Scrapping a Wind Turbine: Policy Changes, Scrapping Incentives and Why Wind Turbines in Good Locations Get Scrapped First.

• Author: Mauritzen, Johannes

1. INTRODUCTION

   The cost of electricity that comes from wind turbines or other renewable energy sources is to a large degree based on capital costs and associated financing costs. The expected lifetime of the turbine is then an important factor in the investment decision. In turn, the decision to scrap a wind turbine becomes particularly important in analyzing the economics of wind power. Yet few empirical studies of wind power scrapping exist. In this article I use non-parametric and semi-parametric survival models and a data set consisting of all wind turbines built in Denmark to analyze the scrapping decision.

   Wind turbines (1) along with other forms of renewable energy production have several properties that make the scrapping decision for these goods different from most other types of productive goods:

   * Low marginal operating costs

   * Importance of geographic placement and infrastructure

   * High rate of technological change

   * High level of government involvement in output price-setting and subsidies

   Though turbines do incur considerable costs for maintenance and repair, there are of course no fuel costs. Once the turbine is built, the real operating margin--defined loosely as the flow of revenue received less the real operating costs like repair and land rent--is likely to be positive. Compared to coal or gas plants, a negative real operating margin is unlikely to be the direct cause of scrapping.

   In a study of Danish wind turbines by Jensen et al. (2002) it was found that of those turbines that were scrapped the largest single reason given for scrapping (40 percent) was to make room for newer turbines--often called repowering in the industry. Only 12 percent were reported to be scrapped due to mechanical defect or due to wear. Jensen et al. suggest that repowering is also the grounds for the scrapping of most of the remaining 47 percent where the reason was not reported.

   The study by Jensen et al. then strongly suggests that an important reason for scrapping a wind turbine is the opportunity cost that results from a combination of scarce land resources and a high rate of technological change. An older turbine operating on a wind-rich location means that one cannot put in its place a newer, larger and more productive turbine.

   Scarce land resources are an especially important consideration for wind turbines since the total energy yield of wind turbines is highly dependent on average wind speeds. A simplified energy conversion formula (2) for wind power is E =[[1]/[2]][PHI][Atv.sup.3] where A is the sweeping area of the blades, [PHI] is a constant and v is the average wind velocity (MacKay 2008). Thus energy output from a wind turbine increases approximately cubically with average wind speed.

   An important result from this article is to show that changes in subsidy policy can have a strong effect on scrapping through an interaction with this opportunity cost. The scarcity of land resources is not primarily about the amount of geographic land available--though this certainly plays a role. Building out the appropriate grid infrastructure is expensive and the planning process of zoning an area for wind turbines can be both contentious and lengthy. Policy can create strong incentives to invest quickly, leading to artificially high land scarcity as there simply is not enough time to go through the planning process of zoning and grid infrastructure.

   The role of land scarcity and opportunity costs leads to some testable implications about the pattern of wind power scrapping. In particular turbines located in better, windier locations will tend to be at a higher risk of being scrapped and on average have a lower lifetime. The simplified idea is illustrated in Figure 1. (3)

   Consider first the top panel in the figure. The vertical height represents the instantaneous cash flow from a turbine while the horizontal distance represents time. The dotted line represents the instantaneous cash flow that could be obtained by investing in a newer turbine. Since in practice technological change and manufacturing improvements have meant the ability to produce larger turbines with higher rated capacity, I draw the line sloping upwards.

   The turbine owner would choose to scrap the turbine at a point at which turbine technology has advanced so that the total expected revenue from the new turbine less the cost of investing in a new turbine is greater than the revenue lost from scrapping the old turbine. In the figure this point is shown by the insertion of a bold vertical line.

   Now consider the effect of lower average wind speed, as illustrated in the lower panel. Assume that lower average wind speed affects the cash flow of old and new turbines proportionally. When replacing a turbine, the foregone revenue from the scrapped turbine is higher for the turbine located in the windier area. However the benefits of installing the bigger turbine are in absolute terms even greater. Assuming that the cost of investing in the new turbine is fixed then the investment in the new turbine, and corresponding scrapping of the old turbine, will take place later in the poor location.

   The figure, of course, represents an extreme simplification of the actual replacement decision. Discounting and the effects of uncertainty are not considered. The role of technological change is in itself complex. I use the term 'technological change' as an umbrella term for several factors like improved engineering knowledge that has allowed for larger turbines over time and the advantages of scaled manufacturing that has also developed over time. But when considering the interaction of technological change, scarce land resources and variation in wind resources, the figure shows the essential elements of the replacement decision.

   To test the predictions I use a Cox regression model on data of Danish wind turbines. Since I do not have data on the actual wind conditions of each location I create a proxy instead--the average annual full-load hours of a turbine. I take the average yearly electricity produced from each turbine and divide it by the rated power capacity of the turbine. This variable can be interpreted as the number of hours per year a turbine producing at full rated capacity would need to operate to equal the actual energy produced by that turbine. This variable, which can be seen as an indicator for capacity utilization, likely reflects the wind resources of a turbine's placement. The results show that turbines with higher average annual full-load hours have a higher hazard (4) of being scrapped.

   I will also show that subsidies and changes in subsidies for wind power play a strong role in the timing of the scrapping decision. The effect of wind resources and associated opportunity costs interact strongly with the implementation and changing of such subsidies. In particular, the announcement of a forthcoming reduction of subsidies creates an incentive to invest quickly. As discussed, planning new wind turbine sites takes time, thus a rush to invest in effect creates its own form of land scarcity. This in turn has a substantial effect on the scrapping decision.

   Subsidies were also introduced to directly encourage the scrapping of older, poorly placed turbines. However, I show that these policies actually have a greater effect on turbines in wind-rich locations. A rigorous analysis of the optimality and efficiency of these subsidies is outside the scope of this paper, but I do want to emphasize that even though the observed effects of the scrapping policies seem to go against the stated goals of the policy, this does not necessarily mean that the end result was suboptimal. In fact, it is likely that the scrapping of older turbines in good locations first is economically optimal.

   An extensive literature exists on the optimal scrapping of a productive good and the economics literature on renewable energy is growing. The literature on optimal abandonment is vast and goes all the way back to Hotelling (1925). It has long been acknowledged that a capital good or project can be abandoned well before it becomes unprofitable. The role of technological change in early investment and abandonment is taken up by Gaumitz & Emery (1980).

   More recently, a large and growing literature exists on the effects of uncertainty in the face of irreversible investment or abandonment--so called real options. Chapter 7 of Dixit & Pindyck (1994) focuses on output and input price uncertainty on the decision to scrap. A related analysis on firm entry and exit with irreversible investments can be found in Dixit (1989). Subsequent work has recognized that technological change is also ex-ante uncertain and can affect the timing of investment decisions. See for example Murto (2007), Huisman & Kort (2000) and most recently Meyer (2011).

   Empirical work on investments in the energy sector under uncertainty have been done by Bockman et al., (2008) for investments in small hydropower plants and by Kellogg (2010) for oil rigs in Texas. Empirical studies of vehicle scrapping are a particularly popular subject and tend to focus on repair and replacement costs and issues of depreciation. See for example Walker (1968), Parks (1977) or Manski & Goldin (1983).

   This article is mainly descriptive in scope. I do not attempt to explicitly estimate or test aspects of optimal abandonment or real options theory. But the results have important implications for studies that do seek to take a real options approach to the investment and scrapping decision of wind turbines, and possibly other renewable energy technologies. Uncertainty around technological advances and government policy should be seen as at least as important a factor as uncertainty around output prices.

   A growing literature on wind power investment and wind power subsidies also exists. In particular, analysis of the Danish market includes Morthorst (1999) who looks at the driving forces of wind power capacity development in Denmark...