“Outcomes” is a buzzword that has in the past few years become popular in the energy sector; so popular, in fact, that many veterans of the industry gag every time they hear it. As with other buzzwords, it gets to the point where people are asking “Does that even mean anything?”
Well, what if we made “outcomes” actually mean something? What if, just as in competitive markets, utilities only made money if they delivered outcomes that their customers valued, and for which they were willing to pay. What if profits were driven, not by a “regulated return on equity”, but by outcomes?
Regulated utilities currently make money by spending as much of the ratepayers’ money as possible on capital projects. More spending = more profits. It is a direct relationship. For years regulatory theorists have wondered whether this provides the right signals for efficient utility operation. It doesn’t, of course, and everyone knows that it doesn’t.
There is no other profit for the distributor. The system is specifically designed so that they can only make money if they invest in capital expenditures, and profit is entirely driven by the amount spent and the long-lasting nature of the assets. (Technically, they can also make money by over-forecasting operating costs, then keeping actual costs lower to retain the net difference, but regulators try to prevent that, sometimes even successfully. Over-forecasting exists, but it is not an accepted part of the system.)
You could legitimately ask the question: do the customers really want to reward utilities with increased profits for higher levels of capital spending? Is that what the customers value – big spending?
Maybe it is time to get rid of return on equity altogether.
To understand the problem, you have to go back to the mid-1800s, when courts and legislatures decided that, if a company is to provide a public monopoly service (at that time, railroads), they should be allowed to recover in regulated prices the cost to provide the service, including a reasonable rate of return on the capital employed.
The basic structure of cost of service regulation, as it developed over those first fifty or so years, thus had two steps. First, you determine how much the utility needs to recover from ratepayers to meet its cost to serve them, called the “revenue requirement”. Second, you split up that cost between rate classes, and between customers within those rate classes, to produce rates. If you add up the rates charged to all customers for a year, the total dollars should equal the revenue requirement. Simple, right?
There is no concept of “profit” in this model. This was actually quite sly, although one can only speculate whether that slyness was intentional. As the cost of service model developed, the idea grew that the cost of capital was just another cost, like any other. You need to pay the employees. You need to pay the suppliers of raw materials. And, you need to pay the debt and equity financiers who provide the capital to build the enterprise.
To calculate the revenue requirement under the cost of service model, you determine three components:
- The annual operating costs. This is the recurring costs of the business, like rent, salaries and wages, etc. This is recovered dollar for dollar from the customers in rates, usually on a forecast basis.
- The capital costs. This is the annual additions to the infrastructure used to provide the service: wires, transformers, buildings, computer software, etc. These expenditures are recovered annually over the period each asset is useful. If a building is expected to last for 50 years, 2% of the cost is recovered each year for those fifty years.
- The cost of capital. Since you don’t recover the capital costs from customers immediately, you have an additional cost to finance them, just like the mortgage on your house. The cost of capital in turn has three parts:
- The interest cost on debt used to finance the capital spending. This debt is typically 50-70% of the cost of the assets.
- The amount equity investors expect to earn on the rest of the financing (preferred or common shares), both in dividends and in retained earnings in the company (called return on equity, or ROE).
- Income tax on the second component, which is considered profits for tax purposes.
Just because the amounts earned on equity are “profits” for tax purposes, and “net income” for accounting purposes, doesn’t change their treatment in the regulatory world. Today, if you deign to call the amounts earned by utilities on their equity capital profits, utility accountants will have hissy fits. Those are not profits, they will say. They are a cost of doing business, in this case the cost of risk capital. (Sometimes they will wear garlic around their necks, to ward off your obvious vampiristic tendencies.) This is all ridiculous, of course, but ROE as a cost is a well-accepted concept in cost of service regulation.
Notice that in the cost of service model prices have no reference to what customers want, need, or value. In the competitive markets (yes, yes, I know I’m oversimplifying), prices are driven by what customers are willing to pay for a product or service. Companies are, for the most part, price takers, meaning that prices are not primarily driven by their costs. The markets tell them their prices. They have to control their costs accordingly, if they want to make a profit.
Not so with utilities. Their prices are pretty well entirely driven by their costs.
The two fundamental problems with cost of service regulation are therefore a) it incents capital spending, and b) it doesn’t reflect customer value.
In the 1980s, Dr. Stephen Littlechild of England (now a Fellow at Cambridge Business School) essentially invented the concept of price-cap regulation. The purpose of the price-cap concept was to decouple prices from costs, but it was really only a baby step in that direction.
Under price cap systems, you still start with an initial cost of service calculation. This step – called “rebasing” – is done in the normal way, with prices for one year set entirely based on the utility’s costs to provide the service. No component of the process involves what the customers need or value.
What separates price cap from normal cost of service is that, in the next two to seven years after the rebasing, prices are based formulaically on a calculation of how costs should rise (or fall, in theory) from the prices set on rebasing. The most common formula has three parts: inflation, productivity, and stretch. Inflation is a measure of cost increases in the economy, whether costs specific to utilities, or calculated on a broader basis. Productivity is typically a calculation, from past experience, of how much economies of scale, technology, government policy changes, and other factors cause utility costs to rise at, on average, a lower (or higher) rate than inflation. Stretch is said to be a ratepayer benefit reflecting the greater business freedom the utility gets under price cap. In practice, it is a calculation of how much room an individual utility has to be more efficient than the productivity level of other utilities. However you characterize it, it is quite small.
The three components of the price cap model bring in a concept called “yardstick regulation”, which means that a utility’s costs are measured, not only self-referentially (i.e. based on what they want to spend), but also by what other utilities in a similar situation have spent or will spend. It adds a more objective component to cost-based regulation.
What is striking is that price cap regulation is still all about costs. Instead of using bottom-up forecasting to calculate costs, price cap regulation uses top-down methods – econometrics, for example – to predict what costs should be for a utility, given where its costs are today. There is an advantage, because this more rigorous approach provides an empirical context, something missing if utilities just say what they need, and expect to get it.
However, it is still entirely about costs.
And what about the things the customers value? Certainly, they are considered by the regulator anecdotally in reviewing the activities of the utility, but they play absolutely no part in the calculation of utility prices. None.
The Real World
Let’s take Ontario as a real world example.
In 2014, Ontario’s seventy-odd electricity distributors recovered about $3.5 billion in charges to their customers (excluding transmission and commodity charges, which added another $13.3 billion, but are a wholly separate and largely unrelated discussion). This distribution bill works out to an average of just over $700 per customer. Of the $3.5 billion of revenues, about $1.7 billion was spent on annual operating costs, $800 million on depreciation of capital assets, and $400 million on interest. The result was net income (before taxes) of about $600 million. Taxes were only $40 million, leaving more than $550 million in net profits.
In that same year, the distributors spent $1.9 billion on new capital assets, which will be recovered, with interest and profits, over an average of 33 years. The cumulative capital base of the distributors (called “rate base”) was about $17.5 billion.
Distributors also paid about $300 million in dividends to their shareholders in 2014, mostly the province of Ontario and various municipalities. This is not treated as tax revenues by the shareholders, even though they are most governments. In this respect, the shareholders are like normal investors, making a return on their investment.
On the gas side, the three regulated gas distributors are privately owned. Between them, they had 2014 distribution revenues from customers of about $2.0 billion (just over $500 per customer on average). This was spent $900 million on operating costs, $450 million on depreciation of capital assets, and $300 million on interest. The remaining $350 million was net income, on which taxes of about $40 million were paid.
Like the electricity distributors, the gas distributors spent lots of money – $1.1 billion – on new capital assets, with a similar average recovery period of around 25 years. Their combined rate base was about $8.5 billion. They paid $200 million in dividends to their private sector owners.
Then we have the electricity transmitters, mainly Hydro One, also regulated in the same way. They collected another $1.6 billion from customers in 2014, made up of $500 million in operating costs, $400 million in depreciation, $300 million in interest, and $400 million in net income, on which about $60 million in taxes were paid. Their $11.0 billion of rate base had additions of more than $1.0 billion in 2014.
If you add up all the “wires and pipes” utilities regulated in Ontario, the numbers are big. Current totals (i.e. for 2016) are roughly:
- Amounts charged in rates to customers: $7.5 billion per year
- Operating costs: $3.3 billion
- Depreciation of capital assets: $1.8 billion
- New capital assets: $4.2 billion
- Total rate base (capital deployed): $40.0 billion
- Interest costs: $1.0 billion
- Profits/Net Income: $1.4 billion
- Taxes paid or payable: $150 million
- Dividends paid: $700 million
(This doesn’t include Ontario Power Generation, a regulated generator, which adds almost as much again.)
On average, customers pay about $1,700 per year for these wires and pipes infrastructure services, yet at no time are the prices for those services based on what the customers want, and the value the customers place on what they are getting. The prices are entirely driven by the costs to provide the services. They are utility-centric, not customer-centric.
What if we changed that?
An Outcomes-Driven Model
With all the yapping in Ontario about “outcomes”, we still don’t use outcomes to set rates. We just talk about outcomes. Nothing more. “Engage” with your customers, give them the outcomes you tell them they want, and by the way let’s continue to set rates based on utility costs. Outcomes delivered, or not delivered? It doesn’t matter to the utility, at least not financially.
The regulator is supposed to act as a “market proxy”, mimicking the effects of a competitive market to control the prices of each monopoly utility. That’s not what happens right now (markets don’t set prices based on budgets or wish lists), but thinking like a competitive market may help us deliver an outcomes-driven approach.
Is there a model that sets rates based on outcomes to customers? The answer is no (although RIIO in Great Britain now tries to move a few steps in that direction), but there could be.
To test out the hypothesis, let’s set up a possible model to see if it can work. It’s easiest to look at this in the context of electricity distribution. You can do this as well for electricity transmission, and for gas distribution, but both are somewhat more complicated. Electricity distribution is simpler, plus there are around seventy distributors, so there is a sufficiently large sample.
Start by getting rid of ROE. Return on equity is not outcomes-driven; it is spending-driven. Get rid of it. Then, with a clean slate, look for a new way of earning profits.
If you want an outcomes-based price structure, two things need to be determined. First, what measurable outcomes do customers want from their energy utilities? Second, what are the amounts customers are willing to pay for those outcomes? Assuming you are able to measure those two things, the customers can become the market.
So, what do the customers want? At the most basic level, we already know that customers want three things more than anything else:
- Connection to the electricity grid at the lowest possible cost.
- Reliability in their electricity supply.
- Customer Service that is efficient, knowledgeable and customer-focused.
Can each of those be measured and valued?
Connection. The first of these, the connection itself, is in some respects the simplest. In Ontario, electricity distributors have an “obligation to serve” which means that they are required to connect customers who are within their service territory. It doesn’t have to be a great connection, but people in Ontario have a right to an electricity connection. (Thank you, Sir Adam Beck.)
The price of the bare-bones connection is not zero. Just as the crappiest cellphone is not free, so too just being connected, before you measure the quality of what you are getting, still has to cost something. What is a reasonable price for a connection, with below standard customer service and reliability? This is likely to be hotly debated, but in the end there is a judgment call to be made here. There is a (theoretical) stripped-down price. Everyone who wants to be connected, pays this.
A logical place to start might be the stripped-down cost of a distribution system, without any profit, and without any shareholder dividends. It is possible to identify the lowest-cost distribution systems in the province (we do this already), and calculate the prices for customers based on those costs, but without the profit component. That’s only one possible pricing approach. The point is it is binary. You want the service? This is your basic cost.
(I would actually suggest using a RIIO-like approach to basic cost, drawing on the concepts developed by Ofgem in Great Britain. That is, establish a total annual spend – totex – that is both capital and operating costs for a bare bones utility, and standardize the calculation of annual revenue for basic totex, with no profits included, but including a risk-free time value of money. In economic terms, this would be a standardized calculation of the low cost frontier. Individual utilities could approach their capex/opex breakdown differently from year to year, and spend more or less than the standard totex, but their basic revenue requirement, without performing on the other things customers value, would be a standard amount. However, that is probably for another article.)
Reliability. Regulators and utilities (and customers, for that matter) have developed a number of metrics to measure the level of reliability of electricity service. These metrics basically measure one of three things: how often is your service interrupted; how long are interruptions; and what is the quality of the power you get when you are not interrupted (e.g. voltage control, or momentary power flickers). Terms like SAIFI, SAIDI, CAIDI and MAIFI – the electricity sector equivalent of speaking Klingon – relate to specific types of reliability measurements.
What distinguishes these and other such reliability metrics is that they can be defined clearly, measured accurately, and compared from one distributor to another. It is also possible to find out from customers how much they value incremental improvements or reductions in these metrics.
For example – and this has been done in many places – one can ask customers of various classes a question such as the following:
“You currently have one electricity outage every eight months, with an average duration of 1.6 hours. How much extra – in addition to your current distribution rates – would you be willing to pay each year for:
- Reducing the average duration of outages from 1.6 hours to 1.2 hours?
- Reducing the frequency of outages from one every 8 months to one every 10 months?”
With sufficiently robust polling, it is possible to develop a customer value curve for these metrics, based on existing frequency or duration, and increments of improvement. Similar questions can be developed that price the bill reductions customers would expect if reliability metrics were to decline by a given amount. (Just a head’s up: it’s not symmetrical.) It should even be possible to identify regional variations in willingness to pay for particular reliability results.
The two other pieces of information the regulator can have relating to reliability metrics are industry average performance, likely adjusted econometrically for different business conditions, and frontier (perhaps measured as top decile) performance. The former identifies the top of the reliability bell curve, while the latter generates a target performance level.
The industry continues to develop new metrics – WPF (worst performing feeder), for example – which with some refinement could also be tested for customer value, and compared between utilities for various aspects of reliability.
Whatever the reliability metrics selected (as verified by customer value analysis), a component of the profits of the distributor can be calculated based on those metrics.
One approach would be to establish a modest price/profit level for achievement of the average industry performance levels for given reliability metrics. Then, the customer value curve can be used to interpolate prices between the average performance level, and the top decile performance level. Similarly, the negative value curve can be used to interpolate prices below the average performance level.
For example, suppose that you determine the value of average reliability for province’s 4.5 million residential customers is $20 per year in profits to the distributors (i.e. $90 million a year). Further suppose that residential customers have said that improvements to reach the top decile would be worth a further $20 a year to them. (e.g. “I will pay an extra $20 a year to have one outage a year instead of two, and 1 hour total instead of 3”, although this is much oversimplified) A distributor with 50,000 residential customers would thus get $1 million per year in profits for achieving average reliability, and that would scale up to $2 million per year as they move towards top decile reliability. The scale would not necessarily be linear. It would depend on the value curves learned from the customer surveys.
The simple result would be that customers incent distributors to deliver results that the customers value. If they want to make a profit, and pay dividends, they have to deliver favourable outcomes for the customers.
In addition, different customer classes can value reliability differently, and pay accordingly.
Customer Service. The same concepts can be applied to customer service metrics. There are already a number of metrics used to assess customer service. More could be added, or they could be modified, by finding out from the customers what is important to them, and how much those things are worth.
For example, how important is telephone response? The answer may differ from one distributor to another, and it may already be changing radically in some places. In the Powerstream franchise area, for example, it may now be true that the value of website functionality – the ability to interact comprehensively with the utility online – is sharply increasing, while the value of improved telephone response times is similarly declining. It may be that customers would be willing to reward the utility more for website improvements than for installing a new IVR. In the Bluewater franchise area, the opposite may be true.
It should be possible to identify a suite of customer service metrics for which customer value curves can be determined. Then, as with reliability, performance at an industry average level could generate a modest profit, while performance at the top decile could generate a substantially higher profit.
Innovation and Distributor Generated Outcomes. In addition, utilities could periodically identify customer outcomes (based on customer engagement or public policy goals or other factors) with specific metrics, and specific profit factors for achieving those metrics. This would work something like DSM incentives in gas, except that in every case the measurement would have to be empirical in nature.
For example, a utility could propose a profit factor if weather normalized average use per residential customer, adjusted for GDP or some other economic factor, declines by a specific amount. The same could be true of other outcomes valued by customers, like connection of renewables, or increased undergrounding, or demand response.
The intention, however, is to encourage the utilities to be innovative. A utility might seek to incent customers to buy, in the market, in-home displays to monitor their energy-use. The utility could then be rewarded by reference to increases in the market share for these products in their service territory. Utilities might join together to offer services, and be rewarded for successful cross-franchise results (such as cost reductions). A utility that comes up with a new product or methodology (for cost control, for example) could be rewarded based on the number of utilities that adopt that model. Utilities could come up with innovative rate structures (prepaid service, for example), or any number of other things that right now are considered fixed in stone.
In each case, the utility would propose an improved outcome for customers, a metric or metrics for measuring it, including all safeguards needed to make it actually reflect the outcome, and a profit factor that reflects the value to the customer. Each would have to be supported by evidence that the customers want to pay for the outcome proposed. The utility would also propose the shareholder investment expected, as well as the split between shareholder benefit and ratepayer benefit (how is the value shared). More shareholder money at risk means more shareholder participation in the benefit, and vice versa.
What’s the End Game?
Over time, the goal would be for utilities to replace all of their current ROE, and even some of their hard cost recovery, with outcomes-driven revenue/profit factors, both short-term and long-term.
The new approach outlined above would replace the current cost-driven revenue requirement with utility revenues made up of the following components:
- Standardized Basic Cost. This is a unit cost per customer for each class based on the bare bones costs from the lowest cost distributor in the province. A utility would recover a location-specific version of this just for providing the basic connection to the customers. This would likely be 60 – 70% of the average revenue requirement today, i.e. $2.0 to $2.5 billion. To the extent that an individual distributor can control their costs better than others, this standardized basic cost would cover a larger percentage of their actual spending.
- Reliabilty. A modest level of recovery would be provided for distributors that maintain the average level of reliability in the province. Based on a customer-driven incremental price curve, this level could increase substantially. For example, the revenue for average performance could equal about 5-10% of average industry revenue requirement under cost of service, and performance at the top decile could equal 15-20% or more of the cost of service equivalent. Reliability metrics would be adjusted between distributors based on exogenous business conditions.
- Customer Service. As with reliability, recovery at a low level would be available for maintaining industry average customer service metrics, again perhaps in the range of 5-10% of current cost of service. The value curve would then be applied to calculate increases in rates available for performance up to the top decile, which might be 15-20% of current cost of service. Customer service metrics would also be adjusted for exogenous business conditions, but would be more likely to see regional and local differences due to different customer value curves.
- Distributor-Specific Profit Factors. Individual distributors, or groups of distributors, can propose additional customer valued services that they wish to offer, with comprehensive supporting evidence on measurement, and what customers are willing to pay for that additional value.
Today, just over 80% of industry revenues come from costs other than profits, and just under 20% come from profits. Under this outcomes-based approach, all of the profit component, and some part of the hard costs, would be recovered only by delivering outcomes valued by customers. No longer could distributors improve their bottom line simply by spending ratepayer money on capital assets. Distributors would be incented to be nimble, and to manage actively.
A distributor that is average in every respect, and does nothing else to improve outcomes for customers, would likely recover their out-of-pocket costs, but would not make a profit and could not pay any dividends. As with the competitive markets, profits would go to those who deliver what the customers want, and value. For a top performer, profits could be substantially higher. Lack of performance would equate to lack of profits.
And, as with the competitive markets, top performers would have an ability to purchase weaker, less profitable companies. The shareholders of those weaker companies, of course, would be incented to sell (no dividends), and M&A values of distributors would reflect their profitability, which would in turn be driven by how well they serve their customers.
Is this the answer? No. (Duh.)
This is an idea, presented to generate debate and further ideas. There are many ways of approaching this, of which this is just one. A number of other jurisdictions are experimenting with changes to rate-making that move in this direction (although for the most part they show great trepidation in doing so). Further, in this, as in any approach to the problem, the devil will most certainly be in the details.
However, the point is a broader one. Maybe it is time we stopped scoffing whenever the word “outcomes” comes up. If we really believe that energy regulation should follow the market proxy concept, perhaps we should start drawing on the goal of “outcomes” to produce prices, and a regulatory paradigm, that really do mimic the competitive markets.
Get rid of ROE, and allow only profits based on customer value.
Walk the walk, instead of just talking about it.
– Jay Shepherd, May 31, 2016