Superannuation funds assign the bulk of their infrastructure allocations to infrastructure equity. Infradebt is an active lender to renewable energy projects. We thought newsletter readers might be interested in the different perspectives – that is debt versus equity – of the same underlying project – using a typical renewable energy project as an example.
While the results of this show that there are some risks that both debt and equity need to worry about, there are also significant divergences.
Let’s start our analysis of a renewables project by looking at it from a whole of project basis. On this basis, the most important drivers of returns are construction costs, revenues and operating costs.
Solar construction costs vary based on site acquisition costs, Engineering, Procurement, and Construction (EPC) arrangements as well as grid connection costs. Often the biggest variant between projects relates to grid connection. Single axis tracking costs range from $1.20-$1.50/W. Fixed tilt solar farms such as the Belectric system, are materially cheaper, at around $0.90-$1.10/W. Although fixed tilt is cheaper, total forecast generation is usually 15% or so less. Wind project construction costs are $1.90-$2.10/W with the most significant variations between sites relating to grid connection and forecast yield.
While costs have fallen substantially compared to even two or three years ago, the pace of construction cost decline has stalled on the back of a lower Australian dollar (most equipment components are effectively priced in USD or Euro). Assuming the AUD stabilises at current levels and looking out longer term, it would be reasonable to expect that costs will continue to fall in real terms through technological improvements. This has important implications for long-term revenue assumptions, as projects commencing today, will end up competing against new entrants with lower capital and operating costs.
Electricity in the National Electricity Market (covering Eastern Australia and SA) is traded via a five minutely auction system, whereby generators bid to supply the prevailing level of demand. Prices are extremely volatile (they vary between -$1,000 and $14,500 MWh). The following chart shows the history of annual average prices for the Victorian pool. Even on an annual basis, volatility is high with a standard deviation of $19/MWh or around 40% of the long-run average.
The majority of projects have some form of offtake in place with varying degrees of credit quality. The latest Power Purchase Agreements (PPA) we have seen are coming in are at the low $40s/MWh (2018 real) for high credit quality counterparties. PPA tenor and indexation arrangements vary. Contracts with full CPI indexation are less common. PPA contract terms have shortened from the 20 years contracts in the era of the ACT government (S&P AAA) reverse auctions, these days PPAs are typically 10 years and in some cases shorter, off-takers may be low investment grade, but increasingly off-takers are also unrated.
Over the last year or two, projects have been entering into lower and lower PPAs offtake prices. Effectively proponents are willing to sacrifice substantial amounts of project revenue (compared to current spot/near term futures prices) for the certainty of having at least a portion of revenues locked in via a PPA. The contracted proportion of generation has also been falling, nowadays it is common for projects to follow a hybrid revenue model – with 40-60% of revenues locked in via a PPA (over say 10 years) and the balance on a merchant basis. Given the shorter tenor of the PPA, on a Net Present Value (NPV) basis, these projects might have less than 20-33% of revenue locked in with a PPA (that is, less than half of debt).
This strategy aims to take advantage of the substantial premium between expected merchant pricing (typical assumptions are long term real spot prices of $70-80/MWH in 2018 dollars) and PPA pricing ($40-50/MWh). For these projects, the PPA is a necessary evil as part of securing debt financing, with the implicit Internal Rate of Return on the merchant portion of the project substantially above that of the contracted portion.
The older wind turbines carry a higher variable cost component when compared to the more modern wind farms we are seeing. In simple terms, maintenance costs are reasonably consistent on a per turbine basis, which means if you double the capacity of the turbine then operating costs per MWh fall substantially. This is reinforced by the higher capacity factors of new taller turbines (which can take advantage of lower wind speeds). Wind operating costs are around $20-30/MWh, and solar $10-15/MWh.
Solar operating costs are lower than for wind and are often significantly lower for large projects compared to smaller projects.
It is difficult to assess WACC/valuation multiples across renewables projects due to the multitude of different merchant price assumptions. An 8% WACC on the assumption that long-term merchant pricing is $100/MWh, is not comparable with to a 6% WACC for a project that assumes $60/MWh.
With this caveat – that assumptions vary massively – fully contracted projects are often on a 5%-6% WACC (reflecting high leverage and low equity returns) where projects with a substantial merchant component have WACCs more in the 7-10% range.
Project Operating Lives
Most projects are assessed on assumed operating lives of 25-30 years. It is usually possible to get major component warranties for 20-25 years. For windfarms, which are more operationally complex than solar, major turbine manufacturers will provide fixed price Operations and Maintenance (O&M) contracts (including lifecycle replacement) running out 25 years. Some equity investors include re-powering assumptions (i.e. running the plant for an extra 5-10 years beyond planned life). Debt investors normally assess projects (and require debt to be fully amortised) over an 18-20 year period.
The practicality of extending plant life is relatively untested (given that most projects are quite new). For wind, one challenge will be that by the time the plant gets to 25-30 years of age, the turbines are likely to have been superseded by new models – which may mean sourcing parts/O&M is expensive compared to new projects.
Concluding Remarks at Project Level
When comparing renewable energy projects with other infrastructure projects, we would make three key points:
Merchant revenues are significantly more uncertain than for other infrastructure asset classes.
Technological cost deflation. Existing projects need to compete against new entrants often with lower construction/development costs. This is a deflationary headwind on project revenues and, hence, asset values.
For projects with a PPA – counterparty credit risk is a key element of transactions – which often doesn’t exist for other infrastructure assets, where either key counterparties are governments (hopefully very low credit risk) or a diffuse set of users (eg ports, toll roads and, hence, individual counterparties don’t usually have a material impact on returns)).
Equity versus Debt – Example of Stereotypical Wind Project
The following section considers a typical wind farm project with a fully contracted 10 year PPA (30 year operating life) and financed with 60% senior debt and 40% equity. The chart below shows the attribution of operating cash flows. In this example, the long term merchant price is $67/MWh real and the PPA price is $45/MWh. The unlevered project IRR is approximately 7% p.a (pre tax).
Using a 7% discount rate, the NPV of PPA versus merchant revenue is split 40%/60%. Investors have approximately 60% of the project exposed to merchant prices. That is, the NPV of the merchant portion of revenues is 1.5x the equity of the project.
A key feature of renewable energy projects – from an equity perspective – is that equity cash flows are quite back-ended. This reflects the fact that revenues are lower during the PPA period and PPA revenues are disproportionately used to paydown debt.
The back-ended nature of equity distributions has three key implications:
Equity valuations will be quite sensitive to level of interest rates. In our example project, the weighted average tenor of the equity cash flows is 23 years (or the 21st year of operations). A 1% increase in equity discount rates would result in a 14% fall in equity valuations.
Equity returns are highly sensitive to assumptions about merchant revenues. For example, if merchant revenues were 10% lower than forecast, then the NPV of equity falls 18% (using an 8% discount rate).
It is important to understand that lower expected merchant revenues can even substantially impact equity during the PPA period via the need to refinance the project’s debt. Typical financing structures for renewable projects is the inclusion of a 20 year “mini-perm” with five yearly refinances. That is, debt is structured as a sequence of five year loans, each ending with a bullet maturity, over a 20 year amortisation profile. This amortisation profile is sized based on the project’s expected revenues. Thus, if merchant revenue falls (or more importantly, is expected to fall) then the target amount of debt outstanding is reduced. This implies a lower debt size at the next refinancing point.
In our example project, if expected merchant revenues post year 10 fell by 20% (from $67/MWh to $53/MWh), this would result in an approximately 18% lower debt size at the first refinancing (assumed to be at the end of year 5) and a substantial equity injection (26% of the original equity amount). The equity IRR would fall from 8.7% to 6.3%. Clearly this is a downside scenario, the converse also applies, an increase in expected revenues would allow an increase in debt size and a substantial one-off super distribution to equity.
This means that even for a fully contracted project, equity returns can be quite sensitive to the long-term outlook for merchant prices, despite being contracted in the short term.
Another important issue from an equity perspective, is that returns are quite sensitive to generation outcomes. This is relevant at both a short time and long-term time scale – for example it is not uncommon to have windy or below average wind years where generation for a windfarm is 10% below (or 10% above) the long term average.
Short term basis. For contracted projects debt is usually sized at 1.25-1.35x DSCR during the PPA period. This means that debt service is sized at 74%-80% of forecast operating cash flow. This means a 10% shortfall in generation (which would result in a 10% fall in revenue) could result in a 40%-50% reduction in equity distributions (or payments could be completely cut off if debt covenants are triggered). Thus, year to year equity distributions are likely to be quite volatile.
Long-term basis. Putting it politely, the history of forecasting the generation yield of windfarms is patchy. Many projects have ended up with long-term generation well below pre-construction forecasts – a Fitch study found that three quarters of wind Projects were operating below their P50 levels, and 43% were lower by more than 10%. A 10% permanent shortfall would reduce the value of equity by 20%.
As noted above, most renewable projects are financed using a mini-perm financing structure. That is, debt is sized on the basis of a notional amortisation profile (typically 18-20 years) with a refinancing every five or so years. This means the legal maturity of the debt at any point of time is five years or less. To mitigate interest rate risk, projects usually enter into interest rate swaps, usually matching the tenor of the PPA (and sometimes longer).
Typical debt structures for renewable projects:
have low interest rate risk – ie most commonly structured on a floating rate basis;
have a much shorter life than equity. The maximum tenor of debt is typically 5-7 years, allowing debt investors get to reset credit margins and debt size every five years. This creates a refinancing risk for equity (and existing debt if the changes are big enough). This substantially reduces the valuation volatility of debt. For example, a 1% rise in credit margins (the equivalent of a 1% rise in equity discount rates – for a debt investor) would only result in a 4-5% fall in debt values (and that is a worst case, if it happened immediately following issuance).
debt investors can be much more sanguine about generation outcomes. Given debt sizing parameters (target DSCRs of 1.35x during the PPA period and 2.0x on the operating period) normal year to year variation in generation can be relatively easily absorbed (it just reduces equity distributions). Similarly, these buffers can absorb reasonably significant long-term forecasting risk (provided there aren’t other adverse events at the same time).
That all sounds great – but what are the key risks for debt investors?
Ultimately the key risks to debt are scenarios (including combinations of adverse events) that result in the value of the project falling below the value of debt. When this happens – equity is no longer available as a buffer – and the pain will fall on debt.
The key risks debt investors need to worry about are:
Substantial construction delays/cost overruns (eg builder insolvency during construction). Most projects are financed with buffers to absorb a 3-6 month delay in construction. In these instances – equity returns will be affected, and debt interest may need to capitalise for a period – but it is unlikely that debt will suffer a permanent loss. More substantial delays/overruns – and in particular a mid-construction contractor insolvency – are likely to be more challenging. Perversely, fully contracted projects which have higher gearing are likely to have a higher risk to debt from construction issues (as higher gearing means a smaller equity buffer). Also, PPAs often have penalties if electricity isn’t delivered by a certain date (resulting in a doubling up of losses).
PPA counterparty insolvency. Insolvency by the PPA counterparty has the potential to have catastrophic impacts on project value. Critical to the impact is whether the PPA contract locks in prices above or below merchant pricing at the time of insolvency – i.e. is the PPA in the money or out of the money? The hit to project value comes in two forms. First, the extent to which current and prospective merchant revenues are lower/higher relative to the PPA price at the time of insolvency. Second, the project will have less locked in revenue and, hence, even if the overall level of cash flows was unchanged, the amount of debt those now more volatile cash flows can support will have fallen considerably (ie DSCR sizing goes from 1.25-135x to 1.90x-2.0x).
Substantial falls in the long-term merchant revenue assumptions. As outlined above, even with a 10 year PPA in place, a relatively substantial proportion of project NPV (around 60%) is exposed to merchant pricing. These merchant sizing assumptions drive the size of the debt outstanding at the end of the PPA. Given this is 10+ years in the future, a key risk for current renewable projects is technological progress driving a substantial fall in levelised costs, and these lower cost new entrants then competing down electricity prices. To illustrate this – current PPA prices are $40-50/MWh (for a 10-15 year contract with a high quality offtaker), while merchant pricing assumptions are often $60-80/MWh (real 2018) in years 10+ of a project. What would happen if future prices turned out to be consistent with current PPA pricing? In this scenario – ie a 40-50% shortfall compared to current merchant expectations – it is possible for the debt outstanding at the end of the PPA to exceed the value of the project (or at least be very close to it). In this scenario, it would be extremely difficult for existing debt to be refinanced and debt may suffer a loss (or forced conversion to equity) through the process.
Another way of looking at the above issues (in particular the last bullet point) is to compare debt and equity returns under consistent merchant price assumptions. Under the base case, in this example, debt is expected to return circa 5% and equity around 9%. In low price scenarios, say $30-50/MWh debt returns would actually exceed those for equity. To get double the debt return and earn a 10% equity IRR, requires a merchant energy price around $70-80/MWh (2018 real). Large energy users are currently locking in energy at a $40-50/MWh price. Time will tell who is correct!
Hopefully this analysis of a renewable energy project through two different lenses has been useful. The key conclusions to be drawn from this analysis are:
Debt and equity – even in the same project – are very different. For renewables, equity is long-duration and back-ended, while debt is short duration with a fixed return.
Which risks matter depends on your perspective. From an equity perspective, there are many factors that can have a meaningful positive or negative impact on returns. For debt, many smaller risks are easily absorbed by the buffer of equity (and that’s why returns are much lower).
For both equity and debt – long-term outlook for merchant pricing can be a critical driver of returns. This is an issue on which there are more opinions than market participants – but given its importance, no investor in renewables can afford to be without their own opinion.