The electricity death spiral

by

Paul Simshauser and Tim Nelson put out an AGL working paper last week that captured a few headlines. The ‘death spiral’ title probably helped! The paper is here.

They raise some interesting points. What if demand for electricity overall is falling but peak demand is rising? Looking at the components of electricity pricing they note that:

… current tariffs for households are about $260/MWh, with power generation (including carbon taxes) accounting for about 36% of the costs, high and low voltage network costs comprising about 46%, environmental schemes about 7% and retail supply costs about 11%.

So network costs are the largest component of electricity prices and these are driven by the ‘peak’. In other words, the transmission and distribution networks are designed and built to meet the peak load. There is ‘spare capacity’ at other times. So if peak demand is rising but ‘average’ demand is falling, this means bigger networks with more spare capacity most of the time. 

In a market, we would expect that the capacity required for peak use would be paid for by the peak users. In ‘off peak’ times capacity would be close to free (as there is an excess supply with spare capacity) with the ‘peak period’ facing high charges reflecting the additional capacity costs the peak users create ‘at the margin’. But that is not the way electricity networks are priced.

The main problem is that peak users do not face higher marginal network charges to reflect the additional costs that they create. Most peak users face a flat charge or a rising block tariff  based on total consumption over a period of time – not peak consumption. Smart meters were meant to solve this by providing real time price signals, but these have been coming for a long, long, time. And it is not clear that smart meters, without smart appliances connected to the internet, are going to do much. Further, a variety of state governments continue to regulate retail electricity tariffs for households, and these tariffs do not reflect the economic difference between peak and off peak network costs.

This creates an interesting problem for government and regulators. One way forward would be to try and set a fixed charge to consumers (both household and business) based on activities and assets that are correlated with ‘peak use’. So, consumers with bigger houses and more reverse cycle airconditioners pay a large annual ‘fixed fee’ as these items are correlated with peak use. The ‘fixed fee’ is used to cover network charges.

OK – this is a pretty ‘second best’ solution. There is a big installed base of ‘big houses’ and air conditioners so the fee would only work slowly over time. Further, it doesn’t focus on peak use, just things that we believe are associated with peak use. So it would catch the pensioner in a big house who can’t afford heating or cooling. So, other suggestions very welcome!

Oh – and why the ‘death spiral’? Well regulators set network charges on the basis of the network owners earning a reasonable return on their investment. And the lights have to be kept on. So a larger peak use means more network investment, higher total allowed returns and higher network charges. But these are ‘averaged’ over all users and time periods, with the higher prices potentially discouraging ‘off peak’ use more than peak use. So the problem gets exacerbated – with higher network charges spread over less average power consumption. Now, this is unlikely to lead to a ‘death spiral’. But it will lead to inefficient network investment and driven by inefficient electricity consumption.

6 Responses to "The electricity death spiral"
  1. What’s preventing charging more for power during peak hours? We’ve smart meters here in NZ that allow for it – we at least get differential charging for day and nighttime use, with no particular reason this couldn’t be split out further.

    Just put in a household meter listing current prices, expected prices an hour out, and let folks shift dryer use to off-peak.

  2. Eric – sounds like NZ is more advanced than Australia. In Victoria, smart meters are being rolled out and there will be an option for customers to move to flexible pricing (if they have a smart meter) from next year.

    For NSW a Fairfax article from late last year states that:
    “In NSW, Ausgrid has more than 400,000 first-generation smart meters installed, with 250,000 customers on ”time-of-use” contracts”.

    So overall smart meters are ‘in progress’ in Australia but the progress appears slow.

  3. ‘Time of use’ contracts don’t solve the problem. I’m currently on one, with electricity prices much higher between 2pm and 8pm on weekdays. So I normally try to reduce electricity use during these hours. However, on the hottest day of the year I’ll still be using my airconditioning. So ‘time of use’ reduces the average daily peak, but not the annual (which is what matters).

    The solution is to have smart appliances which can be turned off remotely and associated supply contracts where, for example, all the airconditioning for a building can be switched of for up to 20 hours (say) per annum (when load peaks).

    If meeting peak costs are really very expensive then such supply contracts could be so cheap as to make it worthwhile for some businesses to shutdown on such days. Even a limited rollout of such appliances could make a large impact.

  4. Meters that record peak, shoulder and off-peak usage separately and corresponding differentiated tariffs have been available in the ACT for at least 4 years.

  5. @Bruce: You’d still refrain from running other discretionary power sources while blasting the AC though, right?

    Our meters aren’t as cool yet as they could be. No facility as yet, for example, for broadcasting power prices. But it’ll come.

  6. I’m surprised no one has noticed that this is where rooftop solar panels fit in. They reduce the peak AND reduce the requirement for network investment because of their distributed nature. Another two arguments for using them. Perhaps some incentives should be based on these considerations?

%d bloggers like this:
PageLines