Prof. Bryan Caplan

bcaplan@gmu.edu

http://www3.gmu.edu/departments/economics/bcaplan

Econ 816

Spring, 2000

 

Weeks 6-7: Monetary vs. Fiscal Policy with Nominal Rigidities

I.                     Nominal vs. Real Rigidities; Nominal Price Vs. Nominal Wage Rigidity

A.                 Many NK models have real rigidities like efficiency wages.  Whatever their validity, they do little to explain fluctuations; all of the problems with RBC also apply to NK models with real rigidities.

B.                 There remain NK models with nominal rigidities.  Both kinds of rigidities make it possible for markets to fail to clear, but nominal rigidities respond quite differently to policy.

C.                Two general classes of nominal rigidities:

1.                  Price (Blinder + others)

2.                  Wage (Akerlof, Dickens, and Perry)

D.                Most explanations apply isomorphically to both price and wage rigidities.  E.g. the Mankiw menu costs model.

E.                 Wage rigidity is both more and less plausible in some views:

1.                  Hall lifetime employment critique of apparent nominal wage rigidity applies much less forcefully to prices.

2.                  ADP psychological view applies much more clearly to wages than to prices.

F.                 RE Models with nominal rigidities are quite possible; main problems:

1.                  Key implication of menu cost model: Big nominal shock will matter less than small shock!  A sufficiently big shock will have no real effect.

2.                  Problem with contracting models: Why aren't contracts made contingent on economic conditions?

3.                  In general, problem can only last until next adjustment period.

II.                   The ADP Model and Its Interpretation

A.                 ADP model just drops RE and appeals to large body of evidence that workers do react worse to nominal cuts than real cuts.

B.                 Does this make sense in terms of my rational irrationality model?  A case could be made:

1.                  Workers with irrational views about the nominal/real distinction are disgruntled, but it is hard for an employer to tell how rational an individual's view is.

2.                  Disgruntled workers are less productive, but lower productivity hard to attribute to any one individual.

3.                  As long as neither one's views nor behavior "give you away," the marginal private cost of sharing popular confusions is low or zero.

C.                Sets up model with monopolistic competition and heterogeneity of industries.

D.                Firm demand equation: .  D is the demand faced by one firm, M is the money supply, p-bar is the general price level, p is the price set by the firm, b is demand elasticity, n is the number of firms.

E.                 Firm output proportional to labor: Q=L.  Total labor force normalized to 1, u is unemployment, so Q=1-u.

F.                 Bargaining equation: wage is weighted average of firm and worker surplus:   where omega-n is "bargained real notional wage per efficiency unit", w-bar is the average nominal wage, a is the index of worker bargaining power, s is the value of unemployed time, and f is the ratio of fixed costs to output value at full employment.  Real wage is competitive when a=0.

G.                This implies that price will be a constant markup over nominal unit labor cost: .  (Implies acyclical wages and flat AD curve).

H.                 From (3), one can derive the equilibrium AS curve by plugging in the equilibrium values: , so  which will be upward-sloping so long as the value of leisure exceeds the real wage.

I.                     Intersection of ADP AD/AS yields the LSRU ("lowest sustainable rate of unemployment") which is only identical to the NAIRU without nominal rigidity.

J.                  Now add heterogeneity and rigidity. 

1.                  Add a random term to each firm's demand, and a random term to the bargaining equation.  (Also build in autoregressive elements). 

2.                  Set up the simulation so that nominal wage cuts can only occur in extreme circumstances: firms with 2 years of consecutive losses can cut wages.  Firms can get 1% cut per year via sneaky compensation cuts.

K.                 Results (LSRU is 5.8%):

Inflation

Unemployment

% Constrained

3%

5.9%

5%

2%

6.1%

10%

1%

6.5%

19%

0%

7.6%

33%

-1%

10.0%

53%

L.                  ADP's out-of-sample predictions for the Great Depression.

M.                Can nominal rigidity be eliminated by forcefully changing expectations? 

N.                 Objections:

1.                  Isn't this really a theory of low nominal GDP growth regimes rather than low inflation regimes?

2.                  Historical counter-example #1: contraction of 1920-21

3.                  Historical counter-example #2: Mild deflation of the 1920's combined with low unemployment and high growth.

4.                  Modern computer industry.

O.                Overall assessment: ADP probably provides the most intuitive case for nominal rigidities, and avoids the strange implications of other theories (e.g. a big nominal shock will have no real effect).  Quantitative claims are much weaker: Plenty of cases of zero inflation or mild deflation coupled with low unemployment, as commentators point out.

III.                  An Abbreviated Monetary History, 1914-1941

A.                 Friedman and Schwartz have probably made the most influential case for the view that nominal factors have real effects.  Now, a brief review of their evidence...

B.                 The Fed coincidentally created just before WWI.  Monetary policy is extremely expansionary during the war years as both nominal and real output rise.  Suddenly reduction in monetary growth - and actual decline in the money stock (-14% growth in 1921!)- coincides with the sharp recession of 1920-1.

C.                From 1921-1929, the stock of high-powered money is approximately constant, although the broader money supply expands moderately, with on average mild deflation.

D.                The Great Contraction, 1929-1933, and the aftermath, 1933-1941.  These will be explored in more detail towards the end of the semester.  Main points of interest:

1.                  The monetary base falls first, but not by that much.

2.                  Almost all of the decline in the money supply ultimately resulted from a fall in the money multiplier, not the monetary base.

3.                  Nevertheless, the monetary base fell first, and initially it fell by more in percent terms than M2.  (I.e., the money multiplier initially increased a little!)

4.                  Most of the monetary contraction did not happen for over two years.

5.                  The money supply growth rate (both base and M2) was high during the first 4 years of Roosevelt's administration, but M2 growth turned negative for a second stretch when reserve requirements were increased in 1937.

Year

Base

% ch

M2

% ch

7/1929

7.123

-

55,416

-

7/1930

6.925

-2.8

54,645

-1.4

7/1931

7.321

5.7

52,581

-3.8

7/1932

7.858

7.3

44,833

-14.7

7/1933

7.891

.4

40,897

-8.8

7/1934

9.422

19.4

44,411

8.6

7/1935

10.707

13.6

49,279

11.0

7/1936

12.322

15.1

54,814

11.2

7/1937

13.385

8.6

56,673

3.4

7/1938

14.778

10.4

55,756

-1.6

E.                 The simplified monetarist story:

1.                  The Fed contracted the monetary base during the first year of the Great Depression, getting the downturn going.  (Dinosaur Keynesian alternative: "autonomous decline in consumption.")

2.                  Once the downturn began, the Fed neglected the function that clearinghouses performed in the pre-Fed days of bailing out temporarily illiquid banks.

3.                  Bank failures multiplied as the downturn continued: partly due to the Fed's malign neglect, partly due to suspicion that the U.S. would abandon the gold standard.

4.                  After the abandonment of the gold standard, monetary policy was expansionary for four years - and output and employment began to rapidly recover from an extremely deep trough.

5.                  The Fed created a second downturn by raising reserve requirements in 1937.  (Dinosaur Keynesian alternative: Roosevelt tried to balance the budget by raising taxes).

F.                 Similar findings over longer time horizon by Boschen-Mills: exogenous monetary tightenings precede recessions.

IV.               The Liquidity Effect: Theory and Evidence

A.                 In real model, one-time monetary expansions have no impact on nominal interest rates; changes in the rate of monetary growth raise nominal rates on a 1:1 basis.

B.                 Most casual observers and practitioners say they see the opposite: Monetary expansions temporarily depress nominal rates.

C.                This is easy to understand with nominal rigidities, and hard to understand from any real perspective.  With nominal rigidities, price level does not instantly rise when nominal money rises, so the market clears along a different margin by reducing the nominal rate until people are again satisfied with their real balance holdings.

1.                  If the short-run effect of monetary expansions is to reduce nominal rates (liquidity effect), but the long-run effect is to raise them (Fisher effect), there is an important empirical implication: High nominal rates are consistent with both very expansionary AND very contractionary monetary policy.  This generates considerable confusion about the direction of monetary policy in inflationary AND deflationary environments.

D.                Empirical studies of the liquidity effect have been more problematic.  Distinguishing cause, effect, and correlation is difficult, and your answer often depends on your procedure for solving the problem.  Policies are frequently endogenous responses to circumstances.

E.                 The narrative evidence (e.g. Boschen-Mills, Friedman and Schwartz) supports the liquidity effect. 

F.                 Other studies (e.g. Bernanke and Blinder [1991]) definitionally treat changes in the federal funds rate as monetary policy shocks, so they have a liquidity effect by definition.

G.                Formal econometric tests frequently fail to detect a liquidity effect unless they assume it.  Leeper and Gordon (1992) is a good example.  They find that if you control for past interest rates, money growth, prices, and output, then the impact of money on interest rates is positive. (?!)

H.                 Possible conclusions:

1.                  Econometric tests have more trouble coping with policy endogeneity than the narrative studies do.

2.                  There is no liquidity effect; a key NK "stylized fact" is not a fact at all.

V.                 Monetary Policy in Models with Nominal Rigidities

A.                 Diverse sources confirm the real effects of monetary policy, including both narrative studies (Boschen and Mills, Friedman and Schwartz), and a wealth of econometric studies (Bernanke and Blinder is just one example).  Disagreement about the best measure of money (interest rates, various monetary aggregates), but these findings are hard to avoid.

1.                  Main problem with econometric studies: Price puzzle.  Again, probably due to endogeneity.

B.                 If monetary policy has real effects given nominal rigidities, what is the mechanism?  I.e., assuming that more spending implies more output, how does more money induce more spending?

1.                  Note: Movement of V could dampen or amplify impact of M on nominal GDP.

C.                The simplest theory (more money-->people feel richer-->they spend more) doesn't work for modern economies because money is such a small fraction of wealth.

D.                The standard ISLM mechanism. 

1.                  More expansionary monetary policy's short-run effect given nominal rigidities is to reduce real rates.  Nominal rates definitely fall, and if expected inflation rises, this means real rates fall more than nominal rates.

2.                  Investment, consumer durables, and other interest-sensitive spending increases as a result.

3.                  Again due to nominal rigidities, the aggregate result of greater spending is increased output.

E.                 Main problem: Empirical evidence for interest rate channel is weak.  Interest-sensitive components are not big enough or sensitive enough.

F.                 Weaknesses with the ISLM story have generated many other possible channels:

1.                  Exchange rates

2.                  Equity markets (q theory; simple wealth effects)

3.                  Credit channels (bank lending; balance sheet and credit market imperfections; note Fisher's debt-deflation theory of the GD)

G.                Main strength of credit channels: With credit rationing, borrowers' interest rates don't have to fall; looser monetary policy may lead to less stringent rationing rather than lower rates.  Explains some anomalies...

VI.               Fiscal Policy in Models with Nominal Rigidities

A.                 Empirical confirmation of the link between fiscal policy and real output is much sparser.  Many take WWII as conclusive proof, but monetary policy was expansionary at the same time, so what can be concluded?

B.                 Again assuming that more spending implies more output, how does more expansionary fiscal policy induce more spending?

1.                  One might be a priori skeptical: All spending has to be financed either with taxes (which reduce spending) or borrowing (which reduces lending to the private sector, which reduces spending).  Note you don't need Ricardian equivalence for this argument.

C.                This a priori argument is however not strictly correct, as the standard ISLM channel for fiscal policy shows.  Looser fiscal policy induces more total spending by raising nominal interest rates; as nominal rates rise, desired money balances fall allowing for more total spending.

D.                Empirical research on this channel has grown quite thin.  Intuitively, it seems like the LM curve would be vertical or nearly vertical, although various money demand studies found more substantial elasticities.  (Remember that most components of the broader aggregates pay interest too, so at least in the long-run what matters is the interest-sensitivity of the monetary base).

E.                 Minimal effort to map out alternative nominal channels for fiscal policy.  Ideas?

VII.              Macro Policy and the Quantity Theory of Money

A.                 The quantity theory of money (QTM) as a theory of money demand: money demand depends solely on income/wealth, not interest rates.  Implies vertical LM.

B.                 QTM as a theory of velocity: People want their money holdings to be simply a constant fraction of their income.  In other words, money demand depends solely on income/wealth, not interest rates, implying a vertical LM.

C.                With vertical LM/constant velocity, note: Fiscal policy doesn't matter even for nominal output!  This seems counter-intuitive to many, but it isn't really; note the previous a priori argument that more expansionary fiscal policy won't change total spending.

D.                Caplan (1999) tests the quantity theory of money, using wars as exogenous policy shifters in order to distinguish cause and effect.  Assembles two pooled time series, dubbed the "narrow" (15 countries, 100 years) and the "broad" sets (about 60 countries, 50 years.

E.                 Caveat: It turns out that it is critical to distinguish foreign wars from domestic wars.  Actually helpful, providing more instruments.

F.                 Econometric problem: Explosive results.  I get around this by assuming that policy works directly only on nominal output, and indirectly on real output.  (Reversing this assumption for fiscal policy to allow RBC alternative doesn't matter).

G.                Results: QTM of money works!  Fiscal policy doesn't move nominal GDP even ignoring all of the practical political and timing difficulties.  Monetary policy affects both nominal and real GDP.