Cost-Benefit Analysis, Rep. McClintock, and the PPIC

Some readers have asked for elaboration on the McClintock/PPIC comparison in the last post.  Tom McClintock wrote this in a recent article in the Sacramento Bee.

as chairman of the House Subcommittee on Water and Power I have announced that all projects – including the Auburn dam – will first be evaluated under a uniform cost-benefit analysis that establishes the amortized cost of construction, and annual operations and maintenance balanced against the value of water, hydroelectricity, recreational leases and flood control protection afforded by these projects.

I like that McClintock is focused on cost-benefit analysis, and I especially like that he is emphasizing the importance of a uniform approach to it.  However, his description of cost-benefit analysis is not correct.  Cost-benefit analysis does not amortize costs into the future and compare them to future benefits.  This approach ignores the time to build, and is problematic when benefits are not smooth.  The correct way to do cost-benefit analysis is to estimate the full path of costs and benefits and discount them back to a single present value.  This is fundamental.  Consult any textbook, government guideline for confirmation.  Even Wikipedia has it right,

Benefits and costs are often expressed in money terms, and are adjusted for the time value of money, so that all flows of benefits and flows of project costs over time (which tend to occur at different points in time) are expressed on a common basis in terms of their “present value.”

If a project has a short build time and a smooth time series of costs and benefits, it isn't a big difference mathematically.  However, if it is a big project with a long build time before benefits appear and those benefits aren't smooth over time (i.e. dams, peripheral canal around the Delta), the error heavily biases the analysis towards making the investment.

So why does McClintock have me thinking about the influential PPIC water reports?

There are two key analysis by the PPIC/Davis group, both originally published in the 2008 Comparing Futures report.  In the one analysis, they evaluate whether a peripheral canal should be built around the Delta, a project long desired by water exporters.  In the other analysis, they evaluate whether Delta levees should be upgraded or repaired after a flood, investments strongly supported by Delta interests.

The PPIC/Davis team does not apply a uniform approach to evaluating these investment decisions. 

When it comes to Delta levee investments, they use the correct present value framework and even account for the lack of benefits during the construction period.  I have no problem with the framework they utilize here.   They conclude that in most cases, investing in and repairing levees is not economically efficient.  Their conclusions depend on the values they assign to benefits, costs, and flood probabilities, and those have values have been challenged by many, but that is outside this discussion of the framework.

When it comes to evaluating the peripheral canal, the PPIC/Davis group uses the incorrect, McClintock style approach that amortizes costs forward to the future for comparison to future benefits.  It is a much easier standard.  The approach ignores a 10-25 year build period when costs are incurred and no benefits are received.  And then, they choose a very distant future year to evaluate benefits (they say 2050, but their 2050 water demand looks more like 2080 or 2100) when benefits of a canal are estimated to be very high, ignoring the fact that benefits will be much smaller at first.

I can forgive Congressman McClintock and staff for not knowing the difference.  After all, ordinary voters are familiar with amortization, not discounted present value; and he is talking about a concept and not making calculations.  But the PPIC group certainly should know the difference, are making influential calculations, should apply a uniform approach.  Instead, they set up an inconsistent framework to evaluate these two investment alternatives, and thereby severely biased their analysis in favor of a peripheral canal before they even input a single number into the models. 

Is this splitting hairs?  Is it a big deal quantitatively? 

Consider a simple example, based on current cost estimates for alternative conveyance, and benefits of a conveyance as calculated in the 2008 PPIC report. 

Assumptions: 50 year analysis period, 5% real interest rate, peripheral conveyance costs $15 billion and takes 15 years to build so $1 billion in costs in years 1-15, operating costs $200 million annually from years 16-50, benefits of the conveyance are $2 billion 50 years from now, increasing by $50 million per year to reflect the growing demand and growing risk of through-delta water supply interuptions from flood.  In other words, I set the benefits at $350 million in year 16, and increased them in a linear fashion to $2 billion 50 years from now.

Results:

Incorrect PPIC/McClintock analysis:  Amortized capital costs are $817 million + $200 million operating costs = $1.017 billion in costs in year 50.  Benefits in year 50 are $2 billion.  Costs are about 50% of benefits.  Build it.

Correct Cost-Benefit analysis (present discount value):  Present value of costs = $12.937 billion.  Present value of benefits = $7.402 billion.  Costs are about 175% of benefits.  Do not build.

That is a very large difference, and it shows the substantial bias introduced by the PPIC/Davis team's incorrect approach.  Of course, you can legitimately argue about the assumed numbers in the example, the point is to show that the error is potentially very important quantitatively in addition to showing bias. 

In summary, the PPIC's analysis of a peripheral canal uses an incorrect framework that is heavily biased towards supporting a peripheral canal.  Importantly, the framework is inconsistent with the much tougher standard they framework they use for evaluating Delta levee investments.  The inconsistency demonstrates substantial bias towards the agenda of Delta water exporters and against in-Delta interests.

Postscript:  Several other posts on this blog demonstrate the PPIC bias in other ways, most notably in the parameters selected for their model (water recycling costs 3x too high, desalination costs 2x too high, vastly understating conservation gains and overstating population growth, etc.)   In this post from over a year ago, I stated

The economic analysis in Comparing Futures suffers from 3 fatal flaws.
1. Grossly overstates future urban water demand and the cost of alternative water supplies.
2. Does not value environmental services or even the market values of recreation and fishing.
3. Ignores established scientific methods for evaluating investments over time which skews their analysis to favor big capital projects like canals.

The post went on to talk about the first 2 flaws, but left an explanation of the last flaw for a future post.  This post finally gets around to it, 15 months late.  I thank Rep. McClintock for providing the inspiration.