Using Clustering Across Mutual and Alternative Funds to Measure Diversification, Lower Bound of Market Size, and Market Efficiency

Abstract

We study clustering of underlying investments across funds in public and alternative markets. Funds often invest in the same company or stock, and we study whether the amount of overlap is indicative of random or clustered investing and what that can tell us. We start with the simplest case, and derive formulas for the overlap distribution of pairs of mutual funds investing in stocks.

We then model clustered investing, generate artificial data with various clustering, and test measures for their efficiency in detecting clustering. We retain two measures: the free parameter of our clustering model, and the relative effective market size assuming random investing. We then apply our measures to an alternative market, using investment data from a large European venture capital fund-of-funds. We study its overlap ratio over time and compute clustering as a measure of diversification. We also compute a lower bound on market size based only on the portfolio without reference to a market survey. We suggest that clustering might be a good measure for market efficiency that is independent of market prices and therefore applicable to all markets including alternative ones.

Keywords: Venture Capital Fund, Private Equity Fund, Diversification, Clustering, Market Efficiency, Venture Capital, Private Equity, Alternative Asset, Fund-of-Funds, Diversification

JEL Classification: D4, D5, G14