Mobile Edition
Print
Get the Mag
Weekly UpdatesThis article analyzes the historical performance of hospitality stocks, taking into account the magnitude and timing of investor capital flows in and out of the hospitality sector. Using historical return data on hospitality common stocks, this article shows that there is substantial market timing in the hospitality industry, with firms issuing equity capital near market highs and retiring capital near market lows. For hospitality investors, market timing translates into a shortfall of 1.5 percent per year over the time period 1962-2006. A value-weighted portfolio of both restaurant and hotel firms earns a lower average return over this time period compared to a similar portfolio of either hotel or restaurant stocks only, because of the same timing issue.
An analysis of the investment returns of hospitality stocks shows first that investors have attempted to time the market for hospitality stocks and second that the timing of investment inflows has been inverse to stock performance. Hospitality firms have issued equity capital near market highs, and they have retired capital near market lows. The effect for hospitality investors of this inverse market timing is a shortfall of 1.5 percent per year in returns, compared to "perfect" timing from 1962 through 2006. This analysis is based on a value-weighted portfolio of both restaurant and hotel firms. Unexpectedly, due to the timing issue, the return of the combined portfolio of restaurant and hotel stocks earned a lower average return over this time period than did a similar portfolio of either hotel or restaurant stocks only.
Keywords: stock returns, market timing
**********
Existing studies of the historical stock market performance of the hospitality sector do not draw a distinction between the performance of hospitality stocks and the returns to investors holding those stocks. The most common approach to analyzing the historical record is to compute average historical returns and then adjust the average returns for risk (see, e.g., Gu 1994; Kim, Matilla, and Gu 2002; Madanoglu, Erdem, and Gursoy 2008). In this approach, however, the returns of hospitality stocks and the returns to hospitality stock investors are implicitly assumed to be one and the same.
That the two returns can potentially be quite different is best understood through an example. Consider a firm that goes public in an initial offering at, say, $20 per share. The stock performs well for the first year, climbing to $30 a share, which leads to a secondary offering in which additional shares are issued. Unfortunately, the second year in the life of the stock proves difficult, and the price eventually falls back to $20 per share. Assuming no dividends in this example, the average buy-and-hold historical return (i.e., the geometric average return) for this stock is simply zero. This correctly reflects the experience of investors who bought at the initial public offering and held for two years: the stock was acquired at $20 per share and it is trading at $20 per share two years later, so investors buying the shares and holding them for two years just broke even. But it is clear that investors as a group have lost money here. Those who bought initially broke even, but those who bought during the secondary offering suffered losses. As this example illustrates, average historical security returns can significantly overstate the returns to investors holding securities if the timing of investor capital flows into securities is such that investors tend to increase their holdings of securities before periods of below-average performance.
My objective in this article is, first, to help portfolio managers and hospitality managers understand and quantify the difference between security returns and investor returns. It turns out that one can account for the difference precisely: a recent study by Dichev shows how to construct so-called dollar-weighted returns to properly capture the timing and magnitude of investor capital flows into, and out of, the stock market (Dichev 2007).
Because they take into account the timing and magnitude of investor capital flows over time, these dollar-weighted returns accurately reflect the experience of investors as a group: dollar-weighted returns place greater weight on periods during which investors have more capital invested. Intuitively, the idea is that security returns should be weighted by the amount of capital invested in order to determine investor returns. In contrast, historical geometric average returns are buy-and-hold returns: they assume equal weighting over time and therefore measure only the performance of buy-and-hold investors. This can produce a skewed view of average investor returns if capital flows in and out of securities are correlated with the performance of securities over time, as the above example shows.
Fortunately, dollar-weighted returns are quite simple to compute. They rely on the familiar concept of internal rate of return (IRR) that is widely used in capital budgeting. The method I advocate here is therefore practical. Dollar-weighted returns are computed as IRRs that treat the stock investment as a capital project. The initial investment and any subsequent inflows (e.g., secondary offerings) enter the calculation with a negative sign, and the final value and any outflows (e.g., stock repurchases) enter with a positive sign.
My second objective is to provide an empirical analysis of the extent to which the newly developed dollar-weighted returns differ from more standard buy-and-hold returns when applied to the hospitality industry. The main results are easily summarized. First, the results reveal that dollar-weighted returns are substantially lower than buy-and-hold returns in the hospitality industry. From 1962 through 2006, the return differential is 1.5 percent per year for all hospitality firms. When the sample is subdivided into hotel firms and restaurant firms, the differential is 2 percent for hotels and 1.7 percent for restaurants. Over the same time period, the return differential is about 0.6 percent per year for the aggregate stock market. This implies that historical returns of hospitality stock investors are lower than existing estimates suggest. Estimates based on average buy-and-hold returns do not reflect the experience of hospitality investors as a group. Second, the fact that dollar-weighted returns are significantly lower than buy-and-hold returns in hospitality stocks means that capital tends to flow into the hospitality sector at the wrong time: hospitality firms tend to raise equity capital after good stock market performance and before poor stock market performance. (1) The magnitude of the effect is such that this apparent timing cannot be attributed to (bad) luck. In fact, the magnitude of the effect is such that a value-weighted portfolio of both restaurant and hotel firms earns a lower average return than a similar portfolio of either hotel or restaurant stocks only, over 1962-2006. (2)
It should be emphasized that the results in this article do not speak to the performance and timing ability (or lack thereof) of any specific class of investors. If some investors who are active traders buy or sell at the wrong time, and therefore underperform, investors taking the other side of the transaction will necessarily fare well, and in the aggregate, these trades cancel out. The capital flows discussed in this article are aggregate flows (across all hospitality investors), and the term timing refers to the timing of capital flows by hospitality firms, such as stock repurchases and equity issues (an example of a study of the market timing ability of specific investor classes is Barber and Odean 2001).
Methodology
The rationale for computing dollar-weighted returns can be laid out intuitively in the above IPO example, where the stock price starts at $20, increases to $30 at the end of the first year, and then falls back to $20 at the end of the second year. For concreteness, let us assume that 10 million shares are offered initially, with another 5 million offered at the end of the first year. How is one to compute the historical return in this case?
A common approach (used, e.g., in mutual fund prospectuses) is to compute the simple average historical return, that is, the arithmetic return. In this case, the return over the first year is 50 percent and the return over the second year is--33.33 percent, so the arithmetic average is 8.33 percent. Clearly, this is not an appropriate measure of historical performance, as it overstates the return of every investor in this stock. Because the arithmetic average return is not an appropriate measure of historical performance, I do not use it further in this article.
The geometric average return is the compounded buy-and-hold return, in this case,
[R.sub.BH] = [[(1 + 50%) x (1 - 33.33%)].sup.0.5] - 1 = 0.
Notice that this correctly reflects the return of investors who bought at the IPO and held the shares for two years, unlike the arithmetic average, which overstates the performance of all investors. However, even the geometric average return does not reflect the experience of investors as a whole, as the experience of those who provided the additional $150 million at the end of the first year is quite different: they lost money. As a group, investors invested a total of $350 million over time only to obtain $300 million at the end of the second year. It is quite clear that the average investor return is negative in this example, even though the average security return is zero. Intuitively, the dollar amounts of investor capital inflows and outflows, and their timing, matter and therefore should enter the calculation explicitly.
Computing investors' overall returns is a simple matter if one views the stock as a capital project. Viewing the stock as an investment project, one can compute the intuitively correct dollar-weighted return as the IRR on the investment. In this case, investors initially put up $200 million. At the end of the first year, they invest another $150 million, and at the end of the second year, their total investment is worth $300 million, so the dollar-weighted return is
FEED