Scottish Government

6. Conservation Auction Performance

Conservation auctions are still in their infancy and data from the field are scarce. Anecdotal evidence on auction performance is often spurious and intuition unreliable, especially for complex resource allocation problems. Recent research has therefore employed experimental methods (laboratory auctions) and computer simulation to investigate the performance of conservation auctions. In this section, we review both experimental and simulation studies of conservation auctions. We also summarise the key findings from auction pilots. We start with a note on different ways of measuring auction performance vis-à-vis fixed-rate payments.

6.1 Measuring auction performance

Figures 4 A and B show how an auction may be more or less cost-effective than a fixed-price scheme for the same given budget. In Figure 3A, bidders shade their bids to a greater extent than in Figure 3B, so that, for the same budget, a smaller number of winners are able to be selected by the auction.

Figure 3A: Discriminatory auction and fixed-price scheme: when the FPS is more cost-effective for a given budget: area _FOECX = area _DOABX but OX_D < OX_F

It is important to understand that the opportunity cost curve is the relevant supply curve when a fixed payment is offered. Then all landholders with opportunity costs below the fixed payment stand to gain from participation in the scheme. The marginal participant is the one whose opportunity cost is equal to the payment rate offered. Thus, under the fixed-price scheme, X _F units of service will be traded at price p _F. The total budget cost thus is represented by area OECX_F.

Under a discriminatory auction scheme, by contrast, the ordered bids (not the opportunity cost curve) represent the supply curve. The auction creates room for bidders to shade their bids above their true opportunity costs and thereby to secure themselves an information rent (Latacz-Lohmann and van der Hamsvoort, 1997; Hailu, Schilizzi and Thoyer, 2005). (The detail of how this happens can be seen by reading Appendix 2.) Bidders are accepted in the order of their bids until the budget is exhausted. The total budget cost thus is represented by area OABX_D. Assuming the same budget as under the fixed-price scheme, X _D units of service can be bought.

The cost-effectiveness of the auction thus depends upon the degree of bid shading. One would normally expect bid shading to be low and the auction to be superior to the fixed-price scheme (as shown in Figure 3B). However, if bidders have learned the bid caps from previous auction rounds, bid shading can be significant, resulting in poor auction performance (as shown in Figure 3A).

Figure 3B: Discriminatory auction and fixed-price scheme: when the FPS is less cost-effective for a given budget: _FOECX = area _DOABXarea and OX_D > OX_F

6.2 Simulation studies

Hailu and Schilizzi (2004) construct an agent-based model to evaluate the long term performance of conservation auctions under settings where bidders are allowed to learn from previous outcomes. The results relating to the impact of learning were already reported in section 4.3.3. The focus here is on auction performance relative to a fixed-payment scheme. The details of the setup used to generate the results are shown in Box 7.

Source: Hailu and Schilizzi (2004)

The results (see Table 6) show that the performance of the auction relative to the fixed-price scheme depends on the level of the fixed price employed in the latter.

Table 6: Performance of discriminatory auction relative to fixed price schemes

Source: Hailu and Schilizzi (2004)

With a fixed price set at 90% of the average opportunity cost, the auction is found to be inferior in terms of efficiency, participation, and the provision of environmental benefits. Looking at the results from fully adjusted bids in round 30, the auction provides lower rates of participation and environmental benefits per dollar of program outlay (by a factor of 0.89). The proportion of net income transfer in programme payments is higher under the auction than under the fixed price scheme. Put differently, the informational rents extracted are higher, by 5%. Thus, the fixed-price scheme with reserve prices set at 90% of average opportunity cost outperforms the auction mechanism in terms of participation and efficiency. The auction mechanism compares more favourably only to fixed-price schemes with higher payment levels ( e.g. column 5 of Table 4), but the efficiency advantage still appears marginal. This is because the higher reserve prices involve built-in net income transfers that are similar to those achieved by bidders who learn to 'game' the auction over time.

In conclusion, the authors issue a cautious message about the cost-effectiveness of multiple-round conservation auctions: with bidder learning the efficiency benefits of single shot auctions do not necessarily extend to repeated auctions. Learning can ensure that bidder prices adjust to extract almost all information rents despite competitive bidding conditions.

6.3 Laboratory studies of conservation auctions

Latacz-Lohmann and Schilizzi (2005) investigate the performance of two auction formats (budget-constrained and target-constrained) vis-à-vis a fixed-payment scheme. The comparison was made with the use of controlled economic experiments described in section 4.3.3. See Box 3 in section 4.3.3 for a summary of the experimental setup.

The outcomes of the budget-constrained ( BC) auction were compared with a fixed-price scheme with the same budget as the auction. The target-constrained ( TC) auction was also compared with a fixed-price scheme, but in order to ensure comparability, the fixed price was set such that the same target was achieved - that is the same number of hectares were bought out. The collaterals for the two auction formats thus are not identical. The findings from the Kiel experiments (Germany) are reported in Table 7.

Table 7: First-round auction performance vis-à-vis fixed-price scheme, Kiel

1 FPS = fixed-price scheme
Source: Latacz-Lohmann and Schilizzi (2005)

First consider the BC auction in the left-hand part of Table 7. With the same budget, the BC auction was able to generate 27 per cent more environmental benefit (nitrogen abated) than the fixed-price scheme. Information rents under the auction were only 67 per cent of the fixed-price scheme. Total opportunity costs were higher, reflecting the higher level of abatement under the auction. Likewise, the TC auction achieved roughly the same environmental outcome (save for a few indivisibilities) with only 80 per cent of the payments of the corresponding fixed-price scheme. These findings highlight the cost-effectiveness of auctions as an allocation mechanism, but also suggest that the choice of auction format has only a minor effect on outcomes - a finding we already flagged in section 4.3.3.

Auction performance deteriorated in the second and third round: bids tended to move up relative to the cost curve, and did so more for low-cost bidders than for high-cost bidders. By the third round, much of the initial advantage of the auction had been lost relative to the fixed price scheme. Bidders had effectively learned the implicit reserve prices and adjusted their bids accordingly. The Kiel experiments were replicated in Perth, Western Australia, to check for the robustness of these results. Very similar results were obtained.

6.4 Evidence from field pilot auctions

The BushTender trial in Victoria (see section 5.3 for details) was the first pilot auction to test the proposition that competitive bidding, compared to fixed-rate payments, can significantly increase the cost-effectiveness of conservation contracting. Stoneham et al. (2005) analysed the bids of the first two bidding rounds and compared these to a hypothetical fixed-price scheme. Drawing on information from the bids, Figure 4 illustrates the cost (= bids) of generating additional units of biodiversity (measured as a biodiversity quality-adjusted unit, or BQ). ⁹ The curves thus represent the supply curves for biodiversity in a discriminatory first-price auction. ¹⁰

Figure 4: Supply curves from BushTender

Source: Stoneham et al. (2005)

As shown in Figure 4, the supply curves for biodiversity are relatively flat over much of the quantity range, but then transform to relatively steep as the quantity of BQ rises. Although it is difficult to compare the results from the auction with other mechanisms, it has been possible to examine how a hypothetical fixed-price scheme would perform compared with the discriminative price auction used in the pilot.

The results are shown in Table 8. For the Northern Victoria (first-round) pilot, a fixed-price scheme would have required a budget of approximately US$2.1 million (almost seven times more than the actual budget) to elicit the same quantity of BQ units as the discriminative price auction. Looked at it another way: for the same budget of around US$325,000, a fixed-price scheme would have given an agency approximately 25 per cent less biodiversity. This discrepancy (700% versus 25% performance gain) clearly reflect the flat shape of the supply curve over much of the quantity range combined with the sharp increase of its slope as the quantity of BQ rises beyond a certain level. Similar results were obtained from the Gippsland (second-round) pilot, although the proportionate increase in cost is less, but percentage fall in quantity is greater.

Table 8: Comparison of fixed-price scheme to BushTender price discriminating auction

Source: Stoneham et al. (2005)

These results should be interpreted with caution. The suggestion here is that the size of the gain may be overstated due to an inappropriate counterfactual comparison. Stoneham et al. (2005) take the bid curve to be equal to the true opportunity cost curve. They argue that the bids shown in Figure 1 are inclusive of information rents that bidders may have included in their bid price. They thus assume that opportunity costs and information rents make up bids. This is correct, but it is not right to claim that the bids also represent true opportunity costs. The standard characterisation of supply curves in economic theory is exclusive of rents. Rents arise if there is a difference between the price received (bid or fixed payment) and the true opportunity costs (exclusive of rents).

Recall Figure 3 for clarification: There are two distinct supply curves, one for the auction (which does include information rents) and one for the fixed-price scheme (which does not include information rents). Simulating the outcome of a fixed-price scheme requires one to relate the price to the true opportunity cost curve. The area between price and opportunity cost curve represents information rent. Relating the fixed price to the bid curve ( i.e. assuming that the bids represent true opportunity costs) leads to an overestimation of the fixed price needed to achieve the same outcome, in terms of BQ quantity, as in an auction. Given the very steep slope of the supply curves in Figure 4 for higher levels of BQ output, this overestimation is likely to be significant. The performance gains from using an auction are thus overestimated.

White and Burton (2005) used data from the Auction for Landscape Recovery (see section 5.4 for details) to benchmark the budgetary cost-effectiveness of the auction to that of an equivalent fixed-price scheme. They show that the cost-effectiveness of the ALR compared to that of a uniform price scheme varies between 315% and 207% in round 1 and 165% and 186% in round 2, depending on whether the fixed price scheme is input-based or output-based (see Box 8). See Table 9 for the full results. White and Burton (2005) also show that comparing BushTender to an output-based scheme would considerably reduce the cost-effectiveness gains of 700% claimed by Stoneham et al. (2003) and Stoneham et al. (2005). It is also clear from Table 9 that auction effectiveness can vary from round to round.

Table 9: Cost effectiveness of the ALR pilot auction assessed against different counterfactuals

Source: White and Burton (2005)