Measuring managerial skill using value added and alpha: evidence from the Korean equity fund market Open Access

Understanding whether mutual fund managers possess genuine investment skill remains a central research question in the asset management literature. A substantial body of research, beginning with Jensen (1968) and followed by numerous studies such as Gruber (1996), Wermers (2000), French (2008), and Cremers and Petajisto (2009), finds that actively managed mutual funds underperform passive benchmarks on a net-of-fee basis. This long-established empirical finding has often been interpreted as evidence that mutual fund managers lack the ability to generate abnormal returns and that capital markets are broadly efficient. Under this view, alphas estimated from factor models represent meaningful indicators of managerial underperformance, and persistent positive alpha would be difficult to justify in a competitive asset management industry.

However, several conceptual and empirical studies complicate this conventional interpretation. First, the equilibrium model of Berk and Green (2004) demonstrates that even if managers possess differential skill levels, investor capital flows will respond by allocating more capital to managers with higher expected alphas. As assets under management grow, decreasing returns to scale drive equilibrium alphas toward zero, making excess returns an equilibrium outcome rather than an indication of no skill. Second, empirical studies including Grinblatt and Titman (1993), Daniel et al. (1997), and Kacperczyk et al. (2005) provide evidence that some managers exhibit stock-picking skill even if their realized alphas are statistically indistinguishable from zero. These findings suggest that alpha may fail to capture true underlying skill, especially in equilibrium environments characterized by performance-sensitive investor flows.

In light of these concerns, Berk and van Binsbergen (2015) propose a fundamental redefinition of skill measurement. They argue that the economically meaningful quantity is value added, defined as the product of a fund's gross alpha and its lagged assets under management. Value added measures the dollar amount of abnormal return a manager creates for investors relative to a benchmark. They document three key empirical results using U.S. mutual fund data: (1) value added is positive on average, (2) value added exhibits substantial persistence, and (3) managerial compensation is highly correlated with value added. These findings imply that percentage alpha should not predict future performance in competitive equilibrium, while value added should.

Despite the influence of this framework, little is known about whether the underlying economic mechanisms apply outside the United States ^[1]. International fund markets differ significantly in regulatory structure, industry maturity, fee competition, and investor behavior. The Korean fund industry represents a particularly striking contrast. Korean equity funds tend to have shorter life cycles, more frequent mergers and liquidations, more volatile investor flows, and a distribution-driven sales structure that often weakens the link between performance and capital allocation. These institutional characteristics raise the possibility that Korean funds may not operate under the equilibrium conditions assumed by Berk and Green (2004). If equilibrium does not hold, then value added may not outperform alpha in measuring skill, and performance-sensitive flows may be insufficient to eliminate abnormal returns. Consequently, alpha remains a robust metric in frictional markets where competitive forces are insufficient to erode abnormal returns.

Our study investigates whether the value added framework proposed by Berk and van Binsbergen (2015) effectively captures managerial skill in the Korean fund market. We address three research questions. First, does value added exhibit persistence and out-of-sample predictive power in Korea? If value added indeed measures skill, funds ranked highly by value added should continue to generate high value added in future periods. Second, does compensation, defined as lagged total assets multiplied by the total expense ratio, predict future value added? The U.S. evidence suggests that compensation reflects investors' expectations of skill; thus, a strong relationship would indicate that Korean investors allocate capital in a performance-sensitive manner. Third, is alpha rendered uninformative about future fund performance, as predicted by equilibrium models? If value added dominates alpha in Korea, this would support the generality of the equilibrium model. Conversely, if alpha remains predictive, the Korean market may diverge from equilibrium dynamics.

Our empirical findings reveal a markedly different picture from that documented in the U.S. mutual fund industry. In the Korean equity fund market, both value added and alpha indicate that fund managers on average possess economically and statistically significant skill. Moreover, past performance, measured using either value added or alpha, exhibits clear and persistent predictability for future value added over horizons of up to five years. This stands in contrast to the U.S. evidence, where alpha fails to forecast future performance and where value added dominates as the equilibrium-consistent measure of managerial skill.

A key empirical distinction is that, unlike the U.S., alpha remains an informative and reliable proxy for skill in the Korean market. We find little evidence of decreasing returns to scale: fund size does not attenuate alpha, and the cross-sectional distribution of alpha does not compress as fund size increases. Consequently, market forces that, in equilibrium, should eliminate abnormal returns appear weaker in Korea. This helps explain why alpha sorts perform nearly as well as value added sorts and why both measures reveal strong persistence in manager skill.

Compensation also predicts future value added, consistent with investors rewarding skilled managers. However, compensation does not fully extract the value created: even after expenses, performance persistence remains visible, implying that part of the value added accrues to investors. Taken together, these findings suggest that the Korean fund industry operates in a setting that is less competitive and less frictionless than the equilibrium environment assumed in Berk and Green (2004). As a result, equilibrium predictions such as the superiority of value added over alpha do not fully materialize. Our results highlight the importance of accounting for local institutional features, market competitiveness, and investor behavior when applying equilibrium-based frameworks to international asset management industries.

The rest of the paper proceeds as follows. Section 2 reviews the related literature. Section 3 describes the data and empirical methodology. Section 4 presents key empirical findings. Section 5 concludes.

The literature on mutual fund performance spans more than 5 decades and offers a wide range of empirical perspectives on whether actively managed funds add value for investors. The earliest strand of research, beginning with Sharpe (1966) and Jensen (1968), documents that actively managed mutual funds underperform passive benchmarks on a net-of-fees basis. These findings, further reinforced by later studies such as Gruber (1996), Wermers (2000), and French (2008), form the foundation of the traditional view that mutual fund managers, on average, do not possess sufficient skill to generate persistent abnormal returns. In this framework, the poor average performance of active funds is often interpreted as evidence supporting the efficient capital market.

However, a substantial body of research challenges the notion that mutual fund managers lack skill entirely. Several studies argue that traditional risk-adjusted returns may fail to capture managerial ability because they overlook the complexity of fund decisions. Grinblatt and Titman (1993) and Daniel et al. (1997) demonstrate that some managers have superior stock selection or style timing ability that is not fully reflected in alphas from factor models. Kacperczyk et al. (2005) further show that funds with concentrated industry bets often outperform, suggesting that skilled managers deliberately deviate from benchmark holdings. These findings imply that skill may exist in dimensions not captured by traditional performance measures, raising questions about whether alpha alone can accurately represent managerial ability.

This tension between evidence for and against managerial skill prompted a theoretical shift in the literature. Berk and Green (2004) develop a rational competitive equilibrium model showing that investor flows respond to managerial skill and efficiently allocate capital across funds. In their model, skilled managers attract capital until diminishing returns reduce alphas to zero. Consequently, zero alpha in equilibrium does not imply the absence of skill; rather, it reflects efficient capital reallocation that equalizes marginal returns across managers. This insight provides a theoretical explanation for why alphas are small and often insignificant even when skill may be present.

Building on this theoretical foundation, Berk and van Binsbergen (2015) propose value added, defined as the dollar value of abnormal performance, as a more economically meaningful measure of managerial skill. They argue that value added should remain informative about skill because investors allocate more capital to skilled managers, increasing the scale at which skill is applied. Their empirical results using U.S. data are compelling: value added is positive on average, strongly persistent, and highly correlated with future compensation. They conclude that value added, not alpha, is the correct measure of managerial skill in competitive asset management industries.

Despite the compelling U.S. evidence, the generalizability of the value added framework remains uncertain. Several international studies indicate that investor behavior and fund industry structures vary significantly across markets. Ferreira et al. (2013) show that fund performance, flows, and fees differ substantially across countries due to institutional, regulatory, and cultural factors. European and Asian mutual funds have weaker flow–performance sensitivity, more pronounced distribution-channel effects, and more heterogeneous fee structures (Huij and Verbeek, 2007). These market features suggest that the competitive equilibrium mechanism assumed by Berk and Green (2004) may not operate fully outside the United States.

The Korean fund industry provides a particularly interesting contrast. Korean funds tend to have significantly shorter lifespans, higher merger and liquidation rates, heavy reliance on bank and securities-firm distribution channels, and investor flows that are more sensitive to marketing efforts than to performance. This institutional environment implies weaker competitive pressures, potentially limiting the degree to which investor flows arbitrage away alpha. Existing Korean studies (e.g. Yoo and Kim, 2012; Ha, 2014) largely conclude that active equity funds underperform benchmarks, but the mechanisms behind this underperformance remain largely unexplored. Importantly, prior work in Korea has almost exclusively relied on alpha-based measures and has not examined managerial skill through the lens of value added.

Thus, two gaps emerge in the literature. First, while value added has been shown to be informative in the U.S., it is unclear whether this finding extends to fund markets with weaker competitive dynamics. Second, the Korean fund industry has not yet been analyzed through the value added framework of Berk and van Binsbergen (2015). Addressing these gaps is essential for understanding whether equilibrium-based measures of skill can be applied universally or whether their validity depends on specific institutional structures.

This paper contributes to the literature by conducting the first comprehensive empirical examination of value added, compensation, and performance persistence in the Korean fund industry. By comparing alpha and value added frameworks in a market that differs structurally from the U.S., we provide new evidence on how institutional frictions, distribution channels, and flow dynamics shape the measurement and interpretation of managerial skill.

This study uses weekly portfolio holdings and weekly return data for Korean equity funds obtained from Zeroin, covering the period from January 2005 to December 2022. Because the original dataset is reported at various frequencies, all information is converted to a weekly frequency using Friday observations. Specifically, net asset value (NAV), total net assets (AMT), and other balance-sheet variables are recorded as of Friday, while weekly returns, alphas, and value added measures are computed using price changes from the previous Friday to the current Friday.

Our sample includes only funds with at least five years of available observations. Funds for which fee information is not reported are excluded. Because master funds are not directly investable, only feeder funds are included. To focus on actively managed products, index funds and ETFs are removed. To mitigate the influence of extreme observations, we trim the sample by removing the top and bottom 1% of funds based on the time-series average of gross alpha; all observations corresponding to those extreme funds are deleted entirely.

Weekly factor returns used for estimating risk-adjusted alphas, including the Carhart (1997) four-factor model, are obtained from FnGuide's DataGuide service. The weekly risk-free rate is constructed by converting the 91-day certificate of deposit rate to a weekly series.

Table 1 presents summary statistics for the funds used in the study. Panel A reports pooled statistics, while Panel B summarizes cross-sectional means after computing each fund's time-series average. NAV, AMT, and compensation are measured in 100 million won units; fund age is measured in years; and total expense ratio is expressed in annual percentage terms. As shown in Panel B, the average NAV is 50.1 billion Korean Won (KRW), whereas the median is only 9.6 billion KRW, indicating the presence of several very large funds. The average fund age is 7.55 years. The average total expense ratio is 1.72%, and mean compensation computed as lagged AMT multiplied by the total expense ratio is approximately 15 million KRW per week.

This study employs both conventional return-based performance measures and the value added metric proposed by Berk and van Binsbergen (2015). We begin by describing standard return-based measures before introducing the value added construction. The net return $Ritn$ represents the return to investors after deducting fees and thus corresponds to the actual performance delivered to investors at time $t$⁠. The gross return $Ritg$ includes fees and is therefore calculated as $Ritg=Ritn+fi,t−1$⁠, where $fi,t−1$ is the total expense ratio at time $t−1$⁠. Gross returns are widely used as a proxy for managerial ability in prior literature. The net alpha, $αitn,$ is computed as $αitn=Ritn−RitB$⁠, where $RitB$ is the benchmark return for fund $i$ at time $t$⁠. Gross alpha is similarly defined as $αitg=Ritg−RitB=fi,t−1+αitn$⁠.

For each fund, the time-series average of gross alpha is computed as $αˆig=1Ti∑t=1Tiαitg$⁠, and has traditionally been used as a measure of managerial ability. However, Berk and Green (2004) argue that gross alpha fails to measure skill in the presence of competitive capital flows, which drive alphas toward zero even when managers possess skill. They propose that the appropriate measure of performance is the dollar value created for investors. Following this approach, this study defines fund value added in monetary units (Korean won). Value added for fund $i$ at time $t$ is $Vit=qi,t−1αitg$⁠, where $qi,t−1$ is the real, inflation-adjusted AMT of the fund at time $t−1$⁠. This represents the real economic value generated by fund $i$ at time $t$⁠. The average value added of a fund is thus $Sˆi=E[Vit]$⁠. Because many funds do not survive for the entire sample period, we compute the cross-sectional average value added by weighting each fund's mean value added by its survival length.

Accurate estimation of alpha requires a benchmark model that captures relevant risk factors. To ensure robustness, we employ two benchmark specifications widely used. First, we use the Carhart (1997) four-factor model to estimate risk-adjusted benchmark returns. The predicted return for a fund under this model represents the portion of performance attributable to systematic risk exposures.

Second, we use the benchmark index designated by each fund. Korean equity funds typically specify a benchmark such as the Korea Composite Stock Price Index (KOSPI), or KOSPI 200 index. In our dataset, 85.57% of the observations use the KOSPI200 index as the benchmark, while the remaining observations are benchmarked against dividend indices or customized indices. Because managers are evaluated relative to these benchmarks in practice, alpha is calculated as the difference between the fund's realized return and its designated benchmark return.

NAV and AMT are deflated using the monthly Consumer Price Index (CPI), normalized to 2020 as the base year. Gross alpha is computed as the difference between the fund's realized weekly return and the benchmark's weekly return, while net alpha is obtained by subtracting the weekly-equivalent total expense ratio.

Risk-adjusted alpha is estimated using a 52-week rolling window regression of the Carhart four-factor model. Specifically, we estimate model coefficients using weekly returns over the preceding 52 weeks, compute the predicted returns using factor realizations, and define the risk-adjusted alpha as the difference between realized and predicted returns. Value added is then calculated by multiplying gross alpha (or its risk-adjusted counterpart) by the real total assets from the prior period.

Figure 1 presents the time-series evolution of fund total assets and the number of funds in the sample. The solid lines depict the 5th, 25th, 50th, 75th, and 95th percentiles of the real value of total assets, while the dashed line represents the logarithm of the number of funds. Fund size, measured in real terms using CPI-adjusted total assets, increased steadily until around 2010 and remained relatively stable thereafter. This stability is important because value added must be stationary for it to serve as a meaningful measure of skill; the relatively stable fund-size distribution suggests that value added is also likely to be stationary. Meanwhile, the number of funds rose rapidly until approximately 2016, followed by a gradual decline, consistent with consolidation and maturity in the Korean fund industry.

Table 2 reports the distribution of average alpha and value added across funds. For each fund, we first compute its time-series average and then report the cross-sectional distribution of these averages, weighted by the number of observations per fund following Berk and van Binsbergen (2015). Gross alpha is defined as the excess return over the benchmark, net alpha deducts the weekly equivalent of total fees, and risk-adjusted alpha is estimated using the Carhart (1997) four-factor model. Value added is the product of lagged total assets and the corresponding alpha, and therefore represents the incremental economic value generated per week. Reported units for value added are 100 million KRW.

Three findings in Panel A are noteworthy. First, the mean weekly gross value added is 0.2099, which annualizes to approximately 2.13 billion KRW—equivalent to about 2.13% of the average fund's total assets. The mean weekly risk-adjusted value added is even larger at 0.4758 (approximately 4.84 billion KRW per year, or 4.84% of assets). This indicates that, on average, Korean equity funds generate economically and statistically significant value for investors. Second, the share of funds with negative average value added is about 22%–24%, implying that most funds create positive economic value. Third, value added exhibits substantial heterogeneity across funds. For example, the median gross value added is only 0.02, far below the mean, and the 75th percentile is 0.0906, still below the mean. These patterns suggest that while most funds generate modest positive value, a subset of funds produces exceptionally large value added, driving the cross-sectional average upward.

Panel B of Table 2 shows the distribution of weekly alpha. The mean gross alpha is 0.0435% (approximately 2.26% annually), and the mean risk-adjusted alpha is 0.0684% (about 3.34% annually). The proportion of funds with negative alpha is below 20% for both measures. Notably, both gross and risk-adjusted alpha exceed the average level of total fees, implying that Korean fund managers generate positive abnormal returns even after accounting for expenses. This contrasts with the U.S. evidence reported in Berk and van Binsbergen (2015), where average net alpha is close to zero.

In summary, both alpha-based and value added-based measures provide strong evidence of managerial skill in the Korean equity fund market. Unlike in the U.S., where alpha typically converges to zero in equilibrium due to competitive capital flows, Korean fund alphas remain positive and exhibit considerable cross-sectional variation. This finding serves as an important foundation for our later analysis of persistence, compensation, and the role of fund size.

This subsection examines whether managerial skill in the Korean fund market exhibits persistence, namely, whether past value creation predicts future value creation. Figure 2 reports the out-of-sample performance of funds sorted into quintiles based on their past gross value added (Panel A) and past risk-adjusted value added (Panel B). Each week, funds are ranked using the previous three years of value added observations and allocated into five equal-sized groups. Funds with insufficient historical observations are included only if they have at least one full year of past data. For each ranking date, the future performance of each quintile is measured from 26 weeks ahead over horizons of one, three, and five years. The resulting weekly series of future value added for each quintile is then averaged across the entire sample.

Across all specifications, the top-quintile portfolios consistently produce the highest future value added. Notably, the magnitude of outperformance remains economically meaningful for horizons as long as five years. For example, when sorted on gross value added, the top quintile generates approximately 0.6 hundred million KRW per week—equivalent to roughly 31 hundred million KRW per year—even at the five-year horizon. This pattern indicates that Korean fund managers who historically generated higher value added tend to continue doing so, suggesting an economically meaningful degree of skill persistence in gross terms.

Figure 3 performs an analogous exercise using alpha-based rankings. Funds are sorted using either their gross alpha (Panel A) or their risk-adjusted alpha (Panel B), and future performance is again measured using gross value added. The results closely mirror those from Figure 2: funds in the highest alpha quintile continue to generate substantially higher future value added than those in lower quintiles across all horizons. This alignment between alpha-sorted and value added–sorted results indicates that alpha, despite its limitations as emphasized in Berk and van Binsbergen (2015), nonetheless contains information about managerial skill in the Korean context.

Table 3 provides statistical confirmation of these patterns. For each horizon, the table reports whether the top quintile's future value added exceeds (1) the bottom quintile, (2) the cross-sectional median, and (3) zero. The rejection rates for the null hypotheses are overwhelmingly high in the value added sorts and remain strong in the alpha sorts as well. In nearly all cases, the top quintile's future performance is significantly positive and significantly greater than both the bottom quintile and the median.

Overall, the evidence shows that managerial skill in Korea exhibits persistence when measured using gross value added and gross alpha, consistent with the economic interpretation that skilled managers continue to create value over long horizons. A noteworthy implication concerns the mechanism behind persistence. The fact that both alpha-sorted and value added–sorted portfolios produce nearly identical patterns suggests that, unlike in the U.S. evidence, alpha does not necessarily lose informational content as funds scale up. This aligns with the descriptive observation that fund size in Korea has not strongly eroded alpha, which in turn explains why alpha-based sorting continues to predict future value creation. Consequently, although Berk and van Binsbergen (2015) argue that value added is the theoretically relevant measure of skill, the Korean results imply that alpha remains an empirically powerful measure in environments where diseconomies of scale are weaker or where market competition is less intense.

Empirical results from the Korean fund market diverge in several important ways from findings in the U.S. literature. In particular, when skill is measured using value added, Korean fund managers exhibit both economically and statistically significant skill, and value added measures display meaningful persistence. These patterns are consistent with the conceptual framework of Berk and van Binsbergen (2015). However, unlike most prior empirical studies on the U.S. market, Korean fund managers also exhibit positive gross alpha on average, and portfolios sorted on past gross alpha produce a monotonic ordering of future value added. In other words, alpha-based skill measures, which typically show little to no persistence in the U.S., continue to identify skilled managers in Korea. Understanding why this difference arises requires a closer examination of fund size dynamics and the underlying market environment.

Figure 4 reports the distribution of alpha across fund size categories. Panel A shows gross alpha, and Panel B shows risk-adjusted alpha. To construct the figure, we divide funds into size intervals of 10 billion KRW (from 0 to 200 billion KRW in real total assets) and compute the mean alpha within each interval. The solid line presents the average alpha for each size category, and the dashed line shows the 95% confidence interval. The most salient feature of Figure 4 is that alpha does not decrease as fund size increases. In the U.S. market, the prevailing interpretation, consistent with the equilibrium model of Berk and Green (2004), is that as a skilled manager attracts assets, diminishing returns to scale reduce alpha toward zero. This mechanism is one of the principal motivations for using value added instead of alpha as a measure of managerial skill. In contrast, the Korean data show only a weak relationship between size and alpha. There is some evidence that medium-size funds earn slightly lower alpha than small funds, but this decline does not persist into the largest size categories, which often exhibit higher alpha.

This pattern helps explain why both value added and alpha reveal positive skill and meaningful persistence in the Korean market. If fund size does not scale sufficiently to the point where diminishing returns eliminate alpha, then alpha remains informative about managerial skill. The absence of decreasing returns to scale suggests that Korean equity funds remain below the capacity constraints typically observed in the U.S. market. Given the liquidity and depth of the Korean equity market relative to average fund size, managers can deploy capital without incurring the severe price impact costs that erode alpha in the Berk and Green (2004) framework. This contrasts sharply with the U.S. empirical literature, where net alpha is typically negative, gross alpha is close to zero, and persistence is weak or nonexistent (e.g. Carhart, 1997; Fama and French, 2010). Because U.S. fund size adjusts competitively to manager skill, alpha loses its informational content, precisely the mechanism formalized by Berk and Green (2004).

Interpreting the divergence between the two markets requires considering institutional and competitive frictions. The Korean fund industry is smaller, less competitive, and more distribution-driven than its U.S. counterpart. Investor flows are less sensitive to performance, and large-scale reallocation of capital toward skilled managers is less common. As a result, abnormal returns are not fully arbitraged away, even for funds that have already grown in size. The supply of skilled managers may also be more limited relative to the size of the investable universe, reducing the severity of decreasing returns to scale.

Taken together, these factors imply that the equilibrium condition underpinning the alpha-to-zero mechanism is not strongly active in Korea. Consequently, alpha retains explanatory power for managerial skill, and persistence in both value added and gross alpha emerges naturally. From a performance evaluation perspective, this suggests that value added and alpha should be viewed as complementary rather than competing measures in the Korean context. Value added continues to capture the economic magnitude of skill, while alpha remains useful for identifying managers whose skill has not yet been fully capitalized into fund size.

In this subsection, we examine how the created value is allocated between fund managers and investors. In other words, when managers generate economic rents through skill, we ask whether these rents accrue primarily to managers through compensation or whether a meaningful portion remains with investors. To address this question, we replicate the analyses in Figures 2 and 3 using net-of-fee measures to sort funds. Specifically, when forming quintiles based on past performance, we sort funds using net value added and net alpha. We then evaluate subsequent gross value added for each quintile to determine whether skill persistence is still present after managerial compensation has been removed.

Figure 5 summarizes the results. Panel A presents future gross value added sorted on past net value added, while Panel B presents future gross value added sorted on past net alpha. The design is therefore fully comparable to Figures 2 and 3, except that the sorting variables exclude manager compensation. In the U.S. market, Berk and van Binsbergen (2015) show that once compensation is removed, persistence disappears entirely: managers capture essentially all of the value they create, leaving no persistent net benefit for investors. In sharp contrast, the Korean market shows that persistence remains even when funds are sorted using net value added or net alpha. This implies that, unlike in the U.S., Korean fund investors retain a nontrivial portion of the value created by skilled managers. Put differently, the compensation structure in the Korean fund industry does not fully extract managerial rents, allowing some of the created value to pass through to end investors. This incomplete rent extraction is likely driven by institutional fee stickiness, as contractual rigidities and regulatory norms prevent managers from adjusting fees in response to superior performance. Unlike in the U.S., where managers can capture surplus by raising fees or rapidly expanding assets, Korean managers face structural barriers to price discrimination. Consequently, a meaningful portion of value added accrues to investors as involuntary surplus, rather than being captured by managers as economic rent.

Table 4 formally tests this interpretation using the same hypothesis-testing framework as in Table 3. The only methodological difference is that the sorting variables are net value added and net alpha, rather than their gross counterparts. Consistent with Figure 5, the results in Table 4 show that (1) the top-quintile funds continue to outperform the bottom-quintile funds, (2) the top-quintile funds' future performance generally exceeds the cross-sectional median, and (3) the top-quintile's future value added remains positive. Together, these findings confirm that skill persistence survives even after deducting total expenses, and provide strong evidence that Korean investors retain part of the economic value created by fund managers.

Finally, Figure 6 investigates whether more skilled managers receive higher compensation. For each quintile formed on past performance, we compute future compensation as the product of lagged total assets and the total expense ratio. Panels A, B, C, and D correspond to the four measures of past performance: gross value added, risk-adjusted value added, gross alpha, and risk-adjusted alpha. As expected, higher-skilled quintiles tend to receive higher future compensation, implying that the Korean fund industry does reward skill to some degree.

Taken together, the evidence from Figures 5 and 6 indicates that compensation in the Korean fund industry responds meaningfully to managers' past performance, yet it does not fully extract the value they generate. Even after accounting for fees, future value added continues to exhibit strong persistence, implying that a nontrivial portion of managerial skill ultimately accrues to investors rather than being entirely captured by managers themselves.

This stands in stark contrast to the U.S. evidence, where managerial compensation absorbs nearly the entirety of value added. The Korean industry therefore exhibits characteristics of a market that is less competitive or less mature, in which investor flows may not fully arbitrage away skill-based rents. In short, while skilled Korean fund managers do receive higher compensation, investors continue to capture meaningful residual value, indicating that the Korean fund market has not yet reached the frictionless competitive equilibrium described in Berk and Green (2004) and Berk and van Binsbergen (2015).

This study re-examines how managerial skill should be measured in the Korean equity fund industry by applying the value added framework of Berk and van Binsbergen (2015). The motivation arises from the possibility that equilibrium-based predictions developed primarily for the highly competitive U.S. market may not generalize to countries with different market structures, investor behavior, and levels of competition. Using a comprehensive dataset of Korean equity funds from 2005 to 2022, we evaluate managerial skill through both value added and alpha to determine whether value added dominates alpha as a performance measure in an international context.

Our empirical findings reveal a markedly different set of patterns compared with the U.S. evidence. Korean fund managers, on average, generate positive and economically meaningful value added, and this performance exhibits persistence over horizons of up to five years. Unlike the U.S. market, however, alpha also remains strongly positive and displays similar persistence, indicating that alpha retains informational content and continues to measure skill in the Korean environment. We further document that fund size does not attenuate alpha, suggesting that decreasing returns to scale are not strongly present. These deviations imply that the Korean fund market does not operate under the same competitive and frictionless equilibrium conditions assumed in prior theoretical models.

The results carry several implications for both theory and practice. For academics, our findings underscore the importance of incorporating institutional frictions, market segmentation, and heterogeneous investor behavior when applying equilibrium-based models internationally. The Korean fund market's partial lack of competitive pressures allows abnormal returns to persist, preventing fund size from fully capitalizing managerial skill and weakening the predictive dominance of value added. Accordingly, our results indicate that value added is not always a superior performance measure, but one whose validity is conditional on market efficiency. In markets where capital reallocation is sluggish and competitive pressures are dampened, traditional alpha retains strong informational content. For practitioners and policymakers, the evidence suggests that performance evaluation frameworks relying exclusively on value added may be misleading in less mature markets. Strengthening performance-sensitive capital flows, enhancing fee transparency, and fostering competitive dynamics may improve the alignment between managerial incentives and investor outcomes.

Finally, a limitation of this study is that the Korean mutual fund market is smaller and has a shorter history than the U.S. market, which warrants caution in interpreting the results as direct contrasts to U.S. evidence or as general tests of equilibrium-based models.