EC341: Mathematical Economics 2:

Mechanism Design and Alternative Games

Organisation

Topic 1: Mechanism design (weeks 1-4: JC)

– Auctions
– General mechanism design
Topic 2: network game theory (weeks 5-6; JC)
– Games played on networks
– Games of network formation
Topic 3: Evolutionary game theory (weeks 7-10; JM)
Review

EC341 2022-3: Auctions 1 2

Materials
There are lecture notes on the Moodle page
– students are encouraged to read lecture notes and slides and to watch videos before
lecture
– Some topics will be recorded and short video lectures
– Some topics will be presented ‘live’
– JC lectures will consist of one hour face-to-face and one hour ‘live’ on Teams per week
– JM lectures are face-to-face
Assessment
– There are four sets of class problems for the module; these will be discussed in seminars –
you may be called on but will not be marked
– A 2000-word essay will count for 20%
– A 2-hour exam (primarily computational) will count for 80%

EC341 2022-3: Auctions 1 3

Outline
Auctions
– Auction basics
– Analysis of standard types
– Relaxing the assumptions
Mechanisms
– Optimal auctions
– General mechanism design

EC341 2022-3: Auctions 1 4

Auctions
Common methods
– Examples
– Elements
– Private values, sealed-bids
– Equilibrium
Symmetric independent private values
Envelope Theorem
Risk Aversion

EC341 2022-3: Auctions 1 5

Auction Reading
– Material covered in
• Krishna, V. (2010) Auction Theory London: Academic Press (Second Ed.) or
• Klemperer, P. (2003) “A Survey of Auction Theory” at: http://www.nuff.ox.ac.uk/users/klemperer/VirtualBook/07_Chapter1.pdf
– Also relevant:
• Optimal (max seller revenue) auction mechanism - Myerson, R. (1981) “Optimal Auction Design” Mathematics of Operations Research,
Vol. 6, No 1, pp 58-74 and Riley, J. and W. Samuelson (1981) “Optimal auctions” American Economic Review, 71(3), 381-392.
• Auctions with risk-averse bidders - Holt, C. (1980) “Competitive bidding for contracts under alternative auction procedures” Journal of
Political Economy, 88(3), 433-445.
• Asymmetric auctions - Maskin, E. and J. Riley (2000) “Asymmetric auctions” Review of Economic Studies, 67(3), 413-438.
• Auctions with budget constraints Che, Y-K and I. Gale (1998) “Standard auctions with financially constrained bidders” Review of
Economic Studies, 65(1), 1-21.
• Auctions with the possibility of resale – Hafalir, I. and V. Krishna (2008) “Asymmetric auctions with resale” American Economic Review
98(1), 87-112.
• All-pay auctions applied to lobbying - Baye, M., Kovenock, D. and C. de Vries (1993) “Rigging the lobbying process: an application of the
all-pay auction” American Economic Review, 83, 289-294.
• The many laboratory experiments on auctions are surveyed by Kagel, J. (1995) “Auctions: A survey of experimental research,” pp. 501-
585 in Kagel and Roth (eds.) The Handbook of Experimental Economics. Princeton, NJ: Princeton University Press.

EC341 2022-3: Auctions 1 6

Common examples
Bids are open (ascending or descending prices)or sealed (first- or second-price); tiees resolved according
to preset rule (e.g. coin toss)
Open bids
– Ascending (English auction) – auctioneer publicly discloses current high bid, raises price (or acknowledges bids) until only one
bidder remains. Winner pays last price disclosed. Strategy = at what price to drop out
– Descending (Dutch or clock auction) – price starts high and reduces until first bidder accepts (or until a minimum level is reached).
Winner (if any) pays the price they accepted. Strategy = highest acceptable price
Sealed bids – each bidder submits a single price bid to the auctioneer; high bidder wins
– First-price – winner pays the amount they bid 1PSB
– Second-price (Vickery auction) – winner pays the second-highest bid 2PSB
– Strategy – how much to bid
Other interesting variants
– Sealed-bid auction with a reservation price - winner is highest bid above (or no-one)
• First-price: winner pays their bid
• Second-price: winner pays the larger of or the second-highest bid
– Auction with entry (or continuation) fees a sealed-bid auction in which every bidder pays a (common) entry fee. Strategy =
whether to enter and how much to bid
– All-pay auction is a sealed- or open-bid auction in which each bidder pays their bid. Models queuing, costly competitions

EC341 2022-3: Auctions 1 7

Values
Auction models assume that each bidder has beliefs about the value of the object to them.
The auction rules determine who wins the object and how much the winner (and perhaps
others) must pay. Utility of winner is (expected) value of object - payment
Usually
– we assume that the value of not winning the object is the same () for all bidders, regardless of who else wins
– the winner is the player with the highest value – but not always.
The values may be private (different for each bidder, known only to that bidder); if not, they
may be common (same for each bidder, but bidders may have different information)
Values are random; the signal received by a bidder about their value may be independent of
the signals received by other bidders, or jointly distributed.
We start with the simplest case: independent private values

EC341 2022-3: Auctions 1 8

Independent private values
Definition 1: a sealed-bid auction with independent private values is where
– is a (finite) set of players or bidders;
– is the set of player ’s possible values – and is the set of vectors of values;
– is the cumulative distribution function over ;
– associates to each vector of bids a distribution on that picks the winner; and
– is the payment function; player pays when the vector is submitted and player is the winner.
– We will sometimes reduce this auction description to
Discussion
– Because the private values are independent,
• the vector of values is drawn according to the product distribution .
• For correlated or affiliated values, we work with a joint distribution ; bidder uses conditional on the signal they received.
– In most auctions, is a degenerate distribution on the highest bidder(s) – but we’ll see that this isn’t always the case
(e.g. when auctioneer’s value payment or when losing bidders’ preferences over winners are counted)
– In most practical cases, is the highest or the second-highest bid, but the general form allows us to model designs
where certain groups or side-conditions are valued, all-pay auctions, profit-share auctions, etc.

EC341 2022-3: Auctions 1 9

More discussion
Sealed-bid auction as a Bayesian Game

Player ’s set of types is

– The set of players is ;

The joint cumulative distribution over types is , which is in the independent values case
–
–
– For each type vector , the state of the world is the state-game where player ’s strategies are
bids and the payoff to a vector is

A full formal description may be very complex (e.g. taking in what players
know about each others’ identities, what they know about who has already
quit (and at what point), etc. We won’t tackle this.

EC341 2022-3: Auctions 1 10

Sealed-bid Auctions
Denote the high bidders when the vector of bids is as

Dutch (in case of ties, winner pays highest price):

First-price:

EC341 2022-3: Auctions 1 11

Strategies and equivalent forms
A pure strategy for player in a sealed-bid auction is a measurable function

If all players use pure strategies , the payoff to player is

Theorem: The open-bid descending auction is equivalent to the sealed-bid first-price auction,
in that both methods describe the same strategic form game, strategy sets and payoff
functions.
– Proof: ’s pure strategies in both auctions are the measurable functions ; a rule telling how to play at
each information set. In the 1PSB, the information sets are private signals ; in the open descending
auction they are the current prices. Also, in both auctions a strategy vector leads to the same
outcome. The winner submits the highest bid (or accepts the highest price) and pays that price. QED

EC341 2022-3: Auctions 1 12

Equilibrium
In view of the Bayesian game interpretation (slide 8), we use Bayesian equilibrium at the interim stage.
– Reminder: in Bayesian equilibrium, each player maximises expected payoff. The expectation can be taken:
• ex ante – before type is known;
• interim – when types are known but opponents’ moves are not; or
• Ex post – when both types and others’ moves are known.
– Let be a strategy vector, where is player ’s bid when their value is . Define be the bids of the other players given their
types and strategies . We are interested in ’s expected payoff from bidding an amount when ’s value is , and the other
players follow the strategies

– This is the expectation over types of other players of the probability wins times the private value minus the ’s expected
payment when or another player wins. is the cumulative distribution of other players’ values.
– The payoff only depends on the bid that the type of player makes, in accord with the usual Bayesian game idea of a Nash
equilibrium among all types of all players
Definition: strategy vector is an equilibrium if for every

• Remark: this describes equilibrium in pure strategies, which may not exist (but an equilibrium in pure strategies is also a mixed strategy
equilibrium)

EC341 2022-3: Auctions 1 13

Some results
Theorem: In a 2PSB (second-price sealed-bid auction), bidding truthfully weakly dominates all
other strategies.
– This assumes nothing about other players’ behaviour, how many there are or their types. Truthful
revelation is useful both for learning about the distribution of true values and for solving the game.
After a 2PSB auction, values are public and a losing bidder might find an acceptable deal elsewhere.
Theorem: In a 2PSB auction, the strategy in which everyone bids truthfully is an equilibrium.
Theorem: In an English auction, dropping out at your true value is weakly dominant.
Remark: In a 2PSB auction, there are equilibria other than truthful bidding. For example,
– suppose 2 players have private values independently and uniformly distributed on
– Let ’s strategy be for all
– Let ’s strategy be for all
– will always win and pay 0; neither player benefits from changing their bid, no matter their value.
Note: in what follows, stands for the probability of an event

EC341 2022-3: Auctions 1 14

Symmetric independent private values case
We assume
– There is only one indivisible object for sale
– The seller will sell the object for any price
– There are bidders, denoted
– [symmetric private values] All bidders have values dawn from the same set , which can be a closed, bounded interval () or
all nonnegative numbers () . Each bidder knows their own value. The random values are independently drawn from the
same distribution (cdf ) over
– [Continuity] for each , is continuous and its density function is both continuous and positive everywhere in
– [Risk neutrality] all buyers seek to maximise their expected profit (value minus payment )
– All these assumptions are common knowledge among the bidders
Remarks
– Since every buyer knows their private value, information (esp. regarding other bidders’ values) has no impact on their
private value.
– In a more general model, the bidders may not know their private values with certainty, information about others’ values
(or signals of value) may lead the bidder to update their beliefs about
– Even after winning, a bidder learns that the others’ values were less than or equal to their own. So expected profit uses
expected value conditional on having won.
– An equilibrium is symmetric if, for any value and any bidders ,. We will abuse notation and write the common strategy of
each bidder as .

EC341 2022-3: Auctions 1 15

A simple example
2 bidders, , – i.e. is uniform.
1PSB (first-price, sealed bid) case
– We will show that is a symmetric equilibrium
– Suppose that bidder uses this strategy, and that bidder has private value and bids an amount . Their expected profit is

• This is quadratic over the interval and reaches a maximum at ; over the interval it is linear with slope .
• The following figures show that the optimum (best reply) is always

𝑣1 𝑣1 1 𝑣1 1 𝑣1
2 2 2 2

Case 2:
Case 1:

EC341 2022-3: Auctions 1 16

Example continued
2PSB (second-price, sealed-bid) case
Truthful bidding () is a symmetric equilibrium (Slide 12)
Which method does the seller prefer?
– Seller’s expected revenue is the expected equilibrium price
– For the 1PSB case, this is

• The cdf of is

• The second equality follows from independence; the third from uniformity
• The corresponding density of is
• The expected price is thus

– For the 2PSB case the seller’s revenue is .

– Note: so in this case
– We already know that , so , the same as for the 1PSB

EC341 2022-3: Auctions 1 17

Comment
This may seem surprising; it might seem obvious that the seller would prefer to sell to the
highest, rather than the second-highest bid, But bidders in the 1PSB auction will submit lower
bids.
This result (and the underlying logic that the seller’s desire for higher revenue and the
bidders’ desires for lower price balance in many auction forms) lies behind the main result of
auction theory, the Revenue Equivalence Theorem.

EC341 2022-3: Auctions 1 18

Equilibrium strategies in a selection of models
Definition: a symmetric equilibrium strategy is monotonically increasing if
Remark: in an equilibrium with monotonically increasing strategies, the winner is the bidder
with the highest private value. Let be the highest value of bidders other than . Since the
distribution of values is continuous, ties can be ignored. The next result picks out one
equilibrium for the case we are discussing (not necessarily the only one)
Theorem: in a symmetric independent private values 1PSB auction, the following strategy is a
symmetric equilibrium

In words, “bid the expected second highest value conditional on yours being the highest.”

EC341 2022-3: Auctions 1 19

Equilibrium strategies, 2 (proof of theorem)
– is monotonically increasing. We know from first principles that for any random variable and any two disjoint
events and

1. Because everywhere, is a continuous random variable with positive density . Let be an interior point of , i.e.,
and, if then . For every s.t. we have

2. Since is everywhere positive and is interior,

– From 1 above, we see that is the weighted average of two numbers, one of which is strictly larger than the other.
Therefore is strictly larger than the smaller of these numbers, which is according to 2 above. Thus as required.
– is continuous. Note that , , so . This shows that the strategy is continuous at . Now consider a positive value. By
definition of conditional expectation

Since is continuous, so is its cdf ; since the density is positive , ; the denominator in the above equation is not 0.
– So is the ratio of two continuous functions of where the denominator is not zero , and is continuous

EC341 2022-3: Auctions 1 20

Equilibrium strategies, 3
– Note: using integration by parts, we get something that will be useful later

– is a symmetric equilibrium strategy. Suppose all bidders use this strategy , We will show that is a best reply for bidder . Consider
the bounded interval case ; . Because is strictly monotonic and continuous, it has a continuous inverse . Bidder , with private value
has an expected utility from bidding of [ indicates that all bidders other than are following the strategy ].
– If bids , the probability of winning and thus the expected utility are (The reason is that ; is strictly monotone increasing so another
bidder will win).
– If bids , the expected utility is .
– Since is everywhere positive, . Hence, over the domain is the product of two positive functions and is thus a positive function.
Therefore the maximum of is attained at some . Now rewrite as follows:

is the set of all possible bids by players using strategy . Assuming is an interior point of this, write . Then , so . Substitution gives

EC341 2022-3: Auctions 1 21

Equilibrium strategies, 4
– To find the bid that maximises we only have to find the point that
maximises . is differentiable over ; its derivative is

– This derivative is only at which is to say

To sum up results for the symmetric, sealed-bid auction with
independent private values:
– In the 1PSB auction, is a symmetric equilibrium
– In the 2PSB auction, is a symmetric equilibrium
– [You should verify that the 1PSB auction for values uniformly distributed on
has a symmetric equilibrium at ]

EC341 2022-3: Auctions 1 22

Equilibrium profits and utilities
Theorem: in both symmetric equilibria on slide 20, a bidder with private value expects to pay

– In the 2PSB auction, a bidder bids , wins with probability and pays . Note that the bid does not affect
payment conditional on bidding, only the probability of winning.
– In the 1PSB auction, a bidder bids , wins with probability and pays the bid. In this case, the bid affects
the amount paid conditional on winning.
Theorem: in both symmetric equilibria on slide 20, a seller expects to get

– The expected payment of a bidder with value is . The seller does not know , so expects this
to be . If this is true for each bidder, the total expected revenue is .

EC341 2022-3: Auctions 1 23

Revenue equivalence
The results above (same expected payments and revenues from symmetric and monotone increasing equilibria for
both types of independent private values sealed-bid auction) are no fluke. We now look at a more general result –
possibly the main result of auction theory.
Let us denote by an auction where the winner is determined by a function (the probability that bidder wins) and the
payment by a function (the payment by each bidder – we do not assume that only the winner pays). Let be a
monotonically increasing strategy. ’s expected payment when all other players use strategy and has true value and
bids (not necessarily ) is written :

We can also write ’s expected surplus (utility) conditional on true value of bidding - or behaving as though ’s value
was . The probability that wins is ; ’s expected probability of winning if they bid is

We can therefore define incentive compatibility as the condition that bidder with value would rather follow than
behave as though they had a different value

If we consider a small deviation to we get the derivative - the ‘truthful’ expected probability of winning.

EC341 2022-3: Auctions 1 24

Revenue equivalence, 2
Theorem (RET): Let be a symmetric, monotonically increasing equilibrium in a sealed-bid symmetric,
independent private-values auction such that:
– [Efficiency] The bidder with the highest private value wins the auction; and
– A bidder with private value expects to pay nothing -
Then

– Because the seller will accept any non-negative price and all bidders have nonnegative values, the object is always
sold. The object is sold to the highest bidder in the symmetric, monotonically increasing equilibrium.
– The RHS of the above equation is independent of the auction method, and thus so is the expected payment by a
bidder with private value . These depend only on the distribution of the private values, so risk-neutral bidders are
indifferent among these auctions.
– By integrating over bidder types as on slide 21, a risk neutral seller is also indifferent.
A more general statement of the RET: Assume each of a given number of risk-neutral potential buyers of a
single object has a privately-known signal independently drawn from a common, strictly-increasing,
atomless distribution. Any auction mechanism in which
(i) the object always goes to the buyer with the highest signal, and
(ii) any bidder with the lowest-feasible signal expects zero surplus
yields each bidder the same the expected payment as a function of her signal and gives the seller
the same expected revenue.

EC341 2022-3: Auctions 1 25

Side-trip: Envelopes
Describes the value function of a parameterized optimization
problem in terms of the objective function
One-line proofs of Shepard’s Lemma (Consumer Theory) and
Hotelling’s Lemma (Producer Theory)
Straightforward proof of Revenue Equivalence and other results in
auction theory and mechanism design
With strong assumptions on derived quantities, it’s trivial to prove,
but we’ll approach it from first principles.

EC341 2022-3: Auctions 1 26

How it works
Consider an optimization problem of choosing to maximise a function of and a parameter :

Define the optimal and indirect value function

(For auctions, is value, is bid, and is expected payoff given other bidders’ strategies)
We’ll give two versions:
– the value of when it exists and
– as the integral of that derivative

𝑉 (𝑡)=max ⁡{ 𝑓 (1,𝑡 ), 𝑓 (2,𝑡), 𝑓 (3,𝑡)}

𝑓 (2 , 𝑡 )
𝑓 (3 ,𝑡 )
𝑓 (1 , 𝑡)

EC341 2022-3: Auctions 1 27

Derivative version
Theorem: pick , and assume exists at .
• If and the right-hand derivative exists, then
• If and the left-hand derivative exists, then
• If and the derivative exists, then
In words, the derivative of the value function is the derivative of the objective function
evaluated at the optimum.

𝑓 ( 𝑦 ,−)
𝑉 (−)
𝑡

EC341 2022-3: Auctions 1 28

Proof (for background and understanding)
If exists, then

For any selection (single-valued function ) from the possibly multi-valued

By optimality, , so

A symmetric argument shows that (when it exists)

If exists, it equals both the left- and right-hand derivatives, so

[This is, of course, obvious if is differentiable in both arguments and is single-valued and
differentiable]

EC341 2022-3: Auctions 1 29

But V need not be differentiable
Recall figure from slide 4, where V is only differentiable ‘most of the time’
𝑉 (𝑡 )=max ⁡{ 𝑓 (1 ,𝑡 ), 𝑓 (2 , 𝑡 ) , 𝑓 (3 , 𝑡 ) }

𝑓 (2 , 𝑡 )
𝑓 (3 , 𝑡 )
𝑓 (1 , 𝑡 )

t
That’s usually good enough (‘kinks’ not at maxima)
Suppose that the range of x is compact and f is continuously differentiable in both arguments.
Then V is differentiable at t and if
– s single-valued or
– is concave or
– maximises
In most auctions, all ‘interior’ types (in terms of valuations) will have unique best responses,
so will usually be differentiable. But this is more than we need (see next slide)

EC341 2022-3: Auctions 1 30

Integral version
Theorem: if

– exists and
– is absolutely continuous
Then for any selection from ( indicates an arbitrary value of )

[Absolute continuity means that is differentiable almost everywhere and so

must be the integral of its derivative, which is wherever that exists]

EC341 2022-3: Auctions 1 31

Absolute continuity
Formal definition: such that for every finite collection of disjoint intervals

Lemma: if is absolutely continuous in and there exists an integrable function such that for
all and almost all then is absolutely continuous
[proof standard, omitted]
Bottom line: if

– exists and
– is absolutely continuous
Then for any selection from ,

EC341 2022-3: Auctions 1 32

Revenue Equivalence Theorem (RET) restated
Assume symmetric private values case, and fix a mechanism:
distributed iid according to continuous, strictly increasing cdf (density ) on
= i’s expected surplus conditional on ;
= probability i wins given
= expected payment by i given in other words,
Incentive compatibility: for all , ; if , we get

Any two mechanisms with the same

–Pi(v) function (e.g. those that give the object to the highest valuation bidder) and
– (e.g. those for which this is 0)
have the same function for each , hence the same functions and the same expected
revenue
Note similarity between IC condition and truthful direct revelation
If object goes to highest bidder,

EC341 2022-3: Auctions 1 33

Step 1: Does the ET hold?
Let be bidder ’s equilibrium strategy
Let be ’s expected payoff in the auction, given type (value) and bid , assuming
everyone else bids their equilibrium strategies
If is an equilibrium strategy, , so
…
…so , which gives the other sufficient conditions
– exists
– Fixing , is linear in , and therefore absolutely continuous
– is everywhere bounded above by
So the integral version of the Envelope Theorem holds

EC341 2022-3: Auctions 1 34

Step 2a: the bidder’s side
We know
For the envelope theorem, we care about at

But we assumed the bidder with the highest type always wins, so this is
The ET then gives

By assumption, , so
The point: this does not depend on the details of the auction, only the distribution of types
And so is the same in any auction satisfying our two conditions

EC341 2022-3: Auctions 1 35

Step 2b: the seller
Since the bidder with the highest value wins the object, the sum of all bidders’ payoffs is

The expected value of this is , where is the seller’s expected revenue

By ET, the sum of all bidders’ (ex-ante) expected payoffs is

So

which again depends only on , not the rules of the auction

EC341 2022-3: Auctions 1 36

Handy tip: using the ET to compute equilibrium bids in various auctions
All-pay auction (high bidder wins, every bidder pays its bid)
– Bidder , conditional on value and bid function expects to get
– By RET, we know that
– Setting these equal, we conclude that
– For instance, if values are uniformly distributed on (meaning ),
Top 2 bidders pay (high bidder wins, top 2 bidders pay their bids)
– Assuming that there3 is an increasing symmetric equilibrium bidding function a bidder with value expects to get (
not highest but others have lower values)
– By RET
– Hence

EC341 2022-3: Auctions 1 37

Using the RET
We can use the RET to find symmetric equilibrium strategies for specific auction types.
As a side-trip, recall from slide 22 that the derivative of the expected surplus function is the
probability of winning by faithful adherence to the equilibrium strategy as assessed by a
bidder knowing only their own private value:

This lets us find the expected surplus by integration

This lets us rephrase the RET as: any auction mechanisms where the highest-value bidder wins
with probability 1 and with the same expected surplus for the lowest−value bidder give the
same expected revenues and payments.
This does not assume that . In the special case where the object goes to the highest bidder we
can further write the expected win probability

EC341 2022-3: Auctions 1 38

Finding the strategies
Theorem: let be a symmetric, monotonically increasing equilibrium in a first-price, sealed-bid symmetric,
independent private-values auction where . Then

– Above, we showed this was an equilibrium for 1PSB auctions. This shows it is unique. (proof omitted)
Theorem: let be a symmetric, monotonically increasing equilibrium in an all-pay, sealed-bid symmetric, independent
private-values auction where . Then

– Proof: in an all-pay auction, each bidder pays their bid, so a bidder with value pays . Since is monotonic and , the
RET (slide 23) applies and .
Example: 1PSB with bidders. Assume values are uniformly iid on . Then and . Using the theorem at the top of the
slide,

– A buyer with private value expects to pay

– Ex ante (before learning ), the buyer’s expected payment is
– The expected revenue is times this:

EC341 2022-3: Auctions 1 39

 Marginal Revenue interpretation
Suppose a monopolist faces a single buyer with unknown value (reservation price) v, drawn from a distribution
(density ) on an interval . If he charges a price of (in this interval), his probability of selling the object (or average sales
over many transactions) is the probability that the customer’s value () is bigger than the price: . Thus the curve is a
demand curve. Working with the inverse demand curve () in the usual way, we can construct the associated marginal
revenue curve:

d (vq (v)) q 1  F (v ) v+
MR (v) MR (q (v))  v  v 
dq(v) dq f (v ) Sales at price v
dv v
Expected revenue
at price v
1-F(v)
v-
f(v)
v++v-
2
Demand

MR

v-

0 1-F(v) ½ q=1-F(v) 1

EC341 2022-3: Auctions 1 40

Marginal Revenue 2

Optimal auctions sell to bidders with highest marginal revenue – ideal is to set
MR = marginal reservation cost of seller. In this way, we can extend the
analysis to sequential or repeated auctions by taking careful account of
marginal opportunity cost of (not) selling today.
Note: bidder with highest MR may not have highest value or highest signal:
– 2 bidders with values iid and uniformly distributed on .
– Marginal revenues are:

(for ’s that both types could observe)

– Therefore, MR is higher for weaker bidder (lower ai)
This may motivate the use of inefficient auctions.
EC341 2022-3: Auctions 1 41