سرمایه گذاری غیر قابل برگشت، قرارداد های اختیار واقعی و رقابت: اسناد از توسعه املاک و مستغلات

We examine the extent to which uncertainty delays investment, and the effect of competition on this relationship, using a sample of 1214 condominium developments in Vancouver, Canada built from 1979–1998. We find that increases in both idiosyncratic and systematic risk lead developers to delay new real estate investments. Empirically, a one-standard deviation increase in the return volatility reduces the probability of investment by 13 percent, equivalent to a 9 percent decline in real prices. Increases in the number of potential competitors located near a project negate the negative relationship between idiosyncratic risk and development. These results support models in which competition erodes option values and provide clear evidence for the real options framework over alternatives such as simple risk aversion.

Over the last two decades, the application of financial option theory to investment in real assets has altered the way that researchers model investment.1 Under the real options approach, firms should apply a higher user cost to new investments in irreversible assets when returns are stochastic, reflecting the option to delay that is lost when investment occurs. The effects can be quite large. For example, Dixit and Pindyck (1994) use simulations to show that the optimal hurdle price triggering new irreversible investment can be two to three times as large as the trigger value when investments are reversible. Yet others argue that competition erodes option values and limits the empirical relevance of the real options framework for many industries. Empirical support for the real options model has suffered from the absence of a clean test to differentiate between real options and more traditional discounted cash flow (DCF) models of investment in which the discount rate depends on risk.In this paper, we address these issues by examining the relationship between uncertainty, competition, and irreversible investment using unique data on 1214 individual real estate projects (condominium or strata buildings) built in Vancouver, Canada between 1979 and 1998. In looking at real estate, we examine an asset class that represents a large component of national investment and wealth and a sector that exhibits great cyclical volatility in investment. Some theoretical papers have argued that real options models have limited power to predict investment in competitive markets. Caballero (1991) suggests that imperfect competition is vital to predicting a negative relationship between uncertainty and investment. For example, competition might mitigate the value of a real option through the threat of preemption as in Grenadier (2002). Trigeorgis (1996) associates increased competition with a higher dividend yield from the underlying asset. When the dividend is high enough, it can induce early exercise by reducing the value of the option to wait. In a similar vein, Kulatilaka and Perotti (1998) argue that firms with a strategic advantage (market power) are in a better position to gain greater growth opportunities when uncertainty is higher. This induces more investment in growth options for this type of firm while companies that do not have a strategic advantage will be discouraged from investing. In response to these critiques, others argue that with the addition of a few realistic assumptions, the value of the option to wait is preserved even with perfect competition. For example, Novy- Marx (2005) shows that competition does not diminish the value of an option to develop in the case of differentiated products such as real estate where locations are never perfect substitutes for each other and sites have varying opportunity costs of development due to differences in the pre-existing use of a site. Leahy (1993) and Dixit and Pindyck (1994) also contend that perfect competition does not necessarily reduce the value of waiting. Existing empirical research supports the existence of a negative relationship between volatility and investment (Downing and Wallace, 2001; Moel and Tufano, 2002; Cunningham, 2006 and 2007). Nonetheless, real options models are not the only models in which one would expect a negative correlation between uncertainty and investment, an issue that is often not discussed in empirical real options research. In fact, if increases in volatility are driven by a greater exposure to non-diversifiable risk, most neoclassical models (such as the familiar capital asset pricing model—CAPM) would predict that greater uncertainty would lead to lower investment through an increase in the investor’s required rate of return. In the case of incomplete markets, even increases in idiosyncratic risk will cause risk-averse investors to reduce investment if they cannot adequately hedge this type of risk. This latter condition is especially likely in the context of real estate, where many investors and developers are small and hold portfolios that are concentrated in a particular local market where they hold great expertise, but where there are no existing methods to hedge local market risk. Our findings described below address both of these issues. We find clear support for the negative relationship between idiosyncratic uncertainty and investment that is a crucial prediction of the real options model. To separate the impact of the alternative models, real options and the CAPM, we decompose the volatility of condominium returns into idiosyncratic and market risk components. As predicted by the real options model, exposure to idiosyncratic risk reduces investment. However, consistent with the CAPM, exposure to market volatility also delays investment to nearly the same extent. A one standard deviation increase in idiosyncratic volatility reduces the probability of development by 13 percent, about the same predicted impact on new investment as a 9 percent decrease in real prices. A similar one standard deviation increase in market volatility reduces the likelihood of investment by the equivalent to a 7 percent fall in real prices. Addressing the debate about how market structure impacts option exercise, we show that competition, measured by the number of potential competitors for a project, reduces the impact of condo return volatility on new investment. Empirically, competition has little direct effect on investment. Instead, competition only matters when interacted with volatility. We show that volatility has a smaller impact on option exercise for developments surrounded by a larger number of potential competitors. In fact, for the 5 percent of all units facing the greatest number of potential competitors, idiosyncratic volatility has virtually no effect on the timing of investment. These findings provide unambiguous support for the models of Caballero, Trigeorgis and Grenadier, which show that competition can erode the value of the option to delay irreversible investment. Finally, the finding that competition only impacts investment indirectly through its correlation with uncertainty provides support for the real options model even in the presence of risk averse owners and incomplete markets. While risk averse owners without hedging opportunities will reduce investment in response to greater idiosyncratic volatility, only a real options model has the additional prediction that option value diminishes with competition. The relationship between competition and real option exercise may help explain the strong pro-cyclical correlation between investment and output. Macro economists have often puzzled over the high volatility of investment relative to output, documented over long periods of time and across many countries (Basu and Taylor, 1999). Variation in competition over the cycle could provide at least one explanation for the excess volatility of investment. Rotemberg and Saloner (1986) and Rotemberg and Woodford (1991, 1992) argue that tacit collusion is difficult to sustain in booms, relative to busts. Our findings suggest that variation in competition can impact investment. Firms might optimally further delay investment in busts when product markets are less competitive, but undertake equivalent investments in booms when they face greater competition. This higher volatility for investment is consistent with the macro evidence. The remainder of the paper is structured as follows. Section 2 provides a review of related work and a discussion of how this paper fits in with the empirical real options literature. In Section 3, we present the empirical specification along with a summary of its theoretical support. We also discuss the impact of various assumptions on the specification with respect to the completeness of capital markets and the unique properties of the real estate market. We present a more detailed discussion of the data in Section 4. The empirical results are presented in Section 5, and in Section 6 we conclude.

The results in this paper support many of the conclusions from the burgeoning theoretical literature on the importance of real options and competition. Our empirical estimates suggest that builders delay development during times of greater idiosyncratic uncertainty in real estate prices and when the exposure to market risk is higher. These findings hold across different time periods. The impact of volatility in our sample is large and statistically significant in most specifications. A one standard deviation increase in condominium return volatility leads to a 13 percent decline in the hazard rate of investment, the same effect as a 9 percent decline in prices. Similarly, our estimates suggest that the hazard rate falls 8 percent when exposure to systematic risk increases by one standard deviation. We also show that competition significantly reduces the sensitivity of option exercise to volatility. Increases in competition appreciably decrease the coefficient on volatility in our hazard rate specification. In fact, volatility has no estimated effect on option exercise for the 5 percent of our sample with the largest number of potential competitors. This finding is fully consistent with Caballero (1991), Trigeorgis (1996) and Grenadier (2002) who argue that competition diminishes the value of waiting to invest. The erosion in value of the investment opportunity due to one’s competitors creates incentives to invest earlier. Hence competitive firms are not able to capture the full benefits to waiting that a monopolist has. This result supports the real options model because the interaction between competition and volatility should not affect the user cost of a reversible investment. This provides clearer evidence in favor of the real options model rather than the alternative that risk averse developers choose not to build at times of greater uncertainty. From a policy perspective, these results have important implications for understanding real estate cycles. An often-repeated claim in the real estate industry is that overbuilding in the real estate industry is due to irrational developers. Grenadier (1996) has suggested a rational basis for the bursts of construction that sometimes occur just as market prices begin to fall, strategic behavior by competing developers in imperfectly competitive markets. We find some evidence in favor of the Grenadier model; holding the level of prices constant, builders appear more likely to build when prices begin to fall. More compelling, however, is the observation that the volatility of returns, exposure to market risk, and competition play important roles in the timing of investment. Builders are especially susceptible to business cycle shocks, as developer bankruptcies rise considerably in recessions. If competition is less pronounced in recessions, real options behavior may lead developers to delay irreversible investments in structures longer than they would in booms when markets are more competitive. Given that changes in investment are an important component in the business cycle, these results suggest that uncertainty and competition may play a role in understanding cyclical movements in investment in real estate and the macro economy.