RDP 7703: Price and Quantity Responses to Monetary Impulses in a Model of a Small Open Economy 2. Previous Research

There is an increasing consensus that, with respect to the issues discussed in this paper, the so-called monetarist debate in macroeconomics is largely a matter of empirical detail. This is explicitly argued, for example, by B. Friedman (1976), and underlies the methodology of Stein (1976a, 1976b, 1977), Modigliani (1977) and others; furthermore, as indicated below, there is significant convergence on the role of money in economic dynamics, at least in long run analyses.

The evidence for the effects of monetary disturbances on the rest of the economy comes largely from studies with U.S. data, although there is some evidence which relates to smaller, more open, economies. The main techniques involved are the business cycle analysis of M. Friedman and his colleagues, single equation reduced form studies, and simulation analysis of both reduced form and structural econometric models.

For a closed economy, which the U.S. is often assumed to approximate, or an open economy with freely flexible exchange rates, the classic view is that, in the long term, a change in the rate of monetary growth normally affects only the rate of growth of prices,^[1] although monetary impulses are believed to also have important short run effects on output. Friedman and his colleagues have analysed the dynamics of monetary change most intensively, and on the basis of evidence from comparison of reference cycles and from single equation results for the U.S. (Friedman and Meiselman (1958), Friedman and Schwartz (1963), Friedman (1970)) it is argued that on average a change in the rate of growth of money produces a change in the rate of growth of nominal and real income after about six to nine months, with a change in the rate of growth of prices occurring after a further six to nine months. This statement is probably best interpreted as referring to significant impact effects, rather than to average lags in the usual sense.^[2] Friedman (1970, 1972) has suggested that in some cases, for example the British experience after World War 2, the longer run effects may take decades to work themselves out, partly because of prolonged use of direct controls. The period for which effects are important is a major question which only detailed empirical work can answer.^[3]

An increase in government spending which is not accompanied by an increase in the money supply is argued, again as a first approximation, to have no long run effects on the price level or on the aggregate level of real activity, although there may be significant short-run effects on both; private spending is “crowded out” by increased public spending, and the fixed money stock constrains the price effects. Friedman stresses that these relationships are far from perfect and that the lags in the effects of changes in monetary growth in particular are “long and variable”.

The evidence from the large U.S. models^[4] is not entirely consistent with these results, especially with respect to the longer run effects on prices. This appears to be partly due to theoretical weaknesses in these models,^[5] and partly because the models are estimated with data which are mostly from years of relatively low inflation.^[6] In general, the large U.S. models seem to give too much output response and too little price response to a monetary or budgetary impulse, relative to both Friedman's results and the implications of simple theoretical models, although Modigliani (1975, p.241) has argued that the F.M.P. model gives results which are consistent with Friedman in “the very long run”, and produces significant price responses in the short run. A particular point of disagreement is about the effects of a sustained increase in real government spending; the large U.S. models typically predict a positive effect on real product for several years and, in some cases, permanently. In strong contrast, the original FRB-St. Louis reduced form model results imply that the peak effect of fiscal policy occurred with two quarters, with “crowding out” occurring within four quarters.^[7] The St. Louis results also imply lags for the effects of changes in monetary aggregates which are much shorter than those suggested by the structural models. The reliability of the results given by both research efforts has been the subject of considerable debate.^[8]

Laidler (1973) fits a three equation model of the U.S. economy in which the only exogenous variable is the change of the stock of money and in which it is assumed that in the long run an increase in monetary expansion only increases the rate of inflation.^[9] Because of the small size of the model its properties can be examined analytically and it is asymptotically stable, approaching the steady state with long cyclical swings. As suggested by Friedman the period taken to converge to the steady state is a matter of decades.

For smaller, more open, economies the assumptions about the exchange rate are crucial. For the perfectly fixed rate case which has received most attention in the theoretical literature, the quantity of money is demand determined in the long run, with the balance of payments adjusting money supply to demand; prices are set in world markets, and income is exogenous.^[10] In the short run, however, there is scope for important domestic price and output effects, a point which is often ignored in empirical work. Thus, for example, the model of Australian capital flows developed by Porter (1974) assumes that domestic price and output effects are unimportant, and appears to demonstrate that a large proportion of domestic monetary expansion is offset by changes, in the capital account in the balance of payments.^[11] The RDX2 model of the Canadian economy^[12] and a small structural model of the U.K. economy^[13] have similar implications if domestic prices and income are artificially held constant. In both of these models, once this assumption is relaxed, domestic price and quantity responses to monetary expansion persist for long periods of time.

There are two related points which arise when domestic responses are allowed. Firstly, if domestic price and output responses are significant, they are likely in the short run to change the demand for domestic assets in a direction which reduces the extent to which the balance of payments offsets the initial change in domestic credit expansion. If the exchange rate comes under threat, however, expectations of a change in the rate can weaken or even reverse this effect; if speculation becomes extremely powerful, the “offset ratio” can exceed one. The second point is that as the exchange rate becomes more flexible, domestic monetary changes become bottled up and closed economy results become increasingly relevant. Under flexible exchange rates an expansion (say) of domestic credit will raise prices and devalue the exchange rate; whether there is an output effect of monetary expansion once an inflation is entrenched to the extent that the exchange rate is continuously devaluing is an important question which has received relatively little attention.

As mentioned above (footnote 3, page 4), there is evidence for the U.S. that threshold and other effects may raise the response of expected inflation to actual inflation, especially when the increase in inflation is persistent. A further point arises from the role of current economic variables – and particularly conditions in the money market – in the generation of price expectations. There is some evidence, for both Australia and the U.S., that either real stock or flow disequilibrium in the money market influences price expectations,^[14] and that this effect is becoming more important in recent times.^[15] If a given monetary expansion has a larger short-run effect on prices, through its effect on expectations, the disequilibrium in real terms generated by the monetary expansion will be smaller, implying a smaller transitory increase in consumption expenditure than it would otherwise.^[16] Furthermore, to the extent that devaluation of the exchange rate does not immediately offset the effect of the domestic price rise, there will tend to be an increase in imports and a reduction of exports as a result of a monetary expansion, and these reactions will reduce any output effect. Any adverse effect of inflation on business confidence and therefore on investment would represent a further reason for a damped output effect, and although such an effect could not be detected with data to 1974(4), it is significant when more recent data are added to the sample period.^[17] There is also the possibility that high and variable inflation adversely affects consumer confidence, although there is a problem of disentangling this effect from the more traditional substitution effect of inflation on consumer demand, and it has not been possible to provide convincing evidence on the net effect in the RBA76 model.^[18]

Although these effects require further analysis, they are possibly of considerable relevance for the extent to which a monetary impulse produces a price rather than an output response.^[19] It is worth noting that if the combination of increased inflation and economic instability reduces real output and raises interest rates, it will increase the excess supply of money for any given quantity of money.^[20] During the resulting adjustment period higher inflation will accompany lower real activity.

In summary, there is some systematic evidence pointing towards a smaller output effect and a larger price effect under conditions of relatively high inflation. This indicates the importance of the climate of price expectations in particular in determining the response to policy changes. Indeed, it is possible that in conditions in which price expectations are more responsive to monetary disequilibrium, price responses could accompany or precede output responses. This illustrates the danger of uncritical acceptance of results from a structure which represents average responses over a period which might be unrepresentative of the period for which the policy change is relevant.^[21] In the simulation analysis presented in Section 4, this point is allowed for to some extent by giving results from two versions of the model, with alternative assumptions about the size of the response of prices to monetary disequilibrium.