RDP 8901: The World Economy from 1979 to 1988: Results from the MSG2 Model 5. Tracking 1979 to 1988

The properties of the model give some indication of the likely performance of the model in tracking the world economy during the 1980s. For example, it has been shown that asset prices can fluctuate substantially, especially for policy changes which involve shifts in expectations about expected future paths of policy.

This section examines the tracking ability of the model given changes in OPEC oil prices, lending to LDC's and fiscal and monetary policies in the major OECD countries. As can be seen from the results in section 4, it is very important to specify the expected future path of policy in simulating the effects of any particular policy change.

The procedure we follow in this paper is iterative. For each period, we first generate expectations of future fiscal policy, based primarily on OECD forecasts of fiscal policy. Monetary policy in each country is then arbitrarily geared towards approximately reaching the realized output gap that occurred in each year as well as attempting to reach observed short term and long term nominal interest rates (working partly on inflationary expectations). Given the output result and the cyclical fiscal deficit which accompanies this, we then adjust the exogenous fiscal instruments until we reach the actual and expected deficits as well as the term structure. This iterative procedure takes some time to converge. Where major fiscal policy announcements are made, these are taken into account.

Table 17 shows the path for the world economy generated by the model for 1979 to 1991. This is found by starting in 1979 and, given the assumed expectations about future policies, solving the model forward to 2019. The model is then solved starting in 1980, inheriting the results from the 1979 simulation, and taking into account any new information from policy announcements and actual realizations of policy variables. This procedure is then repeated until 1988.

The results in table 17 are levels of variables. It should be pointed out that the model is not solved in level form but is solved in deviations from some level. Each new shock implies cumulative deviations from the underlying baseline. To express the results as levels of variables we add the cumulative deviations of variables to their levels in 1978. Implicitly this assumes that had all nominal variables in each country remained at their 1978 growth rates, the model would have generated a path for the real economies in which all real variables would grow at the underlying real growth rates; variables would not change as a share of GDP. The major difference between this procedure and simulating the model using levels of variables is that it ignores the inherited dynamics from 1978 which would (in principle) have some effect on the path for the following decade. We argue that the size of shocks are such that they dwarf the inherited dynamics. A technique for finding the exact model generated base path has been developed on a small version of the model consisting of 3 regions and in this case we found very little deviation of real variables from the 1978 initial shares of GDP, given no shocks to the world economy from 1979. In future work we intend to further develop this technique.

To read Table 17 remember that GDP is the output gap; that is the cumulative deviation from potential output. To convert this to an actual growth rate the change in output gap should be added to the trend growth rate for each economy (e.g. for the U.S. one could use 2.5 percent). For example a move from −8.2 in 1984 to −6.4 in 1985 is a rate of growth of output over this period of 4.3 percent (2.5 percent plus 1.8 percent). In this case the output gap narrows because the U.S. economy grew faster than trend. The other quantities are expressed as percent of GNP for real variables. The real exchange rate is the percentage change from the 1978 value.

The shocks, apart from monetary policy, that we impose on the model are contained in table 18. This table has fiscal deficits, OPEC oil prices and the change in lending to LDC's. The monetary policy stance was solved out by an iterative technique discussed above where the term structure of nominal interest rates was targeted given the fiscal stance.

Perhaps the surprising feature of table 17 is how well the model tracks the features outlined in section 2 above. Note that all arbitrage conditions hold ex-ante, but only hold ex-post if there are no surprises in the world economy. In fact in every year there are policy surprises; some of these are quite large.

The path of the world economy generated by the model in the early years is dominated by the rise in OPEC prices from 1979 to 1981. In 1981 and 1982 the world economy is dominated by a global monetary contraction which is particularly severe in the U.S.. The 1982 recession also reflects an expected fiscal expansion in the U.S. as well as Japanese fiscal contraction. From table 10, in the first year of the announced fiscal expansion output actually falls because interest rates rise before the impact of the future spending increase feed through the economy. This, together with the monetary contraction, explains the 1982 recession as well as the strong U.S. dollar. Both tight monetary policy and expected expansionary fiscal policy, appreciate the real exchange rate because they raise the real interest rate. Long-term nominal interest rates have a tendency to rise because of the rise in long term real interest rates which results from the expected fiscal expansion. This is offset by the lower expected inflation rate from the monetary tightness. The fiscal stimulus in the U.S. economy begins to flow into output by 1984, when real growth is close to 3.4 percent. This is less than the outcome of nearly 7 percent; in the model, the strong growth also spills over into 1985, where the average growth over 1984 and 1985 is approximately equal to that experienced. By 1983 the contractionary German fiscal policy adds further to the Dollar's appreciation relative to the Deutschemark.

Up to 1985 the model tracks very well in terms of broad trends in the data.^[6] From 1985 the performance of the model is not quite as good. It is not clear if this is due to accumulating errors from the earlier 6 years or if some factors not present in the model become important.

In 1985, it is assumed that a shift in expectations about global monetary policy occurs. Both actual and expected U.S. monetary policy becomes more expansionary while actual and expected German and Japanese monetary policy tightens. This causes a large real and nominal depreciation of the dollar relative to the Deutschemark and yen in 1985, 1986 and 1987. In the model, the real appreciation of the dollar relative to the yen by 1985 is 10 percent more than experienced. The fall in the dollar relative to both the yen and the Deutschemark during 1986 and 1987, is close to that experienced, although the level of the yen/dollar rate settles at a rate about 10 to 15 percent higher than actually experienced during 1988. The result from the model for the dollar/Deutschemark exchange rate is much closer to the actual. In addition, the German and the Japanese trade imbalances improve by more than experienced over this period. The tracking of the U.S. trade balance is quite good which suggests that the behavior of the ROECD, REMS , LDC or OPEC regions may be causing the excessive improvement of the Japanese and German trade imbalances. There is also a built-in expectation of Japanese fiscal expansion from 1987 which partly explains the Japanese trade balance turn-around in the model.

Global inflation begins to rise gradually from 1986, although the largest rise occurs in the U.S.. The inflation rise in the U.S. is less than the monetary expansion would predict because of the convenient fall in OPEC oil prices in 1986. In Japan, from 1987, there is a steady rise in inflation, reflecting the strong rise in demand. U.S. inflation by 1988 is 6.4 percent, which is above the level experienced.

The results from 1988 on, illustrate that the monetary policy induced depreciation of the U.S. dollar only goes part of the way to reducing the U.S. trade imbalance. The fundamental reason is that, as was shown in the section on policy multipliers, a monetary expansion depreciates the exchange rate and tends to raise exports, but it also stimulates demand and raises imports with very little improvement in the overall trade balance. Any improvement in the U.S. trade balance reflects the partial fiscal adjustment in the U.S.. The net effect is the prospect of very little adjustment of U.S. trade imbalances up to 1991, given the lack of further U.S. fiscal adjustment assumed in the simulation.

It should also be pointed out that the share markets in the model do not experience the scale of the surge of 1986 to 1987 nor the subsequent crash.

Several caveats should be made about the results from 1988 to 1991. First, the results from 1988 on, inherit a good deal of inertia from any errors from earlier periods, since we do not adjust the model for errors accumulated from 1979. These accumulated errors, after a decade, could potentially be quite large. Our intention in this exercise was to put as much burden as possible on the model to explain history without adjusting for errors. Second, the forecasts assume that there are no significant changes in policy from 1987 on. This assumption is only made for convenience because, as shown in McKibbin and Sachs (1988a), governments are likely to have incentives to change policy, especially if they continue to target the trade imbalances with monetary policy alone or if the recent commodity price surge feeds into inflation, as the OPEC price shock did in 1979–80. A tightening of monetary policy in one major country has a tendency to lead to excessive global tightening due to the international linkages between countries which are commonly ignored by policymakers. The extent of the monetary contraction in 1981–82 in the U.S. surprised many observers because the rest of the world echoed the U.S. policy change which made the global consequences quite severe.

Footnote

Note that in this study, the increases in the U.S. fiscal deficit from 1982 to 1985 are assumed to be permanent. Morris (1988) attempts to track the real exchange rate for the period to 1985 in a small empirical IS/LM model, assuming that fiscal deficits from 1982 were perceived to be temporary and that each year, the continuing U.S. fiscal deficit was a surprise. [6]