ANALYSES OF MODEL DERIVED ISLM, AGGREGATE DEMANDAGGREGATE SUPPLY, AND BP CURVES* The group of the EPA World Model Economic Research Institute Economic Planning Agency * This paper was presented at the Sixth EPA International Symposium How Far Have International Payments Adjustment Made Progress U.S. and Japan)? held on February 89, 1989 in Tokyo.
ANALYSES OF MODEL DERIVED ISLM, AGGREGATE DEMANDAGGREGATE SUPPLY, AND BP CURVES I. Introduction...55 II. The ISLM Curves and the Impact of Fiscal Policy...57 1. The Textbook ISLM Curves and their Determinants...57 (1) The Slope of the IS Curve and its Determinants...57 (2) The Slope of the LM Curve and its Determinants...60 2. The Impact of Fiscal Policy and ISLM Curves...63 (1) Analytical Framework...63 (2) The Size of the Initial Shift of the IS Curve and its Determinants...66 (3) The Slopes of the IS and LM Curves and Transaction Crowding Out...68 (4) Wealthchangeinduced Shifts in the IS and LM Curves and Portfolio Crowding Out...68 (5) Pricechangeinduced Shifts in the IS and LM Curves and Inflation Crowding Out...71 (6) The Overall Impact of Fiscal Policy and its Determinants...74 (7) The Pure LM Curve and the Modelderived LM Curve...79 III. The Aggregate Demand and Aggregate Supply Curves and the Impact of Fiscal Policy...81 1. The Slope of the Aggregate Demand Curve and its Determinants...81 (1) Analytical Framework...81 (2) The Slope of the Aggregate Demand Curve and its Determinants...81 2. The Slope of the Aggregate Supply Curve and its Determinants...84 (1) Analytical Framework...84 (2) The Slope of the Pure AS Curve and its Determinants...84 (3) The Modelderived AS Curve Compared with the Pure AS Curve...86 3. The Impact of Fiscal Policy and the ADAS Curve...90 (1) The Size of the AD Shift and its Determinants...90 (2) Inflation Crowding Out and the ADAS Slopes...90 (3) The Impact of Fiscal Policy and its Determinants...95 IV. The BP Curve and the Impact of Fiscal Policy...97 1. The Slope of the BP Curve and its Determinants...97 (1) Analytical Framework...97 (2) The Slope of the Pure BP Curve and its Determinants...97 (3) The Modelderived BP Curve Compared with the Pure BP Curve...101 53
2. The Impact of Fiscal Policy and the LMBP Curves...103 (1) The Impact on the Overall Balance of Payments and the LM and BP Curves under a Fixed Exchange Rates Regime...103 (2) The Impact on the Exchange Rate and the LM and BP Curves under a Flexible Exchange Rates Regime...106 V. Conclusions... 110 1. Main Conclusions... 110 2. Remaining Issues... 111 54
I. Introduction The purpose of this paper is to explain the differences in the macroeconomic impact of fiscal policy among countries by drawing upon the IS, LM, BP, and other associated curves derived for each of the countries from the Economic Planning Agency s World Economic Model and by examining the differences in the slopes and shifts of these curves. This paper is composed of five parts. Following this Introduction, part II analyzes the differences in the effects of fiscal policy among countries using the framework of the ISLM curves. While similar analysis has been attempted elsewhere by the EPA [1], that work was unable to delineate the pure ISLM curves and their shifts clearly. Thus this paper takes the factors accounting for the shifts (i.e. changes in wealth and prices) as indicative and attempts thereby to clarify the factors accounting for the differences in fiscal multipliers among countries. In part III, an effort is made, using the framework of the aggregate demand and aggregate supply curves, to explain the country differences in the impact of fiscal policy. Although the aggregate supply curve is of only limited significance because of inadequacies in the EPA model s supply block, it does contribute to an understanding of the model s features. In part IV we derive the BP curve. Although there are a number of difficulties involved in deriving the BP curve, this paper demonstrates one possibility and attempts to explain the country differences in the effects of fiscal policy on the balance of payments and exchange rates. Finally, part V summarizes the main conclusions and offers some thoughts on the remaining issues. Based upon a comparison of the differences among the nine countries included in the EPA s World Economic Model (the G7 countries plus Australia and the Republic of Korea), a particular effort has been made to address the following three issues in light of the questions raised at the Fourth EPA International Symposium held in March 1988. At the same time, addressing these questions will, it is hoped, provide background information on the simulations done using the EPA model for session two of this Sixth EPA International Symposium. The first issue is why fiscal expansion has such very different results in Japan and the United States in the U.S. fiscal multipliers get smaller over time, but in Japan they get larger. The second issue is why there has been such a vast difference in price responses to fiscal expansion in Japan and the U. S. prices rising sharply in the U.S. but very little in Japan. And the third issue is to what extent the exchange rate response to fiscal expansion can be explained by the BP curve as derived in this paper. 55
The model used in this analysis is the EPA World Economic Model the third version drawn up in September 1987 [2] and that the simulation period is the latest six s in sample period (197984). 56
II. The ISLM Curves and the Impact of Fiscal Policy 1. The Textbook ISLM Curves and their Determinants (1) The Slope of the IS Curve and its Determinants The IS curve differs depending upon which variables are endogenized and which are exogenized (i.e., depending upon the scope of the endogenized variables). The IS curve that is closest to the textbook IS curve is thought to be that which endogenizes only the final expenditure block except capital stock. Table II1 shows the IS curve derived this way and summarizes the underlying income and interest elasticities of the aggregate demand for the third. As seen, the West German IS curve is very flat and the U.S., Italian, and Japanese curves are also rather flat; while the curves for the other countries are steep, with particularly steep curves described for Australia and the Republic of Korea. These differences are generally explained by the differences in the interest elasticity of the aggregate demand (especially investment), and the income elasticity is thought to have little effect. Thus the greater the interest elasticity, the flatter the IS curve is likely to be; and the smaller the interest elasticity, the steeper the IS curve. Let us see what happens to the slope of the IS curve when the scope of endogenous variables is broadened while the final expenditure block is still endogenous. Table II2 indicates the results in the three successively broader cases of (i) also endogenizing capital stock in the final expenditure block, (ii) additionally endogenizing the income distribution and government blocks, and (iii) additionally endogenizing the balance of payments and financial blocks. Endogenizing capital stock leads to smaller GNP expansion under lower interest rates because of the effect of stock adjustment (namely that the larger stock tends to depress investment) and hence the IS curve becomes steep compared with the case where capital stock is not endogenized. It is generally expected that endogenizing the income distribution and government blocks leads to larger GNP expansion under lower interest rates because of possible increases in disposable income and that the IS curve therefore becomes flatter. However, the decline in interest income resulting from lower interest rates has the reverse effect on disposable income and leads to a steeper IS curve in countries where interest income is endogenized. The United States and Japan are the only countries for which interest income is endogenized in the EPA model. Both countries clearly demonstrate this effect and show marginally steeper IS curves. The curve becomes flatter in the other six countries for which interest income is not endogenized. 57
Table II1 Slope of the Pure IS Curve and Income & Interest Rate Elasticities of the Aggregate Demand (3rd ) Slope of the pure IS 1) Income elasticity Interest rate semi elasticity West Germany Δ 0.54 0.35 Δ 1.24 U. S. Δ 1.13 0.97 Δ 0.49 Italy Δ 1.68 0.83 Δ 0.27 Japan Δ 1.72 0.76 Δ 0.46 France Δ 2.58 0.39 Δ 0.25 U. K. Δ 3.15 0.55 Δ 0.19 Australia Δ 4.50 1.03 Δ 0.073 Korea Δ 5.79 0.73 Δ 0.068 (Notes) 1. The slope of the IS is measured by deviation from baseline in interest rate divided by percentage deviation from baseline in GNP. 2. In Canada, the slope of the IS is not calculated with technical reason 3. Δ indicates minus (this is the same in subsequent tables) 58
Table II2 Changes in the Slope of the IS Curve by the Different Scope of the Endogenized Variables (3rd ) Block and variables endogenized cumulatively Final 1 2 3 4 expenditure block Capital stock Distribution Financial and (exc.capital variables and government the balance of stock variables) block payments block West Germany Δ 0.54 Δ 0.67 Δ 0.33 Δ 0.66 U. S. Δ 1.13 Δ 1.42 Δ 1.46 Δ 1.21 Italy Δ 1.68 Δ 1.92 Δ 0.79 Δ 1.49 Japan Δ 1.72 Δ 1.90 Δ 2.01 Δ 2.12 France Δ 2.58 Δ 2.61 Δ 1.05 Δ 1.62 U. K. Δ 3.15 Δ 3.44 Δ 1.72 Δ 13.45 Australia Δ 4.50 Δ 9.04 Δ 2.23 Δ 13.87 Korea Δ 5.79 Δ 9.46 Δ 2.36 Δ 12.77 (Notes) 1. See notes of Table II1. 2. Figures in column (1) are the same as the slope of the pure IS in Table II1. 59
It is also expected that endogenizing the balance of payments block should lead to an increase in imports and hence a decrease in GNP, thereby yielding a steeper IS curve. However, endogenization of the balance of payments block results in a flatter IS curve for the United States. This is attributed to the fact that GNP actually expands as the effect of a decline in interest payments accruing from lower interest rates outweighs the impact of greater imports. As seen, the slope of the IS curve changes considerably depending upon the scope of variables that are endogenized in the model. However, there is little change in the eight countries relative positions whether one uses the textbook type pure IS curve or the IS curve derived with the broadest endogenization. The fiscal multiplier changes as time passes on after fiscal shock is given. In order to understand the temporal changes in the fiscal multiplier, it is important to look at the temporal changes in the slope of the IS curve. Looking first at the pure IS curve, this gets flatter for the first few s and then becomes steeper for the remaining s a result of the fact that income and interest rates have such a lagged impact on consumption and private investment, that the income and interest elasticities of aggregate demand become greater over time for the first few s, and that these elasticities then peak out and decline in subsequent s as shown in Table II3 1. The same time profile of changing slopes is observed in the IS curve derived with the broadest endogenization. But the scale of changes in the slope is larger as the capital stock adjustment effect gets stronger over the s. (2) The Slope of the LM Curve and its Determinants Unlike the IS curve, the financial block is highly independent from the other blocks. The result is that the slope of the LM curve changes very little even when the scope of endogenous variables is varied. Table II4 gives the slope of the LM curve for the third derived from the full model (except supply block) simulations. The income and interest elasticities of money demand remain the main determinants of the slope of the LM curve. In Canada s case, the LM curve is completely elastic because the shortterm interest rate function (the inverted form of the money demand function) does not include the income variable as an explanatory variable. For the other countries, the curves are very flat for West Germany and rather flat for the United Kingdom and France. By contrast, the curves for the remaining five countries are steep, being particular steep for the Republic of Korea. These differences can be largely explained by the differences in the interest elasticity of money demand, the LM curve being steeper the smaller the interest elasticity. 60
Table II3 Slope of the Pure IS Curve and Income & Interest Rate Elasticities of the Aggregate Demand (Temporal Change) Slope of the pure IS 1) Interest rate semielasticity of Aggregate Demand Income elasticity of Aggregate Demand the flattest 6 th 6 th 6 th West (2) 2) (2) 2) (2) 2) Germany Δ 0.67 Δ 0.43 Δ 0.56 Δ 0.75 Δ 1.17 Δ 1.26 0.26 0.40 0.26 (4) (4) (4) U. S. Δ 1.21 Δ 1.09 Δ 1.14 Δ 0.38 Δ 0.52 Δ 0.50 0.75 1.00 0.97 (5) (5) (5) Italy Δ 4.58 Δ 1.44 Δ 1.52 Δ 0.12 Δ 0.32 Δ 0.32 0.37 0.98 0.99 (3) (3) (3) Japan Δ 4.21 Δ 1.72 Δ 1.84 Δ 0.24 Δ 0.46 Δ 0.46 0.49 0.76 0.74 (5) (5) (5) France Δ 52.85 Δ 2.31 Δ 3.24 Δ 0.06 Δ 0.26 Δ 0.26 0.26 0.42 0.41 (3) (3) (3) U. K. Δ 3.57 Δ 3.15 Δ 3.63 Δ 0.19 Δ 0.19 Δ 0.18 0.17 0.55 0.69 (5) (5) (5) Australia Δ 12.12 Δ 3.34 Δ 4.91 Δ 0.06 Δ 0.07 Δ 0.07 0.66 1.08 1.03 (4) (4) (4) Korea Δ 9.58 Δ 5.41 Δ 6.90 Δ 0.06 Δ 0.06 Δ 0.05 0.27 0.76 0.77 (Notes) 1. See notes of Table II1. 2. Figures in parentheses indicate the in which the slope of the pure IS is flattest. 61
Table II4 Slope of the Pure LM Curve and Income & Interest Rate Elasticities of Money Demand (3rd ) Slope of the pure Income Interest rate LM curve 1) elasticity semielasticity Canada 0.0 0.0 Δ 119.32 West Germany 0.09 0.87 Δ 11.41 U. K. 0.20 1.70 Δ 6.09 France 0.21 21.73 Δ 69.45 Italy 0.38 1.06 Δ 2.67 Japan 0.43 0.68 Δ 2.45 Australia 0.55 1.10 Δ 1.67 U. S. 0.59 1.07 Δ 3.71 Korea 1.38 1.17 Δ 1.11 (Note) 1. The slope of the LM is measured by deviation from baseline in interest rate divided by percentage deviation from baseline in GNP. 62
The temporal changes in the slope of the LM curve are also important for understanding the temporal changes in the fiscal multiplier. Excluding Canada, where the slope is consistently zero, three types may be observed: (i) the United States, France, Australia, and Italy, which become almost consistently steeper over time, (ii) Japan, the United Kingdom, and the Republic of Korea, which start out steep but turn flat after the second, and (iii) West Germany, which becomes almost consistently flatter over time. As such, the LM curve stands in sharp contrast to the IS curve, which starts out flat and then turns steeper in all countries. This difference in the LM curves' temporal progressions can largely be explained by the differences in temporal changes for the interest elasticity of money demand to wit: (i) interest elasticity becoming progressively smaller for group one countries, (ii) interest elasticity being small at first yet growing larger over time for group two countries, and (iii) interest elasticity becoming consistently large for West Germany as shown in Table II5 2. 2. The Impact of Fiscal Policy and ISLM Curves (1) Analytical Framework This section attempts to explain the differences in the effects of fiscal policy among countries drawing upon the differences in the slopes and shifts of the IS and LM curves. The framework is represented by Figure II1. (The basic approach underlying this framework is based on B. Friedman [3] and B. Hickman [4].) IS0 and LM0 in Figure IIl represent the IS and LM curves prior to fiscal shock. With fiscal shock, the IS curve shifts from IS0 to IS1 (hereinafter referred to as the initial shift). If the LM curve is absolutely flat, the point of equilibrium moves from A to C. However, so long as the IS and LM curves have slopes, the point of equilibrium shifts to D through both an increase in interest rates and a decline in GNP. The decline in GNP from C to D is referred to as transaction crowding out. While this analysis holds if prices and wealth remain unchanged, the increase in government expenditures is accompanied by an increase in wealth and a rise in prices. In Figure IIl, the result of this increase in wealth shows up in the IS curve s assumed expansionary shift from IS1 to IS2 due to a wealthinduced consumption increase and the LM curve s contractionary shift from LM0 to LM1 due to a wealthinduced money demand increase. The resultant decline in GNP from D to E is referred to as portfolio crowding out. (It should be noted, however, that portfolio crowding in may occur when the IS curve s expansionary shift outweighs the LM curve s contractionary shift as it does in Figure II1.) 63
Table II5 Slope of the Pure LM Curve and Income & Interest Rate Elasticities of Money Demand (Temporal Change) Slope of the pure LM 1) Interest rate semielasti city of money demand Income elasticity of money demand (1) 6th(2) (2) (1) (3) 6th (4) (4) (3) (5) 6th(6) (6) (5) Canada 0.0 0.0 0.0 Δ 55.31 Δ 183.25 3.31 0.0 0.0 0.0 West Germany 0.11 0.08 0.73 Δ 7.91 Δ 13.30 1.68 0.62 0.89 1.44 U. K. 0.18 0.15 0.83 Δ 5.94 Δ 8.82 1.48 1.70 1.70 1.0 France 0.12 0.20 1.67 Δ 98.89 Δ 94.02 0.95 21.73 21.73 1.0 Italy 0.32 0.39 1.22 Δ 3.14 Δ 2.67 0.85 1.06 1.06 1.0 Japan 0.41 0.34 0.83 Δ 2.72 Δ 2.84 1.04 0.71 0.62 0.87 Australia 0.38 0.52 1.37 Δ 3.22 Δ 2.10 0.65 1.10 1.10 1.0 U. S. 0.26 0.66 2.54 Δ 7.14 Δ 2.90 0.41 1.07 1.00 0.93 Korea 0.80 0.90 1.13 Δ 0.91 Δ 1.55 1.70 0.74 1.17 1.58 (Note) 1. See note of Table II4. 64
Figure II1 Framework of the ISLM under Fiscal Impact Interest rate Model-derived Model-derived GNP (Notes) 1. GNP is measured by percentage change and interest rate is measured by difference. 2. Subscript 0 means standard case, and the others show cases after fiscal impact. 3. For details, see II2(1) in the paper. 65
Likewise, the impact of the rise in prices shows up in the IS curve s contractionary shift from IS2 to IS3 and the LM curve s contractionary shift from LM1 to LM2. The shift from E to B the decline in GNP as prices rise along the line with the aggregate supply curve is referred to as inflation crowding out. (2) The Size of the Initial Shift of the IS Curve and its Determinants The fiscal policy examined here is a sustained increase in real government expenditures equivalent to 1% of real GNP. The horizontal axis in Figure II1 represents the percentage deviation from the base line real GNP. As seen in Table II6, the size of the initial shifts of the IS curve under the fiscal shock differs considerably from country to country 3. The following simple multiplier calculations demonstrate that there will be differences in the sizes of initial shifts caused by differences in marginal propensities to consume, import, etc. Designating the marginal propensity to consume c, the marginal propensity to invest i, and the marginal propensity to import m, when government expenditures increase by Δ G,the increase in GNP will be Δ G/(1ci+m). This simplified fiscal multiplier are shown in Table II6, which can be compared with the size of the initial shift of the IS curve under fiscal shock simulations. This comparison indicates that the scale of the initial shift of the IS curve is generally close to the value for the simplified fiscal multiplier. In the simplified calculations of fiscal multiplier, the marginal propensities to consume, invest and import are assumed to remain constant over time. In contrast, however, macro model simulations capture changing propensities over time. For instance, the lagged impact of income on consumption and private investment gets increasingly larger over time, whereas the capital stock adjustment and a lagged effect of income on imports tend to reduce the fiscal multiplier. Looking at the temporal progression in the initial shift, the size of the shift gradually diminishes over the six s in the United States and Canada, while it grows annually in the United Kingdom. And there is an initial decline in the initial shift followed by increase for West Germany. In the other five countries, grows larger at first (the first three to five s) and then shrinks. An important point that we can derive from examining the determinants of the scale of the initial shift of the IS curve under fiscal shock is that the time profile of the fiscal multiplier is influenced not only by induced changes in interest rates, net wealth, and prices but also by the temporal progression of the initial shift of the IS curve, i.e. the temporal progression of the fiscal multiplier in the 66
Table II6 Comparison between the Initial Shift in the IS and the Simplified Fiscal Multiplier Initial Shift 1) Simplified Fiscal Multipier 1) peak (bottom) 6th peak (bottom) 6th U. S. Canada France Italy West Germany Australia Korea Japan U. K. 1.87 1.58 1.41 1.41 1.31 1.26 1.24 1.09 1.05 monotonic decline monotonic decline (5) 2) 1.62 (5) 2.82 (4) 0.53 (3) 1.49 (3) 1.97 (5) 1.52 monotonic increase 1.61 ( Δ 0.26) 3) 1.24 ( Δ 0.34) 1.59 (0.18) 2.72 (1.31) 0.63 ( Δ 0.68) 1.19 ( Δ 0.07) 1.69 (0.45) 1.46 (0.37) 1.57 (0.52) 1.73 1.38 1.26 1.28 1.27 1.63 1.26 1.46 1.16 (2) 2) 1.84 monotonic decline (3) 1.52 (5) 1.70 (2) 1.39 (2) 1.84 (2) 1.61 (2) 1.71 monotonic increase (Notes) 1. Both the initial shift and the simplified fiscal multiplier are measured by percentage deviation from baseline in GNP. 2. Figures in parentheses indicate peak (or bottom) 3. Figures in parentheses indicate differences between 6th and. 1.67 ( Δ 0.06) 3) 1.19 ( Δ 0.19) 1.40 (0.14) 1.67 (0.39) 1.23 ( Δ 0.04) 1.64 (0.01) 1.55 (0.29) 1.64 (0.18) 1.64 (0.48) 67
simple Keynesian framework where there are no induced changes in interest rates, net wealth, and prices. It is thus worth noting that the country differences in the time profile of the fiscal multiplier reflect the differences in this simple Keynesian fiscal multiplier as well as crowding out. (3) The Slopes of the IS and LM Curves and Transaction Crowding Out Table II7 shows the slopes of the IS and LM curves and the extent of transaction crowding out in the third based upon the EPA model simulations. In general, transaction crowding out is larger the flatter the IS curve and the steeper the LM curve. As seen, however, it is impossible to attribute country differences in the scale of transaction crowding out either to country differences in the slope of the IS curve or to those of the LM curve. But the size of transaction crowding out corresponds with the relative slope of the IS curve to the LM curve. The transaction crowding out is most prominent in the United States, where the slope of the IS curve is quite flat relative to that of the LM curve and 27% of the initial shift is crowded out. There is also greaterthan10% crowding out in Italy, West Germany (due to a quite flat IS curve), Japan (both the slopes of the IS and LM curve are medium), and the Republic of Korea (quite steep LM curve), but crowding out is less prominent in the other four countries. In Canada, the fact that the LM curve is completely flat, for the reasons which have already been explained, means that there is no crowding out at all. The temporal progression for transaction crowding out demonstrates a pattern of increasing at first (in the first two to five s) and then diminishing (see Table II8). These temporal changes in transaction crowding out correspond to the temporal progressions shown by the slope of the IS and LM curves. Canada and Italy are exceptions for this general time profile for the transaction crowding out. In Canada there is none and in Italy the crowding out becomes consistently large. (4) Wealthchangeinduced Shifts in the IS and LM Curves and Portfolio Crowding Out When an increase in government expenditures is financed by issuing government bonds and also accompanied by an increase in real assets due to increased private investment, there is also an increase in wealth (financial net worth defined as the sum of the cumulative government deficit and the cumulative current account surplus and total net worth defined as the sum of financial net worth and capital stock) 4. As a result of the wealth increase, the IS curve shifts rightward and the LM curve leftward. While the leftward shift in the 68
Table II7 Relation between the Size of Transaction Crowdingout and the Slope of the ISLM (3rd ) (1) (2) (3) (4) Initial 1) Transaction 1) Rate of loss Slope 2) Slope 2) (3) Shift Crowdingout By Transaction of the of the (4) Crowdingout IS LM ((2)(1)) U. S. 1.75 0.47 0.27 Δ 1.63 0.59 2.76 Italy 2.46 0.41 0.17 Δ 1.92 0.38 5.05 West Germany 0.79 0.12 0.15 Δ 0.50 0.09 5.56 Japan 1.48 0.21 0.14 Δ 2.55 0.43 5.93 Korea 1.97 0.20 0.10 Δ 13.55 1.38 9.82 France 1.60 0.11 0.07 Δ 2.94 0.21 14.00 Australia 1.49 0.06 0.04 Δ 14.08 0.55 25.60 U. K. 1.32 0.02 0.01 Δ 14.13 0.20 70.65 Canada 1.38 0.00 0.00 0.0 (Notes) 1. Both the initial shift and transaction crowdingout are measured by percentage change in GNP. 2. The slope of the IS and LM is measured by deviation from beseline in interest rate divided by percentage deviation from baseline in GNP. 3. In Canada, the slope of the IS is not calculated with technical reason. 69
Table II8 Relation between the Size of Transaction Crowdingout and the Slope of the ISLM (Temporal Change) Rate of Loss by 1) Transaction Crowdingout Slope of the pure IS 2) Slope of the 2) pure LM Peak 6th 6th 6th (2) 3) (2) 3) (2) 3) U.S. 0.18 0.28 0.25 Δ 1.15 Δ 1.46 Δ 1.99 0.26 0.56 0.66 monotonic monotoni monotonio Italy 0.05 increase 0.25 Δ 5.66 ously flat Δ 1.21 0.32 usly steep 0.39 (2) (2) (2) West Germany 0.17 0.21 0.06 Δ 0.54 Δ 0.38 Δ 1.34 0.11 0.10 0.08 (3) (3) (3) Japan 0.07 0.14 0.08 Δ 5.73 Δ 2.55 Δ 4.14 0.41 0.43 0.34 (3) (3) (3) Korea 0.04 0.10 0.03 Δ 1.33 Δ 13.55 Δ 32.98 0.80 1.38 0.90 (5) (5) (5) France 0.01 0.08 0.07 Δ 17.14 Δ 2.11 Δ 5.53 0.12 0.19 0.20 (2) (2) (2) Australia 0.02 0.04 0.01 Δ 18.28 Δ 12.58 Δ 61.60 0.38 0.52 0.52 (2) (2) (2) U. K 0.01 0.02 0.01 Δ 11.75 Δ 13.83 Δ 15.09 0.18 0.22 0.15 Canada 0.0 0.0 0.0 0.0 (Notes) 1. Rate of loss by transaction crowdingout is measured by transaction crowdingout divided by the initial shift. 2. The slope of the ISLM is measured by deviation from baseline in interest rate divided by percentage deviation from baseline in GNP. 3. Figures in parentheses indicate the in which transaction crowding out is largest. 4. In Canada the slope of the IS is not calculated with technicel reason. 70
LM curve is a factor for crowding out, the rightward shift in the IS curve is a factor for crowding in, such that whether portfolio crowding out or portfolio crowding in occurs depends upon whether it is the IS curve or the LM curve that shifts more. In the EPA model, the five countries showing IS shifts as a result of changes in wealth are the United States, Japan, West Germany, Canada, and Australia. In the third, because the wealth effect on the money demand is large in West Germany and Japan, there is a considerable leftward shift in the LM curve, resulting in portfolio crowding out. Yet in Australia, the United States, and Canada, as the wealth effect on consumption is considerable, the rightward shift in the IS curve outweighs the leftward shift in the LM curve, resulting in portfolio crowding in (as shown in Table II9). In the other three countries (France, the United Kingdom, and Italy), portfolio crowding out occurs because only the LM curve shifts leftward. However, the shift is so slight in Italy that the crowding out is negligible. The temporal progression in portfolio crowding out is such that (i) there is a consistent increase in crowding in in Australia, the United States, and Canada, (ii) there is an initial increase in crowding out followed by diminution in Japan and West Germany (with crowding in finally Occurring in Japan), and (iii) crowding out is negligible throughout the six s in France, the United Kingdom, and Italy. These differences among countries correspond to the differences in the temporal progressions of the shifts of the IS and LM curves (see Table II10). (5) Pricechangeinduced Shifts in the IS and LM Curves and Inflation Crowding Out The LM curve always shifts leftward with a rise in prices caused by an increase in government expenditures. In contrast, however, whether the IS curve shifts leftward or rightward is not determined a priori. This is because the IS curve is subject both to leftwardshift pressures from the declines in real income and real wealth, resulting from the rise in prices, and to rightwardshift pressures from the decline in real interest rates. Looking at the results for the third, the IS curve shifts leftward in the seven countries of Australia, Canada, France, Japan, the Republic of Korea, the United Kingdom, and the United States where the former factors (the declines in real income and real wealth) are dominant. However in West Germany and Italy the IS curve shifts rightward because the latter factor (the decline in real interest rates) is dominant and also because real income increases since the consumption deflator rises less than the other deflators. Because the LM curve shifts leftward in all of the countries covered as mentioned above, inflation crowding out occurs in all seven of the countries where 71
Table II9 Relation between the Size of Portfolio Crowdingout and Wealth Changeinduced Shifts in the ISLM Curves (3rd ) (1) Initial 1) Shift (2) Portfolio 1) Crowdingout Rate of Loss by Portfolio Wealth Changeinduced Shifts in Slope of Crowding out (in) ((2)(1)) IS 2) LM 3) IS LM West Germany 0.79 0.21 0.26 0.11 0.19 Δ 0.50 0.09 Japan 1.48 0.07 0.05 0.08 0.42 Δ 2.55 0.43 France 1.60 0.03 0.02 0.08 Δ 2.94 0.21 U. K. 1.32 0.01 0.01 0.12 Δ 14.13 0.20 Italy 2.46 Δ 0.0 Δ 0.0 0.04 Δ 1.92 0.38 Canada 1.38 Δ 0.18 Δ 0.13 0.19 0.08 0.0 U. S. 1.75 Δ 0.49 Δ 0.28 0.72 0.38 Δ 1.63 0.59 Australia 1.49 Δ 0.44 Δ 0.30 0.47 0.23 Δ 14.08 0.55 (Notes) 1. Both the initial shift and portfolio crowdingout are measured by percentage change in GNP. 2. The shift in the IS is measured by percentage change in GNP. 3. The shift in the LM is measured by difference in interest rate. 4. There is no wealth effect on consumption in France, the U.K., and Italy. 5. There is no wealth effect both on consumption and on the money demand in Korea. 6. In Canada, the slope of the IS is not calculated with technical reason. 72
Table II10 Relation between the Size of Portfolio Crowdingout and Wealth Changeinduced Shifts in the ISLM Curves (Temporal Change) Rate of Loss by 1) Portfolio Crowdingout Wealth 2) Changeinduced Shifts in IS Wealth 3) Changeinduced Shifts in LM peak 6th 6 th 6 th West (3) 4) (3) 4) (3) 4) Germany 0.09 0.26 0.10 0.02 0.11 0.19 0.15 0.19 0.19 (2) (2) (2) Japan 0.04 0.06 Δ 0.11 0.00 0.03 0.31 0.13 0.30 0.64 (3) France 0.00 0.02 0.02 0.06 (3) 0.08 0.08 U. K. Δ 0.00 0.01 Δ 0.03 0.38 Italy Δ 0.00 0.00 Δ 0.02 0.22 Canada monotonic Δ 0.00 decline Δ 0.35 0.01 0.44 0.04 0.13 U. S. Δ 0.03 ( ) Δ 0.43 0.10 1.08 0.15 0.77 Australia Δ 0.02 ( ) Δ 0.79 0.03 0.97 0.13 0.17 (Notes) 1. Rate of loss by portfolio crowdingout is measured by portfolio crowdingout divided by the initial shift. 2. The shift in the IS is measured by percentage change in GNP. 3. The shift in the LM is measured by difference in interest rate. 4. Figures in parentheses indicate the in which portfolio crowdingout is largest. 5. See notes 4, 5 of Table II9. 73
the IS curve shifts leftward. In the two countries where the IS curve shifts rightward, West Germany experiences crowding in because the rightward shift in the IS curve is greater than the leftward shift in the LM curve and Italy experiences crowding out because the leftward shift in the LM curve is the greater (see Table II11). The temporal progression in inflation crowding out shows that inflation crowding out grows almost consistently stronger as prices tend to rise and the leftward shifts of both the IS and the LM curves grow more pronounced over time in the eight countries where crowding out occurs in the third (with the one exception that the effect of the LM curve s leftward shift offsets the effect of the IS curve s rightward shift in Italy). In West Germany, because the IS curve s rightward shift peaks in the fourth, the scale of crowding in is reduced thereafter, and since the IS curve shifts leftward in the sixth, crowding out occurs as in the other countries (see Table II12). (6) The Overall Impact of Fiscal Policy and its Determinants The scale of the fiscal multiplier, which varies considerably from country to country, is determined by the sum of the initial shift and the crowding out as described above. Looking at fiscal multipliers for the third, they are in excess of 1.5 in Italy, the Republic of Korea, and Canada and between 1.1 and 1.3 in France, the U.K., Australia, and Japan, but 0.8 or less in the United States and West Germany. While such country differences in the fiscal multiplier are partly accounted for by the country differences in the size of crowding out (especially the differences in inflation crowding out), much of the country differences in the fiscal multiplier is due to different initial shifts ascribed to the different propensities to consume, invest, and import (see Table II13). With regard to the temporal progression in the impact of fiscal policy, the nine countries covered here are divided into three main groups as shown in Table II14: (i) the United States, West Germany, and Canada, where the fiscal multiplier declines, (ii) France, Italy, the Republic of Korea, and Australia, where it grows at first (for the first two to five s) but then declines, and (iii) Japan and the United Kingdom, where it grows consistently larger for all six s. Whereas crowding out is the major determinant of the time profile of the declining fiscal multiplier for the U.S., France, and Italy (with this being inflation crowding out in the United States and France and transaction crowding out in Italy), the diminution in the initial shift is the primary cause of the declining fiscal multiplier for the four countries of West Germany, Canada, Australia, and the Republic of Korea. It is interesting to note that crowding out takes place and 74
Table II11 Relation between the Size of Inflation Crowdingout and Price Changeinduced Shifts in the ISLM Curves (3rd ) (1) (2) Rate of Price Change Slope of Initial 1) Inflation 1) Loss by induced Shift Crowdingout Inflation Shifts in Crowding out ((2)(1)) IS 2) LM 3) IS LM U. S. 1.75 1.02 0.58 Δ 0.98 0.56 Δ 1.63 0.59 Australia 1.49 0.07 0.44 Δ 0.64 0.44 Δ 14.08 0.55 France 1.60 0.18 0.11 Δ 0.18 0.02 Δ 2.94 0.21 Japan 1.48 0.10 0.07 Δ 0.04 0.18 Δ 2.55 0.43 Canada 1.38 0.06 0.04 Δ 0.05 0.00 0.0 Korea 1.97 0.07 0.04 Δ 0.05 0.44 Δ 13.55 1.38 U. K. 1.32 0.03 0.02 Δ 0.03 0.01 Δ 14.13 0.20 Italy 2.46 0.04 0.02 0.15 0.40 Δ 1.92 0.38 West Germany 0.79 Δ 0.12 Δ 0.15 0.44 0.17 Δ 0.50 0.09 (Notes) 1. Both the initial shift and inflation crowdingout are measured by percentage change in GNP. 2. The shift in the IS is measured by percentage change in GNP. 3. The shift in the LM is measured by difference in interest rate. 75
Table II12 Relation between the Size of Inflation Crowdingout and Price Change induced Shifts in the ISLM Curves (Temporal Change) Rate of Loss by 1) Inflation Crowdingout Price Change 2) induced Shift in the IS Price Change 3 ) induced Shift in the LM 6th 6th 6th U. S. 0.08 0.83 Δ 0.11 Δ 1.25 0.06 0.95 Australia 0.06 0.62 Δ 0.07 Δ 0.73 0.17 0.59 France 0.02 0.27 Δ 0.03 Δ 0.46 0.00 0.03 Japan 0.01 0.19 Δ 0.00 Δ 0.24 0.06 0.21 Canada Δ 0.01 0.18 0.02 Δ 0.21 Δ 0.00 0.01 Korea 0.00 0.13 Δ 0.00 Δ 0.22 0.01 0.36 U. K. 0.01 0.04 Δ 0.01 Δ 0.06 Δ 0.00 0.02 Italy Δ 0.01 0.03 0.05 0.31 0.18 0.61 West Germany Δ 0.01 0.43 0.02 Δ 0.06 0.01 0.15 (Notes) 1. Rate of loss by inflation crowdingout is measured by inflation crowding out divided by the initial shift. 2. The shift in the IS is measured by percentage change in GNP. 3. The shift in the LM is measured by difference in interest rate. 76
Table II13 Differences in the Fiscal Multipliers among Countries and its Determinants (3rd ) Fiscal 1) multiplier Initial Shift Crowdingout Sum Transaction Portfolio Inflation Italy 2.00 2.46 0.46 0.41 Δ 0.00 0.04 Korea 1.70 1.97 0.27 0.20 0.07 Canada 1.51 1.38 Δ 0.13 0.00 Δ 0.18 0.06 France 1.29 1.60 0.32 0.11 0.03 0.18 U. K. 1.26 1.32 0.06 0.02 0.01 0.03 Australia 1.22 1.49 0.27 0.06 Δ 0.44 0.65 Japan 1.10 1.48 0.38 0.21 0.07 0.10 U. S. 0.75 1.75 1.00 0.47 Δ 0.49 1.02 West Germany 0.58 0.79 0.21 0.12 0.21 Δ 0.12 (Notes) 1. Real government expenditure is increased by 1% of real GNP (sustained shock). 2. All figures are represented by percentage change in GNP. 77
Table II14 Differences in the Fiscal Multipliers among Countries and its Determinants (Temporal Change) Fiscal Multiplier 1) Initial Shift 1) Crowdingout 1) Peak 6th 6th 6th U. S. 1.43 monotonic decrease 0.57 1.87 1.61 0.44 1.04 ( Δ 0.86) 3) ( Δ 0.26) 3) (0.60) 3) West Germany 0.98 monotonic decrease 0.26 ( Δ 0.72) 1.31 0.63 ( Δ 0.68) 0.33 0.37 (0.04) monotonic Canada 1.61 decrease 1.45 1.58 1.24 Δ 0.03 Δ 0.21 ( Δ 0.16) ( Δ 0.34) ( Δ 0.18) France 1.37 (2) 2) 1.43 1.01 1.41 (2) 2) 1.59 1.59 0.04 (2) 2) 0.16 0.58 ( Δ 0.36) ( 0.18) (0.54) Italy 1.35 (4) 2.12 1.96 1.41 (4) 2.71 2.72 0.06 (4) 0.59 0.76 (0.61) (1.31) (0.70) Korea 1.19 (3) 1.70 1.42 1.24 (3) 1.97 1.69 0.05 (3) 0.27 0.27 (0.23) (0.45) (0.22) Australia 1.18 (5) 1.51 1.39 l. 6 (5) 1.37 1.19 0.08 (5) Δ 0.14 Δ 0.20 (0.21) ( Δ 0.07) ( Δ 0.28) monotonic Japan 0.96 increase 1.24 1.09 1.46 0.13 0.22 (0.28) (0.37) (0.09) monotonic U. K. 1.04 increase 1.47 1.05 1.57 0.01 (0.43) (0.52) (Notes) 1. All Figures are represented by percentage change in GNP. 2. Figures in parentheses indicate the in which fiscal multiplier is largest, 3. Figures in parentheses indicate the difference between 6th and. 0.10 (0.09) 78
accelerates for Japan and the United Kingdom, but that at the same time the fiscal multiplier grows progressively larger with each for these two countries. The continued expansion in the initial shift is a major factor in the steady growth of the fiscal multiplier. In discussing the temporal progression of the fiscal multiplier, it is thus important to look not only at the temporal progression of crowding out but also at the temporal progression in the initial shift (the simplified Keynesian fiscal multiplier). Based upon this analysis, we can now account for the major differences in the temporal progressions of fiscal multipliers between Japan and the United States i.e., for the model property that the multiplier gets larger over time in Japan but diminishes sharply in the United States. Crowding out occurs in Japan, and the extent of this crowding out grows larger over time. However, this crowding out is more than offset by the simplified fiscal multiplier (the initial shift), which also grows larger over time. By contrast, crowding out is much more pronounced for the U.S., and this crowding out gets also much greater for the U.S. over time than for Japan, while the simplified fiscal multiplier (the initial shift) diminishes for the U.S. over time. The result is a sharp decrease in the fiscal multiplier for the U.S. (7) The Pure LM Curve and the Modelderived LM Curve Finally, it is useful to look quickly at the relation between the pure LM curve and the LM curve derived from full model simulation of fiscal expansion. As shown in Figure II1, the change in wealth caused by fiscal policy provokes a shift in the pure LM curve AD to AE (hereinafter referred to as the modelderived LM1) and the change in prices a shift to AB (hereinafter referred to as the modelderived LM2). Because the changes in wealth and prices both provoke leftward shifts in the LM curve, the curves for modelderived LM1 and modelderived LM2 are progressively steeper than the pure LM curve. Three patterns present themselves (as seen in Table II15): (i) the change in slope is particularly pronounced in Canada and West Germany, where the pure LM is very flat, (ii) the slope of model derived LM2 is over twice as steep as that for the pure LM curve in the United States and Australia, where the LM curve s leftward shift resulting from both wealth and price increases is particularly sharp, and in Japan, where there is considerable LM curve shift resulting from the increase in wealth, and (iii) there is little change in the other four countries. 79
Table II15 Relation between the Slope of the Modelderived LM Curves and the Shifts in the ISLM Curves Caused by Wealth and Price Changes (3rd ) Slope of the LM 1) Shifts caused by wealth change Shifts caused by price change Pure Model Model LM derived derived I S 2) LM 3) I S 2) LM 3) LM 1 LM 2 Canada 0.0 0.05 0.06 0.19 0.08 Δ 0.05 0.00 () 4) (1.20) 5) West Germany 0.09 0.48 (5.33) 0.70 (1.46) 0.11 0.19 0.44 0.17 U. K. 0.20 0.29 (1.45) 0.29 (1.00) 0.12 Δ 0.03 0.01 France 0.21 0.26 (1.24) 0.28 (1.08) 0.08 Δ 0.18 0.02 Italy 0.38 0.40 (1.05) 0.60 (1.50) 0.04 0.15 0.40 Japan 0.43 0.78 (1.81) 0.98 (1.26) 0.08 0.42 Δ 0.04 0.18 Australia 0.55 0.67 (1.22) 1.12 (1.67) 0.47 0.23 Δ 0.64 0.44 U. S. 0.59 0.78 (1.32) 1.91 (2.45) 0.72 0.38 Δ 0.98 0.56 Korea 1.38 1.65 (1.20) Δ 0.05 0.44 (Notes) 1. The slope of the LM curve is measured by deviation from baseline in interest rate divided by percentage deviation from baseline in GNP. 2. The shift in the IS is measured by percentage change in GNP. 3. The shift in the LM is measured by difference in interast rate. 4. Figures in parentheses are the slope of the Modelderived LM 1 divided by the slope of the Pure LM. 5. Figures in parentheses are the slope of the Modelderived LM 2 divide by the slope of the Model1derived LM 1. 6. See notes 4, 5 of Table II9. 80
III. The Aggregate Demand and Aggregate Supply Curves and the Impact of Fiscal Policy 1. The Slope of the Aggregate Demand Curve and its Determinants (1) Analytical Framework The aggregate demand (AD) and aggregate supply (AS) curves can be derived from the fiscal policy simulation. Figure III1 shows the framework for this in conjunction with the IS and LM, with the bottom half of the figure being the same as Figure II1. With an increase in government expenditures, the AD curve shifts from AD0 to AD1 (adding transaction crowding out to the IS initial shift). With the addition of portfolio crowding out or crowding in, the AD curve shifts from AD1 to AD2, finally coming to rest at AD2. If the AS curve is perfectly elastic, GNP would be expected to increase to E, but so long as inflation crowding out occurs, prices go up and GNP decreases, finally coming to rest at equilibrium point B. Thus the slope of the AD curve can be derived by connecting the two points B (derived from analysis of the IS and LM curves in section II [the final equilibrium point taking account of inflation crowding out]) and E (taking account of portfolio crowding out or crowding in). The AS curve is then derived by connecting this point B with the initial equilibrium point A. (2) The Slope of the Aggregate Demand Curve and its Determinants As is clear from Figure IIIl, the determinants for the slope of the AD are, seen in correspondence to the ISLM framework, that the AD will be flatter (i) the greater the leftward shift in the IS resulting from price increases, (ii) the greater the LM leftward shift, (iii) the flatter the IS curve, and (iv) the flatter the LM curve. Table IIIl shows the AD slope and its determinants in the third. As seen, the AD slope is generally explainable by the extent of the shift in the IS curve resulting from price increases. As seen in section II, the IS curve shifts rightward for West Germany and Italy with price increases. As a result, the AD curve has a positive slope for West Germany, which is abnormal, but has a normal (albeit very steep) slope for Italy because the impact of the LM leftward shift is great. Grouping the countries according to the steepness of their AD slopes, the three categories found are: (i) the United States and Australia, which have very flat AD curves, (ii) France, the United Kingdom, Japan, and the Republic of Korea, which have rather flat AD curves, and (iii) Canada, Italy, and West Germany, which have steep AD curves. 81
Figure IIIl Framework of the ADAS with the ISLM Price [ADAS] [ISLM] GNP Interest rate GNP (Notes) 1. GNP and price are measured by percentage change and interest rate is measured by difference. 2. Subscript 0 means standard case, and the others show cases after fiscal shock. 3. For details, see III1(1) in the paper. 82
Table III1 Slope of the Aggregate Demand(AD) Curve and its Determinants (3rd ) Slope of the AD 1) Shifts in Slope of 4) I S 2) LM 3) I S LM U. S. Δ 1.1 Δ 0.85 0.49 Δ 1.63 0.59 Australia Δ 1.2 Δ 0.73 0.50 Δ 14.08 0.55 France Δ 3.4 Δ 0.30 0.03 Δ 2.94 0.21 U. K. Δ 4.9 Δ 0.47 0.12 Δ 14.13 0.20 Japan Δ 4.9 Δ 0.09 0.39 Δ 2.55 0.43 Korea Δ 5.1 Δ 0.14 1.21 Δ 13.55 1.38 Canada Δ 8.3 Δ 0.10 0.02 0.00 Italy Δ 22.6 0.14 0.36 Δ 1.92 0.38 West Germany 5.5 0.71 0.28 Δ 0.50 0.09 (Notes) 1. The slope of the AD is measured by percentage change in GNP deflator divided by percentage change in GNP. 2. The shift in the IS is measured by percentage change in GNP normalized by percentage change in GNP deflator. 3. The shift in the LM is measured by difference in interest rate normalized by percentage change in GNP deflator. 4. The slope of the ISLM is measured by difference in interest rate divided by percentage change in GNP. 5. Δ indicates minus (this is the same in subsequent tables) 83
The temporal progressions for the AD curves show four broad groupings: (i) Japan, France, and the United Kingdom, where it becomes consistently flatter, (ii) the United States and Australia, where it becomes flatter until the fourth and then turns steep, (iii) the Republic of Korea and Canada, where it starts out steep for the first few s and then turns flat, and (iv) West Germany and Italy, where it starts out an inverse slope but then shifts to a normal (negative) slope (see Table III2). These differences can be explained primarily with the temporal changes in the IS shift. 2. The Slope of the Aggregate Supply Curve and its Determinants (1) Analytical Framework This section examines both the AS curve derived from the full model simulation (hereinafter the modelderived AS curve) and the AS curve derived from the supply block simulation (hereinafter the pure AS curve). As seen in Figure IIIl, the modelderived AS curve is obtained by connecting points A and B derived from fullmodel simulation. Because the AS curve tracks the combination of anticipated price levels and optimum output levels, the procedure for deriving the pure AS should be to vary the prices exogenously and then to calculate the corresponding GNP levels. However, given the difficulties of this derivation with the EPA model, the procedure followed was that of varying GNP (including its components) exogenously and then calculating the corresponding price levels. (2) The Slope of the Pure AS Curve and its Determinants The supply block of the EPA model can be shown in the following simplified form. (+) () 1. Labor demand function: LE= f(gnp,w/p) LE LE ( 1'. Unemployment rate (definition): UR = ) LF (+)() (+) 2. Wage function: w= f(p,ur,gnp/le) (+) () (+) 3.Price function: P = f(w, GNP/LE, CU) where LF is the labor force, CU the capacity utilization rate, and indicates the rate of change) 84
Table III2 Slope of the AD Curve and its Determinants (Temporal Change) Slope of the AD 1) Shift in the IS 2) Shift in the LM 3) Flattest (or steep est) 6th 6 th 6 th (4) 4) (4) 4) (4) 4) U. S. Δ 2.27 Δ 1.13 Δ 1.27 Δ 0.33 Δ 0.87 Δ 0.74 0.17 0.51 0.56 (4) (4) (4) Australia Δ 5.07 Δ 1.14 Δ 1.42 Δ 0.16 Δ 0.76 Δ 0.59 0.36 0.53 0.48 monotonously France Δ 10.58 flat Δ 2.02 Δ 0.09 Δ 0.55 0.02 0.03 U. K. Δ 9.18 ( ) Δ 3.88 0.14 Δ 0.33 0.11 0.09 Japan Δ 15.12 ( ) Δ 2.62 Δ 0.02 Δ 0.35 0.38 0.30 (2) (2) (2) Korea Δ 1.98 Δ 9.93 Δ 2.05 Δ 0.05 Δ 0.10 Δ 0.45 1.09 1.28 0.74 (3) (3) (3) Canada Δ 4.17 Δ 8.26 Δ 4.05 Δ 0.26 Δ 0.10 Δ 0.22 0.01 0.02 0.01 Italy 29.70 Δ 19.02 0.08 0.19 0.30 0.36 West 6.65 Δ 4.87 0.26 Δ 0.04 0.07 0.11 Germany (Notes) 1.2.3. See notes of Table III 1. 4. Figures in parentheses are the in which the slope of the AD is flattest (or steepest). 85
Assuming this simplified model, the slope of the AS curve is determined primarily by (i) the reciprocal of Okun s coefficient (formula 1 or 1 ) with the unemployment rate reacting to GNP, (ii) the Phillips curve (formula 2) with wages reacting to the unemployment rate, and (iii) the wageprice spiral (formula 2 and 3) with prices reacting to wages. The larger these three factors, the steeper the slope of the pure AS curve. Table III3 shows the slope of the AS curves and these three factors for the third. As seen, there are three groups: (i) the United Kingdom and Canada, where the AS curves are very flat, (ii) the Republic of Korea, Japan, France, Australia, and Italy, where there is a medium incline, and (iii) West Germany and the United States, where the curve is very steep. The factors determining these slopes differ from country to country. For example, in the two countries with very steep inclines (West Germany and the United States), the steepness is primarily because of the strength of the wageprice spiral in the United States and because of the small value of the Okun s coefficient for West Germany. In contrast, the Republic of Korea and Japan have very steep Phillips curves but large Okun s coefficients, which somewhat offset the steep Phillips curves, resulting in the medium steepness of the AS curve. Regarding the temporal progression for the pure AS curve, it grows steeper over time in all of the countries studied. Among the slope determinants, the wage price spiral becomes very strong from time to time (see Table III4). (3) The Modelderived AS Curve Compared with the Pure AS Curve When the AS curve is derived from the full model simulation, it is expected that the increase in capital stock will be accompanied by an increase in productive capacity under expansionary fiscal shock that, in turn, shifts the pure AS rightward, such that the modelderived AS is flatter than the pure AS. However, this does not happen with the EPA model because the present version of the EPA model does not contain the production function (except for the United States and West Germany). Instead, because the labor demand function includes capital stock as complementary to labor, the rise in capital stock with rising employment pushes the pure AS leftward to result in a steeper modelderived AS curve than the pure AS curve in many countries. In the third, the modelderived AS curve is flatter than the pure AS curve in West Germany because of the impact of the increased productive capacity arising from the increase in capital stock; but the AS curve becomes steeper in the other countries because of the rise in wages arising from the increase in capital stock (see Table III5). 86