Scottish Government

7. Conclusions

Key considerations

Theoretical analysis suggests that competitive bidding can be a powerful means for conservation agencies to increase the effectiveness of public spending for the provision of countryside benefits. The outstanding feature of conservation auctions is their potential to reveal, at least partly, bidders' compliance costs, thereby reducing the information asymmetry between landholder and agency. Conservation auctions also act as a price discovery mechanism for environmental goods and services which have no standard value and which are difficult to cost. They thus allow the parties to deal with the uncertainty about the value of the object being traded. In general, the use of conservation auctions fits well with the trend towards a value-for-money approach that policy has adopted in the provision of public services.

Bidding schemes yield the highest benefits when the conservation agency has little information about landholders' compliance costs, the number of potential participants is large, the contracts offered are homogeneous, and farms are heterogeneous in their compliance costs. The fewer of these conditions apply, the less well an auction will perform relative to a fixed-rate payment scheme.

Theoretical analysis further suggests that, under standard discriminatory price auctions, landholders will shade their bids above their costs of compliance: the optimal bidding strategy is one of overbidding; the auction will not reveal bidders' true opportunity costs. In fact, overbidding is highest for the lowest-cost bidders, whereas the highest-cost bidders will bid closest to their true costs. However, these highest-cost bidders are not those that are usually selected; rather, the lowest-cost bidders are, but they get paid well above their true costs.

Empirical evidence about the performance of conservation auctions is inconclusive. Efficiency gains reported in the literature range from a few per cent to seven hundred per cent. The latter value, reported by Stoneham et al. (2003), should be interpreted with care due to the 'non-standard' way in which it has been computed. We conclude that it is too early to make a robust and reliable assessment of the cost-effectiveness of auctions in agri-environmental management. This is at least true for one-shot auctions.

There is unanimity in the empirical literature that bidder learning poses a substantial threat to the efficiency of multiple-round conservation auctions. Both experimental studies and agent-based simulation studies have confirmed the experience with the US Conservation Reserve Program: when bidders have the opportunity to learn from preceding bidding rounds, they will use that information to update their bids and reap a higher share of the 'surplus' - at the detriment of auction performance. Proposals to combat bidder learning have been made in the literature, but none of these have been tested empirically.

Conservation auctions are complex incentive mechanisms. Their design poses many challenges and there is little practical experience, implying the risk of implementation failure. Auction theory is not well developed for conservation auctions and thus does not provide sufficient guidance for most policy design purposes. Theoretical analysis in mainstream economics is predicated on analytically tractable models, which requires the analyst to make drastically simplifying assumptions that reduce, and sometimes annihilate, their relevance to real policy contexts. Researchers are only beginning to explore practical auction design issues through the use of controlled economic experiments or agent-based computational simulations. Results are fragmentary and do not provide, as yet, a sound and encompassing basis for informing practical auction design.

The potential benefits of auctions come at the cost of likely higher administration costs and higher transaction costs on the side of landholders, although empirical evidence is patchy so far. Complaints from unsuccessful applicants for the Scottish Challenge Funds focused on high costs of bid preparation in connection with uncertain outcomes, resulting in the bidding mechanisms being replaced by locational ( i.e. fixed-rate) premia.

Implications for agri-environmental management in Scotland

SEERAD has indicated that it is considering allocating Land Management Contracts ( LMC) through competitive bidding. LMC is a menu scheme designed to provide a range of measures suited to the diversity of agricultural activity and land types throughout Scotland. Landholders can choose which activities they wish to carry out from the menu (see Table 10), depending on what suits their individual circumstances and plans for future business development. The LMC Menu Scheme is a multiple-objective scheme: it is designed to deliver widespread benefits leading to economic, social and environmental improvement. An extension of the LMC Menu Scheme, which is currently being developed, will focus on the provision of tailored economic, social and environmental benefits. The current Scheme, launched in 2005, puts an upper limit on the total allowance an individual landholder can receive per annum in return for carrying out options from the menu.

We do not consider competitive bidding to be the ideal mechanism for allocating Land Management Contracts - for a number of reasons:

First, there is unlikely to be much variation among landholders in the costs for carrying out the options listed in Table 10. In other words, compliance costs are relatively homogeneous. For bidding to be effective, farms must be heterogeneous in their compliance costs. Second, most of the measures are reasonably straightforward to cost. Put differently, there is little uncertainty about the value of the goods and services being traded. There is thus no need to use an auction as a price discovery mechanism. Third, LMC contracts are heterogeneous. Different landholders will include different combinations of management options in their bids. Bids thus will differ in their contribution to each of the three broad programme objectives (economic, social and environmental improvement). Under these circumstances, bid selection can be a complex task. SEERAD would face the challenge of determining weights reflecting its relative preferences for the different measures or, if measurable, their contribution to the Scheme's objectives. That is, a multi-criteria bid scoring system would have to be devised to aggregate the various dimensions of a bid into one figure representing an estimate of the overall (economic, social and environmental) benefit of each bid. While this is not an infeasible task, in practice it is likely to involve high transaction costs and, because of lack of sufficient transparency in the complex ranking method, the risk of complaints and appeals from unsuccessful applicants.

Whether competitive bidding is the appropriate mechanism for allocating contracts for the extended LMC Scheme (to be launched in 2007) is a matter of debate. This scheme is designed to deliver tailored (rather than widespread) economic, social and environmental benefits. Tailored benefits usually require very specific measures targeted to specific resource settings. To the extent that tailoring involves spatial targeting or the use of strict eligibility criteria, the number of potential applicants may fall below the level required to secure sufficient bidding competition. The smaller the group of potential bidders, the lower is the level of bidding competition and the higher the risk of collusion and strategic bidding. In such circumstances, an auction is not the appropriate incentive mechanism and fixed-rate payments or individually negotiated grants should be considered instead.

Table 10: LMC Menu Scheme: measures and payment rates

Source: www.scotland.gov.uk/library5/agri/lmcsl-00.asp

Similar problems arise when landholders have the choice to enter different tiers of a scheme, reducing the number of bidders per tier. This problem, however, can be addressed through an appropriate bid selection mechanism whereby all bids (from all tiers) are ranked based on the ratio of some overall benefit index to the amount bid. Such a mechanism would allow SEERAD to compare and select bids on a common denominator across tiers, effectively increasing the pool of bidders. This, however, comes at the cost of high transaction costs and the risk of complaints and appeals - as discussed above.

Once again: auctions work best when the number of bidders is high, contracts are homogenous and landholders are heterogeneous in their compliance costs. Of all agri-environmental schemes offered in Scotland, the Organic Farming Scheme is probably the one which best meets these requirements. Reserve prices could be set for different types of land use and different land management systems, e.g. arable land, improved grassland, rough grazing. The risk here is that the current fixed payment rates might act as a point of reference in the bidding process, with bids anchored around them.

A possible way forward

Given the lack of theoretical backdrop and practical experience, we advise a cautious approach to the use of auctions in conservation contracting.

Experimental economics and policy 'test-bedding' seem to offer a way forward. Since 2002 when Vernon Smith received the Nobel Prize for pioneering this new field, experimental economics is taking economists by storm. Economic experiments take place, strictly speaking, under totally controlled conditions in a laboratory. They can have three functions: 1) testing existing theories and their predictions; 2) testing new policy mechanisms or institutions; and 3) exploring beyond existing theory where theory is unable to provide sufficient guidance. In the case of auctions for conservation contracts, roles (2) and (3) are involved.

In a broader sense, of greater relevance to the policy maker, experimental economics can be taken to mean an ordered sequence of investigations starting with some theoretical model and ending with a reduced scale field trial, the last stage before full-scale policy implementation. The steps involved include, in roughly the following order (steps 2 and 3 can interact):

1) theoretical model or hypotheses of bidding behaviour and auction outcomes
2) computer-based simulations with "virtual agents"
3) laboratory experiments with (typically) university students
4) laboratory experiments with stakeholders
5) reduced-scale field pilots

Focusing on steps 3 to 5, the question is often asked whether laboratory experiments are able to accurately predict outcomes in the field, on the understanding that if they do not, laboratory experiments are likely to be of little value. Vernon Smith's response to this legitimate question is this: of course the lab does not accurately predict outcomes in the field, and it was never meant to. Instead, its function is to allow us to make at low cost all the serious and potentially expensive mistakes both in the financial and in the political sense. Lab experiments allow us to narrow the range of errors which can be made in the field and thereby save us a lot of money.

The implication for auctioning conservation contracts is that, given the insufficient guidance provided by theory, any policy initiative in this area should be preceded by carefully targeted and designed experiments. This is called 'test-bedding' policies and has also been called 'wind-tunnelling' or 'glass-housing' policies, depending on whether one has an engineering or an agronomic bent. Such experiments must be carried out by experienced specialists. They represent a stage in the learning process towards successful auction design. Laboratory experiments with stakeholders (step 4 above) has potential advantages in terms of: eliciting the opportunity costs (and heterogeneity in costs) faced by landholders; identifying likely participation rates in an auction system, across different auction formats; and identifying the transaction costs associated with a tender mechanism.

An alternative way forward may be this: rather than using a discriminatory-price auction to allocate conservation contracts in multiple bidding rounds, SEERAD could consider running a one-shot uniform-price auction and use the information gained in the bidding process to devise and calibrate an appropriate fixed-price scheme. The main purpose of the auction thus is to reveal information about compliance costs. As we have demonstrated in section 4.3.1, the incentive structure of a uniform-price auction is such that it induces bidders to reveal their true compliance costs with the bids. Bids could subsequently be analysed to determine appropriate, cost-orientated payment rates for different farm types, land use systems and resource settings. However, if landholders (or more generally bidders) know this is SEERAD's strategy, they will have an incentive to bid strategically to bias the future fixed price in their favour.