Optimal Timing of Clean Technology Adoption under Correlated Price Uncertainty – A Two-Factor Optimal Stopping Approach to Irreversible Investment[before doctoral defense]

Kivonat, rövid leírás

The transition to clean technologies, of which electric vehicle (EV) adoption is the leading example, confronts every prospective adopter with a largely irreversible investment decision under multi-factor uncertainty. The adopter must weigh a known upfront capital cost against a stream of uncertain future operating savings. Acting too early forfeits the option to wait for better conditions; acting too late forgoes years of savings. Standard net present value (NPV) analysis ignores the option value of waiting entirely. The real options literature since McDonald and Siegel (1986) and Dixit and Pindyck (1994) captures irreversibility, but the workhorse specifications either hold the investment cost fixed or let both the project value and the cost follow geometric Brownian motions. The empirical record shows that this simplification is untenable for clean technology adoption. Both sides of the adoption ledger are volatile. The operating benefit of an EV relative to an internal combustion vehicle is driven by energy price spreads that are volatile and non-stationary; the upfront cost premium exhibits secular decline through production learning (Nykvist and Nilsson, 2015; Ziegler and Trancik, 2021), punctuated by commodity-driven spikes (the 2022 lithium episode), supply-chain disruptions, and discrete policy shifts such as subsidy introductions and withdrawals. Most recently, the 2025 expiration of the US federal EV tax credit was followed by a pre-deadline surge and post-expiry collapse in US EV sales, consistent with the delay-and-spike cycle analysed in Chapter 5. Moreover, the two sources of uncertainty are correlated: energy market shocks propagate into both the running benefit and the cost of adoption. Three bodies of literature bear on this problem, but none addresses it in full. An empirical literature documents the stochastic behaviour of energy prices and technology costs. An applied fleet-transition literature models adoption with dynamic programming, stochastic optimisation, and real options (Kleindorfer et al., 2012; Falbo et al., 2021), but treats adoption costs as constant or deterministically declining, leaving the interaction between the two uncertainties unmodelled. A mathematical literature on optimal stopping and free-boundary problems (Peskir and Shiryaev, 2006; Oksendal, 2013; Brennan and Schwartz, 1985; Insley, 2002; Szimayer and Maller, 2007) provides the formal infrastructure, but has not been applied to the correlated clean-technology adoption problem. Within the clean technology adoption literature, no continuous-time real options model jointly captures stochastic operating benefits, mean-reverting fixed adoption costs, correlated two-factor dynamics, and the separation of the planning horizon from the operational lifetime. The dissertation closes this gap. To the best of my knowledge, it is the first continuous-time two-factor optimal stopping model of clean technology adoption in which the unit cost advantage follows an arithmetic Brownian motion, the fixed adoption cost follows a correlated mean-reverting Ornstein–Uhlenbeck process, and finite planning horizons are treated separately from finite operational lifetimes. The topic is timely: despite strong policy support and rapid growth in EV sales, the diffusion of clean technologies has remained slower and more uneven than stated decarbonisation targets imply, and diagnosing whether adopters are waiting rationally or facing structural barriers determines which policy instruments can accelerate adoption.