Overall, I would say he did very well, and the presentation reflects substantial improvement for the Economic Analysis of the BDCP. It was certainly much improved over the June presentation, and I commend David for the enormous amount of work he has completed, and his clear and professional presentation and answers to the many questions. It wasn't perfect, and I have a few new concerns, but they are outweighed by the improvements.
Notable improvements since the last time include:
·
The discussion of the arguably fictional regulatory certainty benefit was minimal, and the enormous estimates of its value were nowhere
to be seen.
·
Also gone was the conclusion statement that
the BDCP is clearly worthwhile to the water agencies since that depended on
the regulatory certainty benefit. [Update: I should note that he did not give any bottom line conclusions, for or against.]
·
And most notably, the BDCP is now saying that
they are going to do a complete statewide benefit-cost analysis after refusing
for years.
While the regulatory certainty benefit was gone, there
were two other notable changes in the benefits analysis that concern me because they are blowing up the
benefits estimates to help justify the tunnels. These are 1) increasing the time horizon
and 2) lowering the discount rate. Both
would draw significant complaints in a peer review, and are notably inconsistent
with the approach used in the California High-Speed Rail benefit-cost analysis. I believe the HSR comparison is critically important because these are the 2 controversial mega infrastructure projects on the state's policy agenda, and they should be analyzed consistently. Time Horizon: Rather than estimating benefits of conveyance only up to 2050 (28 years of operations), Dr. Sunding is now extending the benefits to infinity (yes forever). While cutting off at 2050 was too short, extending the benefits analysis to infinity is erring in the opposite direction.
Here is what the CA HSR B-C analysis says about the same issue (p.8).
Benefits and costs are typically evaluated for a period that includes the construction period and an operations period ranging from 20-50 years after the initial project investments are completed. Given the permanence and relatively extended design life of high-speed rail investments, longer operating periods, and thus, evaluation periods are applicable.I support the use of a relatively long 50 year time span after project completion, but extending benefits to infinity in an infrastructure B-C analysis seems nuts. However, it wouldn't make much of a difference if the discount rate weren't so low, which brings me to the second issue.
For the CA HSR BCA, the evaluation period includes the relevant (post-design) construction period during which capital expenditures are undertaken through 2080...
which is approximately 47 years beyond project completion for the scenario with the longest construction period.
Discount Rate: Dr. Sunding is using a 2.275% discount rate, justified by a recent Army Corps statement that I haven't seen and current, historically low interest rates. This is down from the 3% discount rate used in his initial estimates this summer. While I am sympathetic to using something lower than the traditional 7% rate, this 2.275% assumption could get hammered in a peer-review, even if it has been approved by a federal government agency.
Current interest rates on bonds are being extraordinarily influenced by unprecedented Fed intervention and the lingering effects of a global financial crisis. I don't think you can argue that low U.S. treasury rates in the current market signal that somehow the social rate of time preference or the opportunity cost of public funding (discount rate) has dramatically declined. And what are the appropriate interest rates to guide the discount rate selection? While Federal Treasury borrowing rates are historically low, credit is still extraordinarily tight, and many individuals, businesses and state and local governments are finding it difficult to borrow at any interest rate. The appropriate opportunity costs for the BDCP is connected to who is bearing those opportunity costs, and the federal government has made it pretty clear that they are not paying for the BDCP. Many of the costs are going to be passed through directly to households through rate increases, not added to the Federal debt, so I could make an argument that real interest rates on personal loans (even credit cards) are as relevant as the current rate on Treasuries. Other BDCP costs are slated for the California General Fund through a General Obligation Water Bond. I don't think the opportunity costs of utilizing the state general fund are at historic lows. If anything, the opportunity cost of state funding is at a historic high, and the state hasn't even sold billions of already voter approved bonds due to market and budget conditions - even though interest rates are historically low. California is a state that needed a temporary tax increase (Prop 30) to pass to avoid cutting 3 additional weeks off the school year, is releasing criminals early, and cutting public health and other key services due to lack of funds. Given that the Governor has defined school days as the margin in the state budget, I could argue that the social return on investment for time in school would be the appropriate discount rate (opportunity cost).
Probably more relevant is the other mega-project financed with a combination of general fund dollars and other sources, high-speed rail. CA HSR used a 7% real discount rate for their benefit-cost analysis (p. 6). HSR actually used 4% in their initial draft B-C analysis, but moved it up to 7% in the final draft in response to criticism in the peer review. So, I would suggest that if HSR couldn't get a 4% discount rate past peer review when interest rates were equally low in 2011-12, his 2.275% assumption is going to be a tough sell. Personally, I think 7% is probably too high, and would have recommend running both the HSR and BDCP B-C studies using 3% and 6% discount rates. Dr. Sunding should perform some sensitivity analysis and prove that his results are robust to the HSR assumptions about time horizons and discount rates, and aren't just the result of assuming an infinitely lived project and a super low discount rate.
I am also still a little concerned about his approach to the seismic risk. However, I will note that he toned down the rhetoric about this since the summer, and signaled some openness to considering alternative ways to reduce the consequences of a seismic event. So this area of concern is also more positive and I have hope something good will eventually emerge. I don't think there is any doubt that his analysis will show that the earthquake benefits are much less important to justifying the project than water supply benefits. In fact, it already does.
As always, the criticisms take more space than the compliments, so I should note that I think his work in the critical urban water supply and several other areas is outstanding. I think some of the folks in the audience do not appreciate what is involved here. While I have wanted to smack him occasionally over the years (the feeling is mutual I'm sure), I have a lot of respect for Dave Sunding and remain convinced he is the best choice to lead this very important job.
Really, the biggest key to his benefit-cost analysis of the tunnels is defining a good no-tunnel alternative instead of a comparing the tunnel alternatives to a crappy, no-action, status-quo strawman. The no-tunnel alternative has to be a BDCP option, meaning that it includes the necessary habitat and through-Delta flows and water operations to satisfy the ESA. I see no reason why a no-tunnel alternative can't have comparable "regulatory assurance" benefits, environmental benefits, and even the potential water supply benefits from the "decision tree" as a tunnel alternative. Finally, at least one no-tunnel BDCP alternative, and arguably even the tunnel BDCP alternatives, should include significant seismic levee upgrades.
Jeff,
ReplyDeleteTwo important points:
1. The EWC has submitted a no tunnel alternative to the Delta Stewardship Council. It is on the record and is referred to as "Alternatice 2" in the Delta plan. A more accurAte description is on our web site (www.ewccalifornia.org) as the "Reduced Exports Plan." We have presented it informally at a BDCP-NGO meeting and this month I will send it formally to BDCP.
2. I am concerned that David feels it is appropriate not to include the interest rate on bonds or borrowed monies as part of the CBA. I believe he is wrong that this ia common practice for CBA's. his view is a precursor of the bias that he may build into his CBA.
Nick Di Croce, EWC
Jeff:
ReplyDeleteThanks for sharing your views. You raise some good issues that I would like to chime in on:
Project Life: I'm ok with an "extended time horizon" provided that the analysis addresses the unfolding of material risks during the selected time horizon.
Credit Market Conditions: Agree that today's credit market conditions are not relevant. Instead, what are the likely conditions at the time of investment (which is in the future). One scenario worth considering is pre-2008 conditions under assumption that, by time a project is actually financed, abnormal conditions of today are no longer relevant.
Discount Rate: Would be useful to divide the discussion into the underlying components:
time preference: Treasurey Inflation Protected Securities (TIPS). TIPS are an inflation-indexed, so yield is the "real interest rate" Pre-financial crisis, TIPS traded below 2%--around 1.8%
Expected Inflation: look at differential between treasury yield and TIPS for guidance. Pre-2008 crisis, differential suggested expected inflation around 2.5% (in neighborhood of actual CPI experience post 1982)
Default Risk: this is where a 7% real interest rate is potentially high. If real time preference is 2%, for example, then 7% real interest rate suggests a 5% risk premium. This implies an EXPECTED life of 20 years. In my opinion, the best way to run down this issue is to identify the nature of material risks expressly. For example, if 100-year flood destroys the project, then the default risk is 1%, not 5%.
As one thinks about risk, one can conclude that the default risk premium is not the best analytic method. Instead, include modeling of risk directly. For example, if 100 year flood occurs, it would probably damage the project and require "recovery expenditures". Better to run down this scenario directly and incorporate into analysis of project water supply benefits and cost structure. While the water industry has not used this method, most other industries have. A good reference is "Risk Analysis: A Quantitative Guide", David Vose (Wiley).
Market Risk: this last component is probably not signifcant for water projects. Financial risk is probably a fully diversiable risk and, therefore, no need to add this risk premium to the discount rate.
Keep up the dialogue. Critically important that CA water investment alternatives are properly analyzed
Nick: Thanks for continuing to promote a no-tunnel alternative. I am concerned the EWC no-tunnel proposal is too restrictive on water supplies. I assure you David is right about the interest rate on bonds and BCA, although it is highly relevant to financial feasibility. But the real discount rate he is using is very low. See Rodney's comment.
ReplyDeleteRodney: Thanks for your comments. You make a good case for real discount rates well below 7%, but it also clear from your comments that 2.275% is pretty darn low.
Time horizon and discount rates are key to benefit-cost. David is pushing both assumptions to the defensible limits that help his clients get the answer they want (build the canal). While there is an argument that can be made, it is clear that the assumptions are not consistent with most BCA analysis, nor the recently established state precedent for high-speed rail analysis or the 2008 DWR economic analysis guidelines.
Thus, I am fairly confident that most economists would be uncomfortable with the infinite horizon and 2.275% real discount rate assumptions, and believe that nearly all would agree that at minimum some sensitivity analysis to shorter time horizons and higher rates is needed.
One more thought on Rodney's comment.
ReplyDeleteAlthough technical, I recommend a recent exchange of articles in the Journal of Benefit-Cost Analysis. The lead article by Burgess and Zerbe (Issue 2, April 2011), considers arguments made by some environmental economists for low discount rates like 3%, but rejects them and concludes 6%-8% is the appropriate range.
The next issue includes a pair of comments arguing for lower discount rates, specifically the 3-4% rates often used in Europe. Burgess and Zerbe then counter with a discussion of opportunity costs of displaced consumption/investment and defend the 6-8% rates historically used in the U.S.
I am certainly not an expert in the discount rate debate. My point is that most economic arguments of the discount rate are occuring between 3% and 7% real discount rates. Thus, it is clear that 2.275% used in the most recent BDCP analysis is extremely low and pushing the boundaries.
Given the extremely controversial nature of the $14 billion tunnels, one would hope that it can clear a net present value test using conventional assumptions.
I am not sure why Jeff is referring to the BDCP a $14 billion project. With the habitat conservation components, it is my understanding that the cost is likely to be nearly twice that amount. Unless there is some credible argument that the tunnels could be permitted on their own without the habitat restoration, I don't think $14 billion is the correct number to be discussing.
ReplyDeleteOsha Meserve
Local Agencies of the North Delta/Stone Lakes National Wildlife Refuge Association