I then calculated an average price per 1,000 gallons for a household using 7,500 gallons/month (that's about 250 gallons/day or 930 liters/day), which works out to about 100 gallons/day for a household of 2.5 people -- roughly the same consumption of San Francisco residents.
A simple statistical analysis shows that consumption is positively correlated with fixed costs and temperature and negatively correlated with variable costs and precipitation.
The rationale behind temperature and precipitation is simple: people use more water (on lawns) when they live in arid places.
I cannot think of a good reason for the consumption-fixed cost relationship, except that wealthier cities may spend more on infrastructure for people who use more water.
The negative relationship between price and consumption is obvious (to an economist): people use less water when it's expensive.**
All of these variables impact consumption in different ways, but their weighted impact can be estimated by running a multivariate regression. When I do this,*** I find that consumption and price effects persist, but temperature and precipitation effects do not.
These results are easy to interpret when you add a little common sense. An increase in fixed charges is associated with an increase in consumption, but it's unlikely that raising them will cause consumption to increase. There's an underlying, omitted variable that's driving both results.
The effect of price is more straightforward: higher volumetric prices are associated with lower consumption. Indeed, they probably reduce it.
For more on this, check out American Water's 2012 survey of prices. There, you will see that Las Vegas prices are half San Francisco prices. Where's the most expensive water in the US (in that survey)? Seattle and Portland. That's because the price of water service reflects the price of infrastructure, not water scarcity.
Bottom Line: The law of demand holds. The easiest way to lower water consumption is raise prices.
* Hey, Daly City! Chill out! You don't need 11-step increasing block rates [pdf]. Try two, like San francisco or -- better -- one. Then customers can understand what's going on. Speaking of that, I dropped two cities with budget-based prices, because it's impossible to know the price of water for an average customer. I hope managers are setting them right (and fairly) because no outsider will even know!
** It's more accurate to compare the cost of water service in terms of fixed costs plus variable costs for 7,500 gallons of monthly use, but I just looked at variable charges at that consumption. See the XLSX for service costs per 1,000 gallons.
|reg pcc fc volume temp precip, ro|
|Number of obs||53|
|Prob > F||0.0001|
|Per capita consumption||Coef||t||P>t||[95% Conf.||Interval]|