Kevin Clinton October 2005 Open-economy monetary and fiscal policy Reference Ken Rogoff. Dornbusch s overshooting model after 25 years. IMF Staff Papers 49, Special Issue 2002. 1. What monetary policy holds constant In the open economy model of the last lecture monetary policy was represented just by the interest rate. To complete the model, we have to be more specific than that. We have to say what exactly monetary policy holds constant in the short run (i.e. its operating instrument) and in the long run (its objective). That is, we want to make clear what ceteris paribus (other things remaining the same) means with respect to monetary policy, when some exogenous variable changes (e.g. fiscal policy, foreign demand) in particular, we need a nominal anchor for the system to ensure that nominal variables have some definite equilibrium value (e.g. the price level, and the level of nominal interest rates) and to derive the changes in this value following a disturbance to the equilibrium what a change in monetary policy means in the long run and the short run We saw a symptom of incompleteness in the open-economy model described last week. With static exchange rate expectations, the central bank had no room to change the domestic interest rate, even under a floating exchange rate. This takes the variable supposed to represent monetary policy out of play (the exchange rate in the end is the variable that adjusts). A realistic description of monetary policy is that the instrument controlled by the central bank is the short-term interest rate (this requires a floating exchange rate monetary policy is set by the rest of the world if the exchange rate is fixed) the objective is price stability (in practice a low target rate of inflation 2% in Canada)
2. Exchange rate expectations and asset market equilibrium 2.1 Dornbusch model In our version of the model, the central bank sets the interest rate, following a rule for price stability. Rational expectations of long-run equilibrium The M-F assumption that exchange rate expectations are static is useful for showing how monetary policy (might) and fiscal policy (might not) affect output and employment in an open economy. But it is obviously an oversimplification, since it allows the domestic interest rate no freedom to move independently. The extent to which can the Canadian short-term interest rate vary independently is constrained by high capital mobility and exchange rate expectations, but not so rigidly as that. 25 % 20 15 Chart 1 Canada & US 3 month t bill rates Jan 1971-Dec 2003 Canada 10 5 US 0 0 60 120 180 240 300 360 420 In the real world, the Bank of Canada does change the domestic overnight interest rate to some extent independently of the US federal funds rate. Example: March 2, the BofC cut the overnight rate in Canada for 2.50 to 2.25%, while the federal funds rate stayed the same at 1.00%. The differential is currently quite wide 1 ¼% and it varies. 2
% 6.00 4.50 3.00 Chart 2 Ca na da - US 3 m onth t bill diffe re ntia l Ja n 1971-De c 2003 Average = 1.45 1.50 0.00-1.50 0 60 120 180 240 300 360 420-3.00 Bear in mind is that although these independent movements, they are relatively small in the long run perspective. The major movements in Canadian interest rates over time are quite similar to US rates. Other central banks in open economies are in a similar situation vis-à-vis interest rates in major foreign financial centres. A realistic model would allow some interest rate independence in the short-run but not in the long run. We can modify expectations in the M-F model to recognize this. Dornbusch assumed rational expectations for the exchange rate. That is, the expected value of the exchange rate is the long-run equilibrium value, e 1 in chart 3. At an exchange rate above there is excess demand, and hence an inflationary output gap. Below there is a deflationary output gap. At demand is equal to potential output, and the price level is stable such that monetary policy achieves its objective. 3
Chart 3 Equilibrium = expected exchange rate e IS e 1 y We can draw a chart in (r, y) space showing the long-run full employment equilibrium. Chart 4 Locus of interest and exchange rates consistent with full employment output r FE equilibrium r 1 e 1 e log scale The FE equilibrium line is upward sloping. Starting at full-employment equilibrium, a decrease in r on its own would increase domestic demand, and so create inflation. To restore equilibrium, the price of foreign exchange would also decline. And vice versa for an increase in the interest rate hence the upward slope. Points above FE imply a deflationary gap, points below an inflationary gap. 4
The job of monetary policy is to keep output on the FE line, even as disturbances displace the short-run equilibrium. Uncovered interest parity With perfect capital mobility, the domestic interest rate is equal to the foreign rate plus the expected increase (or minus the expected decrease) in the price of FX: r t = r* t + E t e t, where E t e t is the expected increase in e. 1 This may be written as r t = r* t + E t (e t+1 ) - e t. (1) where E t (e t+1 ) is the expected value in period t of e in t+1. Equation (1) is known as uncovered interest parity (UIP). The expected rate of return is the same, adjusted for expected changes in the exchange rate, regardless of currency. Rapid asset price adjustment The short-run interest rate is set by the central bank, and the exchange rate adjusts instantly see chart. Financial asset markets clear continuously. (Current output prices, and hence the price level, are in contrast are sticky they do not instantly jump from one LR equilibrium position to the next, but moves in steps over a period of time.) Chart 5 US$ price of C$ 1 0.9 0.8 0.7 0.6 1970 1975 1980 1985 1990 1995 2000 2005 1 For precision, note that lower case e is the logarithm of E. The change in a log is approximately equal to the proportional change, i.e. the percentage change divided by 100. 5
We can draw an asset market equilibrium line, AA, in (e, r) space from which there are no deviations, even in the short run. Chart 6 General equilibrium: asset and output markets r AA FE equilibrium r 2 r 1 e 1 e 2 e log scale The asset market equilibrium line slopes up, showing the intuitive feature that an increase in the domestic interest rate is associated with a strengthened currency in the financial markets. At an interest rate above r 1, say r 2, asset market equilibrium requires that the exchange rate rises to a higher value, e 2. People expect the exchange rate to return to the long-run equilibrium value e 1. That is, they expect e to rise. With UIP, the expected increase in e would just compensate for the higher interest rate on domestic assets. (If the interest rate is expected to return to its equilibrium value next period, then the slope of the AA line is minus one.) Example A 100 basis point increase in r for one period causes e to jump one per cent. Investors expect it to drop to its equilibrium value next period. The expected exchange rate gain is equal to the adverse interest rate differential on foreign assets. 6
Evidence According to the theory positive shocks to the Canada-US interest differential cause e to drop. The chart below shows some such positive correlation between the interest spread and e, particularly since the mid-1980s, but it is quite loose. Chart 7 Canada-US interest differential and exchange rate 1.4 1.3 1.2 1.1 1 0.9 0.8 0.7 0.6 0.5 US$/C$ rate and t bill spread Jan 1971-Dec 2004 exchange rate t bill spread (scaled) 0.4 0 60 120 180 240 300 360 420 The problem in interpreting the data is that in the real world there are many things happening at the same time. There is not a simple experiment in which the central bank shocks the interest rate to see what happens monetary policy is itself responding to economic events, including sometimes changes in the exchange rate (e.g. 1992, 1995, and 1998). So we have causality going both ways the identification problem in econometrics. However we can see the effect of the tight money policy of the late 1980s and early 90s in both the interest rate differential (strongly up) and e (up). Likewise, the easing of monetary policy, after the inflation rate was brought down to the target rate of 2%, later in the 1990s, is evident in the steep (if uneven) decline in the interest rate, and the decline in the C$ to record lows. 7
3. Monetary policy action Chart 8 Monetary contraction increased interest rate r AA FE equilibrium r 2 r 1 e 1 e 2 e log scale Monetary contraction example Consider the following story. It might even be a realistic description of certain events. In a widely recognized policy mistake, in period 1 the central bank, unnecessarily worried about inflation, unexpectedly raises the interest rate, from r 0 to r 1. Everybody but the bank realizes that nothing in the economy justifies this. Investors expect all variables to return to initial values in year 2. For UIP (equation 1) to hold, e must immediately rise to e 1, such that its expected depreciation, back to the long-run equilibrium e 1, exactly offsets the interest differential: E 1 (e t+2 ) - e 2 = e 1 - e 2 = r 1 - r 2. The asset market equilibrium line AA 1 for the one-period change in r (and only the oneperiod change in r) has a negative one: one slope (one point e for one point r). The shortrun equilibrium moves along AA. Thus, in the Dornbusch model exchange rate flexibility allows monetary policy to affect r as well as e, whereas it affects only the latter in the M-F model. The increase in both variables is deflationary: output falls below potential; the price level falls. The central bank realizes its error, and in year 2 returns the interest rate to r 1. But in the short run the monetary contraction reduces output and prices. 8
The same logic would apply exactly to a cut in the interest rate; the signs of all the changes would simply reverse. We would a temporary inflationary gap. Clearly monetary policy can have substantial effects on output and prices in an open economy with a floating exchange rate. The change in the exchange rate (the external channel) reinforces the policy change in the interest rate (the domestic channel of transmission of monetary policy). 4. Fiscal policy The effect with Dornbusch expectations are virtually the same as with static expectations: with perfect capital mobility and a floating exchange rate, fiscal policy has no impact on output. As before, in principle an expansionary fiscal policy can shift the IS curve to the right. However, such a shift, from equilibrium, implies that the central bank would raise the interest rate, to choke off the inflationary increase in demand. Chart 13 General equilibrium fiscal expansion r 2 r 1 = r* IS (G 2, e 1 ) IS (G 1, e 1 ) = IS (G 2, e 2 ) Y FE Con sider an increase in government spending, from G 1 to G 2. In itself, this would shift the IS curve to the right. For example, given the initial exchange rate e 1, it might move from IS (G 1, e 1 ) to the dashed line IS (G 2, e 1 ). But this would provoke the central bank to raise the interest rate towards r 2, to protect price stability. This would cause the exchange rate to drop at once. Therefore, the latter IS curve could never actually apply. In the long run, the IS curve has to go back exactly to the starting point, i.e. IS (G 1, e 1 ) = IS (G 2, e 2 ). The fiscal stimulus is entirely undone by a drop in net exports caused by the rise in e, from e 1 to e 2. Perfect capital mobility keeps the interest rate at the world level and, in effect, foreigners finance the change in the budget surplus. In the short run, in the Dornbusch model, it is possible that the interest rate would go above the world level for a while, and that the exchange rate would temporarily 9
overshoot its long-run level in this case both the external channel and the domestic channel crowd out the fiscal stimulus. In any case fiscal policy would have no effect on output. The IS curve in (e, Y) space does shift permanently, to the right, because of the higher level of government spending, and the constant interest rate. At any given exchange rate, aggregate demand is greater with government spending of G 2 than with G 1. The equilibrium exchange rate rises from e 1 to e 2. Chart 14 IS curve (e, y) space fiscal expansion e IS (G 2, r 1 ) e 2 e 1 IS (G 1, r 1 ) y 1 y Chart 15 Fiscal expansion Dornbusch expectations (r, e) space r AA FE equilibrium r 2 r 1 P Q e 1 e 2 e log scale The FE equilibrium locus shifts to the right (and up). The world interest rate is unchanged, so therefore is the equilibrium domestic interest rate. The long-run equilibrium will shift from point P to Q, and so the expected exchange rate (per 10
Dornbusch) rises from e 1 to e 2. Thus the asset market equilibrium line shifts immediately, to go through point Q. There is an immediate drop in the exchange rate sufficient to choke off any effect on the level of output. Following fiscal policy initiatives, appreciation or depreciation of the exchange rate produce exactly offsetting changes in net exports. Notice, however, that the short-run impact of a fiscal expansion looks attractive. First, people receive increased government services or a tax cut. Second, the cost of imported goods and service falls, because of the rise in e. There is a short-run drop in inflation. The policy is not so attractive, however, in the long run. The increased accumulation of debt creates higher interest payments both for the government and for the country as a whole (net exports have declined). Note the twin deficits problem. We saw examples of this dynamic in the US, with the Reagan tax cuts and increased military spending in the early 1980s, and also with Canadian fiscal policy throughout the 1980s. Strong expansion of output for a few years was accompanied by a strong currency and large trade deficits. Eventually both output growth and the currency weakened. In Canada, Paul Martin s budget in 1995 vastly improved the budget position. For a couple years afterwards, however, output in Canada was relatively weak, despite the strong expansion of demand in the US. The drop in government spending was a factor in this weakness. Also the C$ dropped against the US$. The first signs of improvement showed up in net exports. After 1997 output growth picked up to US levels, later still, in 2003 the C$ began rose steeply against the US$. In the end, fiscal retrenchment paid off, but it took a few years to see the good results. Since political horizons tend to be short especially in an election year there is a great political temptation to embark on expansionary fiscal policy. By the same token, cleaning up a fiscal mess deficit reduction has politically unpleasing effects in the short run. 5. Fiscal/monetary mix The above results are sometimes expressed somewhat differently. The M-F model indicates that a shift in the mix of policy towards fiscal expansion and monetary contraction creates an attractive short-run result: increased domestic absorption of goods and services; and a lower rate of inflation. However, in effect it is the expansion of fiscal policy that drives the change and creates longer run problems. In our set-up (with no LM curve), monetary policy has not really changed the central bank just does not finance the government deficit. If the central bank was motivated by a price stability (or low inflation) objective this would be the correct strategy it would not represent a tightening of policy, but consistent pursuit of its policy. 6. Contrast of regimes: flexible versus fixed exchange rate 11
A fixed exchange rate regime implies a very different policy environment. No independent monetary policy under fixed rate There is no independent monetary policy, since the interest rate is fixed in the rest of the world, and the nominal exchange rate cannot change. In the long run the domestic inflation rate is the same as that in the rest of the world. Fiscal policy is effective in the short run but beware debt growth in the long run Since neither the interest rate nor the exchange rate can change, expansionary fiscal policy has strong short-run output effects. There is no offset from currency appreciation as in the floating regime, or from interest rate increases as in the closed economy case. However, this also means that there are large implications for foreign indebtedness, which grow over time. Fixed exchange rate regimes often collapse in a crisis when foreign debt burdens become excessive. Adjustment to exogenous real shocks under fixed exchange rate The processes of adjustment following shocks to foreign demand, etc., are much slower under a flexible exchange rate. For example, following a sharp increase in raw materials prices, the nominal C$ usually appreciates promptly. This helps keep the economy in equilibrium the increased value of the C$ releases resources for use in the booming industries. Under a fixed regime, the required real appreciation of the C$ has to occur through domestic inflation which takes time. A downside shock to demand would involve a particularly prolonged and painful adjustment period a recession because the overall domestic wage and price level is particularly inflexible in the downward direction. 12
Volatility of exchange rates The actual movements in the exchange rate (below) are a source of controversy. Some economists worry about volatility in exchange rates which is not connected to economic fundamentals. Exchange rates often vary for no apparent reason at all. There is a lot of froth in financial markets, exchange markets included. However, the broad movements in the C$, over years, have been in an equilibrating direction e.g. the dismal 1990s would have been even more dismal if the C$ had not depreciated 20% or more. And there is no evidence that weekly or monthly volatility has any significant macroeconomic effects at all. Monetary unions These are fixed exchange rate regions par excellence there is only one money. Examples: Canada, the euro area. In Canada monetary union issues arise in 3 contexts: Provincial fiscal policy in a large province fiscal expansion might increase provincial output in the short run (the fixed exchange rate case) but it has a negative impact on other provinces through effects on interest rates e.g. Ontario 1985-95. Beggar thy neighbour policy. Debate about monetary union with US. Lower transactions costs, and more trade, versus adjustment problems. Quebec secession scenarios the PQ has debated these regimes: the C$; the US$; and a Q$? Re the euro area, note: just one monetary policy for the whole area too soon to judge how effective this is, but there are obvious teething problems at least the Maastrich Stability and Growth Pact puts a 3% limit on government deficit/gdp ratio an effort to stop the beggar-thy-neighbour fiscal policy, which is a risk in a currency union. Again, not clear yet whether this is effective. 13