Авторы: 147 А Б В Г Д Е З И Й К Л М Н О П Р С Т У Ф Х Ц Ч Ш Щ Э Ю Я

Книги:  180 А Б В Г Д Е З И Й К Л М Н О П Р С Т У Ф Х Ц Ч Ш Щ Э Ю Я


5.5.1 The policy ineffectiveness proposition

The new classical policy ineffectiveness proposition was first presented in

two influential papers by Sargent and Wallace (1975, 1976). The proposition

can best be illustrated using the aggregate demand/supply model shown in

Figure 5.3. Those readers unfamiliar with the derivation of this model should

refer to any standard macroeconomics text, such as Mankiw (2003). In Figure

5.3, the economy is initially operating at point A, the triple intersection of

AD0, SRAS0 and LRAS. At point A, in line with equation (5.3), the price level

(P0) is fully anticipated (that is, the actual and expected price levels coincide)

and output and employment are at their long-run (full information) equilibrium

(natural) levels. Suppose the authorities announce that they intend to

increase the money supply. Rational economic agents would take this information

into account in forming their expectations and fully anticipate the

effects of the increase in the money supply on the general price level, so that

Figure 5.3 The effects of anticipated and unanticipated changes in the

money supply on the level of output and the price level

output and employment would remain unchanged at their natural levels. The

rightward shift of the aggregate demand curve from AD0 to AD1 would be

offset by an upward shift to the left of the positively sloped aggregate supply

curve from SRAS0 to SRAS1, as money wages were increased following an

immediate upward revision of price expectations. In this case the economy

would move straight from point A to C, remaining on the vertical long-run

aggregate supply curve with no change in output and employment even in the

short run; that is, money is super-neutral.

In contrast, suppose the authorities surprise economic agents by increasing

the money supply without announcing their intentions. In this situation firms

and workers with incomplete information would misperceive the resultant

increase in the general price level as an increase in relative prices and react

by increasing the supply of output and labour. In other words, workers and

firms would mistakenly perceive this as a real (as opposed to a nominal)

increase in the demand for their services/goods and respond by increasing the

supply of labour/output. In terms of Figure 5.3, the aggregate demand curve

would shift to the right from AD0 to AD1 to intersect the positively sloped

aggregate supply curve SRAS0 at point B. In line with equation (5.3), output

(Y1) would deviate from its natural level (YN) as a consequence of deviations

of the price level (P1) from its expected level (P0), that is, as the result of

expectational errors by agents. Any increase/decrease in output/unemployment

would, it is argued, only be temporary. Once agents realized that there

had been no change in relative prices, output and employment would return

to their long-run equilibrium (natural) levels. In terms of Figure 5.3, as agents

fully adjusted their price expectations the positively sloped aggregate supply

curve would shift upwards to the left, from SRAS0 to SRAS1, to intersect AD1

at point C. It is interesting to note that the former new classical adjustment

process discussed above (from A to C) corresponds to the orthodox monetarist

case in the long run, while the latter adjustment process (from A to B to C)

corresponds to the orthodox monetarist case in the short run, regardless of

whether the increase in the money supply is anticipated or unanticipated. To

summarize, the new classical analysis suggests that (i) an anticipated increase

in the money supply will raise the price level and have no effect on real

output and employment, and (ii) only unanticipated monetary surprises can

affect real variables in the short run.

This strong policy ineffectiveness proposition has major implications for

the controversy over the role and conduct of macroeconomic stabilization

policy. If the money supply is determined by the authorities according to

some ‘known’ rule, then the authorities will be unable to influence output and

employment even in the short run by pursuing a systematic monetary policy

as it can be anticipated by agents. For example, the authorities might adopt a

monetary rule which allows for a given fixed rate of monetary growth of 6 per

cent per annum. In forming their expectations of inflation, rational economic

agents would include the anticipated effects of the 6 per cent expansion of the

money supply. Consequently the systematic component (that is, 6 per cent) of

the monetary rule would have no effect on real variables. If, in practice, the

money supply grew at a rate of 8 per cent per annum, the non-systematic

(unanticipated) component of monetary expansion (that is, 2 per cent per

annum) would cause output and employment to rise temporarily above their

long-run equilibrium (natural) levels, owing to errors in inflation expectations.

Alternatively the authorities might allow the money supply to be

determined by a feedback rule (for example, in response to changes in unemployment

and output). Again changes in the rate of monetary growth which

arise from a known feedback rule will be anticipated by agents, making the

feedback policy rule ineffective. Only departures from a known monetary

rule (such as policy errors made by the monetary authorities or unforeseen

changes in policy) which are unanticipated will influence output.

The policy ineffectiveness proposition can be expressed algebraically in

the following way (see Gordon, 1976). We begin by rewriting the Friedman–

Phelps equation in modified linear form as:

P˙ P˙ (U U ) S t t


t Nt t (5.14)

where St represents an ‘exogenous’ supply shock (with zero mean) and Ut –

UNt represents the deviation of unemployment from its natural rate. Equation

(5.14) can be rewritten as:

U U P P S t N t t


t t 1/( ˙ −˙ ) (5.15)

The structural relationship between inflation ˙Pt and the rate of monetary

growth ˙Mt is given by:

P˙ M˙ D t t t (5.16)

where Dt represents ‘unpredictable’ demand shocks (such as shocks from

the private sector) which also have a zero mean. If ˙Mt

e is the expected rate of

growth of the money supply, the rational expectation of inflation will be:

P˙ M˙ t



e (5.17)

Suppose a Keynesian-inspired monetary authority attempts to control monetary

growth so that it grows at some constant rate (0) plus some proportion

(1) of the previous period’s deviation of unemployment from its natural rate.

In this case the actual rate of monetary growth will be:

M˙ (U U ) M˙ t t Nt t 0 1 1 1 (5.18)

where ˙Mt signifies a random or unanticipated element in monetary growth.

Equation (5.18) indicates that the monetary authorities are operating a systematic

feedback monetary rule which can be predicted by rational economic

agents as it becomes part of their information set (t–1) in equation (5.1).

Rational economic agents will therefore have expectations of inflation based

on the expected rate of monetary growth, shown in equation (5.19).

M˙ (U U ) t


t Nt 0 1 1 1 (5.19)

By subtracting (5.19) from (5.18) we obtain:

M˙ M˙ M˙ t t


t (5.20)

Subtracting (5.17) from (5.16) and substituting from (5.20) we derive equation


P˙ P˙ M˙ D t t


t t (5.21)

Finally substituting (5.21) into (5.15) gives us:

Ut UNt Mt Dt St 1/(˙ ) (5.22)

The important point to notice about equation (5.22) is that the systematic

component of monetary growth, (0 + 1(Ut–1 – UNt–1)), which the government

was attempting to use in order to prevent unemployment from deviating from

its natural rate, does not enter into it. The only component of equation (5.22)

that the monetary authorities can influence directly is M˙ t , the random component

of monetary growth. Therefore equation (5.22) tells us that, in a

Sargent and Wallace world, unemployment can deviate from its natural rate

as the result of unpredictable demand (Dt) and supply (St) shocks or unanticipated

monetary surprises (M˙ t ). Any systematic feedback monetary rule,

by becoming part of economic agents’ information set, cannot cause inflation

to deviate from its expected rate. Only departures from a known monetary

rule (such as policy errors made by the monetary authorities or unforeseen

changes in policy) which are unanticipated will influence output and employment.

In summary, the approach predicts that, as rational economic agents will

take into account any known monetary rule in forming their expectations, the

authorities will be unable to influence output and employment even in the

short run by pursuing a systematic monetary policy. Furthermore, any attempt

to affect output and employment by random or non-systematic monetary

policy will, it is argued, only increase the variation of output and employment

around their natural levels. It can be seen, therefore, that the argument advanced

by new classicists against policy activism is subtly different from

those put forward by orthodox monetarists (see Chapter 4, section 4.3.2 on

the role and conduct of monetary policy).

The policy ineffectiveness proposition that only unanticipated monetary

surprises have real output effects (or what is sometimes referred to as the

‘anticipated–unanticipated money debate’) has been the subject of a number

of empirical studies. Early work, in particular the seminal papers by Barro

(1977a, 1978), seemed to support the proposition. Using annual data for the

US economy over the period 1941–76, Barro used a two-stage method in first

estimating anticipated and unanticipated money growth before regressing

output and unemployment on unanticipated money growth. In general, Barro’s

studies provided support for the view that, while output and unemployment

are significantly affected by unanticipated money growth, anticipated money

growth has no real effects. However, subsequent studies, most notably by

Mishkin (1982) and Gordon (1982a), found evidence to suggest that both

unanticipated and anticipated monetary policy affect output and employment.

Overall, while the empirical evidence is mixed, it does not appear to support

the view that systematic monetary policy has no real effects. Moreover, as

Buiter (1980) pointed out, theoretical models can be constructed where even

fully anticipated changes in the rate of monetary growth can have real affects

by altering the rate inflation and hence the rate of return on money balances

that have a zero nominal rate of return. This in turn will affect the rate of

capital accumulation by changing the equilibrium portfolio composition. It

also goes without saying that fully anticipated fiscal changes, such as changes

in tax rates that alter labour supply and saving behaviour, will have real

effects. ‘Clearly fiscal policy is non-neutral in even the most classical of

systems’ (Buiter, 1980). In non-market-clearing models, where prices are

fixed, anticipated changes in monetary policy will have real effects via the

normal IS–LM–AD–AS mechanisms. In response to the Sargent and Wallace

papers, Fischer (1977), Phelps and Taylor (1977) and Taylor (1980) produced

models incorporating multi-period wage contracts and rational expectations

where monetary policy is non-neutral (see Chapter 7).

In addition, many Keynesians find this whole approach misguided, preferring

instead to explore the possibility that non market clearance can be

compatible with maximising behaviour on the part of all market participants

(Akerlof, 1979). In addition, the idea of stimulating aggregate demand when

the economy is already in (full employment) equilibrium would have been

anathema to Keynes. Why would such a policy ever be considered necessary?

As Frank Hahn (1982, p. 75) has commented, ‘Keynesians were concerned

with the problem of pushing the economy to its natural rate, not beyond it. If

the economy is there already, we can all go home.’