillips curve depends on people's expectations, and therefore tends to shift back to the "natural rate" of unemployment.
D. Policy implication: Create a stable monetary environment, and handle "structural" inefficiencies in the labor market with deregulation.
II. Friedman's Descendants, I: Real Business Cycle theory (Long and Plosser 1983; Mankiw 1989)
A. Lucas famously formalized Friedman's arguments with his "island" model. Expectations formalized as "rational expectations": people make mistakes, but these mistakes have a mean value of zero. Implication: Only unexpected monetary policy has real effects.
B. Problem: The "island" model assumes that agents can cheaply and quickly get local information, but access global information with more difficulty and/or delay. But isn't most "global" information published frequently in newspapers available for a quarter? Especially clear if the relevant global information concerns monetary shocks - official figures for the U.S. published every 2 weeks!
1. Related observation: Perhaps nominal wages don't fluctuate much because they are just installment payments for a long-term contract. (Hall) Labor contracts could be conditioned on news if it were worth it.
C. The resolution: Drop nominal "misperceptions" explanations altogether, along with all discussion of involuntary unemployment. Explain behavior of the real economy using nothing but real variables.
D. More precisely: Treat the Solow model with perfect competition as a theory not merely of growth but of fluctuations too. Fluctuations caused by real shocks, especially shocks to multi-factor productivity.
E. Money only matters for the price level, not for employment, output, or other real variables. Apparent impact of money now has to be re-explained somehow.
F. Real effects of fiscal policy through nominal channels are ruled out. But re-thinking the effects of policy shows that fiscal policy could work through real channels.
1. Most obvious case: Labor taxation.
2. Less obvious: Intertemporal elasticities.
G. In any case, economy without intervention is presumptively Pareto optimal. Non-optimalities can be addressed in standard Pigovian microeconomic terms.
1. Application: Lucas' (1978) analysis of unemployment insurance.
III. Friedman's Descendants, II: New Keynesianism (Romer 1993)
A. To rebut RE "policy irrelevance" arguments, NKs develop a large number of nominal and real rigidity models.
B. Monetary policy matters in the short-run in nominal rigidity models, but not in real rigidity models.
1. Friedman's argument is strengthened by the "time consistency" literature, arguing that the more effective monetary policy is holding expectations constant, the worse the actual equilibrium will be.
C. Role of fiscal policy, if any, in NK models is unclear: Sometimes they borrow from RBC models, sometimes they treat it as a nominal shock, but most of the time it isn't discussed.
D. Other less standard interventions might matter in real rigidity models. E.g. industrial policy to compensate for suboptimal efficiency wage equilibria.
E. Traditional macro policies limited to Friedman's goal of creating a stable macro environment, albeit with more optimism about discretionary policy, and more pessimism about the optimality of the resulting equilibrium.
F. Overall: NKs virtually wind up endorsing Friedman's 1968 analysis in its entirety.
1. Almost exclusive focus on monetary rather than fiscal policy.
2. Adoption of the expectations-augmented Phillips curve and related tools like the natural rate of unemployment.
3. Recognition that monetary policy can at most create a stable macro environment, not solve structural problems.
G. Main differences between NKs and Friedman
1. Friedman tells one story of imperfect labor markets. NKs add many more.
2. Although NKs are more likely to see monetary policy as compensating for market instability, rather than itself being a source of stability (at least in the post-Depression era).
IV. Typologies
A. Typology #1: Classical/Keynesian
1. Standard RBC models are often described as "Classical" or "New Classical" because monetary policy has no real effects (as in Hume's famous thought experiment), all unemployment is voluntary, and laissez-faire is optimal under standard assumptions.
2. New Keynesian models - whether they appeal to nominal or real rigidities - are often described as "Keynesian" because there is involuntary unemployment and government intervention could conceivably be welfare-enhancing.
B. Typology #2: Real/Nominal
1. Alternately, one might group both standard RBC models and NK models with (nothing but) real rigidities into the class of "Real models." In both kinds of models, monetary policy is irrelevant for real variables and all prices and wages are perfectly flexible. Especially as more sophisticated RBC models incorporate various real imperfections (imperfect competition, etc.), the distinction between RBC models and NK models with real rigidities becomes increasingly blurry.
2. NK models with nominal rigidities would in contrast be put into the class of "Nominal models." In all of these models, money matters because of nominal price and/or wage rigidities.
3. Note: In this typology, each category can support the full spectrum of policy positions.
V. Policy Questions
A. What does monetary policy do?
1. Real view: Affects the price level, inflation, and the nominal interest rate.
2. Nominal view: In addition, has short-run impact on employment, output, and real interest rates.
B. What does fiscal policy do?
1. Real view: Government spending affects real interest rates, output, employment, with indirect effects on the price level, inflation, and the nominal interest rate. Taxation has the standard microeconomic impact.
2. Nominal view: Fiscal policy's has the same limitations as monetary policy; it impacts output, employment, nominal interest rates, real interest rates, the price level, and other variables as the ISLM model predicts. The nominal view typically ignores the real channels of fiscal policy, seeing little evidence for inter-temporal labor supply or a significant labor supply response to income taxation.
VI. Underlying Questions
A. Why does output fluctuate around its trend?
1. Real view: Multi-factor productivity ("technology") shocks, other supply shocks (oil crisis), plus fiscal policy shocks. Taste shocks also theoretically possible, although rarely considered.
2. Nominal view: In addition to real shocks, output also responds to unexpected money demand shocks and money supply shocks.
B. Why does employment fluctuate?
1. Both views acknowledge the role of demographics.
2. Real view:
a) Intertemporal labor supply - i.e., movements into the labor force when real wages are exceptionally high, and movements out of the labor force when real wages are exceptionally low. A positive shock to multi-factor productivity increases the marginal productivity of labor and therefore real wages.
b) Sectoral shifts. When the desired composition of output changes, people have to move between sectors. This is measured as "unemployment" even though it is a productive act.
c) Changes in the degree of real rigidities in labor markets (unionization, real levels of unemployment insurance, hiring/firing regulations, etc.)
d) Note Lucas' caveats on the "voluntariness" of unemployment.
3. Nominal view: Unexpected nominal shocks.
C. What determines nominal and real interest rates?
1. Real view: Real interest rates are determined by the supply and demand for present vs. future goods; the nominal rate is just the real rate plus the expected inflation rate (the Fisher effect).
2. Nominal view: Unexpectedly expansionary monetary policy has a short-run tendency to decrease nominal and real interest rates. There may also be nominal interest rate rigidities (or perhaps just rigidities in inflationary expectations) such that the Fisher effect kicks in only with a long lag.
D. What determines the price level and the inflation rate?
1. Real and Nominal views don't disagree much on these issues: It's just money supply and money demand. But the Nominal view sees the process as slower and less even due to short-term nominal rigidities.
VII. Applications
A. The Great Depression
1. This is the standard rebuttal to Real theories in general, and standard RBC models in particular.
2. How could RBC explain it? Technology shocks combined with intertemporal labor supply seems absurd.
3. NK models with real rigidities can have mass involuntary unemployment, but it is not clear why the magnitude of involuntary unemployment would have increased so much during the downturn. (Relief programs appear after the Depression is already severe...)
4. NK models with nominal rigidities can readily appeal to Friedman's explanation for the downturn.
5. Duration of the GD is a problem for all theories, and is perhaps what made Dinosaur Keynesianism plausible to begin with. But as many economists interested in this era are beginning to realize, during the GD there was a significant world-wide move to labor market regulation, pro-union laws, etc. (Unclear if these create nominal or real rigidities, though).
B. Oil shocks
1. This is probably the best example the RBC literature has put forward, and the timing seems right during the 1970's.
2. Showed that exclusive focus on nominal shocks is misguided.
3. But it is still problematic:
a) Were workers really voluntarily quitting because their real wages had fallen?
b) Why aren't labor and energy substitutes? The oil shocks would only decrease labor demand if labor and energy were complements.
c) Shock seems too small to be a full explanation, and the subsequent decline of oil prices makes the puzzle harder. (Olson 1988)
VIII. Efficient Policy and Efficient Institutions (Lucas 1978; Summers 1986; Carmichael 1985; Shapiro and Stiglitz 1985)
A. Two kinds of efficiency debates:
1. Debates about what policies it would be efficient for current institutions to implement.
a) Example #1: Inflation targeting.
b) Example #2: Constant money growth.
2. Debates about what institutions would have the best overall performance.
a) Example #1: Optimal contracts for central bankers.
b) Example #2: Free banking
B. NK is much closer to Friedman's positive views than RBC, but RBC is closer to Friedman's normative views. Still, both frameworks often used to support similar conclusions.
C. Underlying disagreement is often theoretical: Is inefficiency possible? The debate over the possibility of "involuntary" unemployment is an excellent illustration.
1. RBC: "It is unlikely that the unemployed and employers ignore unexploited gains to trade."
2. NK: "It is unlikely that most unemployed people prefer idleness to the market wage of jobs they are qualified for."
D. Such debates only determine whether intervention could conceivable be efficiency-enhancing. Usual Public Choice and knowledge problems still have to be addressed.
1. The relevance of the historical record (Sheffrin 1988).