RDP 8712: Policy Analysis with the MSG2 Model 3. Interdependence in the MSG2 Model

The principal goal of the MSG modelling project is to better understand the international transmission of macroeconomic policies, and to design policies for global economic coordination consistent with the results on interdependence. One of the things which has been impressed on us in the development of the model is our tenuous knowledge of key magnitudes in international transmission of macroeconomic policies. Seemingly innocent changes in parameter values can even change transmission of policies from being positive to negative across countries.

It must be stressed that one of the major reasons for doubts on the direction of interdependence effects is that for many policies, there are several conflicting channels of effects on other countries. As a well known example, a U.S. fiscal expansion tends to raise interest rates in Europe (contractionary), depreciate the European real exchange rate (expansionary), and directly increase European exports to the U.S. because of fiscal-led growth in the U.S. (expansionary). The sum total effect of these channels is ambiguous theoretically, and somewhat elusive empirically. Nonetheless, we make some concrete observations about the sign and magnitude of international transmission effects.

While theoretical papers typically focus on the sign of interdependence effects, considerations about policy coordination or domestic economic planning require information on the magnitudes of effects. Indeed, one of the purported lessons of Oudiz and Sachs (1984) and McKibbin and Sachs (1986) is that the degree of interdependence among the major economies appears to be too limited to require an intricate degree of international policy coordination.

Let us now turn to various aspects of macroeconomic interdependence. We begin with fiscal policies, and then turn to monetary policies.

a. Fiscal Policy Transmission

Various simulation results for fiscal policies in the U.S., Australia and Japan, are shown in Tables 1 to 5. Before discussing the results, it is crucial to understand the experiment that is being undertaken in the tables.

In line with rational expectations models, policy experiments must define an entire future path of policies, and not just a change in an initial year, or even the changes over the time interval of interest in the simulations (1986–1990). In the case of fiscal policy, it is important that permanent changes in government spending be matched at some well-defined point in the future by increases in taxes in order to pay for the government spending. In particular, starting from any initial stock of public debt, the discounted value of current and future government taxes must be equal to the present discounted value of future spending plus the initial outstanding stock of public debt.

In our case, a permanent fiscal expansion has the following characteristics. Government spending rises permanently by one percent of potential GDP. The government spending is distributed over domestic goods and imports, in the same proportions as with private spending. Initially, the tax schedule remains unchanged, with taxes increasing only to the extent that the fiscal expansion raises output and thereby induces an increase in tax collections. In general, the one percent of GDP fiscal expansion causes the public deficit to worsen initially by about 0.9 of one percent of GDP. The deficit is financed entirely by the issuance of public debt, with the money stock (both base money, and implicitly Ml) remaining unchanged. Over time, the debt stock will rise, so that interest servicing will also increase. If the tax schedule is not altered, then the debt will grow explosively fast, and the government's budget constraint will be violated. To prevent this, we assume that taxes rise each period by enough to cover the increasing interest costs on the increasing public debt. In this way, the overall deficit remains fairly constant at about 0.9 percent of GDP, though the primary deficit (i.e. government spending net of interest payments, minus total taxes) eventually turns to a surplus as is necessary to prevent an explosive growth in debt. The fact that the deficit is permanently increased does not lead to an ever increasing debt-GDP ratio because GDP is itself increasing in the long run at the potential growth rate (assumed to be 3 percent in the model for all of the regions in the world economy). Thus, a permanent increase in the deficit of 0.9 percent of GDP leads asymptotically to a rise in the debt of 0.9/0.03 percent of GDP, which is a rise of the 30 percent of GDP.

Consider first the permanent U.S. fiscal expansion, shown in Table 1. To read the table, note the following points. All variables are expressed as deviations from an initial baseline. Output is recorded as a percentage deviation from the initial baseline (e.g. 0.60 percent of GDP in 1986). Consumption, investment, exports, imports, and the trade balance are all reported as deviations from baseline in percent of potential GDP. Thus, in 1986, private consumption rises relative to the baseline by 0.08 of one percent of U.S. potential GDP. Labour demand (i.e. total manhours in the economy) is reported as a percentage deviation from the baseline (e.g. a rise of 0.32 percent in 1986). Inflation and interest rates are reported as deviations in percentage points relative to the baseline (rather than as deviations as a percent of their baseline values). Thus, inflation in 1986 is seen to fall by 0.31 percentage points in 1986, while short-term interest rates increase by 0.53 percentage points (i.e. 53 basis points). The three U.S. bilateral exchange rates are reported as a percentage change from baseline values. Note that a negative value for the exchange rates indicates an appreciation of the U.S. dollar.

Now, let us consider the simulation results for the U.S. fiscal expansion. What should we expect from theory? From the Mundell-Fleming model, we should expect that a bond-financed fiscal expansion, in the presence of perfect substitutability of home and foreign financial assets, should result in a rise in domestic income and an appreciation of the U.S. dollar exchange rate. Indeed, output rises by 0.60 percentage points in the first year, while the dollar appreciates by 2.9 percent vis-a-vis the Australian dollar, and by 3.4 percent vis-a-vis the Yen and 3.1 percent vis-a-vis the ECU (where the ECU signifies the currency basket of the ROECD). The rise in output and the appreciation of the dollar produces a trade deficit, equal to 0.29 percent of GDP in the first year of the fiscal expansion. Note that there is a small fall in private investment, and a small rise in private consumption, in the U.S. The consumption behaviour reflects the forward looking nature of the consumers in this model. The implied future taxes from the permanent fiscal expansion leads to an intertemporal substitution from present to future consumption. Households increase their saving in the present period, in order to finance the future taxes implied in the issue of debt. If all consumers were forward looking, consumption would fall by 1 percent of GDP, exactly matching the rise in government spending. Crowding out would be instantaneous. The assumption that a proportion of the consumers also consumer out of current income, implies that the crowding out takes much longer to be achieved.

The transmission of the U.S. fiscal shock to the different regions, including Australia, is perhaps surprising at first sight. Importantly, the Mundell-Fleming model teaches that the transmission effect of a U.S. fiscal policy expansion on foreign (Australian) output is ambiguous, for the reasons already alluded to. On the one hand, world interest rates rise, which tends to depress Australian income. On the other, the demand expansion in the U.S. tends to raise foreign exports which filters through to foreign income. As described in Bruno and Sachs (1985, chapter 6), and in Oudiz and Sachs (1984), the transmission is more likely to be negative if foreign wages and prices rise rapidly in response to the depreciation of the foreign currencies vis-a-vis the dollar following the U.S. fiscal action. If foreign wages and prices are fixed, then the U.S. fiscal expansion will tend to be positively transmitted. In this model, Australian wage setters are partially forward looking and adjust wage claims for expected changes in consumer prices.

As can be seen from Table 1, the effect of the permanent fiscal expansion is negligible transmission in the first year of the expansion, but then a negative transmission thereafter. As is evident from the table, the negative effects on foreign consumption and investment resulting from higher interest rates, start to dominate the expansionary effects of greater exports to the U.S. by the second year for Japan and the ROECD and Australia. Note that inflation is increased throughout the world following the U.S. fiscal expansion. Most of the inflationary effect abroad arises because the foreign currencies depreciate against the dollar after the U.S. fiscal expansion. By the second year in Australia, Europe and Japan, the U.S. fiscal expansion has a net stagflationary effect, by lowering output while at the same time raising inflation.

Table 2 shows the effects of permanent fiscal expansion in Japan. Note the following important point. The Japanese fiscal expansion has a very small effect on the U.S. trade balance, as a result of the fact that Japan is considerably smaller than the U.S. A one percent of GDP Japanese bond-financed fiscal expansion is seen to appreciate the Yen by about 3.9 percent, and to worsen the Japanese trade balance by about 0.6 percent of Japanese GDP. Overall, the current U.S. bargaining strategy of pressuring a Japanese fiscal expansion, can be seen to have very mixed merit. U.S. output is unlikely to change much, and could even decline in response to a Japanese expansion. The U.S. trade balance would improve by only 0.06 percent of U.S. GDP (about $4 billion) for each increase in Japanese government spending of 1 percent of GDP. On the other hand, the Japanese trade surplus would fall substantially with an increase in Japanese public spending.

Compare in Tables 1 and 2 the employment effects of a fiscal expansion in the U.S. and in Japan. In the U.S. case, labour demand rises relative to the baseline for three years. In the Japanese case, on the other hand, labour demand rises in the year of the fiscal policy change, but then falls to exactly the baseline level in the following years. The difference in behaviour stems from the assumed difference in wage setting patterns in the two countries. In the U.S., nominal wages are set according to a partially backward looking indexation mechanism, which imparts nominal wage sluggishness in the model. In Japan, on the other hand, wages are set in an annual wage cycle, with the wages for the following year targeted, with rational expectations, to hit the labour-market clearing level. In a given year, the labour market can be jolted away from full employment because of unanticipated shocks that occur in the year, but in expectation, the labour market always clears in the out years.

Table 3 shows the results for a permanent Australian fiscal expansion. Output rises by 0.67 percent in the first year and then follows a familiar hump shape, declining to zero by 1995. Interest rates rise in 1986 which leads to an appreciation of 2.3 percent and then a gradual depreciation over a very long horizon. The slow rate of depreciation of the currency reflects the small interest differential which emerges after the first year. Long interest rates rise in the first year but short rates move around reflecting short run changes in demand. Uncovered interest parity holds in this model.^[9] The trade balance deteriorates by slightly more than 0.5 percent of GNP reflecting both a loss in export receipts, due to the stronger currency, and higher imports, due to the lower relative price of foreign goods and higher aggregate demand in Australia. Inflation (defined in terms of the consumer price index) initially falls due to the appreciation and due to little change in domestic prices reflecting cheaper imported goods in the production process. Both aggregate supply and aggregate demand increase in response to the shock. The domestic goods price then adjusts to lead to equilibration. In principle, prices can actually fall if the supply response is large than the demand response.

Tables 4 and 5 show the importance of specifying whether a shock is permanent, temporary or anticipated. In table 4, results are presented for a temporary 1 percent increase in government expenditure in Australia. The policy change is assumed to be credibly announced as a rise in government expenditure for three years from 1986. Comparing table 4 with table 3, several interesting points should be noted. Consumption rises by 0.37 percent in the case of the temporary shock. This is more than for the permanent shock because the forward-looking consumers due not need to increase saving to pay for future tax increase. In fact, to finance the deficit, interest rates must now rise by much more because households are reluctant to save at the original interest rate. Notice also that consumers import more in the case of the temporary fiscal shock because of the desire to maintain consumption in the face of a temporary shock. The current account deteriorates by more because it is used for buffering in this case.

Table 5 presents results for a permanent fiscal expansion in Australia, anticipated in 1986 to occur in 1987. This simulation further highlights the role of forward-looking behaviour. The exchange rate appreciates on the announcement and then appreciates further when the policy is implemented. Long term interest rates rise in Australia, but the short rate initially fall reflecting the inial fall in aggregate demand. The differential in short interest rates, determine the path of the exchange rate and therefore shows why the exchange rate does not fully appreciate in 1986; the relatively lower Australian interest rates in 1986 imply an expected appreciation of the exchange rate. Consumers perceive the future taxes implied by the announcement and therefore cut consumption in 1986. Note also that imports decline in the case of the anticipated shock because the shock is known to be permanent consumers do not attempt to buffer the fall in consumption by borrowing from abroad where they did in the case of the temporary shock. Since the government spending has not come on line in 1986, aggregate demand falls by the fall in consumption. Once the policy is implemented the economy booms with higher short interest rates and output higher than the case in table 3, when policy was not announced in advance.

b. Monetary Transmission

As with fiscal policy, the international transmission of monetary policy has a theoretically ambiguous sign. A domestic monetary expansion tends to depreciate the home exchange rate and to reduce world real interest rates. The exchange rate depreciation shifts demand away from other countries and towards the home country, while the reduction in world real interest rates tends to raise demand in the rest of the world. In the simplified Mundell-Fleming model, in which output prices and nominal wages are fixed in the other countries, the exchange rate effect dominates, so that foreign output falls when the home country increases the money supply. Home monetary expansion is then beggar thy neighbor. In more elaborate models with wage price dynamics, either the exchange rate channel or the interest rate channel might dominate.

Monetary policy is also ambiguous with respect to the effect on the domestic trade and current account balances. Higher domestic money improves international competitiveness by depreciating the home exchange rate. Assuming that the standard Marshall-Lerner conditions hold (as they do in the MSG2 model), this effect tends to improve the trade balance and current account. On the other hand, the fall in interest rates tends to raise investment demand and to lower savings, thereby worsening the trade and current account balances. The overall effect is ambiguous.

Finally, note the magnitude of the effect of a monetary expansion on the nominal exchange rate. It is well known from the Dornbusch (1976) model that the exchange rate will depreciate upon a permanent, once-and-for-all increase in the money supply, but that the size of the depreciation on impact may exceed (“overshoot”) or fall below (“undershoot”) the long-run change in the nominal rate, which just equals the proportionate change in the money stock. If the effect of the exchange rate on domestic demand is large (through the effect on the trade balance), and if the effect of domestic demand on money demand is large (through the income elasticity of demand for money), and if the exchange rate depreciation causes a rapid rise in domestic prices, then it can be shown that home nominal interest rates will tend to rise after the money expansion, and the that the home exchange rate will tend to undershoot its long-run change. If on the other hand, one or all of these three channels is weak, then domestic nominal interest rates will tend to fall after the money expansion, and the exchange rate will tend to overshoot its long-run change.

Let us now examine these effects in the MSG2 model. As seen in Table 6, a one percent U.S. monetary expansion raises U.S. output by 0.58 percent in the first year, and causes the exchange rate to depreciate by one percent relative to the Yen but by less relative to the ECU and Australian dollar. U.S. inflation increases by one-third of a percent, which is far more inflation per unit of demand stimulus than for fiscal policy, because of the opposite direction of effect on the exchange rate (i.e. for fiscal policy, the dollar appreciates, tending to reduce inflation; while for monetary policy, the dollar depreciates, tending to increase inflation). There is a slight negative transmission of U.S. monetary policy to the output of the other countries. Moreover, the U.S. trade balance remains virtually unchanged.

Consider the effects on the direction of trade flows. The U.S. sells more to the rest of the world and buys more from the rest of the world. The other regions divert their own export sales from the non-U.S. market to the U.S. market. Total imports in the rest of the world remain unchanged, but shift in composition to a higher share of imports from the U.S.. Total exports in the rest of the world also remain virtually unchanged, but shift to supply the growing U.S. market, and away from third, non-U.S. markets.

The same pattern of proportionate depreciation of the exchange rate, with little effect on the trade balance of the expanding country, or the outputs of the foreign countries, holds for a monetary expansion in the other OECD regions. This general conclusion is a key one, for it says that in fact floating exchange rates effectively insulate the output of countries from the monetary policies abroad. The U.S. would benefit little on the output side from discount rate cuts in Europe and Japan and may even be hurt.

Table 7 contains the results for a permanent 1 percent increase in the money supply in Australia. In contrast to the U.S. monetary expansion, the exchange rate overshoots its long-run value and depreciates by 1.5 percent in the first year. Output rises by 0.5 percent in the first year and then is gradually crowded out by rising inflation and interest rates. In this case the trade balance improves slightly because the stimulus to exports from the depreciation tends to dominate the stimulus to imports from the rise in aggregate demand.

The results for an anticipated monetary expansion in Australia, believed in 1986 to occur in 1987 and be permanent thereafter, are shown in table 8. The exchange rate depreciates on the news by 0.6 percent and then further depreciates, overshooting its long run value, once the policy is implemented. Output rises before the implementation of the policy because of the improvement in the trade balance due to the depreciation. Interest rates actually rise in anticipation of the shock due to the expected inflationary consequences of the shock and the fall in savings which pushes up real interest rates. The interest differential in Australia's favor reflects the depreciation expected once the monetary expansion takes place.

c. The Gramm-Rudman Package

There is a great deal of evidence that the large U.S. fiscal deficits of the 1980's is responsible for the large trade imbalances in the world economy. The purpose of this section is to examine the implications of a Gramm-Rudman style deficit reduction package announced in the U.S. in 1986 and assumed to be credible. This illustrates the usefulness of using a model such as the MSG2 model, to examine policies which are have long time horizons before implementation.

Results are shown in tables 9 and 10. Here the assumption is that the U.S. reduces fiscal deficits by 0.5 percent of GNP from 1986 to 1991 and then maintains a deficit from 1991 onwards which is 3 percent of potential GNP lower than currently expected. The difference between the two tables is that the first shows the impact without any change in U.S. monetary policy. The second assumes that the Fed desires to avoid any unemployment consequences and therefore credibly targets the unemployment rate with monetary policy. The money rule is found using dynamic optimization techniques discussed in further in McKibbin (1987).

Consider table 9 first. Since Gramm-Rudman involves an anticipated sequence of future deficit reductions in the U.S., the forward-looking properties of the assets markets in the MSG2 model are important in the analysis. As can be seen, the announcement of the future fiscal cuts, raises output in the first period, mainly by reducing long-term real interest rates and depreciating the dollar upon the announcement of the policy. Short interest rates rise because of the fall in aggregate saving resulting from private sector dissaving before the realization of higher public sector saving. In later periods, as the fiscal deficits are actually cut, then the negative demand effects on the economy of the fiscal contraction show up in reduced output and employment. In Australia, the fall in long interest rates stimulates domestic demand sufficiently to offset the negative effect of a deteriorating trade balance. Short rates, real and nominal, initially rise due to the strong growth in domestic demand.

In table 10 we assume that the Fed attempts to dampen the effect of the fiscal policy change on employment. The policy which is followed has an initial monetary contraction in the U.S. followed by a continual monetary expansion. This has the effect of reducing the initial depreciation of the U.S. Dollar but increasing the depreciation by 1990. As can be seen, employment is maintained at the baseline level but at the cost of gradually rising inflation. The U.S. trade balance is seen to improve by 0.8 percent of GNP by 1990 but still well below that required to remove the trade imbalance. The Australian economy is also faced with a growing trade deficit and stronger currency without any change in policy. Investment remains flat but consumption is very strong and is responsible for the growth in the economy.

Footnotes

These results differ substantially from other Australian models such as the RBII model (Edey, Kerrison and Menzies (1987)) and the NIF88 model (Simes (1987), although they are similar in scale to the AMP model discusssed in Murphy (1986). [9]