Nitrogen Trading versus a Nitrogen Tax in a Sensitive Catchment
Environment Canterbury administers a catchment cap on nitrogen leaching into Te Waihora/Lake Ellesmere. About 420 dairy farms in the catchment leach an estimated 4,800 tonnes of nitrogen annually; the catchment load reduction target is 30% by 2035. Leaching is non-point source and highly heterogeneous: modelled losses range from 15 to 90 kg N/ha/year, and the model has a margin of error of plus or minus 25%. Monitoring every farm with lysimeters would cost roughly $18 million per year, versus $3 million for modelled estimates audited on 5% of farms. The council is choosing between (a) a uniform nitrogen tax of $60 per kg N leached above a 20 kg/ha allowance, and (b) a tradable discharge permit scheme with permits grandparented to historical leaching and a cap declining 3% per year.
1. Compare the tax and the permit scheme in terms of cost-effectiveness under uncertainty about marginal abatement costs. Which instrument is preferred, and on what conditions?
Read the worked answer
Set-up. The regulator wants to reduce nitrogen load by 30%. Marginal abatement costs (MAC) are heterogeneous and uncertain; marginal damage (MD) from leaching is uncertain but plausibly steep near the catchment cap because of ecological thresholds such as lake eutrophication. Weitzman's theorem: if the MD curve is steeper than the MAC curve, quantity instruments (permits) dominate; if MAC is steeper than MD, price instruments (taxes) dominate.
Application.
- The catchment cap is a threshold: beyond a certain load, damage rises sharply (algae blooms, fish kills). This makes MD steep, favouring permits for quantity certainty.
- But non-point-source monitoring is costly and imprecise (modelled plus or minus 25%). Permits require measuring each farm's discharge to enforce; with measurement error, the quantity instrument delivers a noisy quantity. A tax on modelled leaching similarly suffers.
- Heterogeneity: farms range 15-90 kg N/ha. A uniform tax equalises marginal abatement cost across farms in the static sense. A permit scheme with trading also equalises MAC, and grandparenting gives an initial allocation.
- Under cost uncertainty, a tax fixes price and lets quantity vary; permits fix quantity and let price vary. If the regulator's priority is not exceeding the cap (irreversibility), permits are safer.
Verdict. Permits are preferred on quantity-certainty grounds given the threshold damage, provided monitoring error is addressed through audited models and a buffer. If monitoring error is so large that permit enforcement is arbitrary, a tax on an observable proxy (such as nitrogen fertiliser purchases) may be more administrable, accepting less precision. A hybrid - a permit scheme with a price floor and ceiling (safety valve) - combines both.
2. Evaluate the choice between grandparenting and auctioning permits, considering efficiency, distribution, and the political economy of the catchment.
Read the worked answer
Efficiency. Grandparenting allocates permits free to historical emitters. It does not change the marginal incentive to abate, because the opportunity cost of using a permit is its market price. So static efficiency is similar to auctioning, provided the permit market is competitive. But grandparenting creates a barrier to entry: new entrants must buy permits, raising their costs relative to incumbents, and it may facilitate strategic behaviour. Auctioning raises revenue that can be used to reduce distortionary taxes or fund monitoring.
Distribution. Grandparenting transfers wealth to existing farmers, capitalised into land values. It is politically easier and compensates those who may have made investments under earlier rules. Auctioning transfers wealth to the Crown; if revenue is recycled to the community or to farmers for abatement, distributional concerns can be mitigated.
Political economy. Grandparenting can buy acceptance; but it rewards historical leaching and may be seen as unfair. A phased transition - free allocation declining over time - is common and can balance acceptability with efficiency.
Verdict. Grandparenting is defensible as a transitional measure, but auctioning (or a hybrid with partial free allocation) is superior on efficiency and equity grounds if the revenue is used well. The risk with grandparenting is lock-in and a barrier to entry that perpetuates the existing farm structure.
3. How does non-point-source monitoring error affect the design of either instrument, and what does it imply for the optimal stringency of the catchment cap?
Read the worked answer
Effect on instrument design.
- Measurement error means the regulator cannot perfectly observe individual discharges. Enforcement is probabilistic. Farms may under-report; audits on 5% of farms create a deterrence effect, but detection probability is low.
- With a tax, error attenuates the response: if the tax is levied on modelled leaching, farms may contest the model, and the effective tax rate is uncertain. The optimal tax rate should be lower than under perfect information to avoid over-penalising farms with high model error.
- With permits, error makes the cap uncertain. The regulator may need a safety margin: set the cap slightly tighter than the target to account for under-measurement, or hold a buffer reserve.
- Imperfect monitoring favours instruments that are cheap to administer: input-based proxies (such as nitrogen fertiliser purchases) are observable and correlate with leaching. But they are less targeted, reducing cost-effectiveness.
Implication for stringency. When marginal damage is steep and measurement is noisy, optimal policy is more conservative (less stringent) than the deterministic optimum, because the cost of exceeding the threshold is high and the regulator cannot fine-tune. Alternatively, invest in monitoring to reduce error before tightening the cap. The choice between a tax and permits should therefore depend on whether the regulator prioritises avoiding a catastrophic exceedance (permits with a buffer) or minimising the cost of error (taxes with a conservative rate).