'BOXPI' is an acronym for the boxed Korea Composite Stock Price Index (KOSPI). Uniquely, KOSPI has remained within an extremely narrow range during 2012{2016 despite global liquidity expansion. This study provides an interpretation of the BOXPI phenomenon from industry portfolio perspectives. We derive a continuous-time industry portfolio index model to describe sector rotation and use it to explain the BOXPI. We also investigate how the determinants of the sector returns differ across sub-periods by employing a Bayesian variable selection method. Our empirical analysis finds that sector rotations have progressed in the Korean stock market according to the stages of economic scenarios. The upper bound of the BOXPI can be interpreted as a consequence of the rotations from cyclical to defensive sectors during that period. As a result of the Bayesian variable selection analysis, we find heterogeneities in the key determinants of returns across the sectors and sub-periods. The lower bound of the BOXPI can be regarded as a result of overall low valuation, that is, low price-to-book ratio, of the Korean stock market in the BOXPI period.

JEL Classification: G11, G12, G14

Keywords: Bayesian variable selection, BOXPI, Korean stock market, Sector rotation​​