The paper discusses potential output, the output gap, and inflation in Korea. The paper explores the information content of potential leading indicators of inflation. A broadly balanced current account has been the suggested norm for Korea over the medium term. The challenge is to help build a more robust bond market that prices risk appropriately. The features of pension schemes in Korea and the problems they face are outlined. The paper reviews pension reform, banking sector, corporate sector, and foreign exchange crises with respect to Korea.


The paper discusses potential output, the output gap, and inflation in Korea. The paper explores the information content of potential leading indicators of inflation. A broadly balanced current account has been the suggested norm for Korea over the medium term. The challenge is to help build a more robust bond market that prices risk appropriately. The features of pension schemes in Korea and the problems they face are outlined. The paper reviews pension reform, banking sector, corporate sector, and foreign exchange crises with respect to Korea.

I. Potential Output, the Output Gap, and Inflation in Korea1

This chapter uses several approaches to estimate the output gap in Korea since the onset of the financial crisis in 1997. It is found that Korea’s real GDP fell almost 10 percent below potential during 1998-99, but that by early-2000, the gap had been closed. Historical analyses of Korea’s growth performance indicate that productivity growth had been slowing during the period leading to the crisis. For the medium term, growth rates of physical capital and labor inputs are expected to slow, but higher growth of total factor productivity arising from structural reforms could keep potential output growth at around 6 percent per year. Econometric tests of the gap model show that the output gap provides useful information regarding future inflation rates, but that the role of import prices may now be larger, perhaps due to the increased openness of the economy.

A. Introduction

1. The Korean economy’s spectacular recovery following the financial crisis in 1997-98 raises the question of the economy’s sustainable rate of growth in the medium term. After contracting by almost 7 percent in 1998, the economy grew by almost 11 percent in 1999, and an estimated 9½ percent in 2000. The rapid recovery raised concerns in the first half of 2000 about whether the economy was in danger of overheating. With the economy projected to slow in 2001, concerns about overheating have been replaced by concerns about a possible recession. Estimates of potential growth and the output gap can help shed light on these issues.

2. Potential output is defined as the maximum output that an economy can produce without generating inflation. In the short run, the size of the output gap provides information on inflationary pressures. In the medium to long run, the estimate of potential growth helps in evaluating the sustainable rate of growth. For policy analysis, for example, the stance of fiscal policy can be gauged by measuring the cyclically adjusted budget balance, which is the actual budget balance corrected for the effects of differences between actual and potential output.

3. This chapter discusses estimates of potential output and the output gap in Korea, projections of the sustainable rate of growth, and the economic relationship between measures of the output gap and inflation. The historical analysis covers the period 1970-2000, while the forward-looking part covers 2001-05. The sharp break that occurred in most macroeconomic series during the crisis period in 1997-99 means that quantitative analyses of the current state of the economy are more difficult than usual. Hence, estimates of potential output and the output gap are subject to considerable uncertainty. Nevertheless, the three approaches used in this study point to broadly the same results.

  • The Korean economy suffered a negative output gap of around 7 percent in 1998, and around 1½ percent in 1999.2 By 2000 Q1, the gap had closed, and real GDP is projected to exceed potential output by around 1½ percent in 2000-01, and to settle to approximate balance in 2002-05. The positive output gap in 2000-01 is modest and is unlikely to spark inflationary pressures.

  • For the medium term, the growth rates of physical capital accumulation and labor inputs can be expected to be slower compared to past trends. This is due to a move away from the system of easy credit for favored corporations and excessive focus on investment and expansion, as well as demographic and labor market developments. However, total factor productivity growth should pick up, due to a restructuring of the economy to become more open, competitive, and market oriented. Hence, the annual growth rate of potential output could be around 6 percent. Total factor productivity growth would account for around one-half of total growth, up from one-third before the crisis.

  • Tests of the relationship between estimates of the output gap and CPI inflation suggest that the output gap estimated using the production function approach provides a useful signal to the monetary authority. That is, when actual output exceeds estimated potential output, an increase in inflationary pressures can be expected. However, the role of supply shocks in the form of import prices has grown, very likely due to the increased openness of the Korean economy.

B. Estimates of the Output Gap

4. Although intuitive in concept, estimating potential output is difficult in practice. A wide range of approaches can be used for estimating potential output, ranging from simple detrending techniques, to more structural approaches, such as the production function approach. However, there is currently no single widely accepted approach. As De Masi (1997) notes, “country-specific circumstances have tended to influence the methodology used.” Useful surveys of the various methods available include St-Amant and Van Norden (1997) and Khatri and Lee (2000). Three approaches were used to estimate potential output in Korea: the Hodrick-Prescott filter (HP), the cubic spline smoothing method (CS), and the production function approach (PF).

Time series sechniques (HP and CS)

5. The HP and CS methods are readily applicable using standard econometric software, but may not produce reliable results (St-Amant and Van Norden, 1997). These two approaches are purely statistical and attempt to decompose a time seies into its permanent and cyclical components. Further technical descriptions of the HP and CS methodologies can be found in Khatri and Lee (2000).

6. As Khatri and Lee (2000) note, HP and CS estimates can become ill-defined at the beginning and end of samples, and also if there is a structural break in the data. In the Korean context, the sharp decline in output in 1998-99 would tend to bias downward the filtered estimates of potential output (the “end-point problem”). To address this problem, the quarterly real GDP series was extended to 2005 Q4, using projections derived from the PF approach (see the following section), and the HP and CS filters were then applied to the extended series.

Production function approach

7. The PF approach is appealing for its close links to growth theory and empirics, but in practice can be very data intensive and require substantial judgment in estimating the trends of the production function’s components. The PF approach used in this study follows the methodology described in Congressional Budget Office (1995). A Cobb-Douglas production function was set up with two inputs, physical capital (K) and raw labor (L):

InYt = InAt +αInKt + (1-α) InLt(3)

where In is the natural log of a variable, At is the level of total factor productivity (TFP), α and (1-α) are the shares of physical capital (K) and labor (L), respectively, and are assumed to be 0.30 and 0.70.3 The level of TFP (At) is derived as a residual, using historical data on real GDP, real gross fixed capital formation (which is used to compute a capital stock series)4, and employment (adjusted by average hours worked). Table I.1 shows the results, while Box I.1 provides a comparison of Korea’s productivity performance over time and across countries.

Table I.1:

Korea: Growth Accounting 1/2/

article image

The shares of capital and labor are assumed to be 0.3 and 0.7 percent.

The contributions to growth may not add up horizontally, due to the aggregation of quarterly data.

Potential output growth

Korea: Total Factor Productivity Growth

Estimating the sources of economic growth has been a veritable growth industry in applied economic research. Due to the record of rapid and sustained growth in output, East Asia has attracted a large share of researchers’ attention. The table below summarizes the results of previous studies of TFP growth in Korea and a selection of East Asian economies.

article image

To facilitate comparison between the pre-1980 and post-1980 subperiods for Korea, the implicit growth rates have been calculated from the original results, where warranted and feasible.

The above results point to the following observations.

  • First, Korea’s productivity performance has been striking. Accumulation of physical and human capital has been rapid, but does not seem to have been so excessive as to lead to decreasing returns, as for instance Young (1992, 1995) concluded in the case of Singapore.

  • Second, Korea’s productivity record can be divided into roughly two phases. In the first phase, from 1970 to around 1980, productivity grew modestly. After the 1980 recession, however, the pace sped up. The results presented in this chapter are also consistent with these findings.

8. To estimate potential output, trend levels of physical capital, labor inputs, and TFP are derived and substituted into the following equation (where asterisks denote trend):

InY*t = In A*t + 0.30*InK*t + 0.70 *InL*t(4)

Trend levels of labor (L*) were calculated by smoothing the historical labor force series, and adjusting it by the nonaccelerating inflation rate of unemployment (NAIRU), which in turn was calculated using a method described in Giorno et al (1995). Trend levels of physical capital (K*) were assumed to be equal to the estimated capital stock series. Trend TFP (A*) was estimated in three steps. First, trend TFP was estimated for the period 1970-1997 by applying an HP filter on the TFP series calculated as a residual using the procedure described above. Second, the series was then extended to 1999 Q4 by assuming that trend TFP grows by 3 percent per annum, the same rate as it did over the period 1982-1997 (i.e., between the 1980-81 recession and the 1998 financial crisis). Finally, the TFP series was extended to 2000 Q4, by assuming a slight increase in TFP growth, to 3.2 percent per annum, to incorporate the effects of the structural reforms adopted in 1998-2000 (increased labor market flexibility, increased openness to foreign direct investment, corporate restructuring, financial restructuring, etc.).

9. The method described above does not incorporate the role of increases in human capital through education (as in Young, 1995) or R&D investment (as in Kim and Mun, 1999). Instead, these components of growth are subsumed in TFP growth. Although their inclusion in this study would have been desirable, a balance had to be struck between comprehensiveness and ease of use for purposes of this study as measurement and projection of these variables can be difficult.

10. The above analysis can also be used to make a reasoned estimate of medium-term growth. In the 1990s, physical capital grew by around 10 percent per year; labor inputs (the labor force, adjusted for average hours worked), by 1.2-1.5 percent; and TFP, by around 3 percent. Assuming that the growth rate of the capital stock slows to some 6 percent (due to the shift away from debt-financed overinvestment and towards deleveraging and improved profitability), labor input growth falls to 1 percent (due to slower population growth and decreases in hours worked), and TFP growth modestly increases to 3.5 percent (due to efficiency-enhancing structural reforms, especially corporate and financial restructuring), potential output growth could be approximately 6 percent per year over the period 2000-05.


11. Table I.2 shows the estimated output gaps using the three approaches for selected years before 1997 and on a quarterly frequency for 1997-2000, along with estimates obtained by researchers at the Bank of Korea and the OECD.5 The three estimated series of the output gap are highly correlated (see Figure 1). All three series suggest that there was a large positive output gap (i.e., excess demand) in 1997—amounting to 2-4 percent of potential output—which turned into a substantial shortfall during the crisis in 1998. The HP and CS series suggest that at the trough in 1998, Korea suffered an output gap that was much worse than during the previous recession that Korea experienced in 1980. All three series also suggest that actual output moved above potential output in the second half of 1999. Given current estimates of growth in 2000 and projections for 2001, it is estimated that actual output exceed potential output in 2000 by around 1½ percent, and by 0.2-1.7 percent in 2001. The output gap is projected to close again in 2002.

Table I.2.

Korea: Estimates of Potential Output Growth and the Output Gap

article image
HP: Hodrick Prescott filter.CS: Cubic spline smoothing.PF: Production function approach.

Year-on-year growth, percent.

For 2000-05, the growth rates are projections.

As percent of estimated potential output.

Kim and Mun (1999), results from their production function estimates.

Scarpetta et al (2000).

article image

C. The Output Gap and Inflation

12. The relationship between the output gap and inflation was also tested. Following Coe and McDermott (1997) and Claus (2000), a simple version of the gap model was used, which relates the change in inflation to the level of the output gap:

Model 1:Δπt =α1 +Σk=0pβ1k GAPtk +1t(1)

where πt is the year-on-year percentage change in the consumer price index, GAPt is the estimate of the output gap, εlt is a stochastic disturbance term, and Δ is the first difference operator. All data are quarterly. A constant is included to avoid imposing the constraint that the noninflationary level of the output gap is exactly zero. If the output gap is a significant determinant of the change in inflation, then an F-test should show that the βlis are jointly significantly different from zero. Some of the βlis could be negative, but their sum should be positive.

13. To incorporate the effect of supply side factors, the following alternative specification was also employed:

Model 2:Δπt =α2 +Σk=0pβ1k GAPtk - IMPt +2t(2)
Figure I.1.
Figure I.1.

Korea: Measures of the Output Gap

Citation: IMF Staff Country Reports 2001, 101; 10.5089/9781451822052.002.A001

where IMP t is the year-on-year percentage change in the import price index denominated in won. This specification assumes that supply shocks originate from the external sector. The two models were estimated over the period 1970-2000, and over the subperiods 1980-2000 and 1990-2000.6 All estimates used four lags. The results are shown on Table I.3.

Table I.3.

Korea: Testing the Gap Model

article image

H0 of non-significance is rejected at the 1 percent level.

H0 of nonsignificance is rejected at the 5 percent level.

H0 of nonsignificance is rejected at the 10 percent level.

HO: The gap variables jointly have no effect on the change in inflation.

Estimates including the import price coefficient start from 1972 Q1.

14. Estimates of Model 1 are modestly satisfactory. In all periods studied, the sums of the output gap coefficients are positive, consistent with the gap model. Furthermore, F-tests indicate that the lagged levels of the output gap (for all three estimated series) are jointly significant in explaining the change in inflation, at confidence levels of 1-5 percent. However, Model 1 explains only a small portion of the total variance in inflationary pressures (i.e., no more than 20 percent in the case of the HP and CS gaps in the period 1990-2000). Furthermore, comparing the two periods 1980-2000 and 1990-2000, the sums of the output gap coefficients are substantially smaller in the latter period, indicating a smaller role in the 1990s for the output gap in explaining inflationary pressures. It is interesting to note that Coe and McDermott (1997), in their precrisis study, conclude that the gap model works well in Korea, as in most other Asian countries. They test the simple gap model, as well as one augmented with money. They find that the gap coefficients are statistically significant and that the models explain around 37 percent of the variance of inflation.

15. The explanatory power of the gap model improves with the inclusion of import prices. Focusing on the period 1990-2000, the adjusted R-squared of the estimated equations increases by around 10 percentage points, compared to the simple gap model which does not include the proxy for supply shocks. In all cases, the import price coefficient is statistically significant. However, the sum of the output gap coefficients shrinks (to zero in the cases of the HP and CS gaps), and F-tests of their joint significance indicate rejection at the 1 and 5 percent confidence levels.7 The results of the estimates suggest that: (a) the role of external supply shocks has grown, and may now be more important than the output gap, and (b) of the three estimates of the output gap, the PF series may be a better indicator of inflationary pressures.

D. Conclusion

16. The three approaches used to estimate the output gap in Korea in 1997-2005 broadly point to similar conclusions. Less than two years after the onset of the crisis in November 1997, Korea’s GDP had recovered to its potential level. Growth was strong in 2000, which would normally be expected to lead to a modest increase in inflationary pressures. However, due to the increased openness of the Korean economy, external price shocks (including from exchange rate movements) are likely to be a more significant source of concern as regards inflation in the short run. To the extent that measures of the output gap provide a useful signal of inflationary pressures, the output gap estimated using the PF method is slightly more informative than the HP and CS series.


  • Claus, Iris, 2000, “Is the Output Gap a Useful Indicator of Inflation?Discussion Paper DP/2000/05, Reserve Bank of New Zealand.

  • Coe, David T. and C. John McDermott, 1997, “Does the Gap Model Work in Asia?” IMF Staff Papers, 44, pp. 5980.

  • Collins, Susan M. and Barry P. Bosworth, 1996, “Economic Growth in East Asia: Accumulation versus Assimilation,” Brookings Papers on Economic Activity, No. 2, pp. 13591.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Congressional Budget Office, 1995, “CBO Method for Estimating Potential Output,” CBO Memorandum.

  • De Masi, Paula R., 1997, “IMF Estimates of Potential Output,” Staff Studies for the World Economic Outlook, IMF.

  • Dornbusch, Rudiger and Yung Chul Park, 1987, “Korean Growth Policy,” Brookings Papers on Economic Activity, No 2.

  • Ghosh, Swati and Aart Kraay, 2000, “PREM Note: Measuring TFP Growth,” Mimeo, World Bank.

  • Giorno, Claude, Pete Richardson, Deborah Roseveare, and Paul van den Noord, 1995, “Potential Output, Output Gaps, and Structural Budget Balances,” OECD Economic Studies 24, pp. 167209.

    • Search Google Scholar
    • Export Citation
  • Khatri, Yougesh and II Houng Lee, 2000, “Malaysia: Potential Output and Inflation,” in Malaysia—Selected Issues, Article IV Consultation Paper, SM/00/147, IMF.

    • Search Google Scholar
    • Export Citation
  • Kim, Chi Ho and Mun So Sang, 1999, “Results of the Measurement of Potential GDP and Inflationary Pressures,” (Translation from Korean) Finance and Economic Research Paper No. 96, Bank of Korea.

    • Search Google Scholar
    • Export Citation
  • Kim, Jong-Il and Lawrence J. Lau, 1994, “The Sources of Economic Growth of the East Asian Newly Industrialized Countries,” Journal of the Japanese and International Economies, 8, pp. 23571.

    • Search Google Scholar
    • Export Citation
  • Kim, Kiheung, 2000, “An Analysis of the Sources of Growth in East Asian Economies and R&D Spillover Effects,” Journal of the Korean Economy, 1, pp. 83107.

    • Search Google Scholar
    • Export Citation
  • Kim, Kwang Suk and Sung Duk Hong, 1997, Accounting for Rapid Economic Growth in Korea, 1963-1995, Seoul: Korea Development Institute.

  • St-Amant, Pierre and Simon van Norden, 1997, “Measurement of the Output Gap: A Discussion of Recent Research at the Bank of Canada,” Technical Report No. 79, Bank of Canada.

    • Search Google Scholar
    • Export Citation
  • Scarpetta, Stefano, Andrea Bassanini, Dirk Pilat, and Paul Schreyer, 2000, “Economic Growth in the OECD Area: Recent Trends at the Aggregate and Sectoral Level,” Working Paper No. 248, Economics Department, OECD.

    • Search Google Scholar
    • Export Citation
  • World Bank, 1993, The East Asian Miracle: Economic Growth and Public Policy, Oxford University Press and World Bank.

  • Young, Alwyn, 1992, “A Tale of Two Cities: Factor Accumulation and Technical Change in Hong Kong and Singapore,” in S. Fischer and O. Blanchard (Eds.), NBER Macroeconomics Annual 1992, Cambridge MA: MIT Press.

    • Search Google Scholar
    • Export Citation
  • Young, Alwyn, 1995, “The Tyranny of Numbers: Confronting the Statistical Realities of the East Asian Growth Experience,” Quarterly Journal of Economics, 110, pp. 64180.

    • Crossref
    • Search Google Scholar
    • Export Citation

This chapter was prepared by Henry Ma. Aung Thurein Win provided helpful research assistance. All calculations are based on data available as of December 15, 2000.


The output gap is defined as the difference between actual and potential output. Hence, a negative (positive) gap implies that actual output is below (above) potential.


The time-varying factor shares that Kim and Hong (1997) calculated from the national income accounts were also tried, but the results were unsatisfactory. One reason is that, for the earlier periods (i.e., the 1970s), the calculated shares may underestimate the share of labor, due to the presence of wage repression. As noted by Dornbusch and Park (1987), during that period: “… politics, certainly … left little room for organized labor and even less for union militancy.”


The perpetual inventory method described in Young (1995) was used. This method requires assumptions to be made regarding the initial capital stock and the depreciation rates (6 percent per annum). However, over a sufficiently long period of estimation (as in this case), the development of the capital stock series becomes relatively invariant to initial assumptions. In fact, the estimated growth rates of capital turn out to be quite close to estimates made by Kim and Mun (1999).


For all three estimates of the output gap in this paper, estimated potential output was compared to the seasonally adjusted real GDP series provided by the Bank of Korea.


Time series tests (not shown) indicate that the null hypothesis of a unit root in the level of output gap can be rejected for all three measures of the output gap, and that inflation becomes stationary after being differenced once.


However, the F statistics are quite close to the 5 percent confidence level, suggesting that the two measures of the output gap may nevertheless provide useful information to monetary policymakers.