The sterling-dollar exchange in the early 1920s
It is well recognised that anticipation of future economic policy changes can have a large impact on the pricing of long-lived assets such as foreign exchange. What is less clear is that incorporating such effects in an analysis can influence our thinking about widely studied episodes in economic history.
The behaviour of the sterling-dollar exchange rate in the early 1920s is a case in point. This period started with the lapse of the wartime peg between sterling and the dollar in March 1919, and ended with return of the sterling to gold which was announced by Winston Churchill, the Chancellor of the Exchequer, in April 1925. The literature on this period has developed into two strands.
One strand has focused on the behaviour of the nominal exchange rate, its determinants, and, in particular, whether it converged to purchasing power parity (PPP).1 Taylor (1992) summarises the findings in the literature and reconciles the contradictory results published earlier in favour of the long-run validity of PPP in the case of wholesale prices
This literature disregards the fact that the British government’s decision to return to the gold standard was at least partially expected, and that this could have impacted the exchange rate. Indeed, it has been taken for granted that the British government would aim to return to gold at the pre-war parity of 4.86 dollars to the pound at some date, although there was uncertainty over when this policy change might occur (Hodgson 1972, Moggridge 1972). As Churchill noted in announcing the policy change (as quoted in The Times on 29 April 1925):
“A return to an effective gold standard has long been the settled and declared policy of this country. Every expert Conference since the war – Brussels, Genoa – every expert Committee in this country has urged the principle of a return to the gold standard. No respectable authority has advocated any other standard. No British Government – and every party has held office – no political party, no previous holder of the office of Chancellor of the Exchequer has challenged, or, as far as I am aware, is now challenging the principle of a reversion to the gold standard in international matters at the earliest possible moment. It has always been taken as a matter of course that we should return to it, and the only question open has been the difficult and the very delicate question of how and when.”
The second strand of the literature recognises that a return to the gold standard was expected and studies the impact of such expectations on the market exchange rate in theoretical models of exchange rate dynamics (Smith and Smith 1990, and Miller and Sutherland 1994).
Surprisingly, no attempts to bridge these two literatures have been made. In a recent working paper, we take a first step towards removing this shortcoming by providing estimates of the probability of a return to gold and how it evolved over time (Gerlach and Kugler 2015).
We start by assuming that changes in the market exchange rate at any point in time are a weighted average of convergence to a fundamental exchange rate given by PPP and the pre-war parity of 4.86 dollars to the pound, with the weights being the likelihood that sterling will be fixed at the old parity. Since the market was below the old parity in the early 1920s, expectations of a return to gold would have appreciated the exchange rate. Estimates of the fundamental exchange rate, which disregard the possibility of a return to the gold standard, and of how the nominal exchange rate converged towards it, may therefore be biased.
We assume that the probability of a return to gold depends on a set of political and economic factors. As ‘domestic’ determinants we include the unemployment rate and dummy variables for conservative governments and strikes in Britain, respectively. The rationale for this choice of variables is that a return to gold would lead to an appreciation of sterling, which would raise unemployment and unsettle the labour market. The state of the labour market and the government’s willingness to endure a rise in unemployment, therefore, became crucial determinants of anticipations of a return to gold.
The ‘international’ explanatory variables include the absolute value of the short-term US-UK interest rate differential, relative US and UK industrial production, and base money. Relative industrial production reflects the business cycle position and the corresponding effects on exports, imports, and the trade balance. The relative monetary base reflects differences in the stance of monetary policy in the US and the UK. For both variables we expect a positive effect on the likelihood that the UK returns to gold given a boom and loose monetary policy in the US, which was on gold. The absolute difference of interest rates reflects a deviation from the gold standard situation of internationally approximately equalised interest rates for which we expect a negative effect on the probability of a return to gold.
Results from a new analysis
We first disregard expectations of a return to gold and test for cointegration between the nominal sterling-dollar exchange rate and the ratio of US to UK price levels. We find that the exchange rate is better explained by relative wholesale prices than relative consumer prices, and that in the latter case there is cointegratation, implying that the exchange rate moved toward PPP over time.
Next, we estimate the full model and draw a number of conclusions. Introducing expectations of a return to gold improves the fit of the exchange rate model, as evidenced by the fact that the explanatory power for changes in exchange rates rises from approximately 0.16 to slightly above 0.40. The estimated speed of adjustment to PPP rises from 13% to 38% per month as expectations of a return to gold are allowed for. Overall, expectations of a return to gold, although disregarded in the existing empirical literature, were an important determinant of exchange rate changes in the sample period we study.
Furthermore, the included determinants of the probability of a return to gold are statistically significant (except the UK unemployment rate) and have the expected sign. The effect of a conservative government on the probability of a return to gold is positive and the effect of strikes in the UK labour market is negative. A looser monetary policy in the US compared to in the UK is associated with an increase in the probability of a return to gold of the pound. The same pattern is found for a rise in relative US industrial production. Both effects are in line with our a priori expectations.
The probability of a return to gold
Of particular interest is the estimated probability of a return to gold. Figure 1 displays the median of, and 90% confidence bounds for, the probability, calculated using boot strapping. From 1920 to 1924 the probability of a return to gold is estimated to be relatively low, between 0.05 to 0.3, with peaks in early 1921 and 1923. During 1924 the probability increases strongly and reaches a peak of approximately 0.72 before slightly falling in early 1925.
Figure 1. Estimated expected probability of the return to the gold parity, dollar/pound, 1920/03-1925/4 (500 bootstrap replications)
Expectations of a return to gold were an important determinant of the sterling-dollar exchange rate in the early 1920. Incorporating them into the analysis suggests much more rapid convergence towards PPP. The probability of the sterling’s return to gold increased from around 15% to over 70% in the second half of 1924, a few months before Churchill announced it in April 1925.
Clements, K W and J A Frenkel (1980), “Exchange Rates, Money, and Relative Prices: the Dollar-Pound in the 1920s”, Journal of International Economics, 10, 249-262.
Gerlach, S and P Kugler (2015), “Back to Gold: Sterling in 1925”, CEPR Discussion Paper 10761.
Hodgson, J S (1972), “An Analysis of Floating Exchange Rates: the Dollar-Sterling Rate, 1919-1925”, Southern Economic Journal, 39, 249-257.
Jenkins, R (2001), Churchill: A Biography, Farrar, Straus and Giroux, New York.
Matthews, K G P (1986), “Was Sterling Overvalued in 1925?”, Economic History Review, 39, 572-587.
Miller, M and A Sutherland (1994), “Speculative Anticipations of Sterling’s Return to Gold: Was Keynes Right?”, The Economic Journal, 104, 804-812.
Moggridge, D E (1972), British Monetary Policy, 1924-1931: The Norman Conquest of $4.86, Cambridge University Press.
Redmond, J (1984), “The Sterling Overvaluation of 1925: A Multilateral Approach”, Economic History Review, New series, 37, 350-371.
Smith, G W and R T Smith (1990), “Stochastic Process Switching and the Return to Gold, 1925”, The Economic Journal, 100, 164-175.
Taylor, M P (1992), “Dollar Sterling Exchange Rate in the 1920s: purchasing power parity and the Norman Conquest of $4.86”, Applied Economics, 24, 803-811.
1 See Hodgson (1972), Clements and Frenkel (1980), Redmond (1984), Matthews (1986) and Taylor (1992).