International comparison of economic performance index: the case of the USA, Japan and Korea.

Lee, Sanghack ; Cheong, Kiwoong ; Suh, Seoung Hwan 等

I. Introduction

Since the economic performance of a nation is reflected in many economic variables, we should consider all relevant economic variables for exact measurement of the economic performance of the nation as a whole. The same is true for the evaluation of economic policies. Suppose, for example, the government plan to increase the new housing supply by 50%. In order to evaluate this policy, we should examine effects of this policy on all relevant economic variables such as national income, price, employment, balance of trade, wage rate, ... etc. The problem is how many variables should be considered in what way.

There are many ways of measuring the economic performance of a nation as a whole. One of widely used methods is to use the "misery index." The misery index measures the state of the economy by summing up the inflation rate and the unemployment rate. While there are many variations of the misery index, they share basically the same advantages and disadvantages. The major advantage of these indices lies on the simplicity of calculation. Since the misery index considers domestic variables only, however, it has the limited power in explaining the performance of the economy when its degree of openness is large.

Another method of measuring the economic performance of a nation is to use the "diamond model." The diamond model evaluates the performance of the economy through the diamond graph, the shape of which is dependent upon the four economic variables of economic growth rate, inflation rate, unemployment rate, and the ratio of current account relative to GDP. The advantage of the diamond model is in considering both domestic and foreign variables. The major disadvantage of the diamond model is the impossibility of the international comparison. In order to perform the international comparison, it is necessary to construct the single index by using four economic variables used in the diamond model.

This paper constructs the economic performance index (EPI) comprising four economic variables of economic growth rate, inflation rate, unemployment rate and the ratio of current account relative to GDP. In constructing the EPI, the structural change of each country will be explicitly considered. By using EPIs of the USA, Japan and Korea, intranational analyses and international comparisons will be performed.

II. Surveys on Methods of Measuring the Economic Performance

There are basically two methods of measuring the economic performance of a nation as a whole. One is the misery index and the other is the diamond model. The misery index is the simple sum of the inflation rate and the unemployment rate.

Let U and p denote the unemployment and inflation rate, respectively. Then, the misery index, M, can be expressed as

M = [absolute value of p] + U, (1)

where [parallel] indicates the absolute value. This index was initially derived from the Phillips curve by A. Okun as a general index representing the healthy state of the economy. Since the unemployment rate is generally inversely related to the inflation rate, as in the Phillips curve, the unusually high misery index reveals that the government failed to solve even one of the two problems of inflation and unemployment. This index is famous for its correct prediction of the likelihood of reelection of incumbent US Presidents after World War II. Most US Presidents with the misery index above 12-13% during their terms of office are found to have failed in the reelection.

Golden, Orescovich and Ostafin (1987) and Wiseman (1992) modify the misery index by using the natural unemployment rate, [U.sub.n]: The variations are as follows.

[M.sub.1] = [absolute value of p] + [absolute value of U - [U.sub.n]] (2)

[M.sub.2] = [absolute value of p] + [absolute value of U - [U.sub.n]](U [greater than or equal to] Un) = [absolute value of p] (U < [U.sub.n]) (3)

The subscripts 1 and 2 in [M.sub.1] and [M.sub.2] are introduced for their distinction from the original misery index. Equation (2) means that, as unemployment rate cannot be below 0, only the difference between the unemployment rate and the natural unemployment rate rather than the unemployment rate itself should be considered in evaluating the economic performance. Equation (3) indicates that the unemployment rate does not affect misery index if it is below the natural unemployment rate.

In the equations (2) and (3), the functional form of the misery index is linear, which implies that the substitution rate between inflation and unemployment rates is constant and unitary. But, according to Zaleski (1990), the indifference curves take the form of the circle centered at the origin. This kind of functional form incorporates the plausible assumption of diminishing substitution rate. We can summarize the modified forms of misery indices as follows:

[M.sub.3] = [p.sup.2] + [U.sup.2] (4)

[M.sub.4] = [p.sup.2] + (U - [U.sub.n]).sup.2] (U [greater than or equal to] [U.sub.n]) = [p.sup.2] (U < [U.sub.n]) (5)

[M.sub.5] = [p.sup.2] + [k(U - [U.sub.n])].sup.2] (U [greater than or equal to] [U.sub.n], k is a constant greater than unity) = [p.sup.2] (U < [U.sub.n]) (6)

The variable k is introduced to reflect the fact that the high unemployment rate is less desirable than the high inflation rate, the evidence of which is based on the empirical results obtained from the US Presidential election [Smyth & Dua (1989)].

Various problems arose in applying the misery index to specific countries. Many attempts were made to solve these problems. For example, Yang (1992) claims that the misery index cannot explain the hyper-inflation situation as in Latin American countries. In order to solve this problem, Yang proposed the generalized weighted misery index.

In relation to the misery index, many empirical researches were made regarding how people recognize the relationship between inflation and unemployment rate. The public indifference map between inflation and unemployment was estimated for New Zealand [Smyth & Woodfield (1993)]. It was found that people were much more concerned with unemployment rate than with inflation rate in case of New Zealand.

Literature on political business cycles generally assumes that the preference of voters on inflation and unemployment rate is concave from the origin. This implies that utility functions of voters take the quadratic form as in the following equation (7), which is conceptually similar to the misery index, [Davis, Hinich & Ordeshook (1970), Nordhaus (1975), MacRae (1977), Smyth, Dua & Taylor (1994)].

S = a + b[P.sup.2] + c[U.sup.2], (7)

Here S denotes the level of voters' utility and a, b, and c are constants. Smyth, et al. (1994) estimate the utility level by using various expected values about inflation rate and unemployment rate.

The diamond model is the only one that simultaneously considers four factors such as economic growth rate, inflation rate, unemployment rate, and the ratio of current account relative to GDP. This model draws the graph of the diamond shape by using aforementioned four economic variables. The larger the size and the more balanced the shape of the diamond, the better the state of the economy. The small and skewed diamond implies that the economy is in unstable state. The disadvantage of the diamond model is that intra-national analyses in time series and international comparisons are virtually impossible since the model evaluates the state of the economy by the graph of each year. To solve this problem, one should integrate four economic variables used in the diamond model into the single index.

III. The Construction of Economic Performance Index

III-I. Theoretical Background

Four economic variables of economic growth rate, inflation rate, unemployment rate, and the ratio of current account to GDP will be considered in constructing the economic performance index (EPI). These variables seem to be most relevant variables that are necessary in evaluating the economic performance of a nation.

Let [X.sub.it] be the value of one economic variable at time t in sector i, (I = 1, 2, 3, 4, and t = 1, 2 ..... N). Here, [X.sub.1t], denotes economic growth rate, [X.sub.2t], is inflation rate, [X.sub.3t], is unemployment rate and [X.sub.4t], is the ratio of current account to GDP, respectively. Since these variables are different in absolute values, normalization is required. The normalized value, Z, is obtained by using the mean and variance of [X.sub.it] as follows:

[Z.sub.it] = [([X.sub.it] - [X.sub.i])/[S.sub.i]] x 5 + 100, (for i = 1, 4) (8)

[Z.sub.it] = [([X.sub.i] - [X.sub.it])/[S.sub.i]] x 5 + 100, (for i = 2, 3) (9)

where [X.sub.i], denotes sample mean of [X.sub.it], and [s.sub.i] denotes standard deviation, respectively. The higher the value of [Z.sub.it], the better the economic performance at time t in sector i. The numbers 5 and 100 in the above definitions do not influence the qualitative characteristics of each variable.

The entire sample period is used in calculating [X.sub.i] and [s.sub.i] in equations (8) and (9). This practice is based on the presumption that there are no structural changes within the sample period. But, it is the matter of empiricism whether there are structural changes within the entire sample period.

When there are structural changes, both means and variances are different across different subsample periods. If this is the case, we should normalize [X.sub.it] for each sub-sample period. Suppose that [X.sub.it], experiences (K-1) structural changes. Then, the entire sample period is divided into K sub-sample periods. Also, let k be the kth sub-sample period and #(k) denote the size of the kth sub-sample period, respectively. Then #(1) + #(2) + ....#(K) = N. Let [X.sub.ik] and [S.sub.ik] be the mean and variance of kth subsample period, respectively. Then, eqs. (8) and (9) can be generalized as follows:

[Z.sub.it] = [([X.sub.it] - [X.sub.ik])/[S.sub.ik]] x 5 + 100 (i = 1, 4; k = 1, 2, ... K; [X.sub.it] [member of] S(k)) (10)

[Z.sub.it] = [([X.sub.ik] - [X.sub.it])/[S.sub.ik]] x 5 + 100 (i = 2, 3; k = 1,2, .... K; [X.sub.it] [member of] S(k)), (11)

where S(k) is a set of sample points belonging to kth sub-sample period. When k is 1, the eqs. (10) and (11) are exactly the same as the eqs. (8) and (9), respectively, implying that there is no structural change. By using [Z.sub.it], the economic performance index at time t, [EPI.sub.t], can be expressed as

[EPI.sub.t] = [[summation].sub.(1,4)] [W.sub.i][Z.sub.it], (12)

where [[summation].sub.(1,4)] denotes the sum over i from 1 to 4. Since [w.sub.i] is the weight in ith sector, [EPI.sub.t] is the weighted average of [Z.sub.it]'s. In case of the simple average, [w.sub.i] will be 1/4. But, we can also assign different weight to each sector through the principal component analysis. Which one is more relevant depends on the empirical results.

III-II. The Estimation of EPI

In this section we estimate EPIs by using data of the USA, Japan and Korea. Proxy variables for economic growth rate and inflation rate are real GDP growth rate and the rate of increase in consumer price index. Sample periods are from 1967 to 2001 for all three countries. Before estimating the EPI, we should test whether there exist structural changes in economic variables. Here, Chow's breakpoint test, BP test, is used. This tests whether the coefficients in regressions are different by subsample periods. Since the differences in means are presumed to reveal the structural changes, we perform breakpoint test based on the regression results obtained by using only the constant term as an explanatory variable.

The test procedure is as follows. From 1970, BP test is recursively proceeded by adding one more year in each time. Reference values are F-statistics and the value of log likelihood and the critical probability is 0.01. If critical values of 1967-1971 are greater than 1967-1970, BP test will be continued by adding one more year, i.e., for 1967-1972. Suppose reference values continuously increase until the BP test of 1967-1980 and decrease after 1967-(1980+k), then 1980 is the first break point (k = 1, 2, ...). This implies that the first sub-sample period is 1967-1979. The same procedure is applied after 1980.

Results of BP tests are summarized in Table 1. In Table 1, F denotes the value of F-statistics, log(LR) is the value of log likelihood and values in parentheses are the probability of that year's not being the breakpoint, respectively. BP test results related to [1987, 1998] in case of Korean unemployment rate imply that the test is proceeded for the entire sample periods with the explicit consideration of two break points. The mean values of each economic variable by fixed sub-sample period obtained by the above method are shown in Table 2. According to Table 2, there are substantial differences in the mean values of each variable by sub-sample periods.

Indices can be easily derived if the sub-sample periods are fixed for each variable. There are various methods of obtaining EPI. In this paper, we use two methods. One method is to use the simple average of [Z.sub.it], and the other method is to use principal component analysis for obtaining the weight, [w.sub.i]

In case of using the simple average, the values of [w.sub.i] in the equation (12) are all 1/4 regardless of i. Table 3 reports the standardized values of [w.sub.i]'s under the assumption that sum of weights obtained by principal component analysis using [Z.sub.it], is equal to 1.

If weights of [Z.sub.it]'s are given, EPI can be obtained by the equation (12). The result is drawn in Figure 1. In Figure 1, the solid line shows the movement of EPI obtained by using the simple average, and the dotted line shows the movement of EPI obtained by using the weighted average. Only small differences can be noticed.

[FIGURE 1 OMITTED]

IV. The Analysis of EPI

IV-I. The Intra-national Analysis of EPI

In this section, intra-national analyses of EPIs of the USA, Japan and Korea will be performed. The primary purpose of the analysis is to examine whether the estimation results of EPIs suitably reflect the economic situation of each country.

According to Figure 1, it is difficult to find the trend of EPI movements in Korea. However, in early 1980s when the social and economic instabilities were very severe due to the Second Oil Shock and domestic political unrest, and when real estate bubble was severe, the value of EPI is estimated to be significantly low. Also the value of EPI drops significantly in 1998, the year right after the foreign exchange crisis. This indicates that estimated EPIs reflect the economic situations in Korea fairly well.

If we divide the sample period of Korea by the periods of political regimes, we can find an interesting pattern of changes in EPIs. The shaded areas in Figure 2 denote the periods of 3rd, 5th and Civilian Government, respectively. Note that there are remarkable differences in values of EPIs across the political regimes. Another point is that the EPI decreases remarkably at the end of each political regime since the 5th government. This indicates that instability in political regime has squarely affected the economic performance of Korea.

[FIGURE 2 OMITTED]

The EPIs of the USA and Japan reflect the real economic situations of the USA and Japan to some extent. According to Figure 1, the EPI of the USA shows the fact that the US economy has been getting worse continuously until the middle 1980's. From the late 1980's through 1990's, in the so-called "new economy" era, the USA has experienced the economic boom for a fairly long period. There was also a summit of US economic boom at the end of 1990's.

On the other hand, in case of Japan, estimated EPI shows the negative impacts of 1st and 2nd Oil Shocks. Figure 1 also shows that Japan experienced a better economy after the early 1980s perhaps due to the bubble expansion. Japanese economy has been deteriorating continuously until the end of 1990's right after bubble burst in the late 1980's. Based on these results, it can be said that the estimated EPIs properly reflect changes in economic situations of the USA, Japan and Korea.

We now compare the EPI of each country with its own misery index defined by eq. (1). (1) The misery indices have values around 10, while the EPIs have values around 100. Higher EPIs would be associated with lower misery indices. Figure 3 reveals this negative relationship between the EPI and the misery index. Note that the scales of the indices in Figure 3 are normalized to show their movements in the same Figure. The degree of negative correlation is not so big, however. The correlation coefficient between EPI and the misery index for the USA, Japan and Korea is, respectively, -0.32, -0.23, and-0.33.

[FIGURE 3 OMITTED]

IV-II. The International Comparisons of EPI's

In this section we analyze the relationship between EPIs of the three countries. The analysis will proceed as follows. Firstly, we will analyze the international relation of each [Z.sub.it] (i=1, ...,4). For example, for i = 1, we will analyze the relation among 3 countries based on the standardized index of the economic growth rate. Secondly, we will analyze the relation of 3 countries based on the estimated EPIs obtained both by the simple and the weighted averages. For this, the variance decomposition analysis will be used. The analysis can show the extent to which specific index of one country is influenced by those of other countries.

The VAR analysis should be performed for the stable variables, that is, integrated with order 0, statistically specified as I(0). Thus, it is necessary to test whether each variable is I(0) or not. Test results are summarized in Table 4. When there exist two test statistics like [Z.sub.2t], of Japan in Table 4, the upper value denotes test statistic about the level variable, the lower value test statistic 1st differenced, respectively. In case of Korea and the USA, all the variables are found to be I(0) at the 99% confidence level. In case of Japan, only [Z.sub.1t], is I(0) and the other variables are found to be I(1). Thus we should use 1st differenced variables for I(1) variables in order to perform VAR analysis.

Results of variance decomposition analysis about [Z.sub.it] are shown in Table 5. Since the values are converging after 5th period in all cases due to the simplicity of time lag structure, only 5th values are reported. We can summarize the results of variance decomposition analysis shown in Table 5 as follows.

Firstly, Korean economic variables are more influenced by those of other countries rather than the other way around. This is perhaps due to asymmetry in the size of the economies. The only exception is the unemployment rate, which of each country is rarely influenced by those of the other countries. This is perhaps because the unemployment rate of each country is largely affected by her usual practice of employment and economic system. Secondly, Korea is mainly influenced by Japan in economic growth and current account, while Korea is relatively much more affected by the USA in inflation rate. Thirdly, the USA is rarely affected by other countries, but is a little affected by Japan in inflation rate and current account. Lastly, Japan is more influenced by the USA, but is affected almost equally by Korea and the USA in case of the current account.

If EPIs are estimated properly, there should not be a big difference between the relationships of EPIs and those derived from each economic variable. In order to test this, variance decomposition analysis is performed using EPIs obtained by the simple and weighted averages. The results are shown in Table 6. According to Table 6, the results of variance decomposition analysis for EPIs estimated by simple and weighted averages are in accordance with those for individual economic variables. Only in case of the USA, the results of variance decomposition analysis for EPIs obtained by simple average are more relevant than those for EPIs obtained by the weighted average.

V. Concluding Remarks

Since a nation's economic performance is reflected in many economic variables, it is virtually impossible to perform proper international comparison of economic performances through the comparison of every relevant economic variable. In order to overcome this, the present paper has proposed the EPI that integrates several economic variables reflecting the economic performance of the nation. The EPIs of the USA, Japan and Korea are estimated and compared both intra-nationally and internationally. Variance decomposition analysis for EPIs estimated both by the simple and by the weighted averages are shown to be in accordance with those for individual economic variables. This indicates that the EPI developed in this paper suitably reflects economic situations of the USA, Japan and Korea.

References

Davis, O.A., M.J. Hinich and P.C. Ordeshook (1970), "An expository development of a mathematical model of the electoral process", American Political Science Review, Vol. 64, 426-428.

Golden, J. M., R. Orescovich and D. Ostafin (1987), "Optimality on the Short-Run Phillips Curve: A 'Misery Index' Criterion, A Note," The American Economist Vol.31, No.2, 72.

IMF, International Financial Statistics, various issues.

MacRae, (1977), "A political model of the business cycle," Journal of Political Economy 85, 239-263.

Nordhaus, W. (1975), "The political business cycle," Review of Economic Studies, Vol. 42, 169-189.

Smyth, D.J. and P. Dua (1989), "The Public's Indifference Map between Inflation and Unemployment: Empirical Evidence from the Nixon, Ford, Carter and Reagan Presidencies," Public Choice, Vol. 60, 71-85.

Smyth, D.J., P. Dua and S.W. Taylor (1994), "Voters and macroeconomics: Are they forward looking or backward looking?," Public Choice, Vol. 78, 283-293.

Smyth, D.J. and A.E. Woodfield, (1993), "Inflation, unemployment and macroeconomic policy in New Zealand: A public choice analysis" Public Choice, Vol. 75, 119-138.

Wiseman, C. (1992), "More on Misery: How Consistent Are Alternative Indices? A Comment," The American Economist, Vol. 36, No. 2, 85-88.

Yang, B. (1992), "Optimality On The Short-Run Phillips Curve Revisited," The American Economist, Vol. 36, No. 2, 89-91.

Zaleski, P.A. (1990), "On the Optimal Level of Macroeconomic Misery: A Comment," The American Economist, Vol. 34, No.2, 90-91.

Note

(1.) We thank an anonymous referee for suggesting this comparison between the EPI and the misery index.

by Sanghack Lee, * Kiwoong Cheong, ** and Seoung Hwan Suh ***

* Sanghack Lee, School of Economics, Kookmin University, Seoul 136-702, South Korea Corresponding author. Tel.: +82-2-910-4546; Fax: +82-2-910-4519; E-mail: slee@kookmin.ac.kr.

** Kiwoong Cheong, School of Business, Keimyung University, Taegu 704-701, South Korea.

*** Seoung Hwan Suh, Department of Economics, Yonsei University, Seoul 120-749, South Korea. We would like to thank an anonymous referee for valuable comments and suggestions. The usual disclaimer applies.

TABLE 1.
The test result of Chow's breakpoint

(USA)

Economic No breakpoint
growth rate

Inflation Year 1983: F = 26.00 (0.000) log LR = 20.34 (0.000)
rate Year 1993: F = 10.80 (0.004) log LR = 9.35 (0.000)
 [Year 1983, Year 1993]:
 F = 13.87 (0.000) log LR = 21.85 (0.000)

Unemployment Year 1971: F = 10.85 (0.000) log LR = 9.95 (0.001)
rate Year 1987: F = 11.36 (0.002) log LR = 10.24 (0.001)
 Year 1997: F = 20.08 (0.001) log LR = 14.01 (0.000)
 [Year 1971, Year 1987, Year 1997]
 F = 13.31 (0.000) log LR = 28.97 (0.000)

Current Year 1985: F = 42.07 (0.000) log LR = 28.76 (0.000)
account/GDP Year 1999: F = 12.54 (0.003) log LR = 10.33 (0.001)
 [Year 1985, Year 1999]
 F = 36.44 (0.000) log LR = 41.55 (0.000)

(Korea)

Economic No breakpoint
growth rate

Inflation Year 1982: F = 43.94 (0.000) log LR = 29.63 (0.000)
rate

Unemployment Year 1987: F = 8.70 (0.005) log LR = 8.19 (0.004)
rate Year 1998: F = 35.23 (0.000) log LR = 19.66 (0.000)
 [Year 1987, Year 1998]:
 F = 32.71 (0.000) log LR = 38.96 (0.000)

Current Year 1977: F = 13.35 (0.001) log LR = 11.89 (0.001)
account/GDP Year 1986: F = 9,98 (0.004) log LR = 9.01 (0.002)
 [Year 1977, Year 1986]
 F = 14.38 (0.000) lo g LR = 22.44 (0.000)

(Japan)

Economic Year 1971: F = 37.37 (0.000) log LR = 26.51 (0.000)
growth rate Year 1974: F = 10.17 (0.003) log LR = 9.32 (0.002)
 Year 1992: F = 21.18 (0.000) log LR = 16.69 (0.000)
 [Year 1971, Year 1974, Year 1992]:
 F = 34.78 (0.000) log LR = 51.59 (0.000)

Inflation Year 1979: F = 29.73 (0.000) log LR = 22.48 (0.000)
rate Year 1982: F = 27.35 (0.004) log LR = 19.19 (0.000)
 Year 1994: F = 15.03 (0.001) log LR = 12.14 (0.000)
 [Year 1979, Year 1982, Year 1994]:
 F = 12.92 (0.000) log LR = 28.39 (0.000)

Unemployment Year 1983: F = 31.17 (0.000) log LR = 23.28 (0.000)
rate Year 1998: F = 66.98 (0.000) log LR = 29.63 (0.000)
 [Year 1983, Year 1998]
 F = 77.80 (0.000) log LR = 61.90 (0.000)

Current Year 1977: F = 9.07 (0.005) log LR = 8.50 (0.004)
account/GDP Year 1984: F = 25.69 (0.000) log LR = 18.75 (0.000)
 [Year 1977, Year 1984]
 F = 19.39 (0.000) log LR = 27.78 (0.000)

TABLE 2.
Average by sub-period

(USA)

Economic 1967-2001
growth rate 2.89

Inflation rate 1967-2001 1967-1982 1983-1992
 5.00 7.09 3.81

Unemployment 1967-2001 1967-1970 1971-1986
rate 6.02 3.95 7.00

Current 1967-2001 1967-1984 1985-1998
account/GDP -1.24 -0.16 -2.03

(Japan)

Economic 1967-2001 1967-1970 1971-1973
growth rate 4.16 10.78 6.83

Inflation rate 1967-2001 1967-1978 1979-1981
 4.05 8.49 5.50

Unemployment 1967-2001 1967-1982 1983-1997
rate 2.43 1.68 2.67

Current 1967-2001 1967-1976 1977-1983
account/GDP 1.72 0.69 0.60

(Korea)

Economic 1967-2001
growth rate 7.77

Inflation rate 1967-2001 1967-1981 1982-2001
 9.52 15.70 4.88

Unemployment rate 1967-2001 1967-1986 1987-1997
 3.82 4.29 2.46

Current 1967-2001 1967-1976 1977-1985
account/GDP -1.76 -6.14 -3.52

(USA)
Economic
growth rate

Inflation rate 1993-2001
 2.60

Unemployment 1987-1996 1997-2001
rate 6.08 4.48

Current 1999-2001
account/GDP -4.08

(Japan)

Economic 1974-1991 1992-2001
growth rate 3.91 1.18

Inflation rate 1982-1993 1994-2001
 1.83 0.18

Unemployment 1998-2001
rate 4.63

Current 1984-2001
account/GDP 2.73

(Korea)

Economic
growth rate

Inflation rate

Unemployment rate 1998-2001
 5.23

Current 1986-2001
account/GDP 1.95

TABLE 3.
Weights estimated by principal component analysis

Korea
Economic
variable [Z.sub.1t] [Z.sub.2t] [Z.sub.3t] [Z.sub.4t]
Weight 0.332 0.350 0.149 0.169

USA
Economic
variable [Z.sub.1t] [Z.sub.2t] [Z.sub.3t] [Z.sub.4t]
Weight 0.371 0.210 0.213 0.206

Japan
Economic
variable [Z.sub.1t] [Z.sub.2t] [Z.sub.3t] [Z.sub.4t]
Weight 0.147 0.368 0.173 0.312

TABLE 4. The result of unit root test

Korea
Economic
variable [Z.sub.1t] [Z.sub.2t] [Z.sub.3t]
Test statistic -5.10 -3.10 4.96

USA
Economic
variable [Z.sub.1t] [Z.sub.2t] [Z.sub.3t]
Test statistic x.52 -4.24 -4.56

Japan
Economic
variable [Z.sub.1t] [Z.sub.2t] [Z.sub.3t]
Test statistic -5.01 -3.36 -2.68
 -6.09 -6.82

Korea
Economic [EPI.sub.t] [EPI.sub.t]
variable [Z.sub.4t] (simple) (weighted)
Test statistic -3.73 -4.80 -4.56

USA
Economic [EPI.sub.t] [EPI.sub.t]
variable [Z.sub.4t] (simple) (weighted)
Test statistic -4.17 -3.99 -4.02

Japan
Economic [EPI.sub.t] [EPI.sub.t]
variable [Z.sub.4t] (simple) (weighted)
Test statistic -3.23 -2.76 -3.03
 -4.58 -5.72 -5.72

* critical value at 1% = -3.64

TABLE 5. The result of variance decomposition
analysis (VDA) for economic variables

Economic growth rate
VDA of Korea

Korea USA Japan
73.6 7.2 19.2

VDA of USA
Korea USA Japan
1.3 97.8 0.9

VDA of Japan
Korea USA Japan
0.1 24.9 75.0

Unemployment rate

VDA of Korea
Korea USA Japan
91.4 7.0 1.6

VDA of USA
Korea USA Japan
7.2 92.2 0.6

VDA of Japan
Korea USA Japan
3.6 8.9 87.5

Inflation rate

VDA of Korea
Korea USA Japan
53.0 32.3 14.7

VDA of USA
Korea USA Japan
0.1 85.8 14.1

VDA of Japan
Korea USA Japan
0.2 21.5 78.3

Current account/GDP

VDA of Korea
Korea USA Japan
56.9 3.9 39.2

VDA of USA
Korea USA Japan
2.0 77.5 20.5

VDA of Japan
Korea USA Japan
7.6 8.8 83.6

TABLE 6. The result of variance decomposition
analysis (VDA) for EPI

EPI(simple average)

VDA of Korea
Korea USA Japan
91.7 6.5 1.8

VDA of USA
Korea USA Japan
7.4 86.1 6.5

VDA of Japan
Korea USA Japan
1.1 12.6 86.3

EPI(weighted average)

VDA of Korea
Korea USA Japan
87.7 11.2 1.1

VDA of USA
Korea USA Japan
12.0 84.6 3.4

VDA of Japan
Korea USA Japan
1.8 18.2 80.0