2 Mar 2012

Mike Young on water rights

Mike Young (who will hold the Australia Chair at Harvard next year) sent a great response to this post.

Please comment.

There are some important issues tangled up in your blog and in DC's comment that need to be clarified.

DC seems to assume a seniority allocation system. Your blog assumes a share system.

DC mentions a requirement for a diversion right to be used (taken) and used beneficially. In beneficial use right systems there is normally a use it or lose it restriction.

In a share system, the share entitlement is defined in perpetuity and there is no requirement that the water be used. I assume that you too assume that the share is a perpetual share but that use conditions could change.

In a share system, the share should always be a right to a share of the amount that can be taken sustainably whether or not the use is beneficial is irrelevant. Any person given an allocation is entitled to decide to never use that allocation.

If you also mean a perpetual share entitlement then there are two options:
  1. A share of the amount that can be diverted or taken

  2. A share of the amount that is diverted or taken less an estimate of the amount that is assumed to return.
To decide which of the two options is preferable, one needs to understand risk assignment and transaction costs.

In option (1) risks and transaction costs are lower as there is no need to monitor the efficiency of water use at the individual user level. Instead, it is only necessary to estimate mean water use efficiency and as efficiency is increased reduce the amount that can be taken per share. If this is done, then those who increase water use efficiency get an advantage until others catch up.

To ensure efficient managers need to make it clear that as the efficiency of water use in the entire system increases allocations per share will be decreased. That is, the risk of declining allocations per share occurring as a result of increasing efficiency is assigned to the share pool as a whole. As a result, there is an incentive and a short term reward for those first move to adopt more efficient forms of use. I have not modelled this formally but I think this means that in the long run system efficiency is likely to be greater than under option (2).

Under option (2), as one increases water use efficiency, the amount that one can take goes down. That is, risk is assigned to the individual and the amount allocated to a technically inefficient irrigator is never reduced when others choose to increase efficiency. The costs of monitoring use, however, are much higher. Moreover, as the exact amount that returns is difficult to measure one must expect lots of arguments. This can be managed by deeming the amount that can be taken but expect arguments.

Either way, the critical bit is that the effects of increasing water use efficiency (return flow reduction) are managed. Problems of the type described in both your blog and in DC's comment assume that the effect is not managed. In a well designed and well managed regime, return flows like changes in rainfall, etc need to be managed. If you assume bad management then you get bad outcomes.

In Australia, where management still needs to be improved, there are examples of both regimes operating. I tend to prefer option (1) but in specific circumstances have recommended option (2) in a ground water systems where increasing efficiency increased salinity which also had to be managed and they already had a quasi-option (2) system in place.

[A compromise approach is to assume 100% efficiency and zero returns. Australia has used this approach and then got itself into political strife because increases in water use efficiency cause declines in the quality of the environment. Explaining to people that the entitlement regime is expected to cause a decline in environmental quality is difficult. In most cases, the sooner the preferred environmental state is reached the easier it is to keep most people happy. Working out the optimal environmental state to maintain at any point in time is a separate issue.]