Stock duration and misvaluation

They conclude not only that short-term investors have a predictable impact on stock prices, but also that the length of time institutional investors tend to remain invested in a company increased slightly between 1985 and 2010, in contrast to the commonly held view that investor time horizons in general are shortening. Their findings fuel the ongoing public discussion of the impact that shorter investment time horizons can have on financial market efficiency, or on the actions of corporate management who may be influenced, or directly incentivised, by one- to two-year stock price movements.

From analysis of US equity prices over a twenty-five year period and the length of time that institutional investors remain invested in individual companies, Cremers et al demonstrate that an increase in the proportion of short-term institutional investors in a stock leads to a corresponding increase in the stock price. The price change tends to reverse over the following two years as the proportion of short-term investors in a stock returns to the average level. This pattern occurs with such regularity that the authors believe it indicates the initial price change is unrelated to a change in the outlook for the company, i.e. it is likely to be a price ‘bubble’ or a “speculative component”, driven by the presence of short-term investors. The price reversal trend appears to be strongest among the most over-valued stocks and stocks with characteristics such as illiquidity that make arbitrage (and thus correction of the pricing anomaly) less likely.

The core of the paper presents statistical evidence supporting the authors’ conclusions, with the analysis focusing on a new method of measuring average investor holding period – stock duration. Stock duration is defined as the average length of time that each investor has held a particular stock (based on quarterly reported institutional holdings), averaged across all investors in that stock and weighted by the size of their holding. It differs from traditional methods of calculating investor holding period, such as share turnover, by considering the holding period for individual stocks not an average for all stocks. This is a critical assumption because in practice there can be a wide range of holding periods between the stocks held by an institution in different portfolios, or even within a single portfolio.

Using the stock duration method, Cremers et al find that institutional investors’ average holding period increased from 1.2 years to 1.5 years between 1985 and 2010. This finding contrasts with conclusions based on share turnover (the number of shares traded divided the number of outstanding shares), which grew by about 300% over the same period suggesting that average holding period decreased, and also contradicts conclusions based on fund turnover, which imply that the holding period remained relatively stable. To try and reconcile this difference the authors classify their data into four categories: banks and insurance firms; independent investment advisors and companies, e.g. mutual funds; pension funds, endowments and foundations (asset owners); and all other institutional investors. They find that pension funds and endowments had the longest median stock duration over the period (1.7 years) followed by banks and insurance companies (1.5 years). However the banks’ stock duration was relatively stable (ranging from 1.2 to 1.8 years), while the pensions funds’ stock duration increased from 0.85 years at the beginning of the period to 2 years at the end.

The authors posit that the increase in both stock duration and in share turnover over the time period studied can be explained by a shift towards indexed investments, which by their nature are longer term and therefore increase stock duration, and an increase in ‘high frequency trading’ (typically intra-day, automated trading), partly driven by a reduction in the overall cost of trading. Since stock duration only considers stocks held for more than one quarter it ignores both high frequency trading and shortterm adjustments to institutional portfolios, and the authors therefore believe that stock duration is a more robust reflection of the underlying trend in holding period by institutional investors.

Having demonstrated that stock duration, as a measure of investors’ time horizon, can predict changes in equity pricing, Cremers et al consider why the temporary speculative components are not removed through normal market arbitrage. They find that the predictive power of stock duration is only evident in stocks that have barriers to arbitrage, such as lower liquidity or higher idiosyncratic volatility. They also find some evidence – though not highly significant - that stocks held by short-term investors may be more likely to be overvalued, while those held by long-term investors are more likely to be undervalued. Again, these results are strongest in the group of stocks with barriers to arbitrage.

In concluding, the authors summarise their key findings as being firstly, the stable or lengthening “stock duration” for stocks since 1985, and secondly, the association between short term investors and temporary equity price distortions. Their results also support academic theories on investor overconfidence, which suggest that short-term investors may be more prone to overconfidence and hence can cause over valuations.