Salinity, Delta Agriculture and the PPIC: Muzzled Dogs Can't Bark

Last month, I made an initial comment on the PPIC/UC-Davis California Water Blog post that salinity effects on Delta agriculture from a proposed peripheral tunnel is a dog that doesn't bark, and the approving post that followed on the "BDCP blog".  My original blog focused on the fact that both blogs deceptively suggested the PPIC had modeled the BDCP proposal, something I have complained about since the original report was released nearly a year ago. 

But that post is about the spin, and this post speaks more to the substance (i.e. boring, but keep reading).  This post explains why I see greater evidence (or hear barking dogs) of potential salinity impacts on Delta ag than the Davis researchers.  Along the way, I will make a policy proposal and a research proposal that would go a long way towards resolving the issue.

The three steps in determining potential economic impacts from salinity are as follows:

1. Predict salinity change from Delta conveyance. 
2. Estimate impact of predicted salinity change on agricultural production. 
3. Estimate economic effect of change in agricultural prodution. 

Step 1: Predict Salinity Change

The PPIC/Davis team claims their study is the most advanced because it used a detailed hydro-dynamic model in the first step, and their results showed little to no change in Delta salinity levels for the conveyance facility and operations they modeled.  If this result is true, then obviously there is no economic impact, and debating the 2nd two steps of the analysis is a waste of time.  I have no reason to doubt the result of the hydro-dynamic modeling, but I know it has left many people puzzled by the surprising results.  I recommend the PPIC/Davis folks do more to explain these surprising hydro-dynamic findings to the public. 

When we looked at this same question for the DPC Economic Sustainability Plan, we wanted to use hydro-dynamic modeling results too but didn't have any.  We asked BDCP to share their current modeling, but they declined.  Instead, we used something that is arguably more important than modeling results, the BDCP's proposed standards for Central and South Delta Salinity. 

And yes, those dogs were barking loudly since the draft BDCP did not set maximum salinity standards in the central and south Delta like they did for other areas.  Furthermore, at the same time, the sponsors of the BDCP, the Department of Water Resources' is currently pushing for a 41% increase in maximum south Delta salinity levels during the growing season.  If DWR thinks their will be no salinity impact like the PPIC says, why are they trying to change the current standards?

So that leads to my policy proposal that follows directly from the PPIC findings of marginal salinity impacts.  The BDCP should commit to maintaining the current central and south Delta salinity standards into the BDCP, and DWR should drop it's current proposal to the State Water Resource Control Board to weaken those standards.  If the BDCP isn't willing to do that, then they shouldn't be promoting the PPIC/Davis findings to advance the tunnels since they will not have shown any faith in the results.

Step 2.  Estimate Impact of Salinity Change on Agricultural Production

In the PPIC/Davis model of Delta agriculture, the salinity effect on crops is assumed to follow the analysis of a paper by Hoffman which DWR is using as its primary support for its proposal to weaken south Delta salinity standards from 0.7 ec (mS/cm) to 1.0 (mS/cm).  Hoffman's paper does not include any actual soil or crop data from the relevant parts of the Delta, but makes some favorable assumptions regarding soil conditions to derive crop yield functions where irrigation water salinity has virtually no effect below a threshold of 1.0 ec.  Under different soil assumptions, the thresholds are much lower.  Hoffman provides very flimsy evidence to support his crucial soils assumption.  Last year, lawyers for Delta farmers showed me maps of Delta soils that contradicted Hoffman, and Delta farmers told me they had experienced crop damage from salts that Hoffman's study assumes could not occur.

The bottom line is that this assumed yield function with a relatively high threshold before salinity impacts occur drive the UC-Davis results since their model programs in the disputed Hoffman view of the Delta.  In other words, they have muzzled the dog in their model so it can't bark at current or predicted salinity levels in the Delta.

In contrast to the programming approach with an assumed salinity yield function with a 1.0 ec threshold, we used an econometric approach in the DPC Economic Sustainability Plan.  This approach doesn't assume a yield function, it uses a detailed field level data set to statistically test whether Delta farmers crop choices are correlated with salinity after controlling for a variety of other factors.  This statistical approach showed a strong correlation between salinity and crop choice in a data set where virtually all the observations were below Hoffman's 1.0 ec threshold.  This econometric approach is preferred by most economists when the data is available (field level data needed for this kind of model is pretty new), and I should point out that the Delta Science Progam's Independent Review Panel praised this modeling in the ESP as "state of the art" and made a positive contrast between it and the programming approach used in the PPIC/Davis study.

With this background, you should see why I laughed at this passage in the recent PPIC/Davis blog post in which they offer an explanation for why their salinity effects are so low.

How do we account for such large cost reductions? Four factors explain the difference...
Second, we conducted the analysis island-by-island, which enabled us to show that the islands at risk are those that grow the lowest value crops. Third, these low-value crops are the most salt-tolerant.

The bold was added by the dog barking at evidence of salinity impacts on Delta crops. 

The UC-Davis researchers ignore what this data says about their model and interpret it as an explanation for why economic impacts are low.  It just so happens that the low-value, salt tolerant crops are already grown in the areas that are near but below the threshold where they assume salinity affects crops. How fortuitous!  So if salinity rises further and bumps against the threshold, there will be little economic impact since these fields already have low value crops. 

Other scientists might look at this data, and begin to question their model assumptions.    

And here is my constructive suggestion for research.  While I think there is pretty compelling evidence in the crop choice data that salinity has a bigger impact than assumed by Hoffman/PPIC/Davis, I don't believe it is the final word.  It would be useful to test these relationships more directly through data collection and research in the field about Delta soils and crop response to salinity.  I made this suggestion at the Delta Science Conference as an area where they could sponsor some field research that would go a long way to settling these questions.

Step 3.  Estimate economic impact of change in agricultural production.

In the final step, the PPIC/UC-Davis team used an IMPLAN model to translate the changes in agricultural output into economic impacts such as lost jobs and income.  This is usually a relatively straight-forward step, but I will note some small inconsistencies.

In the Delta study, they find about 16 lost jobs for every $1 million decline in agricultural revenue.
This is about half the 30 lost jobs per $1 million decline in agricultural revenue that they predicted in 2009 from water shortages.

I see two differences in their Delta modeling that result in lower multipliers that I think are poorly justified.  First, they have not adjusted the production functions to account for contract labor as they did in their studies of water shortages in other regions.  Second, they have modeled the multiplier effects over a very narrow region, compared to the eight county San Joaquin Valley used for other agricultural regions.  Their Delta zipcode region extends just outside the legal Delta boundaries but excludes many important areas where many Delta workers, suppliers, and even farmers are located including Lodi and south and east Stockton.  In addition, it leaves out important shopping areas around the Delta in Lodi, Stockton, Elk Grove, the mall area in Tracy, and Sacramento.  If you miss significant areas where employees live and shop, you are certainly going to underestimate the "ripple" effects in the local economy.

Despite all this criticism, I do agree with the UC-Davis researchers about one of their conclusions regarding Delta agriculture and the BDCP.  As the BDCP is developing (shrinking the conveyance but not the habitat) the issue of land availability is a bigger economic threat to Delta agriculture than salinity.