Submissions

APGA Submission to the AER

Written by APGA | Sep 2, 2022 1:35:00 AM

1.         Executive summary

The Australian Energy Regulator (AER) has set out proposed updates to the Rate of Return Instrument (RoRI). The AER is proposing very few changes. The biggest change proposed is to reduce the term of the risk-free rate for equity from 10 years to a term that matches the length of the regulatory period.

1.1.       Our views on the draft 2022 RoRI

As shown in the table over the page, we are broadly supportive of the draft 2022 RoRI, except for the AER’s proposals to:

  • Reduce the term of the risk-free rate to match the length of the regulatory period
  • Not adopt an approach that automatically updates the Market Risk Premium (MRP) at the same time that the risk-free rate is determined
  • Not use international data to estimate the equity beta.

On the first, in our view, the AER should reconsider its proposal to reduce the term of the risk-free rate. This is needed to ensure that the 2022 RoRI can provide an efficient return on investment that best promotes the NGO.

We are particularly concerned that the AER has sought to interpret its regulatory task in a way that leads it to ignore what investors actually do when establishing their return expectations. It instead adopts a series of artificial assumptions that lead it to conclude that those investors set return expectations only for the length of the regulatory period. There is no evidence of this, only assumption. This is an error, both of logic and in interpreting what is required to promote the NGO.

We explain our concerns in Section 2 and strongly recommend that the AER retain a 10-year term for the risk-free rate proxy. Our views on the other components of the draft 2022 RoRI are considered in Section 3.

We also strongly support Energy Network Australia’s (ENA’s) submission on the AER’s draft 2022 RoRI.[1] Although we reference that submission in places, we do not repeat all the points raised by ENA.

 

Component

Do we support the AER’s proposal?

Our views

Gearing

Yes

·       The evidence supports retaining a gearing level of 60%

Cost of equity model

Yes

·       It is appropriate to continue using the Sharpe-Lintner Capital Asset Pricing Model (SL CAPM)

Risk-free rate

No

·       Continuing to rely on yields on Commonwealth Government Securities (CGS) reflects the opportunity cost of capital

·       But reducing the term of those yields to match the length of the regulatory period is not supported by finance theory or market evidence and will deliberately undercompensate energy networks for their efficient financing costs

Market risk premium

Partially

 

 

·       The AER should consider adopting an approach that updates the MRP at the same time as the risk-free rate using the method proposed by ENA, including a calibrated dividend growth model (DGM)

Equity beta

Partially

·       Reliance on a dwindling sample of Australian comparator firms risks setting an equity beta value that does not reflect the forward-looking risks faced by energy networks

·       The AER should pick up the Independent Panel’s recommendation of considering alternative information, including international equity betas

Return on debt approach

Yes

·       Assuming a 10-year term and BBB+ credit rating remains appropriate

·       There is no strong case to move away from relying on benchmark yields published by third party data providers

Imputation tax credits

Yes

·       No major concerns with gamma

Cross-checks

Yes

·       It is good to see the AER incorporate cross-checks and sensitivity analysis into its decision-making

·       We look forward to the AER improving on this for its final decision on the 2022 RoRI

1.2.       Submission structure

Our submission is structured as follows:

  • Section 2 explains our concern with the AER’s proposal to reduce the term of the risk-free rate from 10 years to a term matching the length of the regulatory period
  • Section 3 provides our views on other components of the draft 2022 RoRI, including the market risk premium, equity beta, cost of debt, gearing, gamma, and cross-checks.

 

 

 

 

 

2.         Term of the risk-free rate

We and many other stakeholders are concerned about the AER’s proposal to reduce the term of the risk-free rate from 10-years down to the length of the regulatory period.

The AER has interpreted its regulatory task in a way that leads it to ignore what investors in regulated energy networks actually require. It uses artificial assumptions to reach a conclusion that such investors should only care about and require compensation for risks over a regulatory period, not beyond it. This conclusion is not support by evidence or finance theory.

The AER’s task is to determine the allowed return on equity. It has chosen to do this using the SL CAPM. Having adopted that model, the AER needs to decide what proxy (e.g., instrument) to use for the (unobservable) risk-free asset and what term to apply to that proxy. The AER has rightly chosen CGSs as the proxy – an instrument widely used by investors. But it has proposed not to adopt the term for that instrument used by investors, instead proposing a term that matches the length of the regulatory period.

In our view, there is no case for changing the term of the instrument used to determine the risk-free rate proxy. The AER should also consider the consequences of making such a change – the biggest being that it would significantly dent stakeholder confidence in the rate of return setting process.

We explain our concern in the sections that follow, starting with the AER’s regulatory task.

2.1.       What is the AER’s regulatory task?

The AER starts its analysis by describing its regulatory task, calling out that it is focused on resetting revenue allowances at each regulatory determination. Although it is undoubtedly true that the AER’s determinations reset revenue allowances, this does not really explain what the AER’s task is when it comes to determining a return on equity allowance.

A better staring point is the National Gas Law, which governs the AER’s task and requires it to promote the NGO when developing the RoRI.[2] As an economic objective, it seeks to:

promote efficient investment in, and efficient operation and use of, natural gas services for the long term interests of consumers of natural gas with respect to price, quality, safety, reliability and security of supply of natural gas.

That objective clearly talks about promoting efficient investment for the long-term interests of consumers. Given the long-term nature of pipeline investment, it seems somewhat obvious that to promote that objective the AER should be concerned about determining a return on equity allowance that promotes efficient long term investment. The implication is that consumers will benefit over the long-term if networks make efficient investment decisions.

This focus on long-term is echoed in decisions by the Federal Court and Australian Competition Tribunal when interpreting the NGO and its electricity equivalent, the National Electricity Objective. For instance, on the latter, the Tribunal said it:[3]

requires prices to reflect the long run cost of supply and to support efficient investment, providing investors with a return which covers the opportunity cost of capital required to deliver the services.

Networks have no power in the market for funds. Therefore, as the AER notes,[4] energy networks are likely to face competitive prices in the market for funds, which means that the efficient price is the market price.

The AER also recognises that it is important to look at what the market requires when promoting economic efficiency. After quoting Alfred Kahn, the explanatory statement goes on to say that: [5]

We consider employing a rate of return that is commensurate with the prevailing market cost of capital (or WACC) is consistent with the NPV=0 investment condition. We also consider economic efficiency more generally is advanced by employing a rate of return that reflects rates in the market for capital finance.

Similarly, in a recent paper for ENA, Professor Schmalensee observes that: [6]

Economic efficiency of course, requires that the allowed rate of return is always commensurate with the return that investors require…At the most abstract level, the regulatory task is conceptually a simple one – determine the return that market investors require and set each period’s allowed rate of return and accounting rate of return to match it.

We agree. The regulatory task before the AER is to adopt a RoRI that will determine the return that investors require. Elaborating a little, this involves first understanding what return investors require and then designing an approach to include in the RoRI that seeks to replicate that as best as possible.

What investors require is not directly observable for the most part. Nevertheless, that is the task at hand. Yet, as we describe below, the key failing of the AER’s decision to base the risk-free proxy it chooses on the length of the regulatory period is that it does not actually seek to demonstrate what investors require. Rather, through a series of artificial assumptions, the AER derives what it thinks investors should require without any evidence that they do.[7] The AER is not an investor itself and so such opinions can only ever give a poor approximation of real-world investors.

Moreover, it is not clear that the AER’s views on what investors should require will have any influence on what investors actually require. For instance, a Canadian pension fund that deploys capital globally, seems highly unlikely to change the rate at which it discounts expected cashflows from a potential investment in Australia because an Australian regulator thinks that it should. If it does not like the projected returns for that investment (e.g., because the regulator sets allowed returns too low), then it will just invest elsewhere.

2.2.       Does the AER ignore what investors in regulated energy networks require?

At the outset, the AER draws an artificial distinction between a commercial context and a regulatory context: [8]

In a commercial context, the term of the required rate of return on an asset relates to the expected investment time horizon for a physical asset or holding period of a corresponding security. In a regulatory context, the term of the allowed rate of return is related to the time period of the allowance (such as the length of a regulatory control period, where the rate of return will be reset at the commencement of each new regulatory control period).

The distinction, however, has no bearing on the return required by investors. It is a matter of fact that investors in regulated gas pipelines operate in a commercial context. They must consider cash flows and risks over the life of the investments they make. They must balance competing demands on their scarce investment funds across a range of potential investment opportunities. Even if the assets they invest in are regulated, they are nevertheless subject to commercial imperatives. Similarly, even if the AER ignores the commercial context, investors cannot and will not.

Yet, it is that distinction that leads the AER into error. By making it, the AER is saying that it is not trying to determine an allowed rate of return that reflects the return required in a commercial context. Rather it is trying to determine an allowed rate of return for a regulatory context; one that is affected by the length of the regulatory period. And by extension, given that investors operate in a commercial context, the AER is saying that it can ignore what returns they require.

In its analysis that follows, that is exactly what the AER appears to do. For instance, the AER explains: [9]

Actual investor valuation practices appear to be consistent with using long-term government bonds. In the case of Australia these are 10-year CGS.

However, we do not estimate the allowed rate of return to be used as a discount rate for a business valuation over a long investment horizon…

And: [10]

our exercise is different from that faced by a market practitioner performing a business valuation. While using 10-year CGS yields in market valuations may be supported by both academic works and market evidence, it is not clear that the same evidence provides support for using a 10-year term for the allowed return on equity in our regulatory context.

The AER does not reference any evidence that investors in a regulated energy network would adopt a term of equity that matches the length of the regulatory period. It has received multiple submissions from businesses that invest in energy networks that make clear that they consider much longer horizons when making investment decisions. This is recognised by the AER.[11]

Rather than evidence, the AER goes on to make a series of simplifying assumptions in the context of a simple valuation model (Dr Lally's, or its own "exploration") that it argues support its conclusion that an investor would require the term of the return on equity to match the length of the regulatory period. [12] But that is not evidence of what investors require and for the reasons explained below the assumptions are erroneous or unsupported in any event.

More importantly, investors are not going to change their required rates of return simply because the AER analysis suggests that they should. Investors face commercial imperatives and must consider risks over the life of the investments they make. Even if the AER were right that investors should set their return requirements based on the length of the regulatory period, it is unlikely that they will take investment advice from the AER on that point.

What is much more likely is that, if the expected rate of return were insufficient because of the AER’s change to the term of the risk-free rate, then investors would simply invest elsewhere. If this means that efficient investment does not go ahead as a result, then consumers would likely suffer in the long-term. This outcome would undermine the NGO.

2.3.       Why does the AER ignore what investors require?

There appears to be two key reasons why the AER decides it can ignore what investors actually do when considering their required rate of return:

  • First, a belief that the relevant interest rate risks are only those that exist within a regulatory period, and that longer term risk is eliminated due to price resets. [13]
  • Second, a mathematical analysis.[14]

Before considering these reasons, we first consider the AER’s observation that investors use a 10-year term for the risk-free rate as a proxy for the theoretically correct approach.

AER observation: 10-years is a proxy

The AER observes that the theoretically correct approach to valuing future cash flows is to discount each cash flow using a discount rate with a term matching the period while that cash flow occurs.[15] The AER goes on to conclude that just because many investors tend to use a 10-year term for the risk-free rate proxy does not mean that they are invariant to the length of time over which returns are to be received.

It is not clear to us that anyone has claimed that investors are invariant in respect of the length of time over which they receive returns.

It is relatively easy to understand why they might use a single discount rate as an approximation. In the AER’s view, all that investors are doing by adopting a single discount rate for all cash flows is simplifying their analysis. This simplification neither requires, nor assumes that investors are invariant about returns in different years. And it certainly does not imply that the correct thing to do is use a term that matches the length of the regulatory period.

An illustration may help.

Illustration: why investors might use a single discount rate?

Say an investor has an investment with a 50-year life and knows that it should – when applying the theoretically correct approach – discount each year of cashflows with a different discount rate. In practice, this may make for a complex investment analysis spreadsheet, especially if there are a lot of cash flows to account for (e.g., 50 discount rates for 50 annual cash flows).

However, it is relatively straight forward to test whether it would be appropriate to approximate that complexity using a single discount rate. The first step would be to set up an Excel spreadsheet that calculates the net present value (NPV) of a notional stream of future cashflows using the theoretically correct approach (i.e., with different discount rates for each cash flow). The second step would be to then use Excel’s goal-seek function to determine a single discount rate that, when applied to all cashflows, gives the same NPV as that derived in the first step.

If this approach were applied using yield curves from August 2022 and cashflows were assumed to be constant over the 50 years, then we get a single discount rate that is similar to the yield on 11-year CGS. This aligns with the 10-year proxy commonly adopted by investors and practitioners.

Cashflows from projects will differ and the timing of when the analysis is undertaken – which affects the shape of the yield curve – will matter. However, in general, even very long-term projects can be proxied with an interest rate close to the 10-year rate. Given this, it is not surprising that investors use this proxy, given its liquidity and general applicability.

Reason one: interest rate risk and regulatory periods

The AER states that the exposure to interest rate risk for firms that are regulated is limited to the length of the regulatory period:[16]

Second, 10-year returns may also contain a term premium to compensate for risks of locking in rates for an extra 5 years. These risks include inflation and interest rate risks. In this case, a 10-year return may be higher (lower) than a geometric average of the prevailing and expected future 5-year returns for 2 consecutive regulatory control periods. However, it does not follow that the use of a 10-year, rather than a 5-year, equity term is warranted when the allowed revenues are reset every 5 years. With 5-year resets, investors in regulated assets do not bear the risks associated with locking in the rate of return beyond a 5-year regulatory control period. Therefore, compensation for these risks is not part of the opportunity cost of equity capital and would not be necessary to attract investors.

The AER is wrong with respect to the risks faced by investors in regulated assets.

Those investors are not ‘locked into’ an interest rate for five years. Investors in regulated assets are usually long-term investors buying shares for their own portfolios, investing indirectly through intermediaries such as pension funds, or investing as asset owners and operators. They face the risk that future interest rates used to determine the allowed rate of return will differ substantially from what is expected; differences that materialise through higher or lower cash flows that result from the formulaic way that allowed returns are calculated within the AER’s building block modelling. Given that risk, investors require a return that is higher than that applying to an asset that has a term matching the length of the regulatory period, such as a 5-year bond.

The simple reality is that equity investors do not receive their investments back at the end of the regulatory period. They are concerned about how the AER may reset allowed returns for subsequent regulatory periods, including in response to changes in interest rates. The AER’s statement above appears to ignore that. [17]

This issue is also addressed in detail in a submission from the Queensland Treasury Corporation (QTC).[18] That submission considers finance theory that holds that the risk of an asset is related to the length of the holding period and not to the frequency of payments. Using that theory, QTC conclude that if equity in a regulated energy network is thought of as a long-term floating rate bond with a coupon that is reset at the start of each 5-year regulatory period, then it would be priced at a significant margin above the yield on 5-year interest rate.

Finally, the AER’s view about interest rate risk implies that any equity investor in an unregulated firm whose prices have some exposure to interest rates could lower their cost of capital simply by encouraging the firm to reset those prices more frequently to reflect changes in interest rates. We are not aware of any evidence showing that investors in such firms adjust their required returns based on the frequency of when prices are reset. Nor are we aware of any evidence that investors in competitive capital markets choose to invest in one firm over another depending on how frequently prices are reset. It would be sensible to test assumptions like that made by the AER against what is observed in a competitive market context to see whether it is valid or not.

Reason two: the maths

A cornerstone of the AER’s justification for setting the term of the risk-free rate to match the length of the regulatory period is a mathematical proof of the NPV = 0 principle and earlier work by Dr Lally.

This has been discussed at great length by the ENA,[19] and in a paper commissioned from Professor Schmalensee,[20] who notes quite clearly that the assumptions underpinning both Lally’s and the AER’s maths are simply not defensible. The Consumer Reference Group (CRG) came to similar conclusions to Professor Schmalensee. [21]

Moreover, we engaged CEG to review the AER’s analysis. CEG concluded that:[22]

  • The AER is wrong to claim that its approach necessarily follows from sound finance logic. It does not. It is also not consistent with the expert panel advice on MRP, including Dr Lally, which considered that the AER’s single method and “set and forget” approach was naïve.
  • For the AER’s approach to be correct very strong assumptions…need to be made about the term structure of the MRP that have no grounding in the theoretical or empirical finance literature and which, if true, would have very unusual implications more generally; and
  • The available evidence is inconsistent with the AER’s assumptions…. Namely, evidence:
    • that investors’ actual practice is to use a long-term discount rate at all tenors (i.e., no term structure in the return on equity); and
    • that term premiums on equity vary inversely with term premiums on government bonds (cancelling out, at the return on equity level, variations in bond term premiums).

Finally, we agree with ENA’s view that, even if the reasoning behind AER’s mathematical analysis were sound – which it is not – it is inappropriate to elevate it above evidence about required returns in the market. Such elevation cannot be held to meet the AER’s legislative obligations.

Conclusions

Given these compounding errors, the AER should reconsider its assumptions and analysis. It certainly should not rely on them to conclude that investors in regulated energy networks set their return requirements based on how frequently the return on equity allowance is reset. In fact, doing so gives rise to consequences that further highlight the problems with the AER’s proposed approach.

2.4.       What are the consequences?

Even if it remains open to the AER to adopt a term for risk-free rate that matches the length of the regulatory period, it should seriously think twice before doing so given the consequences.

  • First, it will lead to more volatile return on equity allowances. As CEG notes,[23] the 5-year yields on CGS are more volatile and more procyclical than 10-year yields. More volatile return on equity allowances will lead to more volatile network charges. More procyclical return on equity allowances will increase the systematic risk faced by investors.
  • Second, it requires changes to other parameters used to apply the SL CAPM. The AER has recognised that the MRP needs updating (albeit in the wrong way) but has not considered equity beta or the return on debt. Prior AER decisions have set revenue allowances using a 10-year term for the risk-free rate. This will be reflected in the share price movements of the firms it regulates and, therefore, any betas estimated using those movements. It will also be reflected in the risk assessments undertaken by debt investors and credit rating agencies.
  • Third, it will lead to return on equity allowances that are, on average, lower than the returns required by investors in regulated energy networks because the AER is wrong about what returns investors require. Such underfunding will undermine the efficiency objective set out in the NGO and will risk inefficient investment decisions being made.
  • Fourth, it implies that firms with low betas (i.e., beta < 1.0) typically have a lower return on equity over shorter horizons than over longer horizons. Conversely, firms with high betas (i.e., beta > 1.0) will typically have a higher return on equity over shorter horizons than longer horizons. There is no obvious rationale for that incongruity.
  • Fifth, it will have real world implications for capital budgeting. If the AER is correct that the cost of equity faced by energy networks reflect terms that match the length of the regulatory period, then those networks should be using a lower cost of equity when making capital budgeting decisions. A lower cost of equity has the potential to materially tip the balance of present value assessments in favour of more capital-intensive network solutions.

These and other consequences are considered further in CEG’s report included as Attachment A.

Given there is no compelling case for change, we strongly encourage the AER to retain its 10-year term assumption. Making a change without a compelling case will undermine stakeholder confidence in the rate of return setting process.

No stakeholder has suggested that the framework the AER uses to consider term is relevant and no new evidence supports it. All that has happened is the AER has reinterpreted the theory and placed less weight on real world observations. There is nothing stopping it from relooking at these matters again in the future when different AER board members and staff are involved or it is facing a different context. Such uncertainty can only ever undermine confidence in the regulatory regime.

 

 

3.         Other rate of return components

Our biggest concern with the draft 2022 RoRI is the AER’s proposal to reduce the term of the risk-free rate from 10 years to the length of the regulatory period (as discussed in the previous section). However, we also have views on other components of the draft instrument, which we set out in this section.

3.1.       Market risk premium

We support the ENA’s submission on the MRP. We do not repeat that submission here.

At the AER’s concurrent evidence session, the experts agreed that there were two types of MRP:

  • An unconditional MRP that reflects the average through time that is not affected by market conditions at a given point in time, and
  • A conditional MRP that does vary through time and changes with market conditions.

The draft 2022 RoRI effectively adopts an unconditional MRP. The AER’s explanatory statement asks for stakeholder feedback on whether a formulaic approach to determining the conditional MRP should be used instead.

In our view, the AER should re-consider adopting a formulaic approach to updating the MRP that combines estimates from the calibrated DGM developed by Frontier and historical excess returns. This is a topic addressed at length by the ENA, and we agree with its conclusions.

We do want to respond to a point raised by the Independent Panel. On page 22 of its report, the panel contrasts the use of judgement in the case of beta to place less weight on more recent estimates with the mechanical updating of the historical excess return estimate. It considers that this creates inconsistency rather than clarity. The panel goes on to note that “A mechanical approach to beta would have resulted in a decrease in beta”.

The Panel is wrong for two reasons:

  • First, there is no inconsistency. Judgement is used to determine both the MRP and equity beta. The difference is how it used.

In the case of MRP, the AER uses its judgement to determine the MRP using the historical excess returns approach. It considered other approaches and different ways of applying that approach (e.g., time periods, data sources) before settling on one. True to that approach, the AER ultimately decides to use the longest period of data available following the introduction of dividend imputation in Australia. These are all out-workings of judgement.

In the case of equity beta, the AER uses judgement to weigh up different estimates of equity beta that cover different time periods, combinations of comparator firms, and estimation methods. Rather than rely on just one estimate or calculation approach, it looks at that information together to arrive at its proposed value of 0.60.

There is nothing wrong with considering a wide range of information, especially where there are concerns about data availability (see our discussion below in section 3.2) or less clarity about what would be a single preferred approach to estimating equity beta. It is certainly not inconsistent with its approach to MRP where it decided that there was sufficient data to apply the historical excess returns approach and that that approach was preferred to all others.

  • Second, it is not at all clear that a mechanical approach to estimating beta would have resulted in a decrease in beta (compared to what the beta would have been if a mechanical approach was used in 2018). It could lead to an increase. Whether such an approach would lead to a decrease or not depends on how it works.

For instance, one mechanical approach could be to place more weight on comparators that remain listed. As Spark Infrastructure and AusNet Services were delisted over the last few years, applying this approach would lead to an automatic increase in equity beta estimates that stems from placing more weight on the one remaining listed Australian energy network, APA.

However, unlike its approach to estimating historical excess returns, the AER has not relied on a single (mechanical) approach to estimating beta in the past. As the panel noted, the AER has used judgement instead. Given this, it is simply not possible to conclude that applying a mechanical approach now would lead to a lower beta as there is no counterfactual to compare to because in 2018 the beta was not determined mechanistically either and thus there is no mechanistic way to update that determination.

For these reasons, the AER should disregard the panel’s criticism that it is being inconsistent. We do not think that it is.

On page 28 of its report, the panel also considered that the time series used to estimate historical excess returns should be adjusted to exclude periods of unusual market circumstances. In our view, such adjustment will be arbitrary, and undermine the use of that method by biasing estimates.

The more recent market volatility related to the Covid-19 pandemic and Russia’s invasion of Ukraine is not the only period of unusual market volatility in the time series used by the AER to estimate historical excess returns. That time series has been affected by the global financial crises, the European debt crises, the Asian debt crisis, the dotcom bubble, and many others.

If we head down the path of adjusting out all time periods that were affected by unique circumstances, then we will end up with a very patchy time series that will make the historical excess return estimates highly unreliable. The resulting MRP estimate will be heavily influenced by what periods the AER decides to exclude – which could introduce bias (e.g., if it is more sensitive to period of high returns than lower returns). By the same logic, the time series data used to estimate other parameters (e.g., beta) could also end up patchy and lead to biased estimates. This should be avoided for all parameters.

3.2.       Equity beta

Concerns with the sample

We remain concerned that the dataset relied on by the AER is too small to reliably estimate the equity beta for regulated gas networks. We made these points in previous submissions and so do not repeat them here.[24]

Instead, we observe that there are two important trends going on:

  • First, the energy transition is driving significant, albeit different, change for gas and electricity networks. Such change will inevitably flow through to share price movements and equity beta estimates as the operating environments faced by those networks changes.
  • Second, the pool of listed Australian energy networks is dwindling. We are left with just APA, which owns unregulated gas pipelines and other assets.

If these trends continue, then the usefulness of old share price movements to estimate a forward-looking equity beta will continue to reduce. Eventually, it simply will not be tenable for the AER to rely on those movements and be confident that it is estimating an efficient, forward-looking, return on equity.

The AER has recognised this: [25]

In this review, a key issue on equity beta is the diminishing number of the Australian comparators we use for estimating beta. This has declined from 3 in the 2018 review to being just one (APA) – Spark Infrastructure and AusNet having recently been de-listed due to takeovers. For the majority of the time period since 2018, we still had data from these 3 firms, but this underlines a challenge to our current approach going forward.

We and other stakeholders have proposed using international data to help keep the equity beta estimates current. The AER should reconsider this when developing the final 2022 RoRI. But, even if the AER is unconvinced that such data is sufficiently comparable to a pure play Australian energy network today, we strongly encourage the AER to keep an open mind as to whether it could be useful in the future just as other regulators within Australia and around the world have done.

This point is picked up by the Independent Panel:[26]

The single biggest data challenge facing the AER relates to the shrinking size of the comparator data set. At the time of submitting this report, this had declined to a single entity. The AER has acknowledged the need to explore alternative sources of information to replace the current direct observations. The Panel recommends that the AER adopts a broad approach to identifying new data sources and remains open to the possibility of combining insights from multiples sources.

We agree entirely. As the quality and relevance of Australian data declines the relative usefulness of international comparators improves.

In our view, the time is right for the AER to place some meaningful weight on beta estimates for those comparators. Failure to do so creates a real risk that the AER’s equity beta estimate is out of touch with the risks now facing Australian gas and electricity networks.

As we have previously noted,[27] incorporating international data will also allow the AER to estimate a gas-specific equity beta for gas networks. Reliance on a dwindling Australian comparator sample has not allowed for this in the past.

Recent market developments

Putting to one side our concerns about the dataset used, we agree with the AER that the recent market developments do not provide sufficient evidence to justify reducing the equity beta. As noted above, this is entirely consistent with the AER’s approach to updating the MRP for more recent information.

There are two reasons for this:

  • First, the apparent reduction in beta estimates for Spark Infrastructure, AusNet Services, and APA are heavily influence by a 2-week period in March 2020 when the world first reacted, on mass, to fears about the Covid-19 pandemic and what this meant for the future. This highly abnormal share price trading period is unlikely to give a true reflection of the systematic risk faced by Australian energy networks.
  • Second, as noted by the AER, beta estimates from international energy firms have tracked in the opposite direction. This should lead the AER to pause before simply adopting the more recent equity beta estimates without adjustment.

Although conceivable that the systematic risk facing Australian energy networks might move in the opposite direction to their international peers from time to time, it is not at all obvious that this is occurring at present. The ‘observed’ difference may, in fact, just be an anomaly in the data whereby the limited Australian sample is giving odd results that are not reflective of the ‘true’ beta at a time when the market has faced a barrage of challenges.

3.3.       Cost of debt

The AER has reached a sensible outcome on the cost of debt. We agree that it should retain its current approach of using indices developed by reputable third parties to estimate the cost of debt for BBB+ rated 10 year debt. We have said this in previous submissions so do not repeat it here.

We do, however, want to respond to a recommendation made by the Independent Panel. On page 45 of its report, the panel recommends that the AER gives “further consideration to using the EICSI as the primary source of data relating to credit spreads and using the Yield Curve approach as the crosscheck”.

In our view, the Independent Panel is mistaken in giving greater weight to the EICSI for two reasons:

  • First, the index is still very much in its infancy. It is too early to say whether it reflects the cost of debt that a benchmark efficient entity would incur. It has been heavily influenced by unique corporate activity, such as takeovers, and it does not reflect the efficient financing practices assumed in the AER’s trailing average approach.
  • Second, using the EICSI in the way recommended by the Panel would have no practical effect on the cost of debt estimates at the present time as the two measures are not statistically significantly different from each other.[28] However, it would lead to a benchmark that incentivises networks to continually find ways to revise their debt financing approach by shortening tenor even if other networks do not. This means that forces other than the efficient debt financing practices may come into play.

On the second point, in the short term, this may be advantageous because the allowed cost of debt for all networks would reduce, lowing consumer prices. But, in the longer term it could lead to significant financial challenges for regulated networks, including:

  • Risk of permanent under funding. The reality is that not all networks face the same financing costs. This is reflected in the different credit ratings faced by Australian energy networks and the costs of debt they have been able to realise. Some networks form part of larger groups that benefit from economies of scale. Others have implied parental support that gives them a ‘halo effect’, while some can access cheaper government debt. Many networks also use debt to finance unregulated business activities and working capital.

This means that some networks will simply not be able to achieve the same reductions that others can. And, if they cannot, then this will lead to permanent under funding of their efficient debt financing costs.

  • Potential for increased refinancing risk and higher volatility. To the extent that such benchmarking incentivises networks to take out shorter tenor debt, then this will increase refinancing risk as debt needs to be refinanced more regularly. This will both increase volatility of the cost of debt allowances (and therefore prices) and expose regulated networks to greater risk of a mismatch between cost of debt allowances and their actual debt financing costs.

For these reasons, the AER should reject the panel’s recommendation.

3.4.       Gearing

We agree with the AER’s proposal to retain gearing of 60%. That value appears consistent with market data and there is no obvious reason to change from the value contained in the 2018 RoRI.

3.5.       Gamma

Although we agree that there is no case for changing the distribution rate and utilisation rate inputs at this time, we do want to pick up a couple of points raised by the Independent Panel.

  • First, the panel quite rightly notes that the AER is being a little inconsistent when it estimates the distribution rate and utilisation rate to 2 decimal places and then combines them to get a gamma estimate to 3 decimal places.

We agree with the panel that the AER should round gamma to 2 decimal places. It should do so in the final step rather than an interim step, which gives a value of 0.57 using the values in the explanatory statement (0.887 x 0.646 = 0.5730… rounds to 0.57).[29]

Even if the AER opts to round gamma to the nearest 0.005, it should do the same with the distribution rate and utilisation rate to ensure consistency. This would give a value of 0.575.

  • Second, the panel raises an important question about the choice of data used to estimate the utilisation rate.

The AER has decided that it will only rely on national accounts data from the Australian Bureau of Statistics (ABS) about the extent of foreign ownership. It explicitly ignores data from the Australian Tax Office (ATO) about the actual use of imputation credits.

In our view, the AER’s reasons for not placing weight on the ATO data is weak, which the panel alludes to. The panel concludes that: [30]

Before ruling out the use of ATO data, the AER should seek to gain a better understanding of the difference between the utilisation estimates based on the ABS and ATO data.

Putting all this together, the Panel recommends that the AER engages further with the ATO to gain a better understanding of any data issues that may have a bearing on the accuracy of this sources of information in order to generate estimates of the Utilisation Rate in which it has greater confidence.

We agree, especially given that equity ownership statistics are not a direct measure of the utilisation of imputation credits (unlike the ATO data).

We encourage the AER to pick up these points when developing the final 2022 RoRI.

3.6.       Cross-checks

We support the ENA submission on cross-checks.

In our view, the AER has gone someway to incorporating cross-checks when developing the draft 2022 RoRI. We have previously encouraged the AER to do so and do not repeat those submission here.

We recognise that this is an evolving area. We have some reservations with how the AER is interpreting the RAB multiple information. We also consider that the AER could improve its scenario testing by looking at a wider range of scenarios, including those that try to ‘break’ the RoRI (i.e., give results that don’t make sense) as a way to better understand its limitations.

We look forward to further engaging with the AER and other stakeholders on cross-checks, including those that could be considered in individual revenue determination processes.

 

Attachment A: CEG Report

 

[1]    ENA, Rate of Return Instrument Review: Response to AER’s Draft Decision, September 2022.

[2]    See Section 30D(3) of the National Gas Law, which says that the AER should only make the RoRI if satisfied that doing so will or is most likely to contribute to the achievement of the national gas objective to the greatest degree.

[3]    See: Application by ElectraNet Pty Limited (No 3) [2008] ACompT 3 at [15]. This quote was endorsed by the Federal Court, including in Australian Energy Regulator v Australian Competition Tribunal (No 2) [2016] ACompT 1, 2, 3 and 4 at [77]-[81].

[4]    See: AER, Draft Rate of Return Instrument: Explanatory Statement, June 2022, p.56.

[5]    See: AER, Draft Rate of Return Instrument: Explanatory Statement, June 2022, p.57.

[6]    Schmalensee, Statement of Richard Schmalensee, Ph.D To the Australian Energy Regulator, 29 July 2022, pp.11–12.

[7]    For instance, at page 94 of the explanatory statement, the AER describes what it considers investors should require: “Our practice of resetting the allowed rate of return on equity at each regulatory determination affects the profile and riskiness of regulatory cash flows. In turn this impacts the expected return investors require.” (emphasis added).

[8]    See: AER, Draft Rate of Return Instrument: Explanatory Statement, June 2022, p.95.

[9]    See: AER, Draft Rate of Return Instrument: Explanatory Statement, June 2022, p.107.

[10]   See: AER, Draft Rate of Return Instrument: Explanatory Statement, June 2022, p.107.

[11]   AER, Draft Rate of Return Instrument: Explanatory Statement, June 2022, p.105.

[12]   For instance, the AER uses expectations theory to conclude that an investor only needs to a 5-year rate if it can expect that the allowed return will be reset in 5 years. It also re-examines the NPV = 0 principle and analysis undertaken by Dr Lally.

[13]   See: AER, Draft Rate of Return Instrument: Explanatory Statement, June 2022, p.100.

[14]   AER, Draft Rate of Return Instrument: Explanatory Statement, June 2022, pp.103,104,109,110.

[15]   AER, Draft Rate of Return Instrument: Explanatory Statement, June 2022, pp.107–109

[16]   AER, Draft Rate of Return Instrument: Explanatory Statement, June 2022, p.100.

[17]   One reason for this may be that the AER appears to focus on the characteristics of risk-free assets (i.e., government bonds as a proxy) and assume that the risks faced by equity holders are somehow similar (e.g., using expectations theory and term premia – p.100). Equity is quite different to bonds, and we are concerned that the AER has glossed over the distinction in its analysis of term.

[18]   QTC, 2022 Rate of Return Instrument: Rate of return information paper and final working papers, 11 March 2022.

[19]   ENA, Rate of Return Instrument Review: Response to AER’s Draft Decision, September 2022, pp.53–63.

[20]   Schmalensee, Statement of Richard Schmalensee, Ph.D To the Australian Energy Regulator, 29 July 2022.

[21]   CRG, Advice to the Australian Energy Regulator: CRG Response to the AER’s December 2021 Information Paper, March 2022, pp.46–58.

[22]   CEG, Critique of AER estimate of a 5-year RoE, September 2022, p.3.

[23]   CEG, Critique of AER estimate of a 5-year RoE, September 2022, p.3.

[24]   APGA, APGA Submission to the AER: Rate of return omnibus papers, 3 September 2021, p.31.

[25]   AER, Draft Rate of Return Instrument: Explanatory Statement, June 2022, p.18.

[26]   Independent Panel, Independent Panel Report: AER Draft Rate of Return Instrument, 2022, p.7.

[27]   APGA, APGA Submission to the AER: Rate of return final omnibus paper and information paper, 11 March 2022, pp.14–16.

[28]   We have previously considered similarities and differences between the EICSI and the benchmark yields published by third party data providers in our submission on the AER’s cost of debt working paper. See: APGA, APGA Submission to the AER: Draft working paper on Energy Network debt data, 14 August 2020, p.6.

[29]   AER, Draft Rate of Return Instrument: Explanatory Statement, June 2022, p.255.

[30]   Independent Panel, Independent Panel Report: AER Draft Rate of Return Instrument, 2022, p.48.