It is commonly known that investors in equity and bond markets are significantly different in terms of risk tolerance, information focus, and institutional composition. In the presence of such asset market segmentation, do better-integrated investors exhibit atypical investment decisions using equity and bond mutual funds within the same fund family (sister funds)? Auh and Bai (2019) document that investors’ holding decisions co-move on equity and bonds of the same issuer (cross-held asset), indicating that there is information spillover. They also show that fund families provide superior investment returns for cross-held assets. This result suggests that information content in the two asset markets is not redundant, and that combining information from both markets provides more complete information for firms. In this paper, we test whether the same phenomenon can be observed in Korea. In contrast to the strong co-movement reported in Auh and Bai (2019), we do not find any meaningful correlation in the holding decisions on cross-held assets. This insignificant correlation remains unaffected even after we consider several possibilities that may obscure the true nature of trading behavior. This stark contrast can be driven by several factors. First, asset market segmentation is particularly severe in Korea, as reported by Yang (2013) and Yoon and Ohk (2014), such that even investors in the same institution are not exposed to any information spillover. Second, the corporate bond market in Korea could be too illiquid for any subsequent holding changes to occur regardless of information spillover. To explore these two explanations, we exploit investment decisions on cross-held assets within a set of mixed funds whose asset class mandatorily covers both equity and bonds. It is reasonable to argue that information spillover is more feasible within an individual fund, and it is therefore a test of the degree of asset market segmentation. We find a strong positive correlation with cross-asset holding change from mixed funds. This result immediately rules out the illiquidity-based explanation because sister funds and a mixed fund face the same level of bond market illiquidity. We further examine whether bond sister funds and mixed funds hold bonds that are significantly different from each other in terms of liquidity-related characteristics (e.g., credit rating). However, we do not find any meaningful differences, and we thus confirm that the differences in the results cannot be attributed to heterogeneous bond holdings between the two types of funds. We further investigate the main driver of co-movement. If co-movement is mainly motivated by information spillover, then the investment performance of the cross-held asset must be at least similar to or better than the same asset holding of standalone funds (non-sister-funds). Surprisingly, our results show that cross-held assets are subject to inferior investment decisions from the perspective of ex-post performance evaluation. This finding suggests that the co-movement is not driven by price-relevant information sharing. In particular, the degree of co-movement is strongly affected by the fund flow: it is only significant when funds experience a large degree of fund flow (more than ±5% flow to the assets under management). The mixed funds therefore appear to scale the whole portfolio up or down in response to the fund flow, mechanically generating co-movement of the cross-held assets. As a final test, we examine the effect of mechanical adjustment of fund flow fluctuations on return predictability. The significance of inferior investment decisions disappears when the co-movement is not driven by scaling behavior (i.e., with a small degree of change in fund flow). However, the investment performance of the cross-held assets becomes worse when decisions are associated with a larger variation in fund flow. These results indicate that scaling adjustment is a dominant factor driving ex-post negative fund returns. Previous research on mutual funds in Korea has mainly focused on equity mutual funds. Our paper provides a natural expansion of the research domain given the recent study of cross-asset holding. In particular, we show that there is no meaningful interaction across different asset markets within a mutual fund family in Korea. Even within a mixed fund, it appears that information from both markets is not actively used beyond what is caused by a mechanical scaling adjustment. These findings provide strong support for substantial segmentation between the equity and bond markets in Korea. Finally, we demonstrate that such non-information-driven holding changes are, on average, associated with worse future returns.