Are the Bank of Korea's Inflation Forecasts Biased Toward the Target?

The Bank of Korea (BoK) regularly publishes the Economic Outlook, offering forecasts for key macroeconomic variables such as GDP growth, inflation, and unemployment rates. This study examines whether the BoK’s inflation forecasts exhibit bias, specifically a tendency to align with its inflation target. We extend the Holden and Peel (1990) test to incorporate state-dependency, defining the state of the economy based on whether realized inflation falls below the target at the time of the forecast. Our analysis reveals that the BoK’s inflation forecasts are biased under this state-dependent framework. Furthermore, we examine a range of bias correction strategies based on AR(1) and mean error models, including their state-dependent variants. These strategies generally improve forecast accuracy. Among them, the AR(1)-based correction exhibits relatively stable performance, consistently reducing the root mean square error.

💡 Research Summary

This paper investigates whether the inflation forecasts published by the Bank of Korea (BoK) exhibit a systematic bias toward the institution’s inflation target. Building on the classic Holden‑Peel (1990) test of forecast unbiasedness and efficiency, the authors introduce state‑dependency by defining the economic state as a binary indicator of whether realized inflation at the time of the forecast is below the target. Using a half‑yearly dataset that spans from the first quarter of 1999 to the second half of 2024, the study matches BoK’s projected real‑GDP growth, CPI‑based inflation, and unemployment rates with their actual realizations from the BoK and Statistics Korea. Inflation targets were 3 % until 2015 and 2 % thereafter; the midpoint of the target range serves as the reference point.

The authors conduct separate Holden‑Peel regressions for the two states (below‑target vs. above‑target) and for two forecast horizons: one‑quarter ahead (h = 1) and three‑quarters ahead (h = 3). The results reject the null of unbiased forecasts in both states. Specifically, when inflation is below the target, the BoK tends to over‑forecast inflation; conversely, when inflation is above the target, it under‑forecasts. This pattern holds for both horizons and contrasts with earlier Korean studies (e.g., Kwark et al., 2011) that found no bias. By contrast, forecasts for real‑GDP growth and unemployment do not display state‑dependent bias, suggesting that the bias is confined to the inflation variable.

To address the identified bias, the paper evaluates four simple correction schemes that rely only on the BoK’s own forecasts and realized values: (1) a global mean‑error adjustment, (2) a global AR(1) adjustment, (3) state‑specific mean‑error adjustment, and (4) state‑specific AR(1) adjustment. The AR(1) models treat the forecast error series as a first‑order autoregressive process, allowing dynamic correction, while the mean‑error models simply subtract the average error. The authors compute the root‑mean‑square forecast error (RMSFE) before and after correction. The AR(1)‑based corrections consistently reduce RMSFE, with the state‑specific AR(1) version delivering the largest improvement—approximately a 12 % reduction in RMSFE when inflation is below the target. Mean‑error corrections also improve accuracy but are less stable across states and horizons.

The paper concludes that BoK’s inflation forecasts are indeed biased toward the target and that a modest, data‑driven AR(1) correction can meaningfully enhance forecast performance. The findings have policy relevance: acknowledging the bias can improve the credibility of the central bank’s communication, and incorporating simple bias‑correction techniques into the forecasting workflow could provide more reliable guidance for policymakers, markets, and the public. The authors suggest future work could explore real‑time estimation of state‑transition thresholds and more sophisticated regime‑switching models to capture the underlying bias mechanism more precisely.