Flexible prices: the monetary model
(5.8) where the convention has been introduced (and will be followed wherever necessary
5.4 The monetary model as an explanation of the facts
rates. In other words, they first forecast the ‘fundamentals’, then form their expecta- tions as to the exchange rate accordingly.
What this implies, of course, is that interest rate changes are not exogenous events, they don’t just happen, they are themselves very much the outcome of changes in expectations.
But now ask yourself another question: what kind of factors are these ‘funda- mentals’ likely to be?
It is very probable that the fundamentals will turn out to include, among others, the very variables that figured in the simple monetary model we sketched in Section 5.1 – relative money stocks and income. Changes in the market’s anticipations with regard to these fundamentals will trigger reassessments of the future outlook for exchange rates and this, in turn, will cause instantaneous fluctuations in the demand for the different currencies and hence in their international values. So we may well have a situation where an event that is seen as foreshadowing a change in the domestic money stock for example, brings about an immediate adjustment in the interest rate and a simultaneous movement in the exchange rate.
This type of ‘bootstrap’ situation is one that will be familiar to anyone who has ever observed the behaviour of a typical financial market when the price of an asset, whether foreign currency, share or bond or commodity, is expected to rise at some point in the future, the rise occurs not in the future, but right now. The very fact it is expected to rise pushes it up immediately.
This type of feedback mechanism will be examined at length later in the book (see Chapter 13). For the moment, all we can do is to note that the association between interest rates and exchange rates is complicated by the fact that, unlike money and income, interest rates cannot seriously be regarded as exogenous variables. Our con- clusions in this case, therefore, ought to be viewed as provisional.
5.4
The monetary model as an explanation of the facts
Before going on to look at the question of how well the monetary model explains the facts, it is worth pausing for a moment to try to guess the answer, because we already have a clue from Section 2.5. Remember that we concluded then that, for all the plausibility of the PPP hypothesis, it came nowhere near to explaining the wide fluctuations in exchange rates in the last decade or so. Since the PPP hypothesis is one of the two central building blocks of the simple monetary model, need we bother to go any further? Surely, if the foundations are rotten, there is no need to go on and look at the rest of the edifice?
This is certainly a persuasive argument. At the very least, it is hard to be anything other than extremely pessimistic about the prospects for using the monetary model to explain the facts. Nonetheless, it is worth proceeding – just about. It is still possible (although unlikely) that the monetary model might work for the following reason.
Suppose the reason for the apparent failure of PPP to explain the facts is that the tests use the wrong price index. Suppose furthermore that the ‘true’ price index (the one that actually measures the prices that agents face) is unobservable. As long as that same unobservable price index figures in the demand for money, we could end up finding that the monetary model fitted quite well in spite of the failure of PPP.12
If this seems a forlorn hope, the reader can comfort himself with the thought that sooner or later, in a book of this nature, we would have to take a look at the broad outline of what has happened to relative money stocks and incomes in recent years. Now is as good a time as any.
Figures 5.8 to 5.10 illustrate graphically the recent history of the UK, West Germany and Japan in this regard. In each case, the variables plotted are domestic
M1 and GDP, relative to the USA, and the exchange rate, in home currency per
dollar. All have been scaled so that 1995 = 100.13
The first thing to notice is that all three exchange rates are considerably more volatile than money stocks and incomes and this is a conclusion that would have been even more forcefully illustrated in the monthly or weekly data, had we pre- sented them.
For the UK, for example, while relative money stock had a range of 55% and rel- ative GDP only 17% (as measured by maximum over minimum), the value of the pound fluctuated by more than 80%. For Germany and Japan, the disparity is even more marked. In both countries, relative money stocks varied by 50% and GDP by just over 20%, while exchange rates moved up and down across a range as wide as 140% in the case of the DM and just under 400% in the case of the yen. If anything, this understates the extent of the disproportion, which would be even more obvious if we examined monthly, weekly or daily data. At these higher frequencies, we would find the comparatively slow, trend changes in income, and even the somewhat more volatile money stock statistics, completely swamped by the sharp day-to-day, week- to-week movements in exchange rates.
166 Chapter 5 · Flexible prices: the monetary model
5.4 The monetary model as an explanation of the facts 167 250 230 210 190 170 150 130 110 90 70 50 1975 Year German M1/USM1 German GDP/US GDP DM per $ 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 500 450 400 350 300 250 200 150 100 50 1975 Year Japanese M1/USM1 Japanese GDP/US GDP Yen per $ 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
Figure 5.10 Japanese money, income and exchange rate, 1975 –99 (1995 = 100) Figure 5.9 German money, income and exchange rate, 1975 –99 (1995 = 100)
Next, look at the trends in the data (where there are any). There was a spectacu- lar rise in UK M1 relative to that of the USA in the late 1970s, followed by a long fall through most of the Thatcher years of the 1980s, yet in those 15 years the exchange rate went through two distinct cycles for which there is no apparent ex- planation here. More recently, the UK money stock has again risen sharply, yet the exchange rate against the dollar has been more stable than at any time since the start of the floating rate era in 1972.
Japan and Germany are in some respects even more problematic. Both the yen and the DM enjoyed a long-term appreciation throughout the first 20 years of the period, broken only by the dollar’s rise in the early 1980s. Yet neither currency’s strength was justified by monetary or real factors. In fact the changes in the two countries’ relative income were so small as to be within the likely margin of error for aggregate data like these, while monetary growth differentials could justify appreciation of no more than 30% for Japan and 20% for Germany at the absolute maximum.
The evidence presented here is not, on the whole, very encouraging. All that can be said in favour of the monetary model (and it is not something to be dismissed altogether) is that it does appear to explain the facts over the very long run, at least in broad outline. This is particularly true of fixed rate periods.
For example, until the 1980s, the West German economy consistently enjoyed more rapid growth than either the UK or USA. It also stuck to a monetary policy that was markedly less expansionary than most other industrialized countries. As pre- dicted by the models we have examined in this chapter, the outcome was a balance of payments surplus and reserve accumulation, as well as long-term appreciation of the Deutschmark, whether in forced realignments, when a fixed rate regime was in force, or more gradually, when floating.
Conversely, for most of the post-World War II period until the late 1970s, the UK suffered from relatively slow economic growth, while its policymakers permitted a rapid expansion in the UK’s money stock – or at least in its domestic credit com- ponent. Again, the result was as predicted – declining reserves and a depreciating currency.
At the same time, Japanese growth averaged several percentage points more than that of the UK or USA, while its money supply expanded at a moderate rate, caus- ing mounting balance of payments surpluses and an appreciating international value for the yen.
Perhaps unsurprisingly, the more marked are the divergences between rates of monetary growth (and hence inflation), the better the monetary model fits the facts. If we were to examine a currency under conditions of hyperinflation (inflation at rates of, say, 100% per annum and above), we would find its rate of depreciation against any of the hard currencies more or less equal to the difference in rates of monetary growth. However, as was noted at the end of Chapter 2, this is simply a reflection of the fact that in these pathological cases PPP has to be broadly main- tained, at least to within limits that are narrow compared to the gap between inflation rates.
However, economists are not in the habit of drawing conclusions on the basis of this kind of cursory examination of a small amount of data. There is a wealth of additional evidence that could be used to cast light on the viability of the monetary 168 Chapter 5 · Flexible prices: the monetary model
Summary 169
model as an explanation of exchange rate changes and there are sophisticated statis- tical and econometric techniques that could be brought to bear on the detailed data. Researchers have, in fact, left no stone unturned in searching for evidence on the validity or otherwise of the monetary model. They have examined annual, quarterly and monthly data.14They have looked at trade-weighted as well as bilateral exchange
rates. They have used narrow measures of money (M0 and M1) and broad measures (M3 and so on). They have extended and complicated the simple model to take account of adjustment delays (‘lags’) of all types and they have used state-of-the-art econometric techniques to test the theory.
By and large, their conclusions have been similar to ours. For example, in a series of influential publications, Meese and Rogoff found little support for the monetary model15in the period since 1973. Studies of individual exchange rates, particularly
the pound/dollar and the Deutschmark/dollar, but also of a number of other hard and soft currencies lead to similar conclusions.
Even where a clear pattern associating monetary growth with depreciation can be found, the relationship is rarely anything like the strict proportionality predicted by Proposition 5.1. It is often even harder to find a link between a country’s national income and the value of its currency (Proposition 5.2), although in some cases this may be the result of the use of inadequate proxy measures in place of GDP.
Worst of all, such relationships that do appear tend to be unstable, with appar- ently satisfactory results over one period either reversing themselves or disappearing altogether in the next period.
5.5
Conclusions
The monetary model combines the quantity theory of the demand for money with purchasing power parity to generate unambiguous conclusions about the effect of changes in exogenous variables on a floating exchange rate, or on the balance of payments, as the case may be. It remains an important benchmark with which to compare other models – not least because as we shall see in Chapters 6 –9 its pre- dictions accord in most cases with their long-run equilibrium results. Moreover, the analysis in Chapters 13, 16 and 17 takes the monetary model as its starting point.
Unfortunately, as an explanation of the facts, the monetary model must be regarded as grossly inadequate in anything but the very long run, which is hardly sur- prising given the failure of PPP. If a model that assumes perfect flexibility of prices cannot explain the facts, we direct our attention in Chapter 6 to one that starts from the assertion that the price level is constant.
Summary
n The monetary model of a floating exchange rate predicts that the domestic cur- rency will depreciate when any of the following occurs:
– the domestic (foreign country) money stock increases (decreases) – domestic (foreign) national income falls (rises)
The depreciation will be proportionate to any increase in the relative money stock.
n The monetary model of a fixed exchange rate predicts that the balance of pay- ments will deteriorate and the home country will lose reserves when any of the following occurs:
– the home country’s domestically generated money stock (that is, domestic credit) increases
– domestic (foreign) national income falls (rises) – the foreign price level falls.
n Devaluation of a fixed exchange rate results in a period of balance of payments surpluses, which comes to an end when the home country’s reserves have risen sufficiently to restore the value of the real money stock to its pre-devaluation level. Once this has happened, the real exchange rate is back at its former level (that is, PPP is reinstated) and the temporary competitive advantage enjoyed by the home country as a result of the devaluation has been completely eliminated by inflation. n Since higher interest rates mean a smaller demand for money, other things being equal, they also imply a higher price level and therefore a lower value for the home country’s currency. However, this conclusion must be qualified by our finding in Chapter 3 that interest rates, far from being an exogenous variable, are likely to reflect the market’s expected rate of currency depreciation or appreciation. A firmer conclusion must await the analysis of how market expectations are deter- mined in Chapters 12 and 13 of this book.
n Since PPP does not fit the facts of the 1970s and 1980s very well it is not surpris- ing that the monetary model turns out to be able to explain the facts only very weakly and even then only over the long run. A cursory examination of annual data for the UK, West Germany and Japan since 1975 is not encouraging and this is broadly the conclusion reached by most researchers in detailed scientific studies of the data.
Reading guide
On the monetary model, the starting point (well worth reading) is Hume (1741). The simplest modern statement of its basic propositions is in Johnson (1977). Other influential work was by Mussa (1976) and Bilson (1979). The analysis of devaluation is due to Dornbusch (1973).
Influential empirical work has been carried out by Frenkel (1976) and Frankel (1979), among others. The last (empirical) word on the monetary model appears to have been said by Meese and Rogoff (1983), which also contains a useful bibliography.
On exchange rates under hyperinflationary conditions, see Frenkel (1977). Web page: www.pearsoned.co.uk /copeland.
Notes
1 In the language of microeconomics, the aggregate demand curve has unitary elasticity in this special case. It is in fact a rectangular hyperbola.
Notes 171
Note that in the present case, the LM curve is vertical, since the demand for money is unaffected by the interest rate. It follows that we are only concerned with the extent to which it shifts horizontally when the real money stock changes. We are also able to ignore the interest rate because the IS curve is implicitly flat, with saving and investment being equated by the interest rate alone, independently of income.
2 Of course, in this simple model the inflation rate is the only variable left for the government of either country to choose.
3 We ignore another complication here which is that, in practice, countries running so-called fixed exchange rates invariably allow some measure of freedom for the rate to move up or down from its central parity (see Section 1.2). The consequences of this will be the subject of Chapter 16. 4 The monetary model is normally seen as dealing with the balance of payments rather than the
balance of trade alone. In the present context, the distinction hardly matters.
5 The issue is clouded by all kinds of other problems, for example the apparent instability of the demand for money in the UK and USA, interest differentials (see Section 5.1.3) and risk pre- miums (Chapter 14), and many others.
6 The sterilization mechanism is also likely to prove self-defeating because of its impact on interest rates.
7 It is an awkward question, of course, to decide how small is small enough for present purposes. On the one hand, the USA, the EU or Japan are obviously far too large to be able to ignore the feedback effects of their policies on the world economy. On the other hand, the vast majority of countries in the world are clearly small relative to world money markets. But what about medium sized economies such as the UK, Canada, for example? The answer is not obvious.
Those readers who are familiar with elementary price theory will recognize a similarity here with the simple model of the perfectly competitive firm. Remember the paradox there was that market price was unaffected by the typical small firm’s output decisions. Although the perfectly competitive firm acting alone cannot influence market price, the small firms in aggregate deter- mine the price by their independently taken decisions. By the same token, we assume here the small country cannot on its own have an impact on the world economy, but the monetary policy of small countries in aggregate (together with the large countries) determines the world price level. 8 We ignore the second-order effect on the money stock of the rise in the sterling value of the UK
reserves when the pound is devalued.
9 In fact, it dates back to arguments about whether devaluation could improve the current account even in the long run. Thus, economists afflicted by ‘elasticity pessimism’ in the 1950s believed that the J-curve was all tail and no upswing.
10 In IS–LM terms, the upward shift in the horizontal IS curve cuts the unchanged LM curve fur- ther to the northeast.
11 A formal demonstration of this fact would be a little messier and more complicated than was the case in Section 5.6 with relative money stocks and incomes. The reason is to be found in our formulation of the demand for money that is linear in natural numbers. Fortunately, there is more or less universal support among researchers for a log linear formulation. In this case, it turns out that exchange rates depend on relative values of all variables, provided that income and interest elasticities are the same in each each country.
12 The possibility referred to here would also imply that demand for money studies using the standard price index were misspecified and should be expected to fit poorly. In fact, they fit reasonably well – certainly far better than PPP models (see the reading guide).
13 The pictures would look very similar if we defined money as M3 (M1 plus time deposits) and, although recalculating on a different base year would change the pictures superficially, it would not alter the general nature of our conclusions. There are, in any case, no sound reasons for preferring one definition of money to another in the present context or any overwhelming arguments in favour of any particular base year.
14 Higher frequency data exist only for the US money stock and for financial variables (exchange