The Quantity Theory of Money Approach
The first stage in the development of orthodox monetarism can be traced
from the mid-1950s to the mid-1960s, and involved an attempt to re-establish
the quantity theory of money approach to macroeconomic analysis, which
had been usurped by the Keynesian revolution. Within the quantity theory of
money approach (see also Chapter 2, section 2.5) changes in the money stock
are regarded as the predominant, though not the only, factor explaining changes
in money or nominal income (see Laidler, 1991).
Orthodox Keynesian analysis (see Chapter 3, section 3.3) emphasized real
disturbances (notably fluctuations in investment and autonomous consump
tion) as the main cause of fluctuations in money income, predominantly in
the form of changes in real income. In terms of the stylized quantity theory
outlined in Chapter 2, section 2.5, Keynes’s General Theory was interpreted
as implying that in conditions of underemployment (which could prevail for
protracted periods) income velocity (V) would be highly unstable and would
passively adapt to whatever changes occurred independently in the money
supply (M) or money income (PY). In these circumstances money was regarded
as being relatively unimportant. For example, in the two extreme
cases of the liquidity and investment traps, money does not matter inasmuch
as monetary policy would be completely ineffective in influencing economic
activity. In the liquidity trap case, an increase in the money supply would be
exactly and completely offset by an opposite change in velocity. The increase
in the money supply would be absorbed entirely into idle/speculative balances
at an unchanged rate of interest and level of income. In the investment
trap case, where investment is completely interest-inelastic, an increase in the
money supply would again have no effect on the level of real income. The
money supply would be powerless to influence real income because investment
is insensitive to interest rate changes. Velocity would fall as the demand
for money increased relative to an unchanged level of income. Readers should
verify for themselves that, in either of these two extreme Keynesian cases
where money does not matter, any change in autonomous consumption, investment
or government expenditure would result in the full multiplier effect
of the simple Keynesian cross or 45° model. Under such conditions, although
the quantity theory relationship (equation 2.16) would be valid, orthodox
Keynesians argued it would be useless in terms of monetary policy prescription.
The quantity theory as a theory of the demand for money
It was against this orthodox Keynesian background that Milton Friedman
sought to maintain and re-establish across the profession what he regarded as
the oral tradition of the University of Chicago, namely the quantity theory of
money approach to macroeconomic analysis (for a criticism of this interpretation,
see Patinkin, 1969). Although the traditional quantity theory is a body
of doctrine concerned with the relationship between the money supply and
the general price level, Friedman (1956) initially presented his restatement of
the quantity theory of money as a theory of the demand for money, rather
than a theory of the general price level or money income.
Friedman postulated that the demand for money (like the demand for any
asset) yields a flow of services to the holder and depends on three main
factors: (i) the wealth constraint, which determines the maximum amount of
money that can be held; (ii) the return or yield on money in relation to the
return on other financial and real assets in which wealth can be held; and (iii)
the asset-holder’s tastes and preferences. The way total wealth is allocated
between various forms depends on the relative rates of return on the various
assets. These assets include not just money and bonds but also equities and
physical goods. In equilibrium wealth will be allocated between assets such
that marginal rates of return are equal. Although Patinkin (1969) has suggested
that Friedman’s restatement should be regarded as an extension of
Keynesian analysis, there are three important differences worth highlighting.
First, Friedman’s analysis of the demand for money can be regarded as an
application of his permanent income theory of consumption to the demand
for a particular asset. Second, he introduced the expected rate of inflation as a
potentially important variable into the demand for money function. Third, he
asserted that the demand for money was a stable function of a limited number
of variables.
A simplified version of Friedman’s demand function for real money balances
can be written in the following form:
M
P
f Y r P u d P e = ( ; ,˙ ; )
where YP represents permanent income, which is used as a proxy for wealth,
the budget constraint;
r represents the return on financial assets,
represents the expected rate of inflation; and
u represents individuals’ tastes and preferences.
This analysis predicts that, ceteris paribus, the demand for money will be
greater (i) the higher the level of wealth; (ii) the lower the yield on other
assets; (iii) the lower the expected rate of inflation, and vice versa. Utilitymaximizing
individuals will reallocate wealth between different assets
whenever marginal rates of return are not equal. This portfolio adjustment
process is central to the monetarist specification of the transmission mechanism
whereby changes in the stock of money affect the real sector. This can
be illustrated by examining the effects of an increase in the money supply
brought about by open market operations by the monetary authorities. An
initial equilibrium is assumed where wealth is allocated between financial
and real assets such that marginal rates of return are equal. Following open
market purchases of bonds by the monetary authorities, the public’s money
holdings will increase. Given that the marginal return on any asset diminishes
as holdings of it increase, the marginal rate of return on money holdings will
in consequence fall. As excess money balances are exchanged for financial
and real assets (such as consumer durables), their prices will be bid up until
portfolio equilibrium is re-established when once again all assets are willingly
held and marginal rates of return are equal. In contrast to orthodox
Keynesian analysis, monetarists argue that money is a substitute for a wide
range of real and financial assets, and that no single asset or group of assets
can be considered a close substitute for money. A much broader range of
assets and associated expenditures is emphasized and in consequence monetarists
attribute a much stronger and more direct effect on aggregate spending
to monetary impulses.
The quantity theory and changes in money income: empirical
evidence
The assertion that there exists a stable functional relationship (behaviour)
between the demand for real balances and a limited number of variables that
determine it lies at the heart of the modern quantity theory of money approach
to macroeconomic analysis. If the demand for money function is
stable, then velocity will also be stable, changing in a predictable manner if
any of the limited number of variables in the demand for money function
should change. Friedman (1968b, p. 434) has postulated the QTM as
the empirical generalisation that changes in desired real balances (in the demand
for money) tend to proceed slowly and gradually or to be the result of events set in
train by prior changes in supply, whereas, in contrast, substantial changes in the
supply of nominal balances can and frequently do occur independently of any
changes in demand. The conclusion is that substantial changes in prices or nominal
income are almost invariably the result of changes in the nominal supply of
money.
In this section we discuss various empirical evidence put forward in support
of the quantity theory of money approach to macroeconomic analysis,
beginning with the demand for money function. Constraints of space preclude
a detailed discussion of the empirical evidence on the demand for
money. Nevertheless two points are worth highlighting. First, although Friedman
(1959) in his early empirical work on the demand for money claimed to
have found that the interest rate was insignificant, virtually all studies undertaken
thereafter have found the interest rate to be an important variable in the
function. Indeed, in a subsequent paper Friedman (1966) acknowledged this.
Buiter (2003a) recounts that Tobin, in his long debate with Friedman, ‘convinced
most of the profession that the demand for money has an economically
and statistically significant interest rate-responsiveness’ (that is, the LM curve
is not perfectly inelastic). This argument was a crucial part of Tobin’s case in
support of discretionary fiscal policy having a role to play in stabilization
policy. Furthermore, in the 1950s and 1960s there also appeared little evidence
that the interest elasticity of the money demand increased as the rate of
interest fell, as the liquidity trap requires. This means that both the extreme
quantity theory and Keynesian cases of vertical and horizontal LM curves,
respectively, could be ruled out. The static IS–LM model can, however, still
be used to illustrate the quantity theory approach to macroeconomic analysis
if both the real rate of interest and real income are determined by real, not
monetary, forces and the economy automatically tends towards full employment
(see Friedman, 1968a). Second, although the belief in a stable demand
for money function was well supported by empirical evidence up to the early
1970s, since then a number of studies, both in the USA and other economies,
have found evidence of apparent instability of the demand for money. In the
USA, for example, there occurred a marked break in the trend of the velocity
of the narrow monetary aggregate, M1, in the early 1980s and subsequent
breaks in the velocities of the broader monetary aggregates, M2 and M3, in
the early 1990s. A number of possible explanations have been put forward to
explain this apparent instability, including institutional change within the
financial system which took place in the 1970s and 1980s. The reader is
referred to Laidler (1993) for a detailed and very accessible discussion of the
empirical evidence on the demand for money, and the continuing controversy
over the question of the stability of the demand for money function.
Friedman (1958) sought to re-establish an important independent role for
money through a study of time series data comparing rates of monetary
growth with turning points in the level of economic activity for the USA. On
the average of 18 non-war cycles since 1870, he found that peaks (troughs) in
the rate of change of the money supply had preceded peaks (troughs) in the
level of economic activity by an average of 16 (12) months. Friedman concluded
that this provided strong suggestive evidence of an influence running
from money to business. Friedman’s study was subsequently criticized by
Culbertson (1960, 1961) and by Kareken and Solow (1963) on both methodological
and statistical grounds. First, the question was raised as to whether
the timing evidence justified the inference of a causal relationship running
from money to economic activity (see also Kaldor, 1970a; Sims, 1972).
Second, statistical objections to Friedman’s procedure were raised in that he
had not compared like with like. When Kareken and Solow reran the tests
with Friedman’s data using rates of change for both money and economic
activity, they found no uniform lead of monetary changes over changes in the
level of economic activity. Later, the issue of money to income causality was
famously taken up by Tobin (1970), who challenged the reliability of the
timing (leads and lags) evidence accumulated by Friedman and other monetarists.
Using an ‘Ultra Keynesian’ model Tobin demonstrated how the timing
evidence could just as easily be interpreted in support of the Keynesian
position on business cycles and instability. Tobin accused Friedman of falling
foul of the ‘Post Hoc Ergo Propter Hoc’ fallacy. He also went further by
criticizing Friedman for not having an explicit theoretical foundation linking
cause and effect on which to base his monetarist claims. The claim was
frequently made that much of Friedman’s work was ‘measurement without
theory’ and that monetarism remained too much a ‘black box’. As Hoover
(2001a, 2001b) has recently reminded economists, correlation can never prove
causation. This problem of ‘causality in macroeconomics’ has led to, and will
continue to lead to, endless arguments and controversy in empirical macroeconomics
(see also Friedman, 1970b; Davidson and Weintraub, 1973; Romer
and Romer, 1994a, 1994b; Hoover and Perez, 1994; Hammond, 1996).
In 1963, Friedman and Schwartz (1963) presented more persuasive evidence
to support the monetarist belief that changes in the stock of money
play a largely independent role in cyclical fluctuations. In their influential
study of the Monetary History of the United States, 1867–1960, they found
that, while the stock of money had tended to rise during both cyclical expansions
and contractions, the rate of growth of the money supply had been
slower during contractions than during expansions in the level of economic
activity. Within the period examined, the only times when there was an
appreciable absolute fall in the money stock were also the six periods of
major economic contraction identified: 1873–9, 1893–4, 1907–8, 1920–21,
1929–33 and 1937–8. Furthermore, from studying the historical circumstances
underlying the changes that occurred in the money supply during
these major recessions, Friedman and Schwartz argued that the factors producing
monetary contraction were mainly independent of contemporary or
prior changes in money income and prices. In other words, monetary changes
were seen as the cause, rather than the consequence, of major recessions. For
example, Friedman and Schwartz argued that the absolute decline in the
money stock which took place during both 1920–21 and 1937–8 was a
consequence of highly restrictive policy actions undertaken by the Federal
Reserve System: for example, reserve requirements were doubled in 1936
and early 1937. These actions were themselves followed by sharp declines in
the money stock, which were in turn followed by a period of severe economic
contraction.
Even more controversial was the reinterpretation of the Great Depression
as demonstrating the potency of monetary change and monetary policy. Friedman
and Schwartz argued that an initial mild decline in the money stock from
1929 to 1930 was converted into a sharp decline by a wave of bank failures
which started in late 1930 (see also Bernanke, 1983). Bank failures produced
an increase in both the currency-to-deposit ratio, owing to the public’s loss of
faith in the banks’ ability to redeem their deposits, and the reserve-to-deposit
ratio, owing to the banks’ loss of faith in the public’s willingness to maintain
their deposits with them. In Friedman and Schwartz’s view, the consequent
decline in the money stock was further intensified by the Federal Reserve
System’s restrictive action of raising the discount rate in October 1931,
which in turn led to further bank failures. In this interpretation the depression
only became great as a consequence of the failure of the Federal Reserve to
prevent the dramatic decline in the money stock – between October 1929 and
June 1933, the money stock fell by about a third. By adopting alternative
policies the Federal Reserve System, they argued, could have prevented the
banking collapse and the resulting fall in the money stock and severe economic
contraction. Friedman and Schwartz further justified their view that
changes in the stock of money play a largely independent role in cyclical
fluctuations from the evidence that cyclical movements in money had much
the same relationship (both in timing and amplitude) as cyclical movements
in business activity, even under substantially different monetary arrangements
that had prevailed in the USA over the period 1867–1960 (for further
discussion of these issues, see Temin, 1976; Romer and Romer, 1989; Romer,
1992; Hammond, 1996).
A more intense exchange was triggered by the publication of the study
undertaken by Friedman and Meiselman (1963) for the Commission on Money
and Credit. Although the ensuing Friedman–Meiselman debate occupied
economists for a lengthy period of time, the debate itself is now generally
regarded as largely only of interest to students of the history of economic
thought. In brief, Friedman and Meiselman attempted to estimate how much
of the variation in consumption (a proxy variable for income) could be
explained by changes in (i) the money supply, in line with the quantity theory
approach, and (ii) autonomous expenditure (investment), in line with Keynesian
analysis. Using two test equations (one using money and the other autonomous
expenditure as the independent variable) for US data over the period
1897–1958, they found that, apart from one sub-period dominated by the
Great Depression, the money equation gave much the better explanation.
These results were subsequently challenged, most notably by De Prano and
Mayer (1965) and Ando and Modigliani (1965), who showed that a change in
the definition of autonomous expenditure improved the performance of the
autonomous expenditure equation.
On reflection it is fair to say that these tests were ill devised to discriminate
between the quantity theory of money and the Keynesian view, so that they
failed to establish whether it was changes in the supply of money or autonomous
expenditure that were causing changes in income. This can be illustrated
by reference to the IS–LM model for a closed economy. In general, within
the Hicksian IS–LM framework, monetary and fiscal multipliers each depend
on both the consumption function and the liquidity preference function.
Equally good results can be obtained using the two equations when income
determination is either purely classical or Keynesian. The classical case is
illustrated in Figure 4.2, where the demand for money is independent of the
rate of interest. The economy is initially in equilibrium at a less than full
employment income level of Y0 and a rate of interest r0, that is, the intersec
tion of LM0 and IS. An increase in the money supply (which shifts the LM
curve from LM0 to LM1) would result in a lower rate of interest (r1) and a
higher level of income (Y1). As the interest rate falls, investment expenditure
is stimulated, which in turn, through the multiplier, affects consumption and
income. In the classical case, empirical studies would uncover a stable relationship
between autonomous expenditure and the level of income, even
though the direction of causation would run from money to income.
The Keynesian case is illustrated in Figure 4.3. The economy is initially in
equilibrium at an income level of Y0 and a rate of interest of r*, that is, the
intersection of IS0 and LM0. Following an expansionary real impulse (which
shifts the IS curve outwards to the right, from IS0 to IS1), the authorities could
stabilize the interest rate at r* by expanding the money supply (shifting the LM
curve downwards to the right, from LM0 to LM1). In the Keynesian case,
empirical studies would uncover a stable relationship between the money supply
and the level of income, even though in this particular case the direction of
causation would run from income to money. In conclusion, what the Friedman–
Meiselman tests appeared to demonstrate was that (i) the marginal propensity
to consume had been relatively stable and (ii) contrary to the extreme Keynesian
view, the economy had not been in a liquidity or investment trap because if it
had the tests would not have found such good fits for the money equation.
An assessment
At this point it would be useful to draw together the material presented in this
section and summarize the central tenets that proponents of the quantity
theory of money approach to macroeconomic analysis generally adhered to
by the mid-1960s (see Mayer, 1978; Vane and Thompson, 1979; Purvis,
1980; Laidler, 1981). The central distinguishing beliefs at that time could be
listed as follows:
1. Changes in the money stock are the predominant factor explaining changes
in money income.
2. In the face of a stable demand for money, most of the observed instability
in the economy could be attributed to fluctuations in the money
supply induced by the monetary authorities.
3. The authorities can control the money supply if they choose to do so and
when that control is exercised the path of money income will be different
from a situation where the money supply is endogenous.
4. The lag between changes in the money stock and changes in money
income is long and variable, so that attempts to use discretionary monetary
policy to fine-tune the economy could turn out to be destabilizing.
5. The money supply should be allowed to grow at a fixed rate in line with
the underlying growth of output to ensure long-term price stability.
The Keynesian–monetarist debate, relating to the importance of changes in
the money stock as the predominant factor explaining changes in money
income, reached a climax in 1970, when Friedman, in response to his critics,
attempted to set forth his ‘Theoretical Framework for Monetary Analysis’.
Until the publication of Friedman’s 1970 paper there existed no explicit,
formal and coherent statement of the theoretical structure underlying monetarist
pronouncements. In opening up the monetarist ‘black box’ for theoretical
scrutiny, Friedman intended to demonstrate that ‘the basic differences among
economists are empirical not theoretical’. His theoretical statement turned
out to be a generalized IS–LM model which helped to place the monetarist
approach within the mainstream position (see Friedman, 1970a, 1972; Tobin,
1972b; Gordon, 1974). This debate represented the ‘final big battle between
Friedman and his Keynesian critics’ before the rational expectations revolution
and new classical economics ‘swept both Keynesianism and monetarism
from center stage’ (see Hammond, 1996). According to Tobin (1981), the
central issue for both macroeconomic theory and policy is the supply response
of the economy to monetary impulses. The division of such impulses
between prices and quantities was referred to by Friedman as ‘the missing
equation’. In Tobin’s view, Friedman’s solution to this problem ‘was not
different in spirit from the wage/price/output mechanisms of mainstream
eclectic Keynesian theory and econometrics’ (Tobin, 1981, p. 36).
In retrospect we can now see that Friedman’s debate with his critics demonstrated
that their differences were more quantitative than qualitative, and
contributed towards an emerging synthesis of monetarist and Keynesian ideas.
This emerging synthesis, or theoretical accord, was to establish that the
Keynesian-dominated macroeconomics of the 1950s had understated (but not
neglected) the importance of monetary impulses in generating economic
instability (see Laidler, 1992a). This was perhaps especially true in the UK in
the period culminating in the Radcliffe Report (1959) on the working of the
monetary system in the UK. According to Samuelson, a leading US Keynesian,
‘the contrast between British and American Keynesianism had become dramatic’
by 1959 because many of Keynes’s admirers in Britain ‘were still
frozen in the Model T version of his system’ (see Samuelson, 1983, 1988;
Johnson, 1978).
The first stage in the development of orthodox monetarism can be traced
from the mid-1950s to the mid-1960s, and involved an attempt to re-establish
the quantity theory of money approach to macroeconomic analysis, which
had been usurped by the Keynesian revolution. Within the quantity theory of
money approach (see also Chapter 2, section 2.5) changes in the money stock
are regarded as the predominant, though not the only, factor explaining changes
in money or nominal income (see Laidler, 1991).
Orthodox Keynesian analysis (see Chapter 3, section 3.3) emphasized real
disturbances (notably fluctuations in investment and autonomous consump
tion) as the main cause of fluctuations in money income, predominantly in
the form of changes in real income. In terms of the stylized quantity theory
outlined in Chapter 2, section 2.5, Keynes’s General Theory was interpreted
as implying that in conditions of underemployment (which could prevail for
protracted periods) income velocity (V) would be highly unstable and would
passively adapt to whatever changes occurred independently in the money
supply (M) or money income (PY). In these circumstances money was regarded
as being relatively unimportant. For example, in the two extreme
cases of the liquidity and investment traps, money does not matter inasmuch
as monetary policy would be completely ineffective in influencing economic
activity. In the liquidity trap case, an increase in the money supply would be
exactly and completely offset by an opposite change in velocity. The increase
in the money supply would be absorbed entirely into idle/speculative balances
at an unchanged rate of interest and level of income. In the investment
trap case, where investment is completely interest-inelastic, an increase in the
money supply would again have no effect on the level of real income. The
money supply would be powerless to influence real income because investment
is insensitive to interest rate changes. Velocity would fall as the demand
for money increased relative to an unchanged level of income. Readers should
verify for themselves that, in either of these two extreme Keynesian cases
where money does not matter, any change in autonomous consumption, investment
or government expenditure would result in the full multiplier effect
of the simple Keynesian cross or 45° model. Under such conditions, although
the quantity theory relationship (equation 2.16) would be valid, orthodox
Keynesians argued it would be useless in terms of monetary policy prescription.
The quantity theory as a theory of the demand for money
It was against this orthodox Keynesian background that Milton Friedman
sought to maintain and re-establish across the profession what he regarded as
the oral tradition of the University of Chicago, namely the quantity theory of
money approach to macroeconomic analysis (for a criticism of this interpretation,
see Patinkin, 1969). Although the traditional quantity theory is a body
of doctrine concerned with the relationship between the money supply and
the general price level, Friedman (1956) initially presented his restatement of
the quantity theory of money as a theory of the demand for money, rather
than a theory of the general price level or money income.
Friedman postulated that the demand for money (like the demand for any
asset) yields a flow of services to the holder and depends on three main
factors: (i) the wealth constraint, which determines the maximum amount of
money that can be held; (ii) the return or yield on money in relation to the
return on other financial and real assets in which wealth can be held; and (iii)
the asset-holder’s tastes and preferences. The way total wealth is allocated
between various forms depends on the relative rates of return on the various
assets. These assets include not just money and bonds but also equities and
physical goods. In equilibrium wealth will be allocated between assets such
that marginal rates of return are equal. Although Patinkin (1969) has suggested
that Friedman’s restatement should be regarded as an extension of
Keynesian analysis, there are three important differences worth highlighting.
First, Friedman’s analysis of the demand for money can be regarded as an
application of his permanent income theory of consumption to the demand
for a particular asset. Second, he introduced the expected rate of inflation as a
potentially important variable into the demand for money function. Third, he
asserted that the demand for money was a stable function of a limited number
of variables.
A simplified version of Friedman’s demand function for real money balances
can be written in the following form:
M
P
f Y r P u d P e = ( ; ,˙ ; )
where YP represents permanent income, which is used as a proxy for wealth,
the budget constraint;
r represents the return on financial assets,
represents the expected rate of inflation; and
u represents individuals’ tastes and preferences.
This analysis predicts that, ceteris paribus, the demand for money will be
greater (i) the higher the level of wealth; (ii) the lower the yield on other
assets; (iii) the lower the expected rate of inflation, and vice versa. Utilitymaximizing
individuals will reallocate wealth between different assets
whenever marginal rates of return are not equal. This portfolio adjustment
process is central to the monetarist specification of the transmission mechanism
whereby changes in the stock of money affect the real sector. This can
be illustrated by examining the effects of an increase in the money supply
brought about by open market operations by the monetary authorities. An
initial equilibrium is assumed where wealth is allocated between financial
and real assets such that marginal rates of return are equal. Following open
market purchases of bonds by the monetary authorities, the public’s money
holdings will increase. Given that the marginal return on any asset diminishes
as holdings of it increase, the marginal rate of return on money holdings will
in consequence fall. As excess money balances are exchanged for financial
and real assets (such as consumer durables), their prices will be bid up until
portfolio equilibrium is re-established when once again all assets are willingly
held and marginal rates of return are equal. In contrast to orthodox
Keynesian analysis, monetarists argue that money is a substitute for a wide
range of real and financial assets, and that no single asset or group of assets
can be considered a close substitute for money. A much broader range of
assets and associated expenditures is emphasized and in consequence monetarists
attribute a much stronger and more direct effect on aggregate spending
to monetary impulses.
The quantity theory and changes in money income: empirical
evidence
The assertion that there exists a stable functional relationship (behaviour)
between the demand for real balances and a limited number of variables that
determine it lies at the heart of the modern quantity theory of money approach
to macroeconomic analysis. If the demand for money function is
stable, then velocity will also be stable, changing in a predictable manner if
any of the limited number of variables in the demand for money function
should change. Friedman (1968b, p. 434) has postulated the QTM as
the empirical generalisation that changes in desired real balances (in the demand
for money) tend to proceed slowly and gradually or to be the result of events set in
train by prior changes in supply, whereas, in contrast, substantial changes in the
supply of nominal balances can and frequently do occur independently of any
changes in demand. The conclusion is that substantial changes in prices or nominal
income are almost invariably the result of changes in the nominal supply of
money.
In this section we discuss various empirical evidence put forward in support
of the quantity theory of money approach to macroeconomic analysis,
beginning with the demand for money function. Constraints of space preclude
a detailed discussion of the empirical evidence on the demand for
money. Nevertheless two points are worth highlighting. First, although Friedman
(1959) in his early empirical work on the demand for money claimed to
have found that the interest rate was insignificant, virtually all studies undertaken
thereafter have found the interest rate to be an important variable in the
function. Indeed, in a subsequent paper Friedman (1966) acknowledged this.
Buiter (2003a) recounts that Tobin, in his long debate with Friedman, ‘convinced
most of the profession that the demand for money has an economically
and statistically significant interest rate-responsiveness’ (that is, the LM curve
is not perfectly inelastic). This argument was a crucial part of Tobin’s case in
support of discretionary fiscal policy having a role to play in stabilization
policy. Furthermore, in the 1950s and 1960s there also appeared little evidence
that the interest elasticity of the money demand increased as the rate of
interest fell, as the liquidity trap requires. This means that both the extreme
quantity theory and Keynesian cases of vertical and horizontal LM curves,
respectively, could be ruled out. The static IS–LM model can, however, still
be used to illustrate the quantity theory approach to macroeconomic analysis
if both the real rate of interest and real income are determined by real, not
monetary, forces and the economy automatically tends towards full employment
(see Friedman, 1968a). Second, although the belief in a stable demand
for money function was well supported by empirical evidence up to the early
1970s, since then a number of studies, both in the USA and other economies,
have found evidence of apparent instability of the demand for money. In the
USA, for example, there occurred a marked break in the trend of the velocity
of the narrow monetary aggregate, M1, in the early 1980s and subsequent
breaks in the velocities of the broader monetary aggregates, M2 and M3, in
the early 1990s. A number of possible explanations have been put forward to
explain this apparent instability, including institutional change within the
financial system which took place in the 1970s and 1980s. The reader is
referred to Laidler (1993) for a detailed and very accessible discussion of the
empirical evidence on the demand for money, and the continuing controversy
over the question of the stability of the demand for money function.
Friedman (1958) sought to re-establish an important independent role for
money through a study of time series data comparing rates of monetary
growth with turning points in the level of economic activity for the USA. On
the average of 18 non-war cycles since 1870, he found that peaks (troughs) in
the rate of change of the money supply had preceded peaks (troughs) in the
level of economic activity by an average of 16 (12) months. Friedman concluded
that this provided strong suggestive evidence of an influence running
from money to business. Friedman’s study was subsequently criticized by
Culbertson (1960, 1961) and by Kareken and Solow (1963) on both methodological
and statistical grounds. First, the question was raised as to whether
the timing evidence justified the inference of a causal relationship running
from money to economic activity (see also Kaldor, 1970a; Sims, 1972).
Second, statistical objections to Friedman’s procedure were raised in that he
had not compared like with like. When Kareken and Solow reran the tests
with Friedman’s data using rates of change for both money and economic
activity, they found no uniform lead of monetary changes over changes in the
level of economic activity. Later, the issue of money to income causality was
famously taken up by Tobin (1970), who challenged the reliability of the
timing (leads and lags) evidence accumulated by Friedman and other monetarists.
Using an ‘Ultra Keynesian’ model Tobin demonstrated how the timing
evidence could just as easily be interpreted in support of the Keynesian
position on business cycles and instability. Tobin accused Friedman of falling
foul of the ‘Post Hoc Ergo Propter Hoc’ fallacy. He also went further by
criticizing Friedman for not having an explicit theoretical foundation linking
cause and effect on which to base his monetarist claims. The claim was
frequently made that much of Friedman’s work was ‘measurement without
theory’ and that monetarism remained too much a ‘black box’. As Hoover
(2001a, 2001b) has recently reminded economists, correlation can never prove
causation. This problem of ‘causality in macroeconomics’ has led to, and will
continue to lead to, endless arguments and controversy in empirical macroeconomics
(see also Friedman, 1970b; Davidson and Weintraub, 1973; Romer
and Romer, 1994a, 1994b; Hoover and Perez, 1994; Hammond, 1996).
In 1963, Friedman and Schwartz (1963) presented more persuasive evidence
to support the monetarist belief that changes in the stock of money
play a largely independent role in cyclical fluctuations. In their influential
study of the Monetary History of the United States, 1867–1960, they found
that, while the stock of money had tended to rise during both cyclical expansions
and contractions, the rate of growth of the money supply had been
slower during contractions than during expansions in the level of economic
activity. Within the period examined, the only times when there was an
appreciable absolute fall in the money stock were also the six periods of
major economic contraction identified: 1873–9, 1893–4, 1907–8, 1920–21,
1929–33 and 1937–8. Furthermore, from studying the historical circumstances
underlying the changes that occurred in the money supply during
these major recessions, Friedman and Schwartz argued that the factors producing
monetary contraction were mainly independent of contemporary or
prior changes in money income and prices. In other words, monetary changes
were seen as the cause, rather than the consequence, of major recessions. For
example, Friedman and Schwartz argued that the absolute decline in the
money stock which took place during both 1920–21 and 1937–8 was a
consequence of highly restrictive policy actions undertaken by the Federal
Reserve System: for example, reserve requirements were doubled in 1936
and early 1937. These actions were themselves followed by sharp declines in
the money stock, which were in turn followed by a period of severe economic
contraction.
Even more controversial was the reinterpretation of the Great Depression
as demonstrating the potency of monetary change and monetary policy. Friedman
and Schwartz argued that an initial mild decline in the money stock from
1929 to 1930 was converted into a sharp decline by a wave of bank failures
which started in late 1930 (see also Bernanke, 1983). Bank failures produced
an increase in both the currency-to-deposit ratio, owing to the public’s loss of
faith in the banks’ ability to redeem their deposits, and the reserve-to-deposit
ratio, owing to the banks’ loss of faith in the public’s willingness to maintain
their deposits with them. In Friedman and Schwartz’s view, the consequent
decline in the money stock was further intensified by the Federal Reserve
System’s restrictive action of raising the discount rate in October 1931,
which in turn led to further bank failures. In this interpretation the depression
only became great as a consequence of the failure of the Federal Reserve to
prevent the dramatic decline in the money stock – between October 1929 and
June 1933, the money stock fell by about a third. By adopting alternative
policies the Federal Reserve System, they argued, could have prevented the
banking collapse and the resulting fall in the money stock and severe economic
contraction. Friedman and Schwartz further justified their view that
changes in the stock of money play a largely independent role in cyclical
fluctuations from the evidence that cyclical movements in money had much
the same relationship (both in timing and amplitude) as cyclical movements
in business activity, even under substantially different monetary arrangements
that had prevailed in the USA over the period 1867–1960 (for further
discussion of these issues, see Temin, 1976; Romer and Romer, 1989; Romer,
1992; Hammond, 1996).
A more intense exchange was triggered by the publication of the study
undertaken by Friedman and Meiselman (1963) for the Commission on Money
and Credit. Although the ensuing Friedman–Meiselman debate occupied
economists for a lengthy period of time, the debate itself is now generally
regarded as largely only of interest to students of the history of economic
thought. In brief, Friedman and Meiselman attempted to estimate how much
of the variation in consumption (a proxy variable for income) could be
explained by changes in (i) the money supply, in line with the quantity theory
approach, and (ii) autonomous expenditure (investment), in line with Keynesian
analysis. Using two test equations (one using money and the other autonomous
expenditure as the independent variable) for US data over the period
1897–1958, they found that, apart from one sub-period dominated by the
Great Depression, the money equation gave much the better explanation.
These results were subsequently challenged, most notably by De Prano and
Mayer (1965) and Ando and Modigliani (1965), who showed that a change in
the definition of autonomous expenditure improved the performance of the
autonomous expenditure equation.
On reflection it is fair to say that these tests were ill devised to discriminate
between the quantity theory of money and the Keynesian view, so that they
failed to establish whether it was changes in the supply of money or autonomous
expenditure that were causing changes in income. This can be illustrated
by reference to the IS–LM model for a closed economy. In general, within
the Hicksian IS–LM framework, monetary and fiscal multipliers each depend
on both the consumption function and the liquidity preference function.
Equally good results can be obtained using the two equations when income
determination is either purely classical or Keynesian. The classical case is
illustrated in Figure 4.2, where the demand for money is independent of the
rate of interest. The economy is initially in equilibrium at a less than full
employment income level of Y0 and a rate of interest r0, that is, the intersec
tion of LM0 and IS. An increase in the money supply (which shifts the LM
curve from LM0 to LM1) would result in a lower rate of interest (r1) and a
higher level of income (Y1). As the interest rate falls, investment expenditure
is stimulated, which in turn, through the multiplier, affects consumption and
income. In the classical case, empirical studies would uncover a stable relationship
between autonomous expenditure and the level of income, even
though the direction of causation would run from money to income.
The Keynesian case is illustrated in Figure 4.3. The economy is initially in
equilibrium at an income level of Y0 and a rate of interest of r*, that is, the
intersection of IS0 and LM0. Following an expansionary real impulse (which
shifts the IS curve outwards to the right, from IS0 to IS1), the authorities could
stabilize the interest rate at r* by expanding the money supply (shifting the LM
curve downwards to the right, from LM0 to LM1). In the Keynesian case,
empirical studies would uncover a stable relationship between the money supply
and the level of income, even though in this particular case the direction of
causation would run from income to money. In conclusion, what the Friedman–
Meiselman tests appeared to demonstrate was that (i) the marginal propensity
to consume had been relatively stable and (ii) contrary to the extreme Keynesian
view, the economy had not been in a liquidity or investment trap because if it
had the tests would not have found such good fits for the money equation.
An assessment
At this point it would be useful to draw together the material presented in this
section and summarize the central tenets that proponents of the quantity
theory of money approach to macroeconomic analysis generally adhered to
by the mid-1960s (see Mayer, 1978; Vane and Thompson, 1979; Purvis,
1980; Laidler, 1981). The central distinguishing beliefs at that time could be
listed as follows:
1. Changes in the money stock are the predominant factor explaining changes
in money income.
2. In the face of a stable demand for money, most of the observed instability
in the economy could be attributed to fluctuations in the money
supply induced by the monetary authorities.
3. The authorities can control the money supply if they choose to do so and
when that control is exercised the path of money income will be different
from a situation where the money supply is endogenous.
4. The lag between changes in the money stock and changes in money
income is long and variable, so that attempts to use discretionary monetary
policy to fine-tune the economy could turn out to be destabilizing.
5. The money supply should be allowed to grow at a fixed rate in line with
the underlying growth of output to ensure long-term price stability.
The Keynesian–monetarist debate, relating to the importance of changes in
the money stock as the predominant factor explaining changes in money
income, reached a climax in 1970, when Friedman, in response to his critics,
attempted to set forth his ‘Theoretical Framework for Monetary Analysis’.
Until the publication of Friedman’s 1970 paper there existed no explicit,
formal and coherent statement of the theoretical structure underlying monetarist
pronouncements. In opening up the monetarist ‘black box’ for theoretical
scrutiny, Friedman intended to demonstrate that ‘the basic differences among
economists are empirical not theoretical’. His theoretical statement turned
out to be a generalized IS–LM model which helped to place the monetarist
approach within the mainstream position (see Friedman, 1970a, 1972; Tobin,
1972b; Gordon, 1974). This debate represented the ‘final big battle between
Friedman and his Keynesian critics’ before the rational expectations revolution
and new classical economics ‘swept both Keynesianism and monetarism
from center stage’ (see Hammond, 1996). According to Tobin (1981), the
central issue for both macroeconomic theory and policy is the supply response
of the economy to monetary impulses. The division of such impulses
between prices and quantities was referred to by Friedman as ‘the missing
equation’. In Tobin’s view, Friedman’s solution to this problem ‘was not
different in spirit from the wage/price/output mechanisms of mainstream
eclectic Keynesian theory and econometrics’ (Tobin, 1981, p. 36).
In retrospect we can now see that Friedman’s debate with his critics demonstrated
that their differences were more quantitative than qualitative, and
contributed towards an emerging synthesis of monetarist and Keynesian ideas.
This emerging synthesis, or theoretical accord, was to establish that the
Keynesian-dominated macroeconomics of the 1950s had understated (but not
neglected) the importance of monetary impulses in generating economic
instability (see Laidler, 1992a). This was perhaps especially true in the UK in
the period culminating in the Radcliffe Report (1959) on the working of the
monetary system in the UK. According to Samuelson, a leading US Keynesian,
‘the contrast between British and American Keynesianism had become dramatic’
by 1959 because many of Keynes’s admirers in Britain ‘were still
frozen in the Model T version of his system’ (see Samuelson, 1983, 1988;
Johnson, 1978).
No comments:
Post a Comment