Content uploaded by Robert Joerin
Author content
All content in this area was uploaded by Robert Joerin
Content may be subject to copyright.
Learning from Financial Markets:
Auctioning Tariff-Rate Quotas in
Agricultural Trade
ROBERT JÖRIN* and YVAN LENGWILER**
Published in:
Swiss Journal of Economics and Statistics, December 2004.
JEL classification: Q17, F13, D44.
Keywords: Import Quotas, Auctions, Collusion, Treasury Bonds.
* Federal Institute of Technology ETH, Agricultural Economics, CH-8092 Zürich, Switzerland,
robert.joerin@iaw.agrl.ethz.ch
** University of Basel, Department of Economics (WWZ), CH-4003 Basel, Switzerland,
yvan.lengwiler@unibas.ch
1. INTRODUCTION
An important result of the Uruguay Round was the shift from the earlier quantitative
restrictions towards tariff-rate quotas (TRQs). A TRQ is a hybrid of a simple tariff and
a simple quota. An in-quota tariff is applied up to a given quantity. All subsequent
imports are then taxed with the higher over-quota tariff. In principle, TRQs do not
impose a quantity restriction like a simple quota would do. Anyone is free to import as
much as they like as long as they are prepared to pay the over-quota tariff. Yet, in
practice, many over-quota tariffs are so high that they have a prohibitive effect on trade
(TANGERMANN, 1996).
The following two substantial advantages of using auctions for allocating TRQs should
not give rise to too much debate. One advantage is in the domain of competition policy,
the other is fiscal. We briefly explain these two aspects.
First, it is obvious that trade restrictions reduce competition. BHAGWATI (1995)
demonstrates that import quotas create more market power than tariffs, and
MCCORRISTON (1996, p. 372) shows how import quotas create oligopsony power. This
is particularly worrisome since many agricultural markets feature a high and even
increasing concentration on the side of the buyers (DOBSON ET AL, 2000; ROGERS and
SEXTON, 1994).1 Therefore, it matters greatly who acquires the quotas. The fact that
auctions are an anonymous allocation device makes them ideal for allocating the TRQs
in a non-discriminatory fashion, which is in line with the Uruguay Round Agreement
(SKULLY, 2001).
Second, because of the two tier tariff, TRQs create quota rents for those who can import
at the lower in-quota tariff. Auctions offer an attractive means for the government to
collect some of this rent, simply because TRQs are not given away but sold (BERGSTEN
ET AL., 1987).
Yet, despite these great advantages, the practice of TRQ administration is far more
conventional: only 42 or 3.1 percent of the total number of 1,379 TRQs were auctioned
off in 2000 (WTO, 2001). One reason why auctions are rarely used in this domain may
be the opposition governments face from current quota holders who would lose from the
move to auctions.
But agricultural markets might also be a particularly tough terrain for auctions. Given
the tendency to monopsonize agricultural markets, the problem of collusion among
bidders in TRQ auctions is likely to arise. DE GORTER (1999, p. 9) explains the
situation as follows: “It is possible for one group to purchase the entire portion of the
right to import (domestic or foreign), and then withhold part of the licenses to maximize
revenues.” It seems that auctions might not work well for allocating TRQs under these
conditions. Yet, the situation described by De Gorter is contingent on the lack of
potential competition and on the specific design of the auction. The first claim, that a
contestable market does not allow for large rents to the incumbents, is quite obvious.
2
1 It is noteworthy that the concentration in primary product markets is higher than in markets for proc-
essed products, where substitutes are generally more frequent. For Switzerland, Table 1 (appendix)
shows the concentration ratios for different degrees of processing.
We substantiate this claim empirically with the cases of the Swiss white wine market
and the market for Parma ham, where the TRQs have been auctioned off since 1997
(Chapter 2). The second claim, that auctions can be designed in a way that inhibits or at
least reduces collusion, is based on experiences that have been made by Treasuries
when auctioning off government debt (Chapter 3). We show how an auction with
variable supply can be used to mitigate the danger of collusion. In Chapter 4 we adapt
this auction design for the sale of the TRQs to the Swiss meat market.
2. CONTESTABILITY AND CARTELS: EXPERIENCE
WITH AUCTIONING TRQS
The virtues of the contestability of markets is well known and need no explanation.
Unfortunately, many agricultural markets today are far from this ideal. Cartels are more
the rule than the expection in these markets. However, despite strong opposition of the
involved groups, some TRQs have been auctioned off in recent years. As for
Switzerland, we discuss first how an originally very screwed up system — that of the
Swiss white wine market — was completely liberalized in only a few years with the
help of TRQ auctions. Second, we give evidence of contestability from bidding results
in the Parma ham TRQ auctions.
2.1. The role of auctions in liberalizing TRQs: the case of Swiss
white wine
The case of Swiss white wine illustrates the different administration methods that have
been used for quotas since import restrictions were first implemented in the 1930s
(HANIOTIS, 1998).
Historical shares: Before the Uruguay Round Agreement on Agriculture was
implemented, import quotas were allocated on the basis of historical shares. As the gap
between border- and domestic prices was relatively wide, large rents were collected by
the quota holders. Strong rent-seeking was characteristic and quotas were considered
“historical rights” of the quota holders.
“First come, first served”: In line with the Uruguay Round Agreement on Agriculture,
TRQs were implemented on 1 January 1996. The Swiss government decided to allow
imports at the lower in-quota tariff on a first come, first served basis. As the gap
between the in- and over-quota tariffs still remained wide, there was a strong incentive
to bring the imported wine through customs as quickly as possible, in order to obtain the
rents. A few big firms were able to organize transport capacities so that the whole
quota was completely used at the end of the first week in January 1996. This rent-
seeking behavior had two distortionary effects: low quality wine displaced high quality
products and overseas wine suppliers were disadvantaged. After the collapse of the
system, all other firms had to import white wine at the high over-quota tariff.
Auctioning TRQs: After these turbulences, the first come, first served system was
replaced by an auction. At the same time, the government expanded the quota amount
3
from 160,000 hl in 1997 to 190,000 hl in 2000. This was a remarkable step towards a
liberalized and market-oriented system. An important element of the auction was the
fact that all interested firms and persons resident in Switzerland were allowed to send
their (sealed) bids to the administration. There is no evidence that bidders successfully
colluded in these auctions. The results of two of these auctions are shown in the
Figure 1. The results reflect the existence of quota rents that were hidden under all the
previous systems of quota allocation. This transparency was a necessary condition for
broad political acceptance of the auction system. Consumers became aware of the fact
that the reason for higher domestic prices was the restriction imposed by the import
quota and not the fact that quota holders had to pay for the right to import. Auctioning
the quotas did not increase the price for imported wine but the rents were transferred to
the government. The TRQ auction put an end to the extensive rent-seeking activities,
and improved access to import rights by breaking up the cartelistic market structures.
The result was a more efficient allocation of TRQs.
Figure 1: Tariff-rate quota auction of white wine by the Swiss Ministry of
Agriculture.
Auction 1997
Swiss Francs per hectoliter
0 20406080
1
10.1-15.0
25.1-30.0
40.1-45.0
55.1-60.0
70.1-75.0
85.1-90.0
100.1-105.0
115.1-120.0
130.1-135.0
145.1-150.0
160.1-165.0
175.1-180.0
190.1-195.0
Auction 2000
Swiss Francs per hectoliter
0 2040608
1
10.1-15.0
25.1-30.0
40.1-45.0
55.1-60.0
70.1-75.0
85.1-90.0
100.1-105.0
115.1-120.0
130.1-135.0
145.1-150.0
160.1-165.0
175.1-180.0
190.1-195.0
0
1 000 hectoliters 1 000 hectoliters
allocated quantity
TRQ = 160 000 hl
average bid
CHF 89.- / hl
allocated quantity
TRQ = 190 000 hl
average bid
CHF 33.- / hl
not accepted bids
lowest accepted bid
CHF 51.- / hl
lowest accepted bid
CHF 18.- / hl
Auction 1997
Swiss Francs per hectoliter
0 20406080
1
10.1-15.0
25.1-30.0
40.1-45.0
55.1-60.0
70.1-75.0
85.1-90.0
100.1-105.0
115.1-120.0
130.1-135.0
145.1-150.0
160.1-165.0
175.1-180.0
190.1-195.0
Auction 2000
Swiss Francs per hectoliter
0 2040608
1
10.1-15.0
25.1-30.0
40.1-45.0
55.1-60.0
70.1-75.0
85.1-90.0
100.1-105.0
115.1-120.0
130.1-135.0
145.1-150.0
160.1-165.0
175.1-180.0
190.1-195.0
0
1 000 hectoliters 1 000 hectoliters
allocated quantity
TRQ = 160 000 hl
average bid
CHF 89.- / hl
allocated quantity
TRQ = 190 000 hl
average bid
CHF 33.- / hl
not accepted bids
lowest accepted bid
CHF 51.- / hl
lowest accepted bid
CHF 18.- / hl
Source: Swiss Ministry of Agriculture.
Auctions as a “Bridge from TRQ regimes to tariff only systems”: When the lowest bid
in the year 2000 TRQ auctions for white wine became relatively small, the government
decided as of 2001 to integrate the TRQ for white wine into a larger global TRQ for
wines of all categories, origins and qualities. Importers can select their wines without
any limits of origin or quality. The new TRQ has not been entirely used because the
consumption of wine is changing significantly towards “less quantity and better
4
quality.” Under these market conditions, the TRQ will not be binding and no danger of
rent-seeking should be expected. Therefore, imports will be allowed again on the first
come, first served basis. The final result after the years of TRQ auctions is a tariff
system that allows imports of wine at the lower in-quota tariffs.
Figure 2: Bid functions of quota holders and potential entrants.
Potential Entrants
TRQ-Auction Parma Ham 1999
0
200
400
600
800
1000
1200
1400
6.0 7.0 8.0 9.0 10.0 11.0 12.0
log of quantity [kg]
CHF / 100 kg
Quota Holders
TRQ-Auction Parma Ham 1999
0
200
400
600
800
1000
1200
1400
6.0 7.0 8.0 9.0 10.0 11.0 12.0
log of quantity [kg]
CHF / 100 kg
cut-off price
632/100 kg
cut-off price
632/100 kg
Potential Entrants
TRQ-Auction Parma Ham 1999
0
200
400
600
800
1000
1200
1400
6.0 7.0 8.0 9.0 10.0 11.0 12.0
log of quantity [kg]
CHF / 100 kg
Quota Holders
TRQ-Auction Parma Ham 1999
0
200
400
600
800
1000
1200
1400
6.0 7.0 8.0 9.0 10.0 11.0 12.0
log of quantity [kg]
CHF / 100 kg
cut-off price
632/100 kg
cut-off price
632/100 kg
Source: Swiss Ministry of Agriculture.
2.2. Evidence for contestability in the Parma ham TRQ auctions
It is reassuring that another agricultural market, the one for Parma ham, does also seem
to be quite contestable. The evidence for this finding stems from a TRQ auction that
was run in this market. The auction was uniform price, and every bidder was allowed to
submit three different combinations of price and quantity. The maximum share
allocated to each bidder was limited to 15 percent of the total TRQ. Figure 2 shows the
bid functions of the quota holders and the potential entrants. The rather high
willingness to pay of the quota holders reflects their special know-how in purchasing
Parma ham from Italy. Retailers who lack this expertise formulated bids that turned out
to be below the cut-off price. But the fact that many retailers continuously participate in
the auctions reflects their latent interest in quotas. These are potential entrants, and it is
only their existence that keeps the incumbents from getting away with lower bids. This
pattern suggests that this market is quite contestable after all.
3. DESIGN OF AN ANTI-COLLUSIVE AUCTION
5
The inherent transparency of auctions stands in stark contrast to the secretive bargaining
and logrolling that is so typical for many existing arrangements. We argue that using
auctions instead of the procedures that are ordinarily used to allocate TRQs not only
improves the efficiency of the allocation (a rather obvious impact), but also provides
new options for agricultural policy, and can have a deep impact on how agricultural
markets function. In this section we entertain the idea that agricultural policy may
benefit from studying financial market institutions.
3.1. Insights from recent theoretical research
Potential collusion in an auction is an obvious headache, especially in markets with few
bidders who know each other well. The same problem, maybe to a lesser extent, causes
Treasuries around the world to worry when they try to sell their government debt. The
most famous case of a breakdown of competition happened in the May 1991 auction of
U.S. Treasury two-year notes. Salomon Brothers was able to acquire control over 94
percent of these notes, and squeezed out large amounts of money after the auction from
traders that had gone short in this note prior to its issue (JEGADEESH, 1993).
Ever since that event (and maybe already before), competition in an auction of a
divisible good, such as Treasury bonds, has been a hotly debated issue. The focus was
initially put on the distinction between uniform price and pay-your-bid auctions. In
both auction formats, bidders are requested to submit multiple price-quantity pairs,
expressing the quantity they are at most willing to purchase at any given price. In the
uniform price auction, the auctioneer sets a cut-off price and all bids at or above the cut-
off are honored, but all honored bids only pay the cut-off price. In the pay-your-bid
auction, the honored bids pay the price bids at which they were submitted.2
If quantity is fixed, it is most likely that bids will have to be rationed. Typically, only
marginal bids are rationed pro rata. Wilson (1979) has shown that the uniform price
divisible good auction of this type has many equilibria, and especially equilibria with
arbitrarily low cut-off prices. The problem is that this auction format invites quasi-
collusive behavior, where all bidders submit very low quantities at high prices, but large
quantities at low prices. BACK and ZENDER (1993) have shown that similarly bad
equilibria (from the seller's point of view) exist if supply is uncertain (in their example,
due to small bidders who submit only quantity bids without specifying a price).
Several designs have been proposed that have the purpose of removing the low price
equilibria. BACK and ZENDER (1993) have concluded that the pay-your-bid auction
format is less vulnerable to the implicit collusion underlying the low price equilibria.
The reason for this finding is that in the uniform price auction bidders are effectively
interested only in a single point on their bid function, namely the one at the cut-off
price. All other points on the bid function can be used for strategic purposes to make
the collusion self-enforcing. This is not true in the pay-your-bid auction because there,
6
2 Prima facie one may think that the pay-your-bid auction harvests a far superior revenue, since the
bidders' rent is taxed away. But such a conclusion is faulty because the bidders' behavior is not
independent of the auction rules (SMITH, 1966).
all prices on the bid function that exceed the cut-off price are payoff relevant for the
bidder.
Another remedy that has been proposed addresses the supply side. The conclusion that
uniform price auctions are particularly vulnerable to collusion is not valid if the seller
can change the offered quantity after observing the bids (LENGWILER, 1999; BACK and
ZENDER, 2001; MCADAMS, 2002).3 The intuition for this result is straightforward: the
Wilson-type quasi-collusive equilibria rely on very steep bid functions that force the
seller to choose a low cut-off price in order to sell the whole quantity that he is
committed to. If the seller has the possibility to reduce the supply, though, he will do so
if bidders submit steep bid functions to ensure a low price. As a result, steep bid
functions do not affect the cut-off price very much, but simply reduce the quantity that
is sold. Yet, this hurts bidders, so that sticking to the quasi-collusive strategy is not an
equilibrium anymore. In other words, the elasticity of the supply limits the elasticity of
the equilibrium bid functions, and thus removes the quasi-collusive low price equilibria.
The ability of the seller to reduce quantity if needed thus entails a powerful strategic
effect that removes the ability of the bidders to collude via steep bid functions.
Other design changes have also been proposed. For instance, KREMER and NYBORG
(forthcoming) have recently suggested a very minor amendment to the standard auction
with supposedly large effects. Instead of rationing only marginal bids, they suggest that
all the successful bids should be rationed instead. They show that this simple change
destroys all low price equilibria. Other designs that depart more fundamentally from
the uniform- and pay-your-bid designs have been analyzed, eg., by AUSUBEL
(forthcoming) and PERRY and RENY (2001). These designs offer the prospect of
generating Pareto efficient allocations — something that the standard auctions we have
discussed so far cannot deliver. Yet these designs have not been tested until now, and
we feel that more laboratory experiments and small field experiments are needed before
we would recommend them for application in the substantial market that is the subject
of our study. So, in order to avoid being too adventurous, we constrain ourselves to
mechanisms that have already been used in real markets.
3.2. Empirical and experimental evidence
There is a lot of experience with pay-your-bid and with uniform price auctions with
fixed supply. There is also much experience with uniform price auctions with elastic
supply. We review the empirical evidence with respect to these auctions in the
following. We start with some recent experimental evidence and then move to
empirical evidence using government auction data.
SADE, SCHNITZLEIN, and ZENDER (2003) run a series of experiments to shed light on the
relative performance of the different mechanism. They study uniform fixed supply
(UFS), pay-your-bid fixed suppy (PFS), and uniform variable supply (UVS) auctions.
7
3 LICALZI and PAVAN (forthcoming) is a bit different. They study the implications if the seller commits
to some increasing supply schedule before the auction. They show that this device avoids low price
equilibria as well.
They run some experiments with students that are new to the material, but they also run
some sessions with professionals from the Treasury bond market (professional traders
on both sides of the market). For both groups of subjects, the experiments yield the
following ordering in terms of revenue: UVS ≥ UFS > PFS. This is partially good news
and partially bad news for theorists. First of all, the inequality UFS > PFS is a striking
reversal of the prediction of BACK and ZENDER (1993) and is much more in line with
FRIEDMAN's (1959) intuition.4 The weak inequality UVS ≥ UFS is at least consistent
with the theory of the effects of making supply elastic, but it is not a strong verification
of these predictions either.
There is also plenty of data on real auctions, namely the auctions for various
government assets, mostly Treasury bills and bond. In one of the following studies the
asset is foreign currency, in another it is gold. In sum, the results in terms of generated
revenue are inconclusive. Some studies find an advantage of the uniform auction
(TENORIO, 1993; UMLAUF, 1993; FELDMAN and REINHART, 1995; MALVEY and
ARCHIBALD, 1998), others find evidence that the pay-your-bid auction worked better for
the seller (SIMON, 1994; HORTAÇSU, 2002).
The auctions that are studies in these papers all have an ex ante fixed supply. Luckily,
we can test the variable supply auction as well, because there are at least two Treasuries
that explicitly determine the sold quantity only after observing the bids. These
Treasuries are the Swiss and the Finnish. KELOHARJU ET AL. (2002) study empirically
the performance of the Finnish Treasury auctions and find no evidence for low price
equilibria of the Wilson-Back-Zender type. HELLER and LENGWILER (2001) study the
Swiss Treasury auctions and also find no evidence for excessive profits of the bidders.5
In the case of Switzerland, this may be surprising because two large bidders regularly
acquire a large share of the total issue. There is, however, significant potential
competition in the form of a large number of smaller banks that have the right to
participate in the auction but typically do not. Excessive profits of the incumbents
would incite these potential rivals to enter the auctions and quickly compete away any
excessive profits. The lack of collusion may be even more surprising in the Finnish
case, because there, participation in the auction is restricted to primary dealers (between
five and ten entities during the sample of the study), so there is no imminent potential
competition.6 Yet, collusion does not survive in this market either. This is so despite
the fact that competition is restricted and the auctions keep being repeated on a regular
basis, which should ease collusion.7 Yet, as an empirical fact, the design seems to work
fine. We conclude that the ability of the seller to restrict supply after the fact is a potent
antidote against collusion.
8
4 See also CHARI and WEBER (1992).
5 The cut-off prices do not significantly differ from the pre-auction secondary market prices of the
bonds that are about to be issued.
6 In the longer run, there could of course still be potential competition from banks that acquire primary
dealership status.
7 The Folk Theorem of repeated games tells us that many more payoff combinations (including "quasi-
collusive allocations") can be supported as equilibria in an infinitely repeated game.
3.3. The Swiss Treasury auction rules as a model
The Swiss Treasury has a long experience with selling government debt through an
auction with variable supply. The rules they use seem to work well in an environment
which is characterized by a potentially significant concentration of bidders. After all,
the two largest Swiss banks typically buy between two thirds and three quarters of an
issue. Yet, no events of market cornering or excessive profit taking have been observed
in these auctions. For this reason we will use the rules set out by the Swiss Treasury as
a model for TRQ auctions. We begin by explaining the rules in some detail, and then
briefly discuss an example of one such auction.
The auctions take place on a bimonthly basis. Shortly before the auction, the Treasury
announces the characteristics of the bonds that are to be issued, such as time to
maturity, coupon, and callability. The Treasury also states the maximum number of
bonds that will be issued. Almost always, however, this maximum supply is so large
that this constraint is not binding. The Treasury also reserves the right to cancel an
auction if it does not deem the bids satisfactory. The circle of bidders is restricted to
institutions holding accounts with the Swiss National Bank (SNB), currently some 400
entities. Under the current regulation, these are all banks based in Switzerland. All
bidders are treated equally, i.e. there are no primary dealers.
The bidders are invited to submit as many price-quantity bids as they wish.8 The bids
have to be in by 12 o'clock on the day of the auction. Until spring 2001, the bids were
submitted by fax, then the Treasury switched to a proprietary electronic platform. After
all the bids have been submitted, the Treasury decides on the cut-off price. With this
cut-off price, the Treasury also decides on the quantity that is sold; it is simply the
amount of bids that have been submitted at or above the cut-off price. The Treasury
typically chooses a point on the bid function where it is the flattest, or maybe one price
tick below. Figure 3 depicts the aggregate bid function of a sample Treasury bond
auction, as well as the Treasury's choice. We see here clearly how the Treasury and the
aggregate bids together effectively determine the price and the quantity
simultaneously.9 This system has several advantages:
1. If demand for the bonds is weak, few will be sold. The system automatically adapts
the supply to the demand of the bidders.
2. As a consequence, price volatility is reduced.
3. The fact that the quantity is determined only after the bids have been submitted has
far reaching strategic effects. As discussed before, this is a potent measure against
collusion.
9
8 In addition to submitting price-quantity pairs, bidders may place pure quantity bids without a price.
These unpriced bids have to be less than CHF 100,000 each. Unpriced bids are relatively
unimportant. They account for about 7 percent of the total amount sold in an average Treasury bond
auction.
9 In this example, the bond that was auctioned then had ten years time to maturity and a 4 percent
annual coupon. The Treasury chose a cut-off price of 103.75, which is equivalent to a yield to
maturity of 3.55 percent, and sold bonds with a total face value of CHF 1.191 billion.
This third property is of particular interest for the application we have in mind. Clearly,
in many agricultural markets there is great scope and opportunity for monopsony power.
Yet, the flexibility of the auctioneer to restrict the supply makes it very difficult for the
bidders to enforce a low price, and thus ensure a large profit.
Figure 3 : Auction of Swiss treasury bonds (February 2002).
103.4
103.5
103.6
103.7
103.8
103.9
104
0 200 400 600 800 1000 1200 1400 1600
cumulated bids (Mio. CHF)
price
allocated amount
CHF 1'191Mio.
cut-off price 103.75
Source: Swiss National Bank.
4. RULES FOR TRQ-AUCTIONS: THE CASE OF THE
SWISS MEAT MARKET
4.1. The present system: rent-seeking, bargaining, and a
government sanctioned cartel
There is not much competition in the Swiss meat market. But the bargaining and
collusion between professional market participants does not take place in smoke-filled
rooms. To the contrary, in Switzerland, the market for meat is run by a government
sanctioned interest group called Proviande.10 Farmers, butchers, traders, importers, and
10
10 www.proviande.ch
consumers are members of this institution.11 The government regulates the market by
adapting the import quantity periodically to the changing conditions on the market.
Prior to this decision, the market players represented in Proviande bargain and make a
recommendation to the Ministry of Agriculture. Of course, the Ministry remains
nominally in charge, but typically does not care to deviate from Proviande’s
recommendation.
The stated goal of Proviande is to ensure a “high quality and competitive meat
business,” to “facilitate cooperation among the representatives of the meat industry,”
and to “represent the meat industry” (presumably in the policy debate). These stated
aims make clear that Proviande is a classic cartel.
Proviande allocates import quotas to its members proportional to the amount of meat
and cattle they buy from the domestic producers. This system is known as “prise en
charge” system, under which tariff quota access is contingent on the purchase of
domestic products. The Swiss competition commission has repeatedly criticized this
system, and similar criticism has been expressed in the Uruguay Round negotiations
(WTO, 1996). The system violates fundamental principles of competition policy and
has enhanced the process of concentration in the meat industry (see Table 1 in the
appendix). A recent analysis by ABDULAI (2002, p. 679) shows that price transmission
in the Swiss pork market is asymmetric, “in the sense that increases in producer prices
that lead to declines in marketing margins are passed on more quickly to retail prices
than decreases in producer prices that result in increases in the marketing margins.” We
claim that the use of TRQ auctions lead to a more competitive and transparent meat
market.
4.2. The new system: auctions with variable supply
Instead of the bargaining and logrolling that takes place within Proviande, in the new
system that we propose, the Ministry of Agriculture allocates TRQs as a result of an
auction with variable supply. We propose the following sequence of events:
1. Following the principle of contestable markets, entry barriers should be as small as
possible. As a result, all firms interested in purchasing import quotas are free to
submit bids. The auctions have to be announced to a large enough audience, and
sufficient time has to be reserved between the announcement and the deadline for bid
submission.
2. Based on the received bids, the Ministry of Agriculture allocates the TRQs. Because
the total quantity is not fixed in advance, the system gives the Ministry a certain
leverage by which it can trade off quantity versus price. As an example, the Ministry
could, if it wishes to do so, implement a domestic price target through its supply
decisions. Suppose bidding in one auction was stronger than expected (the result of
an optimistic expectation of the meat importers for meat demand). This is reflected in
the top half of Figure 4 by the “optimistic” demand (DH).12 To support the price
11
11 The influence of consumers is marginal: only 2 of the 14 members represent consumer organizations.
12 We assume the simple model of a small importing country; t1: in-quota tariff, t2: over-quota tariff;
pW : world market price, pIMP = pW + t1 ; S : supply.
level P*, the Ministry would expands the aggregate amount of TRQs from Q0 to Q*.
Conversely, if bidding is unexpectedly weak (low demand DL), the Ministry would
reduce the total amount of TRQs, as shown in the bottom half of Figure 4.
Of course, the Ministry is constrained by the WTO access commitments. It cannot
arbitrarily reduce the total amount of TRQs. This constraint is not binding, however, on
an auction by auction basis, but rather over longer averages and for larger classes of
various kinds of meat. The Ministry should therefore not face any real constraint from
this side. The WTO constraint will become relevant only if bidding is weak for an
extended period of time or for many kinds of meat at the same time. In that case, the
Ministry will simply satisfy all bids and the price of the TRQ will drop to zero (or the
lowest submitted bid price). Alternatively, if this situation emerges, the Ministry could
also decide to drop the need to have TRQs for importing meat altogether until the next
auction.
The specific, practical implementation needs to be adapted from the way Treasury bond
auctions are administered. For instance, we do not consider the exact formal
requirement of bid submissions to be of any significance. No electronic platform is
needed and simple written bids are perfectly acceptable.13 One might also think that the
frequency of the auctions would have to be adapted. It would not make much sense to
auction TRQs which are valid only for e.g. two months. To allow for sensible planning
by the importerts, TRQs should be valid for at least six months. Alternatively, TRQs
with one year validity could also be auctioned at higher frequency, with overlapping
periods of validity. For instance, there could be a December auction for importing meat
from the following January on during the next twelve months. There would also be a
March auction for importing from April on, again for twelve months, and so on.14
12
13 In fact, any sophisticated requirements for bid submissions would only constitute an entry barrier.
14 Such an overlapping structure would have the additional advantage that the Ministry would be more
at ease with reducing supply significantly in one auction if colluding bidders try to enforce a low price
by submitting steep bid functions. Because some TRQs are already on the market (from previous
auctions), the Ministry can reduce the supply in any particular auction without running a great risk of
causing shortages.
Figure 4: Import regulation under a regime of auction with variable supply.
Import expansion in case of optimistic assessment
Price PW + t2
S*
P
H
P*
D
H
PIMP
t1
PW
Quantity
Q
0
Q*
Import reduction in case of pessimistic assessment
Price
S*
Q
0
Quantity
D
L
P
L
PIMP
Q*
P*
t1
PW + t2
13
PW
4.3. Benefits of the new system…
One of the most important benefits of this system is certainly that the considerable
private information and expertise of the importers affects the total amount of TRQs that
are issued. It is quite obvious that an auction that gives bidders an incentive to bid close
to their true expectation of the future development of the market is much better suited to
predict the aggregate need for TRQs than the negotiations that take place within
Proviande today. Bidding not in accordance to the best estimate of market development
is potentially very costly for a bidder: if he bids too high, he is likely to make a loss; if
he bids too low, he runs the risk of not being allocated sufficient quota and lose the
business altogether. But the auction not only regulates the total amount of TRQs, it also
allocates the TRQs to those bidders that are most likely to make the best use of them.
The bidders that have the best outlook for gainful importing will bid highest, and will
therefore receive the largest share of the TRQs. Another advantage is that the rents that
the quota holders were able to reap will be transferred to the government. This form of
taxation is distortion free because the price of the TRQs is simply a measure of scarcity.
Moreover, as we argued before, the variable supply feature constitutes an important
strategic weapon against collusion. Finally, because it makes the aggregate amount of
TRQs flexible, it offers the Ministry a way to smoothly phase out a tariff-quota regime
in favor of a tariff-only regime.
4.4. …and limits
There is one desirable property of the equilibrium allocation that the system we propose
does not deliver, and that is efficiency in the sense of maximizing aggregate welfare.
The strategic situation induces bidders to misrepresent their true demand to some
extent. Thus, it is possible, even likely, that in equilibrium some bidders get more,
some less TRQs than what is optimal for them, given the price they had to pay and their
market expectation. The system will almost surely deliver a less than perfectly efficient
allocation. But there is no doubt that the allocation it delivers is much closer to
efficiency than the arbitrary allocation that the current practice produces. The new
system, we think, is good, but it is not perfect. And we would like to argue that we
should not allow perfection to be the enemy of enhancement.
But the prefect (efficient) mechanism exists! We mentioned earlier that recently some
mechanisms have been designed that should produce an efficient allocation in theory
(AUSUBEL, forthcoming; PERRY and RENY, 2001). We think that these are very
interesting possibilities, and yet we shun from blindly applying theory to politically
sensitive markets such as agriculture. Despite its lack of welfare maximization, the
auction we propose has the undeniable advantage that it is based on a true and tried set
of rules, which is a priceless asset in this domain.
14
5. CONCLUSIONS
Most TRQs today are allocated in an intransparent and hardly efficient way. In many
cases, TRQs are not much more than sinecures for their owners. Agricultural policy
measures, which were intended to protect domestic producers, have effectively given
rise to arbitrary beneficiaries and do no longer contribute to the original goals.
Moreover, the inefficient allocations of the TRQs imposes a significant deadweight loss
on society.
We argue in this paper that auctions offer an attractive way out of this cul-de-sac. The
specific auction design we propose is inspired by a method that has been used very
successfully for auctioning off government debt. The variable supply auction offers
several benefits. First, it allocates TRQs to the users with the highest willingness to
pay, and therefore, one hopes, to the most efficient users. Second, it adapts the total
amount of quotas according to market needs. Third, auctions divert rents from the
quota holders to the government and constitute a non-distortionary form of taxation.
Fourth, the allocation of TRQs is perfectly transparent. If an importer does not receive
a sufficient quota, it is not because he was not nice enough to his fellow cartel members,
but simply because he did not bid high enough. Fifth, the auction design we propose
gives the administration a flexible bridge for slowly phasing out import restrictions in
favor of a tariff-only regime. Finally, the variable supply feature of the auction is
designed to keep the potential of bidders for collusion in check.
The implementation of this new system is currently being debated in Switzerland.
Given the many advantages that these auctions offer, we are eager to observe their
functioning in practice.
REFERENCES
ABDULAI, AWUDU (2002), “Using Threshold Cointegration to Estimate Asymmetric
Price Transmission in the Swiss Pork Market,” Applied Economics, 34, 679–687.
AUSUBEL, LAWRENCE M. (forthcoming), “An Effcient Ascending-Bid Auction for
Multiple Objects,” American Economic Review [www.ausubel.com/auction-
papers/efficient-ascending-auction-r.pdf].
BACK, KERRY and JAIME F. ZENDER (1993), “Auctions of Divisible Goods: On the
Rationale for the Treasury Experiment,” The Review of Financial Studies, 6, 733–
764.
BACK, KERRY and JAIME F. ZENDER (2001), “Auctions of Divisible Goods with
Endogenous Supply,” Economics Letters, 73, 29–34.
BERGSTEN, C. FRED, KIMBERLY ANN ELLIOT, JEFFREY J. SCHOTT, and WENDY E.
TAKACS (1987), Auction Quotas and United States Trade Policy, Institute for In-
ternational Economics, Washington D.C.
BHAGWATI, JAGDISH N. (1965), On the Equivalence of Tariffs and Quotas, in: Baldwin,
R.E. et al. (eds.), Trade Growth and the Balance of Payments, Chicago.
15
CHARI, VARADARAJAN V. and ROBERT J. WEBER (1992), “How the U.S. Treasury
Should Auction Its Debt”, Federal Reserve Bank of Minneapolis Quarterly Review,
Fall, 3–12.
DE GORTER, HARRY (1999), “Market Access, Export Subsidies and Domestic Support
Measures: Issues and Suggestions for New Rules,” World Bank Conference on
Agriculture and The New Trade Agenda, October 1–2, 1999.
DOBSON, PAUL, ROGER CLARKE, STEPHEN DAVIS, and MICHAEL WATERSON (2000),
“Buyer Power and its Impact on Competition in the Food Retail Distribution Sector
of the European Union,” paper prepared for the 4th INRA-IDEI Conference,
Toulouse [www.idei.asso.fr/Commun/Conferences/Food%20Processing/
Juin2000/%20Papiers/Dobson.pdf].
FELDMAN, ROBERT A. and VINCENT REINHART (1995), “Flexible Estimation of Demand
Schedules and Revenue Under Different Auction Formats,” IMF Working Paper
95/116 (November).
FRIEDMAN, MILTON (1959), “Testimony in Employment, Growth, and Price Levels,”
Hearings before the Joint Economic Committee, 86th Congress, 1st Session, October
30, 3023–3026.
HANIOTIS, TONI (1998), “Auctions as a Trade Instrument,” in: Cottier, T. and
Mavroidis, P.C. (eds.), State Trading in the Twenty-First Century, The University of
Michigan Press.
HELLER, DANIEL and YVAN LENGWILER (2001), “Should the Treasury Price-
Discriminate? — A Procedure for Computing Hypothetical Bid Functions,” Journal
of Institutional and Theoretical Economics, 157, 413–429.
HORTAÇSU, ALI (2002), “Mechanism Choice and Strategic Bidding in Divisible Good
Auctions: An Empirical Analysis of the Turkish Treasury Auction Market,”
Manuscript, University of Chicago [home.uchicago.edu/~hortacsu/
ttreas.pdf].
JEGADEESH, NARASIMHAN (1993), “Treasury Auctions Bids and the Salomon Squeeze,”
Journal of Finance, 48, 1403–1419.
KELOHARJU, MATTI, KJELL G. NYBORG, and KRISTIAN RYDQVIST (2002), “Strategic
Behavior and Underpricing in Uniform Price Auctions — Evidence from Finnish
Treasury Auctions,” Nota di Lavoro 25.2003, Fondatione Eni Enrico Mattei
[papers.ssrn.com/sol3/papers.cfm?abstract_id=301922].
KREMER, ILAN and KJELL G. NYBORG (forthcoming), “Divisible Good Auctions — The
Role of Allocation Rules,” Rand Journal of Economics.
LENGWILER, YVAN (1999), “The Multiple Unit Auction with Variable Supply,”
Economic Theory, 14, 373–392.
LICALZI, MARCO and ALESSANDRO PAVAN (forthcoming), “Tilting the Supply Schedule
Enhances Competition in Uniform Price Auctions,” European Economic Review.
MALVEY, PAUL F. and CHRISTINE M. ARCHIBALD (1998), “Uniform-Price Auctions:
Update of the Treasury Experience,” Technical Report, Office of Market Finance,
U.S. Treasury.
16
MCADAMS, DAVID (2002), “Modifying the Uniform-Price Auction to Eliminate
‘Collusive-Seeming Equilibria’,” Manuscript, MIT [www.mit.edu/~mcadams/
papers/mupa.pdf].
MCCORRISTON, STEVE (1996), “Import Quota Licenses and Market Power,” American
Journal of Agricultural Economics, 78, 367–372.
PERRY, MOTTY and PHILIP J. RENY (2001), “An Effcient Multi-Unit Ascending
Auction,” Manuscript [www.ma.huji.ac.il/~motty/papers/ascend.
pdf].
ROGERS, RICHARD T. and RICHARD J. SEXTON (1994), “Assessing the Importance of
Oligopsony Power in Agricultural Markets,” American Journal of Agricultural
Economics, 76, 1143–1150.
SADE, ORLY, CHARLES SCHNITZLEIN, and JAIME F. ZENDER (2003), “Competition and
Cooperation in Divisible Good Auctions: An Experimental Examination,”
Manuscript, AMD Conference, Milano [www-bus.colorado.edu/faculty/
Zender/papers/auction-16aug03.pdf].
SKULLY, DAVID W. (2001), “The Economics of TRQ Administration,” ERS Technical
Bulletin No. 1893, April [www.ers.usda.gov/publications/tb1893/].
SIMON, DAVID P. (1994), “The Treasury’s Experiment with Single-Price Auctions in the
Mid 1970’s: Winner’s or Taxpayer’s Curse?” Review of Economics and Statistics,
76, 754–760.
SMITH, VERNON L. (1966), “Bidding Theory and the Treasury Bill Auction: Does Price
Discrimination Increase the Bill Price?” Review of Economics and Statistics, 48,
141–146.
TANGERMANN, STEFAN (1996), “Implementation of the Uruguay Round Agreement on
Agriculture: Issues and Prospects,” Journal of Agricultural Economics, 47, 315–337.
TENORIO, RAFAEL (1993), “Revenue Equivalence and Bidding Behavior in a Multi-Unit
Auction Market: An Empirical Analysis,” Review of Economics and Statistics, 75,
302–314.
UMLAUF, STEVEN R. (1993), “An Empirical Study of the Mexican Treasury Bill
Auction,” Journal of Financial Economics, 33, 313–340.
WORLD TRADE ORGANIZATION (1996), “Trade Policy Review Body: Review of
Switzerland,” Press Release TPRB/31 and Summary of Secretariat's Report
[www.wto.org/english/tratop_e/tpr_e/tp31_e.htm].
WORLD TRADE ORGANIZATION (2001), “Tariff Quota Administration Methods and
Tariff Quota Fill,” Background Paper by the Secretariat, May 2001.
WILSON, ROBERT (1979), “Auctions of Shares,” The Quarterly Journal of Economics,
93, 675–689.
SUMMARY
17
Most tariff-rate quotas (TRQs) around the world are allocated in rather intransparent
ways. Yet, according to the fundamental WTO principles, TRQs are supposed to be
allocated in a transparent, equitable and non-discriminatory manner. We argue that
auctions offer a promising way to achieve these goals, and at the same time enhance the
efficiency of the allocation. Moreover, they provide the means to slowly phase out a
quota regime in favor of a tariff-only regime. The biggest challenge in this endeavour is
to find an auctioning procedure that operates also well in situations otherwise prone to
collusive bidding behavior. We draw from experiences made in government debt
auctions to come up with a system that would be resistant to collusion.
ZUSAMMENFASSUNG
Die Zollkontingente in der WTO werden weltweit nach Kriterien verteilt, die nicht
transparent sind. Gemäss den Prinzipien der WTO sollten Zollkontingente jedoch den
Importeuren nach einem Verfahren zugeteilt werden, das transparent ist und niemanden
diskriminiert. Wir zeigen, dass die Versteigerung dieses Ziel erreicht und eine
effizientere Verteilung der Kontingente gewährleistet. Versteigerungen können über-
dies auch dazu dienen, ein bestehendes Quoten-Regime durch ein reines Zollsystem
abzulösen. Die wichtigste Herausforderung besteht nun darin, ein Versteigerungs-
verfahren zu finden, das auch in einer Situation von nur wenigen Bietern funktioniert.
Aufgrund der Erfahrung mit der Auktion von öffentlichen Anleihen des Bundes
schlagen wir ein Verfahren vor, das gegen Kollusion unter den Bietern resistent ist.
RÉSUMÉ
La plupart des contingents tarifaires dans le monde sont distribués de manière assez
opaque. Les principes fondamentaux le l’OMC stipulent pourtant que les contingents
tarifaires soient distribués de manière transparente, équitable et non discriminatoire.
Nous considérons que les enchères sont une voie prometteuse pour atteindre ces buts;
pareillement pour améliorer l’efficience de la distribution des contingents. En plus elles
sont un moyen pour passer sereinement d’un système de contingents à un système pur
de tarifs. Le grand défi de cet effort est de trouver une procédure de mise à l’enchère qui
fonctionne aussi dans des situations normalement prédisposées à un comportement
collusoire d’offres. Nous nous inspirons d’expériences faites d’enchères des emprunts
de la Confédération Suisse pour proposer un système résistant à la collusion.
18
APPENDIX
Table 1: Concentration ratios for raw, intermediate and processed product
imports to Switzerland in 1998.
Tariff-rate quota Number of Concentration
importers ratio CR4 in %
Processed products 36
Red wine 1000 10
White wine 500 17
Dried ham 83 33
Dried meet 51 40
Sausage 97 40
Corned Beef 30 53
Fontal (cheese) 59 58
Intermediate products 67
Potatoes (for consumption) 84 61
Poultry 86 67
Eggs (for consumption) 25 74
Raw products 90
Frozen vegetables 40 66
Wheat 29 67
Wheat durum 19 77
Milk powder 17 84
Lamb & goat meet 18 90
Potatoes (for processing) 6 93
Loins (beef) 6 98
Eggs (for processing) 7 98
Seed potatoes 8 98
Slaughterhouse by-products 6 99
Veal 2* 100
Pork 2* 100
Beef for Buendnerfleisch 2* 100
Source: Report from the Federal Council, dated February 24th 1999, on
custom tariffs measures 1998 (allocation of tariff-rate quotas).
* The " Viehbörse" imports on behalf of its members (approx. 1 500 butchers);
The "GVFI", Association for the Import of Cattle and Meet, imports on behalf
of large whole sellers
CR4: market share of the biggest 4 firms
19
