Risk-adjusted discounted cash flow methodology is leading to a radically revised costing hierarchy between power generation technologies. Hydro could soon be considered the most valuable of them all, explains Tim Sharp
Alder dam
THE eventual outcome of an astonishing debate on power generation investment risk that is only now gathering momentum seems likely to strongly favour hydro power. If so, gone would be the conventional view of hydro as a rather inflexible, highly capital intensive, technology that, with questionable environmental credentials, will only ever play a marginal role in overall power supply.
In its place would come a perception of hydro as, on average, the cheapest large-scale source of electricity ahead of nuclear, coal and gas that, because it would also be a major contributor to overall macroeconomic stability, would be seen as the most valuable power generation technology of all. Only typically smaller-scale geothermal, biomass and wind would be cheaper on average. Their macroeconomic impact would therefore also be more modest.
The new view emerges from a reconsideration of the conventional discounted cash flow (DCF) methodology now used to estimate lifetime levelised kWh (llkWh) costs for all power generation technologies. Proposed by Tyndall Centre visiting fellow Dr Shimon Awerbuch at Sussex University’s Science and Technology Policy Research Unit (SPRU), the new method for the first time in the power sector incorporates the evaluation of forward cost and price risk into conventional DCF by applying well-accepted financial risk analysis theory to a much more detailed DCF. This risk-adjusted DCF is sometimes also known as ‘component DCF’.
The immediate impetus for the debate derives from power sector reform. In essence, private sector power generators are exposed to very much greater risk than were state-owned utilities. Moreover, privately owned generators must respond to such risk in much more definite, systematic ways.
But in addition to this overriding factor, the constantly expanding range of available power generation technologies means that much more detailed and more accurate financial analysis than before is required to correctly distinguish between alternative investment options. And because the presence of forward risk significantly raises present value cost expectations over project lifetimes, while simultaneously lowering those for revenue with major implications for profit, rigorous theoretically based risk analysis is essential for any modern power generation investment decision.
Remarkably, power sector investor response to this need has so far been both slow and vague. When reform was first launched it was naively, but widely, believed that the market itself would automatically identify appropriate technologies. There was thus no need to do anything! Even now, most discussion fails to differentiate between two basic types of risk known in finance theory as systematic and non-systematic risk, respectively.
But things are beginning to change. A 24 November 2003 international-energy-agency (IEA) publication, Power Generation Investment in Electricity Markets, briefly discusses three different approaches to power generation investment risk analysis, namely real options, probabilistic assessment and risk-adjusted DCF. It also mentions a July 2003 Massachusetts Institute of Technology (MIT), US, study into future nuclear power prospects that also uses risk-adjusted data, albeit the basis for such adjustment is unclear.
Unfortunately, the IEA report concedes that the real options approach that anyway applies only to peaking plant (hydro and gas) ‘has achieved little acceptance by power generation investors to date’ mainly because in this context it is unreliable. Meanwhile, probabilistic assessment delivers such general information – specifically, ‘a range of probabilities that a [risk-adjusted] investment would be profitable’ – that it is only marginally useful to an investor. Moreover, both approaches depend on prior assessment of forward fuel cost and power price risks.
Consequently, risk-adjusted DCF that both directly assesses these risks and incorporates them into the familiar conventional DCF framework is so far the only game in town. Significantly, it alone brings finance theory directly to bear on what is after all a finance problem. It is also the approach that encompasses Awerbuch’s work and leads to the astonishing conclusions above.
Admittedly, the world of power generation financial risk is highly complex. For instance, the two basic types of risk have many different elements and consequences (Table 1). And because of the confusion, non-systematic risk is commonly assumed to cover all risks. Dealing with such things as cost overruns, strikes, fuel supply disruption, regulatory change and market risk, it is admittedly by far the more dramatic and therefore noticeable of the two types. But it is also by definition mitigable (diversifiable) and can be hedged and is therefore largely manageable – a more or less known cost. By contrast, almost wholly invisible and so far wholly ignored systematic risk – because it is embedded in general economic conditions and can therefore be neither avoided nor diversified – is unpredictable in every respect except its inevitable presence. Consequently, because its magnitude in terms of eventual project profit or loss can be as great as that of any of the non-systematic risks, and because it is wholly undiversifiable, it necessarily represents any project’s greatest risk.
Specific weight
A further complication is that each of the risks in Table 1 impacts each generation technology to a different (and by project, variable) degree (Table 2). Consequently, each risk has a very specific weight in each project’s overall cost and revenue projection that in turn means that any given power generation opportunity probably requires several comparative (by technology and market) financial projections to arrive at the most competitive option. Because an entirely new field of activity (risk analysis) is being addressed and because so far unused techniques (risk-adjusted DCF) are being employed to address it, it would be unusual if a few surprises did not emerge.
Fortunately, plenty of surprises occur. For example, from Table 2, hydro’s very large size, very high cost/kW and high regulatory risk translate into very great exposure to the construction/finance and regulatory/political cost risks identified in Table 1 as well as significant market risk in highly volatile short-term power markets. All these risks are highly visible diversifiable non-systematic risk with little or no impact on systematic risk. They make hydro appear from a superficial risk perspective to be the least attractive technology option in the table. No surprises so far. But hydro also exhibits very low operating cost and no fuel costs that translate into low undiversifiable systematic risk for O&M and fuel. When these factors are correctly valued according to finance theory and incorporated into hydro’s cost and revenue balance, the financial picture changes dramatically.
Conversely, Table 2 makes gas appear the most attractive (lowest-risk/cost) option because only its fuel risk is high. But because this attracts a very heavy systematic risk penalty that, unlike hydro, carries through to the project hurdle rate (the internal rate of return that the project investor requires in order to invest) gas is in financial terms much less attractive than it first appears and as it is still widely assumed to be.
In fact by the new standards, wind is clearly the best option in Table 2 with nuclear and coal more or less equal and somewhere in the middle. Comprehensive, accurate risk allocation and valuation therefore dramatically change current perceptions.
As already implied nobody yet uses these techniques. Instead since privatisation the entire power generation sector has costed and priced a clearly highly risky activity with techniques that in fact ignore risk altogether, yet still happen to incorporate a mangled version of it into their figures anyway. This situation has arisen simply because present power generation investors have adopted without modification a public sector llkWh costing methodology (conventional DCF) that did not need to consider risk. This was because governments were able to pass all their power generation risks on to a captive market in mandated electricity prices.
Under conventional DCF, whose basic purpose is to determine the actual cost and hence price of the first and subsequent kWh delivered to the grid from a very long life power plant, the approach separately sums all the various expected project costs – capital, O&M and fuel – over the project lifetime and then bundles them into a single grand total lifetime cost. This composite sum is then discounted at a rate that is the same for all technologies within any given jurisdiction to reflect present value at the project’s commissioning date. By adjusting this result according to total estimated net lifetime output, a levelised (average) lifetime cost per delivered kWh is achieved. This method is still used by the entire power sector as well as by such organisations as the IEA.
Very clearly, it does not specifically incorporate risk. But the discount rate is in financial theory a measure of future risk. So the standard discount rate for all power generation technologies in any country, typically between 5–10%, that is applied in conventional DCF in fact incorporates a theoretically wholly arbitrary estimate of overall project financial risk. Although this estimate cannot distinguish between technologies conventional DCF originally worked quite well.
Rigid method
It was successful because it originally applied to only two financially very similar power generation technologies. These were highly capital intensive, zero fuel cost hydro and almost equally capital intensive, very low fuel cost coal. But the methodology is also quite rigid so that as rising fuel costs gradually absorbed a larger share of total costs in coal-fired plant and then oil-firing, it progressively became less accurate. This process experienced a dramatic advance with the oil price shocks of the 1970s and 1980s and became particularly pronounced with the advent of low capital cost, high fuel cost natural gas-fired generation. It was further magnified by the increasing volatility of fuel prices, particularly for gas, that obviously impacts future fuel price risk without affecting any of the other project risks.
As long as power generation was a public sector activity, the by-now glaring inaccuracies in fuel-based generation costing did not show up. But when power sector reform took place, risk had to be specifically addressed. And when non-fuel renewable power generation technologies emerged at about the same time, the financial profiles of all the different technologies, but particularly fuel-based versus non-fuel, were obviously so very different that a single bundled discount rate for all technologies could only mask major financial differences that in a competitive market context must be distinct.
Risk-adjusted DCF therefore modifies the conventional DCF process and applies financial risk analysis theory to it. In essence, risk-adjusted DCF applies different discount rates to each individual lifetime cost category within each technology according to its riskiness as perceived by the theory. Thus actual expected project lifetime capital, O&M and fuel costs by technology are summed and discounted separately before being bundled as total expected project present value. This present value total then becomes the basis for deriving the llkWh cost.
Because the discount rate has a very strong influence on this final figure, the very much more detailed, accurate, theoretically defensible procedure produces dramatically different llkWh costs between technologies. Very obviously, a non-fuel technology such as hydro can be expected to turn out to be significantly cheaper than most fuel-based technologies. Consequently, as required by a competitive market, each technology is clearly financially distinguished from the others.
Closer examination of the methodology demonstrates its power. Specifically, the new approach is based on widely accepted capital asset pricing models (CAPM) within finance theory that incorporate the risk that forward prices or costs will deviate systematically from their projected values (see box, right). Since high forward revenue risk (the risk that at some future date one will receive less than expected) obviously requires high discount rates to correctly reflect the likelihood of only low receipts (low present value), it follows that high forward expense risk (the risk that at some future date one will have to pay more than expected) requires low discount rates to correctly reflect its high present value.
And although in the context of kWh costing, the precise appropriate fuel discount rates are debatable, the point here is that forward fuel price risk attracts a high present value and therefore high llkWh cost – so that non-fuel hydro and other renewables obviously avoid the expensive fuel costs altogether. Therefore, because hydro’s levelised kWh cost must support only the already low debt equivalent costs, it can be expected to be much more competitive than fuel-based technologies even when its capital cost is much higher. Table 3 puts all this into perspective by quantifying these differences for a range of power technologies in a European context. Several points should be noted.
First, columns two and four, ‘ideal’ and ‘empirical’, respectively, show llkWh costs derived from low and high forward fuel risk boundaries, respectively. For example, the applied fuel discount rates for gas-firing are 3.9% and 1.3%, respectively, compared to a standard highly unrealistic (impossibly safe) 7.0% in column one. Actual real world levelised costs for all fuel-based technologies will therefore lie somewhere between the two estimates, probably closer to column four than column two. Consequently, the estimates for all non-fuel technologies are the same in columns two and four. In effect, they reflect merely the absence of the distortion imposed by the arbitrary standard-across-technologies 7.0% discount rate in column one.
Therefore ‘large hydro’ is risk-adjusted cost competitive with all fuel-based technologies in column four and with all fuel-based technologies except coal in column two. In the real world, because column two costs are unrealistically low and on this basis coal’s costs are only marginally below those of hydro, hydro can be seen as competitive with all technologies except geothermal, biomass and wind in all risk-adjusted situations.
Second, the great deviations from convention occur among those technologies whose risk-adjusted fuel costs constitute the greatest share of overall project lifetime cost. Thus low capital cost, high fuel cost and high fuel price volatility gas-firing is worst affected while high capital cost, less cost-volatile coal is roughly half as badly affected. Because nuclear has only a low fuel cost with almost zero forward cost volatility risk, it becomes significantly more expensive under risk-adjusted DCF but maintains its costs across columns two and four. Column four offers a new risk-adjusted merit order that virtually turns the conventional one on its head.
A radically revised costing hierarchy between technologies is however only one of three potential outcomes from applying financial risk theory to the power sector. The second of these, relying on the observed fact that risk has financial value, would in the context of available energy sources develop sector-wide portfolios that would deliver least cost power at the national level. Formal non-fuel renewables-based portfolio management might achieve only modest economic gains in a system that was already fully dispatched according to the risk-adjusted merit order in Table 3 column four but it could clear the way for quite extraordinary sectoral gains in cases where, as at present, the main fleet continues to be costed and dispatched traditionally.
In such cases, main fleet fuel-based power costs are low. Using conventional DCF, renewable generation costs are high. But most renewables consume no fuel and therefore have no fuel risk. And if this is factored into an otherwise conventional DCF portfolio management equation, the net effect would be to lower overall system generation costs even though main fleet costs are already low and the renewables themselves, valued conventionally, are expensive. In effect, the lower risk-adjusted costs inherent in zero fuel risk renewables would lower overall system generation costs.
Significantly, this opportunity is so practical it is already being piloted in Morocco, Mexico and India with UNEP, UK Foreign and Commonwealth Office (FCO) and Basel Agency for Sustainable Energy (BASE) funding. If it proves out, large hydro, especially in India, would become the pre-eminent technology among renewables to achieve the greatest degree of portfolio cost reduction.
But the third outcome is potentially the most exciting of all. In this case, the focus of attention is not just on financial risk but on the observed fact that within the general economy, fuel prices vary negatively in relation to all other prices. Thus when fuel prices fall, the rest of the economy benefits and all other prices and earnings pick up. Conversely, when fuel prices rise, the rest of the economy is negatively affected and other prices and earnings fall.
Considered on its own, little can be done to mitigate what in macroeconomic terms is a hugely expensive negative correlation. But if zero-fuel risk renewables are brought into the equation, they can effectively damp both cycles. Thus when fossil fuel prices started to rise, zero fuel renewables would become even more competitive. As in consequence they became more widely adopted, they would soften the impact of rising fuel costs on the rest of the economy so that the resulting slump was less severe.
Revolutions
Since such macroeconomic cycles are believed to cost individual economies billions of dollars for each recession and since much of the loss would be avoided by switching to renewables, they would have a dramatic economic impact far beyond that deriving from their immediate kWh costs – even though these in a properly risk-adjusted system would anyway be lower than those from fossil fuels. Again, large hydro as the largest renewable would play a commensurately greater role in damping such cycles and would thus be macroeconomically more valuable than other non-fuel technologies.
There are so far no indications when this revolution will begin. Most investors and the IEA are still trying to muddle through without distinguishing between systematic and non-systematic risk. Until this is resolved, very little progress can be made. But two things are certain. First, power generation investors widely accept they urgently need rigorous, theoretically based risk analysis techniques in order to distinguish successfully between a constantly growing range of options. Second, as the IEA tacitly admits, risk-adjusted DCF is so far the only viable technique on offer.
| Capital asset pricing |
| The capital asset pricing model (CAPM) was first devised in the 1960s. It attempts in part to estimate the price at which a contract for the future delivery of a product or service would trade in the marketplace today. Since investors are acutely aware of risk, this must be factored into the calculation as the risk that projected values will deviate systematically from expectations. Such risk is expressed as a discount rate. Because, based on their experience, investors would buy into the contract at the agreed discount, this rate can be said to function as a best-available proxy for changes in value over time. |