INTRODUCING DEUTSCHE BANK'S
NEW REGIME SHIFT BAROMETER
Prepared by Michael R. Rosenberg
Deutsche Bank AG
Foreign Exchange Research Department
(January 17, 2003) There has been considerable discussion in recent days whether President Bush's newly announced fiscal stimulus proposal would lend support to the dollar in 2003. Those expecting the dollar to respond favorably point to the prospect of a fiscally induced rise in U.S. interest rates, which should help attract additional flows of overseas capital into the U.S. Those expecting the dollar to respond unfavorably point to the prospect of a fiscally induced rise in U.S. economic activity, which should contribute to a further widening of the already sizeable U.S. trade deficit.
According to the Mundell-Fleming model of exchange-rate determination, whether the dollar rises or falls will depend on whether the induced inflow of capital dominates the deterioration in trade or vice versa. The divergent paths that the dollar could take are depicted in Figure 1. As shown, what matters for the dollar is whether the overall U.S. balance of payments (net capital inflow minus the trade deficit) improves or deteriorates.
Figure 1
The Mundell-Fleming Model Transmission Mechanism
How Changes In Fiscal Policy Affect Exchange Rates
Source--Rosenberg.
On balance, we tend to side with those who believe that the dollar will
respond unfavorably to a new round of fiscal stimulus initiatives. Consider
first what happened to the dollar in response to the 2001-2002 fiscal stimulus
initiative. The U.S. fiscal stance, as measured by the annual change in
the U.S. cyclically adjusted budget balance as a percentage of GDP, turned
highly stimulative in 2001 and 2002. Indeed, the magnitude of the 2002
fiscal stimulus initiative was one of the largest, if not the largest single-year
U.S. stimulus effort, in the past 40 years.
One might have thought that such a powerful fiscal stimulus effort would have pushed U.S. real interest rates up sharply and, in turn, would have boosted the dollar's value up in tandem. But this did not happen. Instead, U.S. real interest rates, as measured by the yield on U.S. Treasury Inflation Protection Securities (TIPS), actually fell sharply in 2001-2002 because weak private-sector demand in the U.S. more than offset the stimulative effects of the U.S. fiscal initiative. The softness in total demand contributed to general downward pressure on U.S. real interest rates, particularly in 2002, and as U.S. real interest rates moved lower relative to real interest rates in Euroland, the dollar moved lower in tandem.
Contributing to the dollar's weakness in 2002 was the marked deterioration in the U.S. trade position (see Figure 2), due in part to the fiscally induced rebound in U.S. economic activity. Overall, using the Mundell-Fleming framework, the dollar moved lower in the past year because the fiscal stimulus contributed to a widening of the trade deficit, and, somewhat unexpectedly, because U.S. real rates failed to rise in response to the stimulus initiative. With U.S. real rates falling instead of rising, net private capital inflows to the U.S. fell instead of rising in response to the stimulus initiative, and as Figure 3 shows, the dollar fell accordingly. The dollar's decline in 2001-2002 can thus be traced to the combined deterioration of the U.S. trade and capital accounts. And as shown in Figure 4, the dollar's value is now falling in response to the deteriorating trend in the U.S. basic balance of payments.
Figure 2
U.S. Merchandise Trade Balance
(1992-2003)
Source--Datastream.
Figure 3
Net Private Capital Flows And The U.S. Dollar
(Net Foreign Purchases of U.S. Agency Bonds, Corporate Bonds, And Equities)
Note--12-month moving average of capital flows.
Source--Datastream.
Figure 4
Modified U.S. Basic Balance Of Payments And The U.S. Dollar
Note--Net foreign purchases of U.S. Agency/Corp. bonds and stocks less the U.S. trade deficit.
Source--Datastream.
Looking ahead to 2003, a new round of fiscal stimulus by the Bush administration
would almost certainly lead to a further deterioration of the U.S. trade
position. If the fiscal stimulus keeps the U.S. economy growing at a 3
percent + clip, one can expect U.S. import growth to rise by around 8 percent
on a year-over-year basis in 2003-2004, while export growth should rise
by only half that amount (since Europe's and Japan's fiscal hands are likely
to be tied in the next few years). If so, the U.S. current-account deficit
should widen by roughly $75 billion in 2003 to around $575 billion and
then widen another $75 billion in 2004 to around $650 billion. As a percentage
of GDP, the current-account deficit would be approaching 6 percent of GDP
in 2004, from 5 percent today.
Everything else being equal, such a widening of the deficit would have to be construed as being negative for the dollar. Unless of course, U.S. capital inflows were to rise dramatically in 2003-2004 to such an extent that it more than offset the expected deterioration in the U.S. trade position.
But how realistic is it to expect such a dramatic improvement in U.S. net private capital inflows? The U.S. already absorbs a hefty 71 percent of world net foreign savings to finance its record current-account shortfall (see Figure 5). If the U.S. current-account deficit were to widen further, is it realistic to expect that the U.S. would be able to attract a still larger slice of world net foreign savings? We believe the chance of that happening is not very high, particularly given that U.S. interest rates are simply too low relative to interest rates overseas.
Figure 5
U.S. Absorbs 71 Percent Of World Net Foreign Savings
(Shares Of 2001 World Current-Account Deficit)
Source--IMF Estimates.
Normally, when a country runs a larger and larger current-account deficit
it needs higher interest rates relative to rates overseas to attract the
necessary inflow of capital to finance the widening external imbalance.
This is true for all countries, even for the U.S. Indeed the trend in the
U.S. current-account deficit as a percentage of GDP has moved closely in
tandem with the trend in the U.S./German long-term yield spread, at least
up until recently. Historically, when the U.S. current-account deficit
as a percentage of GDP has risen, U.S./German long-term yield spreads had
to move higher to help the U.S. finance its widening current-account imbalance.
But in the past couple of years, the trend in the deficit/GDP ratio and
the trend in U.S./German yield spreads have diverged significantly, with
the deficit widening and the yield spread narrowing. It is this divergence
which we believe lies behind the dollar's recent weakening trend.
Even if U.S. yields were to start rising again and yield spreads started moving in favor of the U.S., it is highly unlikely that spreads would widen enough, at least initially, to attract the requisite inflow of capital to finance the mammoth deficits that we are projecting for 2003-2004. It is our view that U.S./foreign yield spreads have a long way to go before they will be wide enough to stop the dollar's hemorrhaging.
Consider what has happened in just the past three months. Yield spreads have actually started moving in the dollar's favor, yet the dollar has continued to slide. Between October 2002 and January 2003 the U.S./Euroland 10-year government bond yield spread swung nearly 60 basis points in favor of the U.S., yet the dollar fell steadily versus the Euro over that period. The U.S./Japan government bond yield spread swung over 80 basis points in favor of the U.S. between late September 2002 and January 2003, yet the dollar fell steadily versus the yen. This suggests to us that structural, and not cyclical, forces now lie behind the dollar's weakening trend and that yield spreads will need to widen further, most likely considerably so, before the dollar will be able to stabilize.
To get a better handle on whether cyclical or structural forces are driving the dollar's value at any particular point in time, we have constructed a new indicator, which we call the Regime Shift Barometer (RSB). The RSB gauges the extent to which regime shifts are giving rise to long-run structural changes in the dollar's value by determining the proportion of the dollar's trend that is being driven by changes in yield spreads (which tend to be cyclical in nature) and the proportion being driven by factors other than yield spreads.
The RSB is derived by first constructing an index to capture the cumulative yield pickup that U.S. money-market rates offer over comparable Euro money-market rates. We then construct an index of the U.S. Dollar/Euro spot exchange rate. We then divide the spot-rate index by the cumulative-yield-pickup index. When the dollar's value is moving more or less in line with the trend in yield spreads, the spot-rate/cumulative-yield-spread index will tend to move sideways. If the dollar's value is rising relative to the trend in yield spreads, the spot-rate/cumulative-yield-spread index will rise, and vice versa.
To help determine whether the dollar is undergoing a trend rise or fall, we compare the index with its 12-month moving average. The difference between the spot-rate/cumulative-yield-spread index and its 12-month moving average is referred to as our Regime Shift Barometer (RSB). If the RSB lies more than one standard deviation above its 12-month moving average, we interpret this as evidence that a structural increase in the demand for dollars is presently taking place. If the RSB lies more than one standard deviation below its 12-month moving average, we interpret this as evidence that a structural decline in the demand for dollars is presently taking place. If the RSB lies between +1 and -1 standard deviations away from its 12-month moving average, we interpret this as evidence that cyclical rather than structural factors are presently dominating exchange-rate movements. The trend in the Regime Shift Barometer is plotted in Figure 6.
Figure 6
U.S. Dollar/Euro Regime Shift Barometer
(Standard Deviations From RSB 12-Month Average)
Source--Datastream, DB Global Markets Research.
When regime shifts occur in the FX markets, large exchange-rate movements,
independent of relative interest-rate movements, tend to occur. As shown
in Figure 6, the RSB suggests that between November 1999 and November 2000,
a period of great dollar strength, a large part of the dollar's gains was
structural, not cyclical in nature as the dollar moved higher relative
to the change in U.S./Euroland interest-rate spreads. Since May 2002, and
carrying into the early part of 2003, the RSB has been indicating that
a structural, and not a cyclical decline in the demand for dollars is presently
underway, with the dollar losing value despite the change in U.S./Euroland
interest-rate spreads.
This is the first evidence of a structural decline in the demand for
dollars since the fall of 1998. Prior to that, one has to go back to 1994
to find a period of structural dollar weakness. A modest rise in U.S. interest
rates relative to Euroland rates is therefore unlikely to alter the dollar's
structural weakening trend, and given the expected deterioration in the
U.S. basic balance of payments in 2003-2004, this new structural downtrend
in the dollar looks set to persist for a long time to come.