Part 7 in a continuing series on CBA -- Atrazine continued

I've decided to finish the story on atrazine before going on to other
parts of the report and the generalities that can be drawn from them.
Atrazine and its mis-analysis in this case forms a fascinating example
of what generally goes wrong with environmental CBA.

To review, CBO scored atrazine and alachlor at $4 billion per cancer case
prevented, when atrazine was only regulated for non-carcinogenic effects
-- giving a more accurate cost/benefit of $10 million per year for an 
unknown number of cases of cardiac, kidney, and liver damage nationwide.
Why did this happen?

Now, CBA proponents often say that one can check a CBA simply by looking
through the numbers and references in the document.  That is most often
not true.  For example, in this particular case, the source for the cost
per cancer figures is not given (except as from EPA).  So I called the 
author of the CBO report and learned that the numbers were 
from EPA's "Regulatory Impact Analysis" or RIA series -- a Regulatory Impact
Analysis must be prepared for each rule.  So the CBO CBA was really
prepared from a previous CBA done by EPA (for the Phase II SOC
Rule, in the case of atrazine and alachlor).  The author of the CBO report
claimed that the RIA didn't say much about non-cancer effects.  The author
had not looked at the Federal Register notice, which justified why atrazine
was regulated.

So I made the journey to the EPA SDWA docket room the next time my job 
took me near there and found the RIA (note that this process would be pretty 
much impossible for someone who didn't live near Washington, DC; as far as 
I know these documents aren't available anywhere else, though I suppose one 
could FOIA them).  The RIA answered a number of questions.

The reason that atrazine and alachlor are always mentioned in one line in
the CBO report is because they are both "corn chemicals" -- pesticides 
used on corn -- and therefore tend to occur together.  To understand why
this matters, I'll have to digress into a short paragraph on treatment
methods.

Readers of this series should understand that there isn't a separate
treatment method for each chemical.  For every SOC the treatment method
is about the same; mostly granular activated carbon is used.  If you
have to install a treatment unit on a water source for one SOC, it removes
the other SOCs as well.  The cost of compliance with SDWA is mainly a 
function of installing and operating treatment units.  So, if two chemicals
occur together, it makes sense to figure out the cost of treatment for
them as a group; breaking down costs for them individually doesn't make
much sense.  That is why a single set of cost and benefit numbers were
calculated for atrazine and alachlor.

Note the following paragraph (EPA, _Regulatory Impact Analysis of Proposed
National Primary Drinking Water Regulations for SOCs_, April 1989, p. 1-3):

"     An outstanding conclusion of the occurance analysis, is that atrazine
has a number of interesting, policy-relevant characteristics.  Atrazine
is one of the highest volume agricultural chemicals currently in use.  It
is also one of the most highly mobile compounds in both ground and surface
waters.  Available data indicate that it occurs at detectable levels in 
nearly 30 percent of both ground and surface water systems in areas where
it is used.  Atrazine may therefore be of value as a 'tracer' that can
serve as an indicator of potential contamination from other agricultural
chemicals."

So now we know why atrazine and alachlor are calculated together; alachlor
treatment is almost a "freebie" that you get with atrazine treatment.  To
a first approximation, then, the costs of treatment for this pair can be
assigned to atrazine.  Note that alachlor is a probable carcinogen; that's
where the cancer cases avoided from this pair come from.

So where did the discussion of the benefits of atrazine treatment go?  The
RIA in question has a short section on the economic benefits of a margin
of safety for drinking water, and of induced efficiency improvements in
the water industry, but it then immediately jumps into an analysis of
cancer cases avoided.  As far as I could tell, there isn't a single sentence
in the RIA benefits chapter that talks about non-cancer health effects.

So my only recourse was to again pick up the phone; this time I called the
EPA office that had produced the RIA.  According to the staffer in charge
of atrazine regulation (among other chemicals), the non-cancer effects
had not gotten into the RIA because they were non-quantifiable.  EPA
doesn't know the dose-response relationship, as previously noted.  Nor is
epidemiology good enough to separate out cases on this kind.  So, since no 
number of cases could be calculated, the economist who wrote the RIA simply 
ignored the entire subject.

This kind of mental blind spot has occured in every example of environmental
CBA that I've seen.  What can not be quantified is ignored.  In effect,
anything that is too poorly understood to have a number assigned to it has
that number implicitly represented in the analysis by a zero (in this case,
zero non-cancer cases).  This is, of course, the opposite of good scientific
practise.  Uncertaincy simply can not be represented by dismissing real
effects in this way.  

Now, remember that this is a very simple subject for CBA.  All that is 
involved is effects on human health from chemicals in drinking water --
a relatively highly studied subject.  If this is what happens for a chemical
that was regulated for non-cancer effects in the first place, imagine the
confusion of the economist confronted by anything with an ecosystem effect!  
Such a CBA is worse than worthless; it represents a reduction in the
amount of available information, or perhaps merely an added source of
disinformation.

That's about all I have to say on atrazine.  Next, giardia, though it will
have to wait until next week.  

--------------------------------------------------------------------------

Appendix: Costs of regulating atrazine

In my investigation I also determined a number of things about how the cost
of regulating atrazine was calculated.  I've put these in an appendix since 
they're not nearly as interesting as the part on the benefits of atrazine 
above.

The costs of regulation are divided into costs for treatment (the majority
of the cost), costs for monitoring for contamination, and the cost of
state regulatory programs.  Cost for treatment are capital costs (annualized
over 10 years, I beleive, though perhaps 20 -- I couldn't find the number in 
the RIA), plus operating and management costs.  Thus a large part of the
cost depends on the discount rate adopted.  In the introduction to this
series I mentioned that this example of CBA avoided discount rate questions;
I might have known that they would turn up.

From the RIA, p. 4-5 footnote:

"    For purposes of Regulatory Impact Analysis, it is appropriate to 
compare costs and benefits on the basis of the social discount rate.  This
is defined as a 'risk-free,' 'inflation-free' rate that represents the
pure opportunity cost of capital to society.  It is appropriate for benefit-
cost comparisons because the risk premiums and inflation premiums 
embedded in market interest rates have no counterpart on the benefit side
of the ledger.  The value of the benefits should be compared to the pure
opportunity costs of the resources that must be expended to obtain them.
Results are presented for social discount rate assumptions of 3.0 percent
and 5.0 percent.  The market interest rate, here assumed to be 10 percent,
provides a more realistic estimate of the total annualized cost that will
actually be faced by water systems as they approach financial markets
to raise the funds needed to support compliance efforts."

According to the CBO study author, a 7.0 percent discount rate was used in
calculating the CBO numbers.  That is the basis for the remark I made
earlier about the costs assigned to atrazine/alachlor being too high.

Though in fact, it looks like that isn't the only factor involved.  As far 
as I can tell, the CBO cost estimate for atrazine & alachlor did not include 
monitoring or regulatory costs -- only treatment costs.  Once monitoring and 
regulatory costs are added back in, the cost for regulation of atrazine might 
be slightly higher than $10 million/year even if a lower discount rate were 
used.  These costs were apparently not added in to the atrazine number 
because EPA only calculated them for categories of chemicals, not for 
individual chemicals like atrazine.  The RIA gives the total cost for 
monitoring of all pesticides, herbicides, & PCBs at $24.9 million per year, 
and the total cost or state regulatory programs at $14 million per year 
(for both IOCs and SOCs).

I'm not that interested in figuring out whether CBO's two problems in cost
estimation cancelled each other out.  Suffice it to say that the cost is
probably somewhere around $10 million per year for atrazine and alachlor
combined.

Disclaimer: This as well as previous posts by me on this and all other 
subjects represent my personal opinions, not my employer's.  This CBA
analysis is a project I undertook because of my own personal interest, not
because I was paid to do so.