Payback for a 2,000 litre water tank: about four years

Update: I have a crow to eat. Playing citation hide-and-seek I find that the original benchmark included purchase, a pump, plumbing and installation. It's here: http://www.marsdenjacob.com.au/cms/index.php?option=com_content&task=view&id=53&Itemid=63 Thanks to Stilgherrian for getting the ball rolling (below). I have added other thoughts at the end of the post. 

Why do I let myself get sucked in like this? I could ignore this stuff on a Sunday, then someone points me at a dodgy statistic swallowed without chewing by an uncritical journalist, and away I go.

The dodgy statistic in question is this:

“Restrictions could also encourage ''inefficient'' measures such as rainwater tanks, which cost up to $2000 but held only $4 worth of water at 2011 prices, the report said.”

It's in this story about water pricing, and because the Sydney Morning Herald doesn't practice source-linking, you have to do a bit of Googling to find out that the report in question is this one, from the Productivity Commission (the quote is on page 184):

“For instance, a common 2000 litre household rainwater tank costing about $1500 to $2000 holds about $4 worth of water at current mains water prices.”

It's a very misleading statistic in two ways. First, I can't find any 2,000 litre water tank that costs that much – even an expensive under-deck tank can be had for about $1,300. Second, it assumes that the water in the tank is a static resource – hence the snitty crack about it containing $4 worth of water.

Sydney's price of water in 2011 was $2.103 per kilolitre (hence “about $4 worth”). If I run a fairly simple model, and use a $640 purchase price for the tank (from here), it's not a bad deal.

Here are my unrealistic assumptions:

1. The turnover of the tank is sufficient that each “fill” gets used.
2. The catchment is 8 meters x 8 meters (the roof area above my head right now).
3. 1mm of rain on 1sqm is 1 litre (this is assumed by every online calculator I see).
4. I used the Bureau of Meteorology Observatory Hill average rainfall here.

Under those conditions, that tank would deliver about $163 worth of water each year – paying for itself in around four years. And it's this that the Productivity Commission dismisses (with unsourced data and zero analysis) as “inefficient”.

Let's reverse the calculation. What do you have to get out of a $640, 2kl tank for it to achieve payback in five years? About 5 Kl per month – which means filling and emptying the tank 2½ times each month, or the equivalent of six days with better than 10mm rainfall per month.

Sure, there are other on-costs. Sure, this isn't a complete study. But even if I'm out by fifty percent, we get six-to-eight years of payback time for a water tank.

Addendum: As noted, the cost includes installation. However:

1. It's still silly for the PC to treat a water tank as $4 worth of static water.
2. Payback doesn't happen overnight, but it does happen.

I'd be interested to see what a better model would show!