This morning, I woke up to be extraordinarily irritated by this infographic from “energy expert” Chris Yelland on

Thanks (?)
Thanks (?)

It is POWERFULLY misleading.

*explodes in frustration*

I am all for stating the severity of a problem – but overstating it is just so irritating. Especially as it gets the racist trolls all frothy at the fingertips.

Consider what he is effectively saying:

  • Nominal Gross Domestic Product is around R3.5 trillion.
  • Growth is forecast to be around 2.1% for this year (so R74 billion) AFTER taking into account the load-shedding saga.
  • But the economists don’t know what they’re talking about.
  • At the rate we’re going, 12 months of Stage 1 Load Shedding is going to cause a R240 billion shortfall.
  • So we’re actually heading for a 7% contraction.
  • And all you need is 20 days of Stage 3 load-shedding before we’re in deep recession not including the normal Stage 1 and Stage 2 load-shedding stories.

These bullet points above?

They are a reasonability check.

The estimates here are not reasonable.

So Where Do The Numbers Come From?

It’s actually quite simple:

  1. In the late 2000s, the National Energy Regulator published an assumed “Cost of Unused Energy” (COUE) figure of R75 per kWh (here).
  2. Chris Yelland upped it to R100 per kWh (6 years of inflation, I guess?)
  3. He then took 10 hours of load-shedding per day, for 20 days of the month, and multiplied it by the amount of electricity not-being-supplied at each stage of loadshedding.
  4. 1,000 kW per hour in Stage 1
  5. 2,000 kW per hour in Stage 2
  6. 4,000 kW per hour in Stage 3

Seems reasonable so far.

Except for two things:

  1. What is this Cost of Unused Energy figure? and
  2. Awkward assumptions

The COUE (Cost of Unserved Energy)

From my brief rummage through the interwebs, there are two reasons that the COUE is calculated.

Primary Reason: The Expensive Short-Term Supply Story

This reason is the one that is more in line with what the Department of Energy actually thinks COUE is (see here).

So you know how Eskom has reserve facilities and the ability to generate spare electricity with more-expensive diesel? Or, at least, you remember how it used to have those back in 2007 before we ran out of capacity?

Well, as it turns out, they did. Which meant that each time Eskom faced an emergency maintenance issue, they had to ask themselves this question:

Do we generate more electricity, or do we let the lights go out unplanned?

The trouble with that question:

  1. They know what the cost is to generate more electricity.
  2. But what is the cost of letting the lights go out?

Solution: the COUE figure.

It’s a parameter to try and work out how much it would cost customers to not have electricity (by surprise) for a bit.

Secondary Reason: The “Do We Build More Power Plants?” Question

There are some reports that suggest that the COUE is used to determine the feasibility of long-term infrastructure investment.

To my mind – this is a bit crazy. Why would you use the short-term cost of unplanned outages to determine the feasibility of long-term infrastructural investments? You’d surely rather be using some estimate of the long-term cost of not-having-enough-electricity-for-long-periods. But this could just be some confusion of terms/definitions.

Either way, getting a COUE figure is difficult

Obviously, the cost to a household of not having electricity for a day is not the same as the cost to a factory. And these are both equally not the same as the cost to a farmer.

Which is the quick way of saying: a general number is almost impossibly unreliable, even if you found a way to measure it.

And as for the measurement itself, the international best practice predictor is to go out to some electricity consumers and say:

Hey there, client! Quick q: let’s say that you got a surprise power cut for an hour. What d’you reckon is the cost to you of not having that electricity, including the cost of any repairs you might have to make to damaged machinery?


I am not joking about this. That is the approximate method.

The other option is to take GDP for a particular year and divide it by the total number of kW hours used. Which gets you to a cost of around R10 per kWh here in SA – which is sort of considered the lower bound of the COUE range (but it’s also unreliable because of multiplier effects and so on).

The real point is: the COUE figure is prone to overstatement. Both because of how it’s measured, but also because higher COUE figures are politically expedient (the higher the COUE figure, the more often Eskom has to prevent power interruptions).

So it’s really not a great estimate for the cost of load-shedding.


Assumption 1: The Passive Economy

Allow me to use an analogy here (I used it in a Moneyweb comment this morning, so forgive me if it’s repetitive):

  • If you get caught in a freak rainstorm, then you get wet, and that’s a cost; but
  • If you are in a country where it rains frequently (if not always predictably), then do you just wander around getting rained on, or do you wear a raincoat and carry an umbrella?

You obviously wear a raincoat and carry an umbrella (unless you like getting rained on).

But given the load-shedding situation that’s due to be around for the next three years at least – Chris Yelland is assuming that people are just going to sit back and be cut off.

Does that sound right?

Hell no. They’re buying generators and inverters and converting their ovens to gas.

Of course they’ll be more expensive to operate, and some jobs will be lost, and some productive processes will be abandoned. But it’s not going to be like living through a surprise power cut indefinitely – because it’s not going to be a surprise. It’s going to be three years of planning.

Assumption 2: One man’s cost is a cost to the economy

Only: one man’s cost is another man’s income.

If money is spent on repairs, then there is a repairman that’s going to be receiving that money.

If a generator is installed, then there are generator suppliers and installation electricians that are going to be paid for it.

That variety of cost is economically-neutral in a lot of ways.

The better approach is to say: “one man’s lost production is a loss to the economy.”

But even then, the lost production in a completely unexpected electricity supply shortage is so much more than the lost production from load-shedding:

  1. When the blackout is completely unexpected, everything goes down.
  2. Under load-shedding, the lost production is only in those industries where the cost of self-generated electricity renders the production unprofitable AND it is not possible to delay or modify the production process to take into account periods without power.

Assumption 3: Load-shedded households are resulting in the same levels of lost production as industry


Not really true now, is it?

If any of the load-shedding is just turning off all the geysers and tumble-dryers that are happily pulling on the grid during the day, then there’s not really “lost production”.

It’s frustrating.

But it’s not the same as turning off factories and mines.

So as I see it, pay no attention to this infographic.

Things are not nearly that dire.

UPDATE: and if you want some evidence of the above, check out this article from Business Report.

Rolling Alpha posts opinions on finance, economics, and the corporate life in general. Follow me on Twitter @RollingAlpha, and on Facebook at