ظرفیت و عامل اثر زمان بندی در مدیریت پرتفولیو فعال

Capacity constraints limit the profits of some investment strategies, while other strategies are more scalable. We develop a dollar-weighted return measure that parses the factor timing by investors and a strategy’s capacity constraints. We find that actively managed funds exhibit significant capacity and timing effects, while index funds display only timing effects. A portfolio’s liquidity, investment style, and distribution policy are important in explaining variation in capacity constraints. The analysis demonstrates that capacity and timing effects are important in analyzing portfolio manager skill and the cost of active investing.

Active portfolio management is a search for alpha in which the portfolio manager seeks to identify investment opportunities that more than compensate for their risks. To generate alpha in a portfolio is to exploit a “mispricing” through the lens of theoretical equilibrium models. Indeed, one tenet of most economic equilibrium models is that the profit-seeking actions of market participants compete away these abnormal profit opportunities. In this sense, there is an implied capacity constraint to any active portfolio management strategy: as more dollars seek out the same alpha-generating opportunities, those opportunities are depleted. This paper empirically examines the existence and nature of capacity constraints in active portfolio management. Open-end mutual funds present an opportunity to examine potential capacity constraints because investors have the ability to add to or withdraw cash from the fund throughout the fund’s existence. We begin by observing that the reported returns of an open-end mutual fund generally differ from the realized returns that each shareholder experiences during their investment period in the fund. This difference arises from two primary sources. First, a “timing” effect results from the factor timing of the individual shareholder’s investment (or disinvestment) in the fund shares. Second, a “capacity” effect arises from the return the fund is able to earn on the incremental dollar investment in the fund’s underlying strategy. In this sense, a fund’s return can be considered a function of the underlying return-generating technology (i.e., the portfolio manager’s “skill”) and the interaction of capacity constraints inherent in the return-generating technology with the size of the assets employing that technology. We derive a dollar-weighted average performance measure as a means to decompose the impact of the size of assets under management on fund performance into timing and capacity effects. In the model, the timing component reflects any correlation in the timing of fund flows and the realizations of a multi-factor model of expected fund returns. After controlling for timing, any residual difference represents a fund-specific effect arising from the correlation of flows and the underlying active strategy’s “alpha.” If managers fall short of their benchmark returns when exposed to flow, then we interpret this as an impact of capacity constraints in active portfolio management. Relying on a database of open-end domestic equity mutual funds, we show that both capacity and timing effects are economically significant and distinct drivers of performance, averaging negative impacts of 50 and 70 basis points, respectively, per year across the sample. Variation in capacity effects is driven by investment style and the capitalization of the active strategy’s underlying holdings. Fund policies which encourage or inhibit flows also matter. Front-end loads suppress both timing and capacity effects. Management fees (excluding 12b-1) are significant in explaining capacity, while marketing fees (12b-1) explain timing. Passively managed funds (i.e., index funds) display only the timing effect of fund flows; they show no significant capacity effects. This paper is structured as follows: Section 2 motivates the analysis and reviews the related literature. Section 3 develops a methodology to parse the difference between dollar-weighted return and time-weighted returns into timing and capacity components. Section 4 describes the data and empirical methods, while Section 5 presents the results for the timing and capacity effects and cross-sectional analysis. Section 6 offers a summary and conclusions.

This paper develops a dollar-weighted average return measure that allows the parsing of portfolio returns into timing and capacity components. Our timing component captures the extent to which flows are correlated with the fund’s underlying benchmark factor-model returns. The capacity component measures the degree to which the portfolios manager’s “alpha” relative to the benchmark index is related to the size of assets deployed in the alpha-generating technology. In actively managed funds, we find both capacity and timing effects, while passively managed funds display only timing effects. The capacity effect is related to the liquidity of the fund’s holdings and is more negative for funds that are sensitive to market liquidity, such as funds that focus on smaller-capitalization stocks. The timing component is linked to investment style, being more negative for growth as opposed to value funds. Management fees (excluding 12b-1) are significantly related to increased capacity effects, while marketing fees (12b-1) are associated with greater timing effects. In contrast, front-load fees act to suppress both effects, due to either effective broker advice or to the liquidity cost they impose on fund investors. Our analysis shows that the parsing of timing and capacity effects is critical in understanding how flows and the size of assets under management impact the performance of portfolios and the experience of investors. Capacity constraints in an active strategy influence the performance in both time- and dollar-weighted measures as the assets under management of the strategy change. Our results suggest that biases in estimates of manager skill arise from capacity constraints that are in addition to trading cost or impact cost of flows that are acknowledged in the extant literature. Our results also shed light on the differences between active and passive (index) portfolio manager performance in that active management faces capacity constraints to alpha-generating technology (French, 2008). Timing effects measure the “passive” component of flow-induced costs. Although this component is determined largely by the actions of fund investors, it can be influenced by a fund sponsor’s marketing policies. Our capacity component is a more direct measure of the fund manager’s ability to dynamically respond to fund flows. Consistent with this characterization, we find that capacity effects are strongly related to the liquidity of a fund’s holdings. Our parsing of the difference between dollar- and time-weighted returns provides a useful starting point for any future modeling of fund performance in light of the capacity constraints imposed by fund flows. Further exploration of the forces that drive the behavior of investors is necessary to determine the optimal fund policy for managing the timing of fund flows. Similar to the gains from prior empirical research that decomposes fund performance (Wermers, 2000), our capacity component can be used as the basis for the development of measures of fund manager skill after adjusting for the exogenous actions of fund investors. Our results suggest that capacity effects transcend the open-end fund vehicle and apply more generally to active portfolio management. That is, even closed-end funds or actively managed separate accounts might be affected if the total assets under management for the strategy’s underlying alpha-generating technology have placed the strategy near its capacity constraints. Our analysis shows that flow and capacity policies of a portfolio manager or fund sponsor can affect the performance of a portfolio or its investors. For example, an active fund with flow constraints could have less of a combined effect than a passive fund without flow constraints. Closures of actively managed funds are consistent with sponsors’ recognition of capacity constraints to active management that we have measured. We also believe that our results can shed light on issues related to “factor-capacity” constraints. In this paper, we find that portfolio strategies that face little or no capacity constraints, such as passive strategies, have differences in dollar- versus time-weighted returns, though they show no average fund-specific capacity effect. While we have interpreted the difference in dollar- and time-weighted returns to a multi-factor benchmark as a “timing” component, we note that this effect could be a market-wide or factor-related capacity constraint. In this case, if the size of assets under management becomes large enough across funds within the same market segment (represented by the same exposure to the benchmark factors), then those asset prices could be bid up to the point that the average returns to those factors are decreased. Given that our focus is on the capacity constraints in the alpha-generating capability of active managers, we have not attempted to distinguish between a “timing” effect and a “factor-capacity” constraint and we leave this issue for future research.