This Selected Issues paper discusses the fiscal policy in the Korean business cycle, and examines the usefulness of the equations for inflation forecasting at horizons consistent with the Bank of Korea 's inflation-targeting framework. It analyzes the causes and macroeconomic consequences of Korea's dual labor market; discusses the government's reform proposals; and the nonbank financial sector restructuring to date.

Abstract

This Selected Issues paper discusses the fiscal policy in the Korean business cycle, and examines the usefulness of the equations for inflation forecasting at horizons consistent with the Bank of Korea 's inflation-targeting framework. It analyzes the causes and macroeconomic consequences of Korea's dual labor market; discusses the government's reform proposals; and the nonbank financial sector restructuring to date.

II. The Phillips Curve and Okun’s Law—Do They Work in Korea?1

The Phillips Curve and Okun’s Law are each now over 40 years old. This chapter examines the usefulness of these equations for inflation forecasting at horizons consistent with the Bank of Korea’s inflation targeting framework. These simple equations appear to forecast inflation reasonably well, provided real GDP growth can be forecast accurately.

A. Introduction

1. Korea has enjoyed low to moderate inflation for the last two decades. Korea had bouts of high inflation in the 1970s, but since 1982 the CPI inflation rate has remained below 10 percent (y/y) and has averaged 4½ percent. The Asian crisis led to the most notable inflation volatility in recent years, as headline CPI inflation jumped to 9.5 percent (y/y) in February 1998 following the collapse of the Korean won. But just one year later the inflation rate had fallen to only 0.2 percent (y/y) due to the effects of the sharp recession.

uA02fig01

Headline CPI Inflation

(Year-on-Year, Percent)

Citation: IMF Staff Country Reports 2004, 045; 10.5089/9781451822144.002.A002

Source: CEIC database.
uA02fig02

CPI Inflation and Target Bands

(Yaar-on-Year, Percent)

Citation: IMF Staff Country Reports 2004, 045; 10.5089/9781451822144.002.A002

Source: CEIC database and Bank of Korea.

2. Inflation has largely been within the Bank of Korea’s target ranges in recent years. An inflation targeting (IT) monetary policy framework was adopted in 1998. From 1998–2003, the target range has been set on an annual basis, e.g., the 2003 target range of 2–4 percent was set at the end of 2002. Initially, the target was defined in terms of the headline CPI, but since 2000 it has been based on the core CPI, which excludes noncereal agricultural products and selected petroleum products.2

3. From 2004, the inflation target was shifted to 2½–3½ percent over the medium-term. Amendments to the BOK Law adopted in 2003 included measures to enhance the BOK’s independence and accountability, along with the shift from targeting inflation in the next year to a medium-term policy focus. The BOK had long advocated this amendment because monetary policy takes time to affect the inflation rate, e.g., BOK research finds that the optimal horizon for monetary policy is four to nine quarters (Kim, 2002). Although the target range has been narrowed, there is the potential for increased flexibility from year-to-year, as the goal is set in terms of average inflation over 2004—06. The BOK plans to continue enhancing its inflation forecasting following this change in the law, but as the BOK had previously maintained its own medium-term target, at 2½ percent, it does not necessarily imply a substantial change in monetary policy formulation practices.

4. This chapter presents a simple model of Korean inflation. The objective of the model is to produce inflation projections for use in the IMF World Economic Outlook (WEO) and to aid assessments of the stance of monetary policy. Considering the optimal horizon for monetary policy, and the WEO requirement for quarterly projections over eight quarters, the main focus for evaluating the model is the forecasting performance over a one to two year horizon.

B. Theory and Empirical Specification

5. The Phillips curve is the workhorse model of inflation in OECD countries (Richardson et al., 2000). Originally an empirical regularity between unemployment and wage growth (Phillips, 1958), the Phillips curve has become one of the most widely studied macroeconomic relationships. It can be derived from a variety of theoretical models, e.g., the imperfectly competitive markets for goods and labor in Layard et al. (1991). The core feature of the Phillips curve is that the inflation rate (π) rises relative to inflation expectations (πe) when unemployment (u) is low and vice versa:

π =πeαu

6. Core CPI inflation excluding public service charges is chosen as the dependent variable. The core CPI is the inflation measure targeted by monetary policy, and it is substantially less volatile than the headline CPI while still covering 88.4 percent of the CPI basket. However, public service charges are a major component of the core CPI, with a weight of 17 percent, and their impact on the core CPI can be substantial at times. For example, in 2002 a reduction in telecommunication fees lowered the core CPI by almost 1 percentage point.

uA02fig03

Public Service Charges and Core CPI Inflation

(Year-on-Year, Percent)

Citation: IMF Staff Country Reports 2004, 045; 10.5089/9781451822144.002.A002

Source: CEIC database and staff calculations.

7. A range of issues are examined in estimating the Phillips curve equation:

  • Alternative indicators of demand pressure are explored. To reflect different transmission channels, a range of variables are tested in addition to the unemployment rate, including: capacity utilization (called the average operating ratio in Korea) in the industrial sector; real GDP growth; consumption growth; employment growth; and estimates of the output gap. 3

  • Nonlinearities in the effects of demand pressure are tested. Research in other OECD countries has found some nonlinearities in the Phillips curve (Clark and Laxton, 1997). The estimation sample includes the Asian crisis, where the cyclical deviation was much larger than in other cycles, so a failure to allow for nonlinearities could distort the estimated parameters.

  • External and supply shocks are also tested. Phillips curves for open economies typically allow for effects from foreign prices and the exchange rate, while oil prices are a common supply shock variable. The New Keynesian Phillips curve literature also includes wages or unit labor costs, but these were not significant in Korea.

  • Inflation expectations are represented by a backward-looking autoregressive process. This is the most common approach, but some recent research allows for a mixture of forward- and backward-looking expectations, although the estimated weight on the forward terms is typically modest (Debelle and Vickery, 1998). A more sophisticated representation of inflation expectations in Korea, including the role of the inflation targets, is left for future research.

uA02fig04

Demand Pressure Indicators

(In percent)

Citation: IMF Staff Country Reports 2004, 045; 10.5089/9781451822144.002.A002

Source: CEIC database and staff calculations.

8. Okun’s Law provides a useful link between unemployment and GDP growth. As part of each WEO round, the Fund’s Korea team produces quarterly projections for GDP which can be used as an input into the inflation forecasts. In particular, the projections for GDP growth (Δy) can be used to project the unemployment rate(u) using a variant of Okun’s Law (Okun, 1962):4

u =βu1 + θ(L)Δy

This equation is intended to serve the narrow purpose of generating relatively short-run projections of unemployment. It lacks any factors that might explain variations in the equilibrium unemployment rate, but this may not be problematic in Korea, where changes in the trend level of unemployment do not appear to be significant relative to other OECD countries.

C. Properties of the Estimated Equations

9. A simple Phillips curve was found to have substantial explanatory power for Korean inflation. With a large number of candidate variables for demand pressure and external and supply shocks, an automated specification search in PcGets was used to find an initial specification and to identify outliers (Krolzig and Hendry, 2001). The resulting model included terms for unemployment, capacity utilization, and real GDP growth, but no other variables besides the exchange rate were found to be statistically significant. No tests of this initial equation indicated misspecification, but capacity utilization had the wrong sign (negative rather than positive). Dropping this variable also made real GDP growth statistically insignificant, which led to the very simple equation presented in the Annex, with only unemployment and the exchange rate driving inflation. This equation has a standard error of 0.3 percent for the quarter-on-quarter inflation rate, similar to the results of Richardson et al. (2000) for the United States and Germany.

uA02fig05

Actual and Estimate for Core CPI ex. Public Charges

(q/q, s.a., Percent)

Citation: IMF Staff Country Reports 2004, 045; 10.5089/9781451822144.002.A002

Source: CEIC database and staff calculations.
uA02fig06

Contributions to Core CPI Inflation

(Percent, s.a, q/q)

Citation: IMF Staff Country Reports 2004, 045; 10.5089/9781451822144.002.A002

Source: CEIC database and staff calculations.

10. The estimated Phillips curve has a number of interesting features:

  • The Phillips curve was found to be broadly linear. A variety of transformations of the unemployment rate were tested, but they did not offer a significant improvement in explanatory power. No evidence of a change in the NAIRU was found using tests for parameter constancy or a model with a time-varying intercept.

  • As a rule-of-thumb, there is a “one-for-one” link between the inflation rate and the unemployment rate over a two-year horizon. A permanent 1 percentage point rise (fall) in the unemployment rate leads to a ¾ point fall (rise) in the year-on-year inflation rate in the first year, and a fall (rise) of about 1 percentage point from the second year on.

  • Lagged inflation was not very important. The only significant lags were the second and third quarters, and the restriction that they have the same coefficient was readily accepted. The sum of the lagged coefficients is only 0.38 compared with the “accelerationist” restriction of unity that is commonly imposed. Such a low weight on past inflation may be rational during the sample period when inflation has been moderate on average, and the authorities are considered to be committed to maintaining low inflation (Sargent, 1971).5

  • The pass through of exchange rate changes into inflation was found to be relatively fast but not very large. The exchange rate of the won to the U.S. dollar was preferred statistically to the nominal effective exchange rate, perhaps indicating some stickiness in the price of traded goods in U.S. dollar terms. A 10 percent depreciation of the won against the U.S. dollar is estimated to raise the core CPI excluding public charges by 1.1 percent in the long-run. Some 60 percent of this effect is realized after two quarters, and three-quarters within one-year.

uA02fig07

Phillips Curve is Broadly Linear

(Data from 1990Q2 to 2003Q4)

Citation: IMF Staff Country Reports 2004, 045; 10.5089/9781451822144.002.A002

Source: CEIC database and staff calculations.
uA02fig08

Impact of Shocks on Core CPI ex. Public Charges

(Percent change, q/q)

Citation: IMF Staff Country Reports 2004, 045; 10.5089/9781451822144.002.A002

Source: CEIC database and staff calculations.

11. Okun’s Law accounts for most fluctuations in Korean unemployment, as presented in detail in the Annex. Real GDP growth in the current and previous quarter has the greatest impact on unemployment, but there are also significant effects from growth two and three quarters ago. The coefficient on the lagged dependent of unemployment is significantly different from unity, at 0.84, so the equation shows a strong mean-reversion tendency. For example, four quarters after a shock to unemployment only 49 percent of the shock remains, and after eight quarters only 24 percent remains.6

uA02fig09

Unemployment Rate Actual and Estimate

(Percent)

Citation: IMF Staff Country Reports 2004, 045; 10.5089/9781451822144.002.A002

Source: CEIC database and staff calculations.

12. Unemployment responds asymmetrically to real GDP growth, as found in other OECD countries. The square of real GDP growth rate was found to be highly statistically significant, implying that unemployment responds more a fall in GDP than to a rise in GDP. A 2 percent fall in GDP raises unemployment by a peak of 0.36 percentage points after four quarters, while a rise of the same magnitude reduces unemployment by only 0.12 percentage points. Similar asymmetries are found in other OECD countries (Harris and Silverstone, 2001; Viren, 2001). This behavior is consistent with the finding that job destruction is more highly cyclical than job creation, and a variety of microeconomic models have been used to explain this finding (Andolfatto, 1997; Campbell and Fisher, 2000).

D. Asymmetry and Monetary Policy Formulation

13. The two equation system shows asymmetric inflation responses to shocks to real GDP. While inflation responds symmetrically to changes in unemployment, the asymmetric response of unemployment to growth implies that negative growth shocks reduce inflation by more than inflation rises following a positive growth shock. For example, after two years, the cumulative impact on the price level of a 2 percent fall in the level of GDP is negative 0.44 percent, compared with a 0.13 percent rise in the price level due to a 2 percent increase in the level of GDP.

uA02fig10

GDP Shocks Have Asymmetric Effects on Inflation

(Core CPI, y/y, percent)

Citation: IMF Staff Country Reports 2004, 045; 10.5089/9781451822144.002.A002

Source: CEIC database and staff calculations.

14. The importance of this asymmetry for monetary policy formulation depends on the uncertainty in the outlook for real GDP. To help illustrate the implications of the inflation-growth asymmetry, we can take the year-on-year inflation rate eight quarters ahead as a guide to current monetary policy formulation. Simulations of shocks to the quarterly real GDP growth rate indicate that if the cumulative impact on GDP is modest, e.g., 1–2 percent after eight quarters, the asymmetry is less pronounced than if the cumulative impact is 3–4 percent. Assuming that projection errors for the level of GDP eight quarters ahead have a normal distribution, it is found that the associated distribution of inflation projection errors is close to normal if the standard deviation of GDP projection errors is modest, e.g., 1 percent. However, if these errors have a standard deviation of 2 percent, the inflation projection errors become negatively skewed. It appears that at times of a rise in uncertainty about the outlook for growth, the downside risks to inflation rise more than the upside risks, suggesting a more accommodative stance might be appropriate than is implied by the central projections alone. Further research will be necessary to more fully explore the importance of this asymmetry for monetary policy formulation in practice.

uA02fig11

Impact on Year-on-year Inflation Rate After 8 Quarters

(Percent)

Citation: IMF Staff Country Reports 2004, 045; 10.5089/9781451822144.002.A002

Source: CEIC database and staff calculations.
uA02fig12

Distribution of Core CPI Inflation Projection Errors Due to Normally Distributed GDP Projection Errors 8 Quaters Ahead

(Year-on-year, percent)

Citation: IMF Staff Country Reports 2004, 045; 10.5089/9781451822144.002.A002

Source: CEIC database and staff calculations.

E. Forecasting Performance

15. Out-of sample dynamic projections support the structural stability of the equations. For both equations, the last three years were removed from the estimation sample, and estimates from 1990–2000 were used to make projections for 2001–03, a period which included sharp swings in real GDP growth.7 The results are:

  • Projections of quarter-on-quarter inflation rates from the Phillips curve were virtually unbiased, and had an RMSE of 0.16 percent for quarterly inflation rates. The implied projections for year-on-year inflation rates in 2002–03 had an RMSE of 0.42 percent.

  • Unemployment projections from the Okun’s Law equation were above actual by a modest 0.14 percentage points on average, with an RMSE of 0.21 percent.

uA02fig13

Out-of-sample Projections of Core CPI ex. Public Charges

(q/q, s.a., Percent)

Citation: IMF Staff Country Reports 2004, 045; 10.5089/9781451822144.002.A002

Source: CEIC database and staff calculations.
uA02fig14

Out-of-sample Projections of the Unemployment Rate

(Percent)

Citation: IMF Staff Country Reports 2004, 045; 10.5089/9781451822144.002.A002

Source: CEIC database and staff calculations.

16. Nonetheless, inflation forecasting performance is highly dependent on the accuracy of growth forecasts. A set of dynamic projections was conducted within the sample to evaluate the forecasting performance of the two equation system.8 For each year from 1991–2002, three sets of four-quarter and eight-quarter simulations were calculated. The benchmark is a naïve model for inflation, with a trend inflation level and an autoregressive deviation from this trend.9 Using actual GDP data to make the projections, the two equation system was substantially more accurate, with the standard deviation of errors at 0.7–0.8 percent, compared with 1.3–1.4 percent for the naïve model. However, this accuracy was lost if the two equation system relied on GDP projections from a similar naive model.

Accuracy of Projections of Core CPI ex. Public Service Charges

(Standard deviation of forecast errors in y/y percent change)

article image

17. Accurately projecting GDP growth in Korea is challenging. Even excluding the Asian crisis period, Korean real GDP growth displays relatively high volatility. The manufacturing sector is dominated by cyclical and export-driven industries (e.g., semiconductors, automobiles, and shipbuilding), and shocks to these sectors have tended to drive broader economic activity. The higher uncertainty attached to GDP projections in Korea relative to other OECD countries also implies higher uncertainty in the inflation projections from these Phillips curve and Okun’s Law equations.

uA02fig15

Distribution of Real GDP Growth Rates, 1970-2003

(Quarter-on-quarter, seasonally adjusted, percent)

Citation: IMF Staff Country Reports 2004, 045; 10.5089/9781451822144.002.A002

Source: CEIC database and staff calculations.

F. Current Model Projections

18. Assuming a sustained recovery in real GDP growth, the model projects a decline in unemployment and a modest rise in inflation over the next two years. At the time of writing, IMF staff projects that real GDP growth would recover from an estimated 2.9 percent in 2003 to 5½ percent in 2004 and 5¼ percent in 2005.10 These real GDP projections were used as an exogenous variable in projections for unemployment and inflation in 2004–05. It is also assumed that the exchange rate remains at W 1,182 per U.S. dollar, which was the average exchange rate in 2003 Q4.11

uA02fig16

Unemployment Rate Projections

(Percent)

Citation: IMF Staff Country Reports 2004, 045; 10.5089/9781451822144.002.A002

The two equation model projects:

  • Unemployment to decline from 3.6 percent in 2003 Q4 to 3.3 percent in 2004 Q4, then to remain around 3.3 percent in 2005.

  • Inflation rates in the core CPI excluding public services rise from 2.8 percent y/y in 2003 Q4 to 3.2 percent y/y by 2004 Q4, and 3.4 percent by 2005 Q4.

19. These projections are subject to significant uncertainty. By the end of 2005, the standard error of the unemployment projections is 0.3 percent, so the 95 percent confidence interval for the unemployment projections is quite wide, at 2.7 percent to 3.9 percent. Similarly, for year-on-year inflation, the standard error is 0.7 percent by end 2004 and 0.8 percent by end 2005, implying 95 percent confidence intervals as wide as 3 percentage points.

uA02fig17

Inflation Rate Projections

(Core CPI ex. Public Charges, y/y, percent)

Citation: IMF Staff Country Reports 2004, 045; 10.5089/9781451822144.002.A002

Source: CEIC database and staff calculations.

G. Conclusion

20. Okun’s Law and the Phillips Curve appear to be useful empirical tools in Korea, Okun’s Law fits the Korean data quite well, making it a useful way to generate projections for the unemployment rate from projections for real GDP growth, A simple Phillips curve was found to have similar explanatory power for Korean inflation to that found in other OECD countries. Simulations of the Phillips curve suggested a “one-for-one” rule-of-thumb, where a lasting increase (decrease) in the unemployment rate of 1 percentage point leads to a 1 percent fall (rise) in the year-on-year inflation rate after two years. Combining these two equations provides a simple means to project inflation developments. An asymmetry in the inflation effects of shocks to real GDP was found, which may have implications for monetary policy formulation at times of high uncertainty about real GDP growth prospects.

21. Nonetheless, inflation projections using these tools are subject to significant uncertainties. While the Phillips Curve and Okun’s Law do appear to work in Korea, the forecast analysis indicates that projection standard errors widen over time, to become quite substantial at the 1 to 2 year horizon most relevant to the formulation of monetary policy. Moreover, these standard errors do not account for the significant uncertainty in the outlook for real GDP growth, while public service charges are another source of variability in the core CPI outside the scope of the model. Altogether, even with these tools, the technical challenges of monetary policy formulation in Korea remain considerable.

ANNEX Phillips Curve and Okun’s Law Estimates

1. Data and Notation

Most data are sourced from the CEIC database, with CEIC variable names in square brackets.

article image

2. Phillips Curve for Inflation

Dependent: Δln(P) Sample: 1990 Q2–2003 Q4, 55 observations, 6 parameters12

article image
S.E. = 0.30 percent R2 = 0.814 DW=1,64

Specification Tests [p-value]

article image

3. Okun’s Law for Unemployment

Dependent: U Sample: 1987 Ql to 2003 Q3, 67 observations, 6 parameters

article image
S.E. = 0.17 percent R2 = 0.987 DW = 1.95

Specification Tests [p-value]

article image

References

  • Andolfatto, D., 1997, “Evidence and Theory on the Cyclical Asymmetry in Unemployment Rate Fluctuations,” Canadian Journal of Economics, 30, August, pp. 70921.

    • Search Google Scholar
    • Export Citation
  • Campbell, J. and Fisher J., 2000, “Aggregate Employment Fluctuations with Microeconomic Asymmetries,” The American Economic Review, Vol. 90, No. 5, pp. 13231345.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Clark, P. and Laxton, D., 1997, “Phillips Curves, Phillips Lines and the Unemployment Costs of Overheating,” IMF Working Paper No. 97/17.

    • Search Google Scholar
    • Export Citation
  • Debelle, G. and Vickery, J., 1998, “Is the Phillips Curve a Curve? Some Evidence and Implications for Australia,” The Economic Record, 74, December, pp. 38498.

    • Search Google Scholar
    • Export Citation
  • Harris, R. and Silverstone, B., 2001, “Testing for Asymmetry in Okun’s Law: A Cross-country Comparison,” Economics Bulletin, Vol. 5, No. 2, pp. 113.

    • Search Google Scholar
    • Export Citation
  • Kim, Yang Woo, 2002, “Optimal Horizons for Inflation Targeting in Korea,” Bank of Korea, Economic Papers, Vol. 5, No. 1.

  • Koopman, S., A. Harvey, J. Doornik and N. Shephard, 1995, STAMP 5.0 Structural Time Series Analyser, Modeller and Predictor, Chapman and Hall.

    • Search Google Scholar
    • Export Citation
  • Krolzig H. and Hendry, D., 2001, “Computer Automation of General-to-Specific Model Selection Procedures,” Journal of Economic Dynamics and Control, 25, pp. 831866.

    • Search Google Scholar
    • Export Citation
  • Layard, R., S. Nickell, and R. Jackman, 1991, Unemployment, Oxford University Press.

  • Okun, A.M., 1962, “Potential GDP: Its Measurement and Significance,” in Proceedings of the Business and Economics Statistics Section, American Statistical Association.

    • Search Google Scholar
    • Export Citation
  • Phillips, A.W., 1958, “The Relation Between Unemployment and the Rate of Change in Money Wage Rates in the United Kingdom, 1861–1957,” Economica, 25, pp. 28399.

    • Search Google Scholar
    • Export Citation
  • Richardson, P., L. Boone, C. Giorno, M. Meacci, D. Rae, and D. Turner, 2000, “The Concept, Policy Use and Measurement of Structural Unemployment: Estimating A Time Varying NAIRU Across 21 OECD Countries,” Economics Department Working Paper No. 250, OECD.

    • Search Google Scholar
    • Export Citation
  • Sargent, T., 1971, “A Note on the ‘Accelerationist’ Controversy,” Journal of Money, Credit and Banking, 3, August, pp. 72125.

  • Viren, M., 2001, “The Okun Curve is Nonlinear,” Economics Letters, April, pp. 253257.

  • Yu, Mansik, 2001, “Inflation Targeting in Korea: Tentative Evaluation and Policy Implications,” in Inflation Targeting: Theories, Empirical Models and Implementation in Pacific Basin Economies, November 15–18, Bank of Korea, Seoul.

    • Search Google Scholar
    • Export Citation
1

This paper was prepared by Craig Beaumont (APD). I thank Bank of Korea staff for comments on an earlier version of the paper at a seminar in Seoul on November 10, 2003 and Junhan Kim, BOK Research Department, for comments on this version.

2

The target was set in terms of the annual average inflation rate, i.e., the percent change in the arithmetic average CPI index in January-December of one year compared with the previous year.

3

The quarterly output gap was estimated by a Kalman filter approach, treating the gap as an unobserved autoregressive component, using STAMP 5.0, Koopman et al. (1995).

4

Okun’s Law is a relationship between the cyclical components of GDP and unemployment. The variant estimated in this paper focuses on actual unemployment and real GDP growth. To allow for possible changes in the NAIRU or potential growth, various trends are tested in the estimated equation.

5

BOK survey data on inflation expectations supports the hypothesis that expectations are quite inertial. Chart 2 of Yu (2002) shows data on average inflation expectations in 1998, 1999, and 2000, and these are little changed despite the rise in inflation during this time.

6

This strong mean reversion may reflect a combination of two forces: (i) unemployment affecting real wages and thus the strength of employment growth relative to real GDP growth; (ii) a discouraged worker effect, where the long-term unemployed exit the labor force, reducing unemployment. A more structural analysis of labor demand and supply would be needed to evaluate the relative significance of these two factors.

7

The projections are dynamic, using projected values for lagged dependent terms, so the standard error bands widen over the forecast horizon. This effect is more important in the Okun’s Law equation due to the large coefficient on lagged unemployment.

8

Dynamic simulations use actual values for exogenous variables, but projected levels of any endogenous variables including lags of the dependent.

9

The naive models were estimated as a structural time series equation within STAMP 5.0, Koopman et al. (1995).

10

These projections incorporate preliminary official estimates of real GDP growth in 2003 Q4 of 3½percent q/q, which implies a high rate of annualized growth of 15 percent.

11

No explicit monetary policy assumption is made. The real GDP growth projection assumes that monetary policy remains supportive of economic recovery.

12

Data on the core CPI excluding public service charges is available since 1985, and allowing for lags, the sample for estimation of the Phillips curve could begin in 1986. When the chosen specification is estimated over 1986–2003, the parameters are similar and statistically significant. However, the residual volatility in the late 1980s is much higher, at 0.56 percent compared with 0.29 percent from 2000 on. This period, which coincides with a boom in the housing market, was dropped from the sample.

Republic of Korea: Selected Issues
Author: International Monetary Fund
  • View in gallery

    Headline CPI Inflation

    (Year-on-Year, Percent)

  • View in gallery

    CPI Inflation and Target Bands

    (Yaar-on-Year, Percent)

  • View in gallery

    Public Service Charges and Core CPI Inflation

    (Year-on-Year, Percent)

  • View in gallery

    Demand Pressure Indicators

    (In percent)

  • View in gallery

    Actual and Estimate for Core CPI ex. Public Charges

    (q/q, s.a., Percent)

  • View in gallery

    Contributions to Core CPI Inflation

    (Percent, s.a, q/q)

  • View in gallery

    Phillips Curve is Broadly Linear

    (Data from 1990Q2 to 2003Q4)

  • View in gallery

    Impact of Shocks on Core CPI ex. Public Charges

    (Percent change, q/q)

  • View in gallery

    Unemployment Rate Actual and Estimate

    (Percent)

  • View in gallery

    GDP Shocks Have Asymmetric Effects on Inflation

    (Core CPI, y/y, percent)

  • View in gallery

    Impact on Year-on-year Inflation Rate After 8 Quarters

    (Percent)

  • View in gallery

    Distribution of Core CPI Inflation Projection Errors Due to Normally Distributed GDP Projection Errors 8 Quaters Ahead

    (Year-on-year, percent)

  • View in gallery

    Out-of-sample Projections of Core CPI ex. Public Charges

    (q/q, s.a., Percent)

  • View in gallery

    Out-of-sample Projections of the Unemployment Rate

    (Percent)

  • View in gallery

    Distribution of Real GDP Growth Rates, 1970-2003

    (Quarter-on-quarter, seasonally adjusted, percent)

  • View in gallery

    Unemployment Rate Projections

    (Percent)

  • View in gallery

    Inflation Rate Projections

    (Core CPI ex. Public Charges, y/y, percent)