This is a revised version of a paper published initially in an article in a German journal called Kredit und Kapital, 1976 following the 1975 Konstanz Conference. In 1999 it was republished in Laidler, David (Ed), The Foundations of Monetary Economics, Volume II, An Elgar Reference Collection, 1999. Footnotes and diagrams have been omitted and minor changes made solely to improve the flow.
This essay aims to put some recent developments in monetary economics into perspective by drawing our what we have learnt - or relearnt - from the literature about the structure of the economic system. Following a very brief historical overview the methodology of much recent analysis is illustrated by presenting a very simple one good, one asset model designed to illustrate one or two important relationships. The way in which the simple model provides suggestions about relationships in larger, more complicated, and presumably more realistic models is shown. The analysis here represents a picture of my state of mind at about the time my PhD was being finalised.
As noted here, the basic insights about the role of money in the economic system as given by this class of model has been tested, and appear to be quite robust. Rigorous testing of economic theory of the sort discussed is a time consuming and expensive business, however, and this essay discusses an alternative way in which the basic insights of modern monetary theory can be applied. The two good, one asset model used in quite a bit of recent writing is introduced, and used to analyze an important policy problem, the shifting trade-off between inflation and unemployment.
Recent developments in the monetary economics of the open economy have been initiated largely by H.G. Johnson and R.A. Mundell, but they are harking back to a much older tradition exemplified in the writing of David Hume. Hume stressed the tendency for the balance of payments to vary to equalise 'somehow' the supply of money with the demand for money. Hume was agnostic about the possibility of understanding in detail the 'hundred canals' whereby excess money flowed out of small open economies, but was not agnostic about whether this would occur. An analogy can be drawn with the proposition that, in a closed economy, doubling the money stock will, eventually and approximately, double the price level. Prominent modern monetary economists, notably Milton Friedman, appear similarly agnostic about understanding in detail how the process occurs.
Hume's analysis also includes the effect of an increase in domestic activity and prices in closed economies, and noted that these effects only occurred until the adjustment process had settled. The causation in a small open economy ran from increased money to domestic output, to domestic prices and raising imports, reducing exports and reducing a nation's reserves. This line of thought eventually leads toward 'a global world of one price', still only a tendency but a strong tendency among modern trading nations.
As recently as the 'Treatise on Money' this view was standard amongst economists. Keynes included in the Treatise a lengthy discussion of the effects of Spanish treasure on European prices, acting like a profit inflation and taking a long time to work its way through system, and his other historical examples illustrate the point equally clearly.
When the Depression of the 1930s shifted the emphasis to quantity adjustment, inflation theory lost favour and models of balance of payment adjustment downplayed price affects and substituted relatively mechanical multiplier analysis. This change in theoretical emphasis occurred at the time when macroeconometric model building became feasible, and has had a and long lasting influence on the large models used for forecasting and policy analysis.
In the British case, to take as an influential example, beliefs about zero or low price elasticities in international trade and payments were invoked to explain the continued balance of payment deficit following the devaluation of the British pound in 1967, and served to distract attention from the enormous domestic monetary expansion at the same time. Ironically, when the International Monetary Fund insisted on quite severe monetary contraction in 1969 as a condition of further assistance, British economists asserted that the system was at last moving up a postulated 'J-curve' in the balance of payments, i.e. finally responding to the devaluation induced change in relative prices. I wrote an article headed 'How the British invented the J-curve', which naturally failed to find a journal.
The post-war resurgence of inflation of course led to renewed interest in monetary economics. In the open economy case the modern contribution has stressed direct world price effects far more than Hume and his successors. For example, modern monetary models focus on the tendency for prices to move together throughout the world without the world without necessarily involving trade or capital flows. Direct price effects (including a tendency for interest rates to equalise) lead to a view of the world with more rapid adjustment of prices envisaged by Hume and earlier writers. Much of the initial work in this area is presented in Johnson and Frenkel (1975), and consists of simple theoretical exercises and preliminary empirical tests of the basic propositions suggested by the theories.
One important contribution is Johnson's widely cited article 'The Monetary approach to Balance of Payments Theory' (1972) which focuses on the long run in which domestic inflation rates equal the world inflation rate, domestic inflation rates equal world interest rates and real income can be assumed to be given, and in this model an exogenous increase in domestic credit leads to an offsetting decrease in international reserves. (As someone once said, what is obvious often depends on how fast you think.) Such models ignore the short run dynamic responses of prices and incomes which are of vital concern to the policymaker, and provide no reconciliation with the standard payments or inflation theory. Some of these gaps are now being filled, however, especially with respect to the short run domestic effects of monetary disturbances, and this emphasis becomes especially important when exchange rates are more flexible.
Jonson and Kierzkowski (1975), Laidler (1975), Mussa (1974) and Parkin (1972 and 1974b) emphasise respectively the effect of disequilibrium in the money markets on goods markets, the interrelationships of price and quantity adjustment, interest rate effects and the dynamic effects of differing rates of domestic credit and unequal rates of growth in different in different sectors in models with traded and non-traded goods. Much of this work is surveyed and extended by H.G. Johnson (1975), who introduces non-monetary assets ('bonds') and further disequilibrium effects. These modifications reintroduce domestic effects of monetary disequilibrium and help to reconcile new effects with existing views of adjustment effects. Even in the fixed exchange case induced price and output effects tend to soften the effect of monetary disequilibrium on the balance of payments, at least in the short run.
Standard neo-Keynesian econometric models have incorporated monetary effects operating through interest rate changes, wealth effects on consumptions and 'availability' or credit rationing effects. Significent interest rate effects on investment are fairly well accepted, but appear to operate with fairly long lags. Wealth effects are usually empirically minor and somewhat suspect theoretically , with a high proportion of money in modern economic systems consisting of inside or interest bearing money. (Similarly the wealth embodied in government bonds may be offset by future tax liabilities.) Availability effects also seem theoretically and empirically unimportant.
A possibly more important channel for the influence of monetary disturbances is what should perhaps be called 'the disequilibrium real balance effect' recently emphasised in balance of payments effect. David Hume's brilliant analysis was resurrected by Archibald and Lipsey (1958).
This is a model with one asset (money) and one good. It shows the choices made by an individual who cannot influence goods prices or a small open economy which cannot influence prices in the world economy. In this model the consumer uses his stock of money to smooth the path of consumption over time, for example when there is a temporary drop in his real income, due for example to a crop failure.
Gradually, when income resumes its normal level, the money stock is restored to its original level. This example illustrates in the simplest possible model the role of money as a buffer stock in an uncertain world, not unrelated to the traditional concept of the precautionary motive for holding money.
Variants of this model allow for permanent drops of income, in which the original income drops and does not recover. Income staying down signals the need to reduce money and will end with both consumption and the money stock reaching a new, lower level.
This point can be generalised to a model in which both prices and quantities are set by individuals with important monopoly powers in the short run. In such models, which following Arrow's (1959) seminal contribution many people see as capturing a crucial feature of the economic system, prices and quantities are influenced importantly in the short run by their past history through expectation-generating mechanisms. Accumulation and decumulation of money balances will signal the need to adjust expectations and therefore the prices and quantities proximately controlled by the economic agent. This point is emphasised by Laidler (1974).
An essential feature of this type of analysis is its attempts to come to grips with disequilibrium, although the word disequilibrium should not be read as implying that economic agents are off profit or welfare maximising paths. Rather, it is assumed that economic agents are constrained by available information and in particular that it may well be optimal to allow money balances to diverge from the levels that would be held if the system was in a full long-term equilibrium with all variables at their steady state levels.
If adjustment proceeds in the way described here, the gap between desired and actual money balances will occur in equations representing decisions about other economic variables, and the resulting decisions to markets influenced by monetary disequilibrium will themselves tend to indirectly eliminate the monetary imbalance.
The resulting dynamic analysis is in strong contrast to much standard monetarist analysis which assumes that markets - and in particular the money market - clear very rapidly. It is also in contrast to much conventional analysis which merely assumes that partial adjustment mechanisms are persuasive and adds the lagged dependent variable to every behaviourial equation.
Empirically, there is considerable support for the influence of adaptive expectations of various types on price adjustment, but much less work has been done exploring the possible effects of monetary disequilibrium on price and quantity adjustment. Forthcoming empirical results provide considerable support for the direct role of 'monetary disequilibrium' , and in view of its unfamiliarity it is worth discussing these results briefly here.
Before going to results, however, I note that much of this work has been done at the London School of Economics and more recently at the Reserve Bank of Australia and the International Monetary Fund. Distinguising features of the powerful econometric systems developed by C.R. Wymer are that hypotheses are tested in a manner that allows for all of the simultaneous interactions in a structural macroeconomic model, with due allowance for the fact that economic data is discrete whereas most economic models are specified to be continuous. Long term steady state properties are imposed on the model before its parameters are estimated.
In estimation Jonson (1975) found a powerful and significent disequilibrium effect on aggregate British expenditure over the past century. At about the same time, Bergstrom and Wymer (1975) found a similar effect on bond yields in a quarterly model of the UK economy. With similar post war quarterly data, Knight and Wymer (1975) found a similar effect with Canadian data, and Jonson, Moses and Wymer found an equally strong result in a model with quarterly data for Australia. In addition, introducing monetary disequilibrium into the price equation in the Australian model created another strong result.
Our interpretation of this result was as follows. Use of a person's or a company's monetary account as a buffer stock was a well established result by then, and in our view using the same vector in a goods price equation made sense as a signalling device. This result has now been found in many similar models. Jonson's 1975 model, looked at in retrospect, contained a weaker version of the inflation effect, perhaps unsurprising given that the UK economy from 1880 to 1970 was largely on the gold standard.
Leaping ahead to 2020, my work with Clifford Wymer with British data from 1855 to 2014, with data mainly provided by the Bank of England, showed a significant inflation effect. In this model we were brave enough to include an equation for share prices and its parameter on monetary disequilibrium is far larger than that on goods inflation. This work will be demonstrated in the last chapter in this book, and holds great relevance for the tools of monetary policy.
Returning to the mid 1970s, the parameter on the goods price equation added materially to the inflation effect on the overall Australian model. The result suggested that the direct partial effect of 10% monetary disequilibrium raised the annual inflation by 2 % - ie from 2% to 4%. There were many effects which tended to increase this effect and feedback effects which tend to decrease it, but the direct monetary effect overall produces a stronger and faster effect on inflation than in the model without it.
The simple one good, one asset model (with appropriate steady state values imposed) means that monetary policy that raise inflation will eventually produce the standard 'monetarist' result of increased inflation. The simplest models however do not throw much light on one of the model of the 1970s, the so-called Phillips curve. The simultaneous rise in both inflation and unemployment was the issue of the day.
Discussion by the water cooler looked at a variety of factors, including: increased uncertainty associated by higher and more variable goods inflation reducing business investment; a tendency for increased wages to rise faster than prices plus productivity, and the two oil price hikes. Our first attempt to model these effects involved starting with a two product one asset model. To add interest and plausibility, the two products were traded goods and non-traded goods, and the asset was, you guessed it, money.
Fellow graduate student Henryk Kierzkowski, a fine trade theorist, helped me to grope my way through the two good, one asset model. We had much fun with curved trade-off frontiers, shifting trade off functions, usually lower, and a more interesting set of monetary effects. We assumed that government decided to raise the proportion of non-traded goods paid for by raising money supply. The rise in spending on non-traded goods raised their price, meaning there would be a higher price level in total, as prices of traded goods are assumed to be fixed by international markets. The rise in the price level reduces the real value of money, thus reducing ('crowding out') private demands for both traded and non-traded goods. A greater trade deficit means a further reduction in money supply. If standard stability conditions apply, the end result is a solution in which the balance of payments deficit equals the budget deficit and in which output has returned to a full employment level on the original transformation curve. This cannot be a steady state solution because the continuing budget deficit means that the country will eventually runs out of reserves. At some stage the government will either have to devalue the currency or cut its spending. In the latter case the process discussed will leave the economy at its original steady state results.
The key point is this. The result of this scenario is a higher rate of inflation and lower output for the entire for the entire period that the economy is in disequilibrium, which sounds like a self inflicted Phillips curve wound to me. If one adds the more general disasters listed above - increased uncertainty associated by higher and more variable goods inflation reducing business investment; a tendency for increased wages to rise faster than prices plus productivity, and the two oil price hikes - it is easy to explain a shifting and less well-behaved Phillips curve.
Working with Henryk the brilliant trade theorist made my head hurt, and I must confess I decided to use the econometric model for real policy analysis. As I shall report later, I eventually used the quarterly Australian model for some real analysis, and the result was effective for policy and also effective for ending my career as a economic policy man.
I finished my 1975 article with two final points. The first was about whether it was a 'generalised asset effect' or a 'monetary effect' that was the important force effecting consumption, bond rates and goods inflation.
Detailed asset inflation models seem to work well but 'have been singularly unproductive in throwing light on the dynamics of inflation and output adjustment'. Even if one is committed to a multi-asset view of the world, monetary disequilibrium is a good proxy for asset disequilibrium if the composition of portfolios adjusts relatively quickly. While fast portfolio adjustment may have been the case in the period to 1939, post WWII direct controls of various sorts have been endemic.
This point aside, the strongest argument for the importance of money comes from insights of its role as a buffer stock and signalling advice. Put simply, life is complicated and money is easy to use. My hypothesis is that people struggle to rebalance their portfolios, pay for the food and other daily calls on their wealth with money and occasionally check the state of their bank deposits, and adjust spending on other things if they do not like the result of checking.
I then finished by gnawing away at an issue that almost always had bothered me. How to reconcile the simply 'monetarist' approach to economics that created an almost infinite number of 'demand for money' equations.
Typically DM = a( Md - M/p) or an even simpler formulation of infinitely fast adjustment DM = D(Md/p).
Clearly 'monetary disequilibrium' is in dramatic contrast to such a model. It just seems obvious to me that monetary disequilibrium makes for sense than money does its simple adjustment approach, either with lags or instantaneously. Having obtained a 'reasonable' estimate of money demand (Md) with such an approach what is to be done with it? Add it to the demand function for consumption, and if so, in relation to what benchmark? Logic says to the actual money demand relative to money supply, and we are back to Hume's intuitive solution. Add a supply function for money, and you are on the way to a viable model.
I finished this article feeling I had sorted out a great weakness in standard economic models. The approach that featured monetary disequilibrium creates a telling economic force, handles well the effects of monetary policy in allowing monetary supply to be strapped into the story and the various estimates of the effects of excess or deficient money tells us a lot about the channels (Hume's 'canals') whether is a fixed or floating exchange rate economy we are analysing.
It was time to put this theory into action.