INTENSITY AND TIMING OPTIONS FOR INVESTMENTS IN A LESS THAN PERFECTLY COMPETITIVE MARKET CHU, Yongqiang and SING, Tien Foo Department of Real Estate National University of Singapore Date: April 14, 2004 Abstract The traditional real options models invariably assume that firms are price taker on one hand, and they behave, on the other hand, like monopolists who face no competition and competitive entry of new players in the future. The assumptions are relaxed in this paper by firstly modeling the real estate development market in a monopolistic framework, and substituting the exogenous rental or price variable with a more general economic shock variable. The market rent in our model is endogenously defined in the demand and the cost functions of firms with a simple linear monotonic production technology. The capacity choice in our model is not a continuous and incremental process. Instead, the developer will have to make the timing and the intensity decisions simultaneously when the development option is exercised. In additional to demand uncertainty, the numerical results show that the rental sensitivity to housing demand will also have an inverse effect on the deferment option values of a real estate project. Both the timing and intensity decisions are sensitive to the demand factors, which is represented by the rental sensitivity variable. Key Words: Real Options, Optimal timing, Optimal Intensity, Real Estate Development *Corresponding Author. Email rststf@nus.edu.sg. Address: Department of Real Estate, National University of Singapore, 4 Architecture Drive, Singapore 117566. We wish to thank the comments by Prof. Abdullah Yavas, Yuming Fu, and other participants at the research workshop at NUS on 12 April 2004. INTENSITY AND TIMING OPTIONS FOR INVESTMENTS IN A LESS THAN PERFECTLY COMPETITIVE MARKET Introduction The timing and intensity of investment are two important decisions faced by investors in the process of selecting their production capacity. Real estate development is a typical capital investment process, where a developer possessing a developable land has to decide on when to start developing the land and how much to build on the land, given the fact that zoning dictates the particular use type permitted on the land. The timing question has been extensively studied in the real option and optimal stopping time literature (McDonald and Siegel, 1986; Titman, 1985; Clarke and Reed, 1988; Williams, 1991, et. al.). Pindyck (1988) first examines the capacity choice issue in an irreversible investment framework. When investment is not perfectly reversible, he found that firms hold less capacity than they would when the demand is uncertain. In his model, the “lumpiness” of capital investment is ignored. The firms are, therefore, assumed to have the flexibility of increasing their capacity as and when the needs arise. This flexible capacity choice assumption may not be realistic in the real estate development process. Firstly, there are statutory planning and technical constraints on the maximum permissible density of development on a land parcel. Secondly, when the foundation and floor plate of a building are set, it will be costly to reverse the intensity decision on the land. The costs incurred are sunk. The investment is though not strictly irreversible, any expansion of the land use capacity in the future would, however, come at a high cost. Therefore, the intensity and timing decisions of a development are important decisions, which are often made simultaneously in a real estate development process (Williams, 1991; Capozza and Li, 1994). In other words, the development intensity must be decided at the time the development option is exercised. In most of the earlier real option models, prices or rental cash flows are used to proxy investment payoffs. The price generating processes are assumed to follow specific 1 exogenous stochastic processes (Williams, 1991; Clarke and Reed, 1988; Capozza and Helsley, 1989 and 1990; Capozza and Sick, 1994). Under this assumption, firms are price takers who could not influence the price process in the future. The assumption is only valid in a perfectly competitive market, where there exists multiple buyers and sellers that will set the equilibrium price. However, in most of the earlier real option models, the firm or the option holder is always assumed to be a monopolistic one, who does not face any market competition. Given the monopolistic right, the firm’s investment decision will always have significant influence on the price behavior of the product. The firm is known to be a price-setter. The exogenous price function is, therefore, rather restrictive in the real option model setting with only a single monopolistic firm. The solution of the real option models offers at best a partial equilibrium option estimate for a perfect competitive market. The real estate market with the characteristics of product segmentation and high barrier of entry for new players is truly not a competitive market. Few big players, instead, dominate the market and establish significant monopolistic influence on the supply and pricing of the products. The price setting behavior of the developers is not well represented by the exogenous price assumption in most of the real option models on development timing. To reflect the monopolistic power of the developer in a sub-sector of real estate market, we model the development timing options using a more general demand shift variable in place of the stochastic price function (Pindyck, 1988). In this model setting, the developer is a monopolist, and yet his/her investment decision is not restricted by the exogenous price function, but is driven by the demand shift variable. The optimal development intensity of land parcels is another issue examined by Williams (1991) and Capozza and Li (1994). In their models, they assume that the developer can optimally pick the development density of a land at the time of exercising the investment option. The optimal density is independent of the cost and price functions, and it is fixed once determined. We relax the assumptions in Williams (1991) and Capozza and Li (1991) models by modeling the density as an endogenous 2 variable in the price and cost functions.1 By imposing the microeconomic structure to the price and cost functions with monotonic elasticity of scale, the capital choice and development timing decisions are no longer separable. The optimal density decision affects the boundary conditions, which in term affect the estimation of the development timing option premium in the proposed model. Therefore, the optimal density and the optimal timing of a development will have to be simultaneously and jointly solved in the proposed real options models. The paper is organized into six sections. Section II reviews the relevant real options literature in optimal timing and intensity of investment. Section III explicitly specifies the timing option and capacity choice problems with necessary assumptions. The theoretical model and its respective analytical solutions are derived in section IV. Based on a set of parameter inputs, the numerical analyses for the proposed optimal timing and intensity of investment models are conducted, and the comparative statics are summarized in section V. Section VI concludes the findings. Literature Review The paper of McDonald and Siegel (1986) was one of the earliest works that studies the optimal timing of investment in an irreversible project. In their model, they assume that the cost and value of the project follow exogenously defined continuous time stochastic processes. They analyze the basic problem of timing of investment under the uncertainty, and showed that the timing options do have value under uncertainty. The investment rule derived from the traditional NPV method is inadequate when values and costs of the project are not deterministic. The firm will invest only when the benefits of undertaking the project exceed the costs by a positive amount that is large enough to compensate the investors for giving up their timing options. 1 Unlike in Williams’s (1991) model, there is no statutory zoning restriction that imposes a maximum permissible density of the land in our model and the model by Capozza and Li (1994). 3 The typical optimal timing model of McDonald and Siegel (1986) has subsequently been extended with different model assumptions and applied to evaluate investors’ behavior in different investments. Titman (1985) employs a simple binomial tree method to explain why deferment of development of parking lots that are located in an exclusive residential neighborhood is a feasible option when the market price is uncertain. Williams (1991) then recast the urban property development problem by modeling both the project cost and cash flows of the property as stochastic processes. They show that developers can optimally decide on the density and timing of development of the lands to maximize the market value of the land in a perfectly competitive market framework. Clarke and Reed (1988) and Sing (2000) also apply the real options framework to analyze the optimal timing problem in vacant developable lands with different assumptions on the stochastic variables. They again confirm that price uncertainty (Clarke and Reed, 1998), or rental uncertainty (Sing 2000) of the developed property defer the land development decision, because the value of the vacant land and its embedded timing options increase proportionately higher than the developed property value. In the urban land pricing model, Capozza and Helsley (1989) develop a simple dynamic land price model that is composed of four additive components: the value of agricultural land rent, the cost of conversion, the value of accessibility and the growth premium. Capozza and Helsley (1990) then extend the urban land pricing model to a stochastic city in which the household income and land rent follow stochastic processes. Using the first hitting time approach to determine the optimal conversion of agricultural land to urban use, the model shows that the rental uncertainty delays the conversion of agricultural land to urban use, reduces the equilibrium city size and also imparts an conversion option premium to agricultural lands. The growth premium also increases the price of agricultural land at the boundary of urban land by a premium that is higher than its opportunity cost. Capozza and Sick (1994) further improve the land pricing model by adding the spatial and temporal risk structure to the model. By assuming that the cash flows on land to follow additive diffusion process, they found that the price 4 agricultural land awaiting for conversion increases with the growth and unsystematic risk of the urban rents, but decreases with risk aversion. They found that uncertainty premium related to urban rents will have significant positive effects on both urban and agricultural lands. In the earlier land pricing model, capacity choice and intensity of investment is fixed and constant. Pindyck (1988), on the assumption that firms can continuously and incrementally expand the capital, shows that firms would hold less capacity in a market with volatile and unpredictable demand. The opportunity cost of exercising the option to increase the marginal unit of capital increases with uncertainty in demand. Williams (1991) extend the capacity choice option to a model involving optimal development density decision by including an exogenous density variable to the real estate development option model. In his model, the optimal timing and optimal density of development are driven by the uncertainties in both revenue of developed property and the cost of development. Capozza and Li (1994) extend the density choice to modeling the decision of converting vacant land to urban use. In an optimal stopping time framework where the decisions to choose the timing and density of conversion of vacant land are made simultaneously by developer, they found that stochastic density of development increases the hurdle rate and delays development. Compared to Williams (1991) and Capozza and Li (1994) models, where the stochastic rents are exogenously defined, we follow Pindyck’s (1988) specification of a stochastic market demand variable, which is a better reflective of capacity and timing of investment decisions in a not so perfectly competitive real estate market. As a typical real estate market is more akin to an oligopolistic or monopolistic market by nature, the developer is unlikely to be a price taker as represented by an exogenous rent or price function. Therefore, we assume a stochastic economic shock as the state variable that drives the density and timing options in our model. In Pindyck’s (1988) model, capacity expansion is a continuous and incremental process. Whereas, in our case, real estate project is lumpy, the assumption of marginal increases in density from time to time is 5 not practical. We restrict our development intensity to a discrete process, which has to be made simultaneously with the development timing decision. There are other real option literature that examines issue relating to time to build options (Majd and Pindyck 1986; Sing 2001), interest rate options (Ingersoll and Ross, 1992), entry and exit decision of firms (Dixit, 1989; Trigeorgis, 1991) and real estate investment in oligopoly market (Grenadier 1996, Wang and Zhou 2002). Problem Specification Real estate market is generally characterized by its heterogeneous and immobile nature of its products. Developers with sufficiently large capitals and strong track records could exert monopoly power in a particular segment of markets. In our model with few homogeneous developers, the market is not perfectly competitive. The monopolistic influence of developers on the supply would indirectly determine the market prices and rental cash flows of developed property. Therefore, the exogenously defined stochastic price and rental functions are not feasible in our framework. The underlying driver of the investment decision is represented in our model by a stochastic economic shock variable that follows a multiplicative diffusion process. To set up the framework of our model, we assume that the developer has already acquired a vacant land. The decision faced by him is “when” and “how much” to develop on this parcel of land. When dealing with the first problem of deciding on when is the optimal time to develop, the decision is not a straight forward binary choice of “now or never” as predicted by the traditional net present value (NPV) rule. When the market is volatile, the simple rule of “NPV greater than or equal to zero” is no longer applicable. Instead, many real options models (McDonald and Siegel, 1986; Clarke and Reed, 1998; Sing 2000; et al) have suggested that the developer should only initiate the development when the profit exceeds the cost by an amount that is commensurate with giving the options of waiting for another period. 6 The question of intensity of development if ignored in the real options models will lead to under-estimation of the urban land prices (Williams, 1991; Capozza and Li, 1994). The flexibility given to the developer in adjusting the density of development will interact in a positive and significant way to the conversion option premiums especially when the market value for the urban land use is volatile. In a monopolistic market framework, the intensity of development will have a direct influence on the price and rental cash flows of the developed real estate. Therefore, unlike in Williams (1991) and Capozza and Li (1994) models, the density variable in our model is defined as endogenous functions of demand and supply of the undeveloped lands. The functional forms of the demand and supply equations, which are dependent on the density variable, will have significant effects on the timing options. As real estate investment is lumpy and indivisible, the sequential expansion and marginal increase in capacity investment processes like those advocated in the capacity choice (Pindyck, 1988) and the time to build literature (Majd and Pindyck, 1986; Sing, 2001) may not be feasible. We, instead, follow the assumptions in Williams (1991) and Capozza and Li (1994) models, which require that the intensity decision will have to be made concurrently and simultaneously when the timing of development is decided. The Model Assuming that the developer faces the following inverse demand function: R = YD(Q ) (1) where R is the project cash flows, D(Q) is a concave market demand that increases at a decreasing rate with respect to the density of new developed property built on the vacant land, (Q), that is [ D ' (Q ) < 0 ], and Y is the exogenous economic shock that is assumed to follow a geometric Brownian motion given below: dY = µdt + σdw Y (2) In equation (2), µis the expected drift rate of Y, σ is the volatility of the incremental change in Y, and dw is the incremental change of the standard Wiener process. The risk 7 variable, σ, in equation (2) is a constant firm specific idiosyncratic risk, which is not related to the systematic market risk as in Capozza and Sick (1994). The developer’s project cost is deterministic, i.e. it is independent of any stochastic driver. It can be defined as a convex function of development density given below: C = C (Q ) (3) and is subject to a positive first order derivative constraint of [ C ' (Q ) > 0 ]. Like the earlier models by Williams (1991), the ownership of a vacant land give the landowner a call option, when exercised, gives him a right to claim on the cash flows generated from the developed property built thereon. Therefore, the vacant land value can be modeled as a development option, [V], with an infinite maturity life, and it can be defined as a function of the stochastic shocks variable, V = V (Y ) (4) Based on Ito’s Lemma, the incremental change in the option value over a short time interval, [dt], can be expanded as follows: 1 ∂ 2V ∂V ∂V )dt + σ dw dV = ( σ 2Y 2 + µY 2 2 ∂Y ∂Y ∂Y (5) Since the firm specific idiosyncratic risk can be fully diversified away, we could simplify equation (5) to the following ordinary differential equation (ODE) form: 1 2 2 ∂ 2V ∂V σ Y + µY − rY = 0 2 2 ∂Y ∂Y (6) We assume that there exits a trigger value of [Y*], at which the development will commence as long as the market shock exceeds Y*. The optimal time at which the development occurs can be represented as [ T = inf[T ≥ 0, Y (T ) = Y * ], and at time T, the value of the development option is simply the net discounted future cash flows of 8 the project, which is given as follows: V (Y * , q) = D(q )q − ( r − µ )τ * e Y − C (q) r−µ (7) where τ is the time to build variable, which is assumed to be constant over time. At time T when the development option is exercised, the developer will also have to simultaneously decide on the optimal development density, q, which will maximize the value of the vacant land as defined in equation (7). The optimal density could be determined by taking the first order derivative of equation (7), and equating it to zero, i.e. [ ∂V = 0 ]. ∂q To solve the optimal density of development, we need to explicitly define the functional structure of demand and cost equations by generalizing the two the equations with the following specifications: D (Q ) = a − bQ (8) C (Q ) = c + dQ (9) Based on equations (6) to (9), the optimal density of development on the subject vacant land, where [ q ≥ 0 ], can be defined as follows: q* = a d − 2b 2 KbY * where K = (10) e − ( r − µ )τ r−µ The optimal value of the development option at time T in equation (9) can be redefined as: ⎧ ⎫ ad a 2 KY * d2 * = + −c− V ( Y ) ; for (q * ≥ 0)⎪ ⎪⎪ * ⎪ 4b 2b 4bKY ⎨ ⎬ ⎪ ⎪ ⎪⎩V (Y * ) = 0; otherwise ⎪⎭ (11) 9 The two parts of the value functions for different Y, and the optimal payoff value for the land with embedded development option, V(Y*), corresponding to equation (11), are shown in Figure 1. After determining the optimal density of development, equation (11) will be used as the traditional value matching conditions in solving the optimal timing of development. The smooth pasting condition that will ensure the development will occur at time T is given as follows: V ' (Y * ) = a2K d2 − 4b 4bKY *2 (12) A natural value-absorbing boundary condition when [Y=0]. At this point, the developer will never exercise the option to start development and the option value is worthless, which is defined as follows: V ( 0) = 0 (13) The ODE (6) is a homogenous second order differential equation, which has the following general solution form: V (Y ) = B1Y β1 + B2Y β 2 (14) where β 1 and β 2 can be solved from the following quadratic equation: 1 2 σ β ( β − 1) + µβ − r = 0 2 (15) Thus, we obtain: β1 = β2 = − ( µ − 1 / 2σ 2 ) + ( µ − 1 / 2σ 2 ) 2 + 2rσ 2 σ2 − ( µ − 1 / 2σ 2 ) − ( µ − 1 / 2σ 2 ) 2 + 2rσ 2 σ2 >0 (16a) <0 (16b) Based on the value absorbing constrain, [ B2 = 0 ], and B1 of equation (14) could be solved subject to the upper constraint in equation (11), where [q* ≥ 0]. The following system of equations: 10 β B1Y * 1 = B1 β1Y * a 2 KY * d2 ad + −c− * 4b 4bKY 2b β1 −1 = (17a) a2K d2 − 4b 4bKY *2 (17b) The analytical solution for B1 and Y* can be given below: a 2 KY * B1 = 4b (1− β1 ) + ad d2 − (c + )Y *( − β1 ) *(1+ β1 ) 2b 4bKY (18) There are two groups of solutions for Y*, which correspond to value functions 1 and 2 in Figure 1 respectively: (ad + 2bc ) β 1 + (ad + 2bc ) 2 β 1 − a 2 d 2 ( β 1 − 1) 2 Y = * (19) a 2 K ( β 1 − 1) (ad + 2bc ) β 1 − (ad + 2bc ) 2 β 1 − a 2 d 2 ( β 1 − 1) 2 Y* = 2 2 a 2 K ( β 1 − 1) (20) Subject to the minimum development density constraint, [ q * ≥ 0 ], the solution in equation (20) will be truncated in the option payoff function. Therefore, the option premium of the timing and intensity of development options for the vacant land can be computed based on equations (18) and (19), ⎧V (Y ) = B1Y β1 ⎪ ⎨ a 2 KY d2 ad + −c− ⎪V (Y ) = 4b 4bKY 2b ⎩ for Y < Y * ⎫ ⎪ ⎬ * for Y ≥ Y ⎪ ⎭ (21) Numerical Analyses Based on a base case scenario for a typical real estate development project that contains the following input parameters as summarized in Table 1: 11 Table 1: Input Assumptions for Numerical Analyses Input Parameter Instantaneous drift of economic shock Volatility of economic shock Fixed cost Variable cost (psm) Demand parameter 1 Demand parameter 2 (psm) Risk free interest rate Time-to-build (year) Base Value 0.06 σ = 0.2 c = $5x106 d = $200 0 a = $5000 b = $0.5 r = 10% τ =3 We numerically analyze the effects of changes in volatility of economic shocks and demand parameter on the trigger value, the optimal intensity and also the value of the development options. A. The Volatility Effect Figure 2 shows the standard real options prediction (Dixit and Pindyck, 1994), which suggests a positive relationship between the volatility and the trigger value of the economic shocks, which can be represented by ∂Y * > 0 . When the future market ∂σ environment is uncertain, the developer will be less incline to commence the development earlier. The higher trigger value for the development options is translated into higher payoffs that are high enough to not just offset the development costs, but to compensate the developer for giving up the waiting options. The results imply that development activities will be relatively lower when economic uncertainty in a market is high, and developer will likely to wait for clearer signals in the market before taking any action to start development. The volatility in the market increases the trigger value, which in term motivates the ∂q * developer to increase the level of intensity of the development, i.e. > 0 (Figure 3). ∂Y * The result is consistent with that in Capozza and Li (1994), which predicts that 12 developer will delays the development when the economic is highly volatile, and when the economic shock triggers the development option subsequently, the development will likely to be more intensive and higher in its density. The intensity premium further enhances the premium required for the developer to exercise the development options earlier. The higher intensity premium adds to the development option premiums in a highly volatile environment. Figure 4 shows the development option value under different volatility of economic shocks. By drawing a radius line from the origin, the payoffs that trigger the development option can be determined as the tangency points that equate to the optimal development timing option premium. Again, there is no surprise to the results that show a positive volatility and value of development options. B. Demand Effects The demand effects, which are represented by the rental sensitivity to market demand, [b],2 on the developer’s decision on the optimal timing and density of development are numerically evaluated. In equation (8), the demand is an inverse function of density of development. Thus, [b] measures the inverse multiplier effects of quantity of demand in the market on the prices. Since the firms in the monopolistic framework in our model are not a price taker, the demand is likely to respond to the changes in the rent and quantity variables. The numerical analyses of the effects of [b] on the standard real options indicators will likely to shed new light to the real options literature that invariably assume a perfectly competitive market with exogenous price functions. The positive relationship between the trigger value and b in Figure 5 indicates that when the rental sensitivity to market demand is low, a higher trigger value is required 2 [b] in equation (8) can be considered as an inverse measure of the rental elasticity of demand, which indicates a lower elasticity effects when the b is high for any given level of rent. This rental sensitivity variable, which we prefer to term b in this paper, is a good proxy for our analysis of the effects of monopolistic market power on the development options. 13 for the developer to exercise the development option. In other words, in a market when the rental sensitivity is low, developers are less likely to start new projects and the development activities in the market are likely to be depressed. The deferment options become more valuable when the market rent is less sensitive to changing demand. There is also a positive effect of [b] on the optimal intensity of development as shown in Figure 6. When [b] increases, i.e. the rent sensitivity of market demand is low, developers will likely to develop a project that is lower in density than it would otherwise be in a normal market condition. This result is consistent in a monopolist market, which predicts that low price elasticity will result in lower outputs. The convex lines in Figure 7 show a negative relationship between [b] and the development option values. The result shows that low rental sensitivity contributes in a positive way to increasing the development option value. In a monopolist market, developers can influence the demand and price of the product, and the sensitivity of the product rent or price to demand will also add to the value of the development options. When demand uncertainty is high and rental sensitivity is low, the waiting options become highly valuable. The development activities will be significantly curtailed and the city growth will also likely be slow. C. Comparative Statics The comparative statics in Table 2 summarize the relationships between various input parameters with the trigger value, optimal intensity and also the development option value. 14 Table 2: Comparative Statics Trigger Value (Y*) Optimal Intensity (q*) Instantaneous Drift ( µ ) - - Value of the Development Option + Fixed Cost (c) Variable Cost (psm) (d) Demand Parameter 1(a) Risk Free Interest Rate (r) Time-to-Build (year) ( τ ) + + + + + + + + - Input Parameter Conclusion The optimal timing problem has been extensively studied in the real options literature. However, most of the models make two restrictive assumptions with respect to the price function and the market structure. On one hand, the models assume that the investors are monopolists who face no competition from future entry, and they are not price taker for their products. On the other hand, the models invariable specify the price function as an exogenous stochastic process, and the price changes are independent of the demand shocks in the market. The proposed real development option model in this paper reconciles the two assumptions by providing a more general framework to analyze the optimal timing and optimal intensity decisions faced by real estate developers. While maintaining the monopolistic structure of the real option model, which is consistent with the characteristic of many real estate sub-markets that are dominated by few large and financially strong developers, we relax the rigid assumption of an exogenous defined rental function. Instead, we allow the market rent to be endogenously and jointly defined in an inverse demand function and a cost function. Like Williams (1991) and Capozza and Li (1994), we also further extend the model by allowing the intensity of development to be variable. Unlike in Pindyck’s (1988) marginal approach, where capital investment is a continuous process embedded with a series of compound options, the optimal timing and optimal intensity are joint decisions in our model. They will 15 have to be solved simultaneously when the development option is exercised. Our numerical results of the volatility effects are not inconsistent with those observed in the standard real option literature, which predicts a positive relationship between volatility of demand shocks and the trigger value and the development option value. However, the effects of market demand factor, which is represented by the rental sensitivity variable, [b], shed new insight to the findings in optimal timing and intensity options in the literature. The results show that when rental sensitivity to the market demand is low, we would expect the development activities to slow down and also the scale of development, if undertaken, will likely be smaller. The deferment option value is not only dependent on the demand shock volatility; it is also negatively related to the rental sensitivity to the market demand. For a market that is insensitive to the changes in demand, we would expect the development activities to be depressed and the city growth to be slower. There are several extensions that can be made to the existing model framework proposed in the paper. Firstly, the cost function, which is assumed to be deterministic, can be assumed to follow a realistic stochastic process. Secondly, the monopolist scenario can be further extended to cover oligopolist market using a duopoly model like those proposed by Grenadier (1996 and 2001). In his duopoly real options papers, the output is again exogenously determined. This assumption could be relaxed by allowing the output to be endogenously determined in both the real option and Stackelberg competition frameworks. References: Capozza, D.R. and R. Helsley, “The Fundamental of Land Prices and Urban Growth,” Journal of Urban Economics, 1989, 26, 295-306. Capozza, D.R. and R. Helsley, “The Stochastic City,” Journal of Urban Economics, 1990, 28, 187-203. Capozza, D.R. and Y. Li, “The Intensity and Timing of Investment: The Case of Land,” The American Economic Review, 1994, 84(4), 889-903. 16 Capozza, D.R. and G. Sick, “The Risk Structure of Land Markets” Journal of Urban Economics, 1994, 35, 297-319. Clarke, H.R. and W. Reed, “A Stochastic Analysis of Land Development Timing and Property Valuation,” Regional Science and Urban Economics, 1988, 18, 357-381. Dixit, A., “Entry and Exit Decision under Uncertainty, Journal of Political Economy, 1989, 97(3), 620-638. Dixit, A. and R. Pindyck 1994, Investment under Uncertainty, Princeton University Press, Princeton, NJ Grenadier, S.R., “The Persistent of Real Estate Cycles”, Journal of Real Estate Finance and Economic, 1995, 10, 95-119. Grenadier, S.R. “The Strategic Exercise of Options: Development Cascades and Overbuilding in Real Estate” Journal of Finance 1996, 51, 1653-1679. Grenadier, S.R. 2000, Game Choices: The Intersection of Real Options and Game Theory, Risk Books, London, UK. Ingersoll, J.E. and S. Ross, “Waiting to Invest: Investment and Uncertainty,” The Journal of Business 1992, 65(1), 1-29. McDonald, R. and D. Siegel, “The Value of Waiting to Invest”. The Quarterly Journal of Economics 1986,101 (4), 707-728. Pindyck, R., “Irreversible Investment, Capacity Choice, and the Value of the Firm” The American Economic Review, 1988, 78(5), 969-985. Pindyck, R., “Irreversibility, Uncertainty, and Investment” Journal of Economic Literature 1991, 29(3), 1110-1148. Sing, T. F., “Irreversibility and Uncertainty in Property Investment” Journal of Financial Management of Property and Construction 2002,7(1), 17-29 Sing, T. F., “Time to Build Options in Constructing Processes”, Construction Management and Economics, 2002, 20(2), 119-130. Sing, T. F., “Optimal Timing of a Real Estate Development under Uncertainty”, Journal of Property Investment and Finance, 2000, 19(1), 35-52. Titman, S., “Urban Land Prices under Uncertainty”, The American Economic Review, 1985, 75, 505-514. Trigeorgis, L., “Anticipated Competitive Entry and Early Preemptive Investment in Deferrable Projects”, Journal of Economics and Business 1991, 43, 143-156. Williams, J.T., “Real Estate Development as an Option”, Journal of Real Estate Finance and Economics, 1991, 4, 191-208. 17 Figure 1: Optimal Payoff Function for Development Options, V(Y*) Figure 2: Volatility Effect on Option Trigger Value Trigger Value (Y* 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00 0 0.1 0.2 0.3 0.4 0.5 0.6 Volatility of Economic Shocks (σ) 18 Optimal Development Intesit (q*) Figure 3: Volatility Effect on Optimal Intensity 4800 4700 4600 4500 4400 4300 4200 4100 4000 0 0.1 0.2 0.3 0.4 0.5 0.6 Volatility of Economic Shocks (σ) 250 σ=0.05 σ=0.1 σ=0.2 σ=0.3 σ=0.4 200 5 ($*10 ) Development Option Value, V(Y* Figure 4: Value of the Development Option 150 100 50 0 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 Economic Shocks (Y) 19 Figure 5: Effects of Rental Sensitivity of Demand on Optimal Timing * Trigger Value, (Y) 0.20 0.16 0.12 0.08 0.04 0.00 0 0.2 0.4 0.6 0.8 1 1.2 Demand Factor, (b) Figure 6: Optimal Intensity of Different Price Sensitivity of Demand Optimal Intensity of Development (q*) 25000 20000 15000 10000 5000 0 0 0.2 0.4 0.6 0.8 1 1.2 Demand Factor, (b) 20 1200 b=0.2 b=0.4 b=0.6 b=0.8 1000 800 600 * 5 V(Y ) ($*10 ) Development Option Value, Figure 7: Development Option Value 400 200 0 0 0.04 0.08 0.12 0.16 0.2 Economic Shocks (Y) 21