Conventional economic wisdom is that, at least in the short and intermediate runs, exchange rates behave as random walks.2 Does this mean that exchange rates are unrelated to fundamentals such as current account deficits and fiscal or monetary policies? Most economists would argue that, although the bulk of empirical work supports the view that the exchange rate (ER) appears to behave as a random walk in the short and intermediate runs, it is related to fundamentals in the long run.3 Does this mean that predictions about fundamentals are irrelevant for predicting the behaviour of ER in the short and intermediate runs? I believe this is not necessarily the case. There exist time periods during which the impact of policy or other fundamentals is sufficiently strong to produce discernible impacts on the ER even in the short and intermediate runs.
This claim is illustrated here by examining the behaviour of the Euro-Dollar (EUR/$) rate between summer 2006 and summer 2008 against the background of the divergent interest rate policies of the Federal Reserve (Fed) and of the European Central Bank (ECB). Relative magnitudes of current account positions and of the stances of monetary policies are both fundamentals that, as a matter of theory, are expected to affect the ER. But they impact it with markedly different lags. Current account deficits normally lead to depreciation of the home currency only if they persist for sufficiently long periods of time. By contrast, as any trader in foreign exchange markets knows, a change in the central bank (CB) policy rate or an anticipation of such a change impact the exchange rate immediately.
2. The impact of policy rates on the exchange rate – some underlying conceptual considerations
Other things the same a decrease in the Fed’s target rate raises the demand for Euros as international investors rush to take advantage of the relatively better yields in the Euro area. This induces a “carry trade” in favour of the Euro and tends to push the EUR/$ rate up, or equivalently, to cause a depreciation of the $.4 By the same logic, an increase in the Fed’s target rate is expected to push the EUR/$ rate down (an appreciation of the $). The well known uncovered interest rate parity (IRP) condition summarizes those factors, by stating that the interest rate differential between the US and the Euro area has to equal the expected rate of depreciation of the $.
Since it abstracts from risk premia and from differences in preferences for $ and for Euro denominated short term assets one cannot expect the IRP condition to hold exactly. It is, nonetheless, a useful benchmark since it implies that an increase in the spread between the Fed’s and the ECB policy rates should be matched, to some extent, by a decrease in the expected rate of depreciation of the $ (or equivalently, by an increase in its expected rate of appreciation). Given a relatively immobile longer term expected EUR/$ rate this implies that the actual dollar price of Euros goes down, (the EUR/$ rate goes down) when the spread goes up.5 Those considerations imply that a ceteris paribus increase in the Fed’s discount rate should trigger an appreciation of the $.
An additional short run complication is that the process described above already takes place a few days before an actual change in a policy rate is implemented or is formally announced. There are at least two reasons for this. First, both the ECB and the Fed try to be transparent about the upcoming interest rate decision in their communications with the public. Between interest rate decisions and up to a week before an actual decision the Fed and the ECB send signals to the financial community about the likely upcoming choice of target rate through board member speeches and other devices. Although imperfect , those signals do carry relevant information about the upcoming interest rate choice, and Fed and ECB watchers pay attention to them.6 Second, participants in financial markets have a broad understanding of the objectives of both central banks and use forecasts about the upcoming state of the economy and of inflation to form informed guesses about upcoming monetary policy decisions.
Two important consequence of the immediately preceding discussion are that: 1. Upcoming changes in policy rates are partly anticipated in advance, 2. The spot rate changes at the moment the news about an upcoming policy change hit the market.
3. Evidence on the the Euro-dollar rate and the spread between the policy rates during the two years ending in June 2008
During the first third of May 2006 the Euro- Dollar ER hovered around 1.28 $ per one Euro. At that time the spread between the Fed’s discount rate and the ECB target rate for the minimum bid rate on the main refinancing operations was 2.5 percent. Two years later, at the beginning of June 2008, the Euro- Dollar rate had climbed to around 1.58 while the spread had decreased to minus 2,5 percent.
This dramatic change in the spread reflects the divergent policies of the Fed and of the ECB. During this period the Fed’s discount rate first climbed from 5 percent to 5.25 percent and then stayed at this level between June 2006 and September 2007. From that point and on, the discount rate was reduced in a series of variable steps until it reached a floor of two percent at the April 30 2008 meeting of the FOMC. By contrast, over those same two years, the target rate of the ECB experienced a gradual and monotonic upward trend that reached a 4.25 level at the July 3 2008 meeting of the Governing Council. In parallel the Euro appreciated vis a vis the Dollar from less than 1.30 to a bit less than 1.60 (an appreciation of over 20 percent).
Figure 1 plots weekly averages of the Euro- Dollar rate in the weeks just preceding a change in either the Fed’s or the ECB policy rate against the spread that became effective in the subsequent week as a consequence of the rate change.7 The reason for matching the spread that becomes effective in a given week with the average Euro-Dollar rate of the previous week is that the change in spread is normally partially anticipated a week or two in advance and the spot rate adjusts as soon as the information about the upcoming change hits the market.
A change in policy rates is usually anticipated in advance for two reasons. First, participants in financial markets have a broad understanding of CB objectives concerning inflation and economic activity and use this information in conjunction with current forecasts of those variables to form educated guesses about the outcomes of upcoming policy meetings. Second, during the inter-meetings period and up to a week prior to the meeting, Board or Council members emit noisy but informative public signals about the likely outcome of the upcoming interest rate decision.
Figure 1 shows that there is a clear negative relation between the Euro-Dollar spot rate and the spread in policy rates between the Dollar and the Euro. As this spread gradually decreased over the two years period between summer 2006 and summer 2008, the Dollar depreciated against the Euro by over twenty percent. Regression analysis confirms that this negative relation is significant even when one controls for other potential factors by including a time trend. In particular, a regression of the logarithm of the Euro-Dollar rate on a monthly time trend and the spread yields significant coefficients for both of those regressors and an adjusted R-squared of about 0.96.8 The coefficient on the spread between the policy rates is significant at the 0.001 level.
From the beginning to the end of the period the Dollar depreciated by about 30 cents per one Euro (from around 1.28 $ to about 1.58 $ for one Euro). The regression implies that the change in the spread caused about half of this depreciation and that factors captured by the time trend were responsible for about one sixth.9 Those results are consistent with the view that, during the period under consideration, changes in the target policy rates of the Fed and of the ECB exerted non-negligible effects on the spot Euro-Dollar rate and that actual changes in target policy rates affected the spot rate about a week or two in advance.
An additional factor not captured by the regression includes verbal allusions to coordinated interventions. Given the spread such allusions, particularly when they are issued by a group of countries, have the potential to move the Euro-Dollar rate for a few days and even more. A case in point is the discussion of possible intervention at the meeting of the G7 in Japan a month or two ago.
Can those results be extended to other periods? Obviously, the Euro-Dollar rate is impacted by a variety of additional factors like the price of oil, the current account balance, fiscal policy, relative productivity shocks and relative changes in market structure. However, except possibly for the price of oil, most of those operate mainly in the long run with substantial lags.10 By contrast, the evidence presented here suggests that the choices of target policy rates by central banks are likely to be important determinants of the short and intermediate term behaviour of the ER. Since they operate along with additional factors they are not always as strongly in evidence as in the period considered here. But this evidence suffices, in my view, to support the more general conclusion that changes in expectations about the behaviour of relative policy rates are important determinants of the spot rate in the short and intermediate terms.
4. The impacts of asynchronization of business cycles and of the different institutional structures of the Fed and of the ECB on the behaviour of the Euro-Dollar rate
Having established that changes in the actual and anticipated spreads between policy rates have a systematic impact on the Euro-Dollar rate it is of interest to probe what are the more basic determinants of this spread. Two classes of factors are important in this context. The first includes the expected relative positions of the US and of the Euro area within their respective business cycles and their expected relative rates of inflation. The second relates to differences in the institutional organization and the mandates of the Fed and of the ECB.
The Fed has a dual mandate — to maintain a high level of economic activity and to maintain price stability. The ECB has a single mandate – to maintain price stability. In practice, both institutions try to stabilize both inflation and economic activity. As a consequence their policy rate choices react to forecasts of both inflation and of the output gap.11 Hence, the spread is partly determined by differences in forecasts of inflation and of economic activity between the US and the Euro area. Thus, when policymakers at the Fed expect a recession in the US while policymakers at the ECB have no such expectation for the Euro area, the interest rate spread between the US and the Euro policy rates is likely to go down.
Such a process was actually in evidence since September 2007 following the eruption of the subprime crisis in the US. Although the crisis also spread internationally it impacted the US economy for several quarters before affecting real growth and exports in the Euro area. As illustrated by Figure 1, since then the spread indeed went down substantially (from 1.25 in June 2007 to minus 2.25 in July 2008).
Although both the Fed and the ECB are flexible inflation targeters it is quite likely that the ECB puts a higher relative weight on the stabilization of inflation than the Fed for several reasons. First, unlike the Fed its formal mandate includes only price stability. Second it carries with it the conservative tradition of the Bundesbank. Finally it is formally responsible to the European parliament rather than to a particular government in the Euro area.
By contrast the Fed is formally responsible to Congress and relatively more susceptible to direct political influence on the part of the executive branch. Hence, given a structure of shocks that leads to similar states of their respective economies the ECB is likely to respond more strongly to inflation threaths than the Fed. This has obvious implications for the future behaviour of the spread, and through it on the Euro-Dollar rate. This consideration will be particularly important in case the increases in the price of oil and of primary commodities persist into the future.
5. Epilogue: The impact of the early August monetary policy actions and pronouncements of the Fed and of the ECB on expectations and the spot rate
On August 5 the FOMC decided to leave the discount rate unchanged at 2 percent. Two days later (on August 7) the ECB followed a similar strategy and left its target rate unchanged at 4.25 percent – so the spread did not change. In spite of that, the $ appreciated by a huge 4.5 cents (about 3 percent) against the Euro from just prior to the Fed’s decision on Tuesday till the market close on Friday, August 8 2008. Without additional information this may appear like a puzzle.
But this seeming puzzle is largely resolved when one recognizes that forex markets move whenever information that had not been available prior to the two central bank decisions becomes available. Since neither the spread, nor the two rates changed one may wonder what this new information was and how it reached the markets.
The answer is that the new relevant information was contained in the statement and press conferences accompanying the Fed’s and the ECB’s decisions in conjunction with freshly released economic data. In particular, the Fed’s statement and Trichet’s press conference led the financial community to believe that the Fed has become relatively more concerned with inflation and that, in parallel, the ECB has become relatively more concerned with economic activity and exports. Those changes in beliefs led the financial community to raise its subjective probability that, in the foreseeable future, the spread is headed downward in spite of the fact that current rates stayed put. This led to an immediate and substantial appreciation of the Dollar.
(1) Dan Zeltzer provided efficient research assistance.
(2) Meese R. and K. Rogoff (1983), “Empirical Exchange Rate Models of the Seventies: Do They Fit Out of Sample?”, Journal of International Economics, 14, pp. 3-24, April.
(3) An example is Rogoff K. (1999), “Monetary Models of Dollar/Yen/Euro Nominal Exchange Rates: Dead or Undead?”, The Economic Journal, 109, pp. F655-F659, November.
(4) Here the EUR/$ rate is measured as the number of $ per one Euro.
(5) In conventional formulations of the IRP condition the two interest rates and the expected rate of depreciation (or appreciation) are all determined simultaneously as functions of exogenous variables. However, since both the Fed and the ECB are flexible inflation targeters that gear their policies mainly to attain appropriate combinations of inflation and of economic activity their policy rates are taken as exogenous to the EUR/$ rate.
(6) This view is supported by the fact that speeches by Fed Board members and ECB Council members move the EUR/$ rate at least for several days.
(7) For example, at the April 30 2008 meeting the FOMC reduced the discount rate from 2.25 to two percent while the policy rate of the ECB remained at 4.00 percent. As a consequence, following the April 30 meeting, the spread fell from minus 1.75 to minus 2.00 percent. The new, -2.00 percent spread, is matched with the average Euro-Dollar rate in the trading week ending on Friday of the previous week (April 25 2008).
(8) Letting t be time measured in months, the estimated regression is: lnEUR/$ = C + 0.033t – 0.0303 (SPREAD)
(9) The remainder is due to factors unexplained by the regression.
(10) The recent positive correlation between the Euro-Dollar rate and the price of oil led some observers to claim that there is reverse causality from depreciation of the $ to the price of oil. In my view, it is likely that causality operates in both directions. An important, yet unanswered, policy question concerns the relative magnitudes of the direct and reverse causalities.
(11) In short both are flexible inflation targeters.