MICROECONOMICS:

IMPERFECT COMPETITION
MONOPOLY
 A monopolist is a firm which has (in pure theoretical terms)
100% market share.
 It faces a downward sloping Demand (AR) curve and can
choose to provide at any point along that curve.
 The monopolists’ profit maximisation problem is much like
the one facing the producer in a perfectly competitive
market
 MR = MC for profit max, as before
 The big difference is that prices are NOT FIXED anymore
 We can describe the price as a function of output.
Π = P(Q).Q – TC (Q, r, w)
 Differentiate wrt Output and set to 0
MONOPOLY
 Barriers To Entry
 A monopoly is created because firms cannot enter the
industry because barriers to entry (or they may not enter
because they may not be able to leave easily due to
barriers to exit).
 Barriers to entry are either Artificial or Natural;
 Natural barriers to entry largely arise from Increasing
Returns to Scale – such that the MES is at such a high level
of production only one firm can really achieve it.
 Artificial barriers may arise from legal process (patents and
copyrights) or lobbying for regulation for protectionist
measures.
MONOPOLY
 Marginal Revenue
 Under perfect competition, we assumed that AR = MR = constant. However, since price is now a
function of output, we can no longer assume that AR = MR.
 MR = d[P(Q).Q] / dQ
 Using the Product Rule:
 d[P(Q)]/dQ . Q + P(Q). [dQ / dQ]
 MR = P + Q.[dP / dQ]

 Because price falls as quantity rises, dP/dQ < 0, therefore MR < P (and as P = AR, MR < AR).
 In addition, we can show that P > MC:
 MR = P + Q.[dP/dQ] = MC
P= MC – {Q.[dP/dQ]} ; P > MC by abs[dP/dQ].Q
 Finally we can also show the link between MR and Elasticity:
 MR = P[1 + (Q/P).(dP / dQ)]
 MR = P[1 + 1 / PeD]
 PeD is inelastic, PeD > -1, MR < 0
 PeD is elastic, PeD < -1, MR > 0
 PeD is unitary elastic, PeD = - 1, MR = 0; All these results make intuitive sense.

 This connection between MR and elasticity can also show the degree of monopoly power in an
industry via the Lerner index:
 L.I = (P – MC)/ P = -1 / PeD
 And also show the degree of mark up in an industry:
P/ MC = PeD / [1 + PeD]
MONOPOLY
Price
 Solving the profit maximisation problem for
the monopolist yields the optimal output Q*,
and the optimal price, P*
 Because TR > TC at that level of output, the
Monopoly Rents
firm earns supernormal/abnormal profits (or
monopoly rents), shown by the red square. DWL
 In addition, there is a Deadweight Welfare
Loss at that level of output, given by the black
P*
triangle.
 The DWL arises with the basis of comparison

being a perfectly competitive market; which AC=MC

would produce at Q1.
 Moving from Q1 to Q*:
 Consumer surplus falls by the amount of the black
triangle and red rectangle. MR AR
 Producer surplus rises by the red triangle.
 Therefore there is a net loss of the black triangle.

 The diagram makes a simplifying assumption Q* Q1 Quantity

that AC = MC, but the results still hold if they
face an upward sloping MC curve.
DEMAND FOR INPUTS
 Assume that, despite being a price maker in the final goods market, it is a price
TAKER in input markets.
 The demand for inputs can therefore be derived traditionally (using Q = F(K,L)):
 Π = P[F(K,L)].F(K,L) – (rK + wL)
 First Order Conditions (using chain + product rules):
 dΠ/dK = MPK.[P[F(K,L)] + MPK.[dP/dQ . F(K,L)] – r = 0
 MPK.{[P[F(K,L)] + [dP/dQ . F(K,L)] } = r
 MPK . MR = r;
 MRPK =r
 dΠ/dL = MPL.[P[F(K,L)] + MPL.[dP/dQ . F(K,L)] – w = 0
 MPL . MR = w
 MRPL = w
 The Marginal Revenue Product of a unit of output is the extra revenue gained from a one unit
increase in the input under consideration. In a perfectly competitive model P = MR, thus
MRPi = P.Mpi
 We can also apply Shephard’s lemma to monopolists – evaluate the Total Cost function at the
optimal quantity Q*, and take the derivative with respect to the input pay:
 L* = d{TC(Q*,r,w)} / dw
PRICE DISCRIMINATION
 A firm is price discriminating if it sells a homogenous good at differing prices.
 Three degrees:

 First degree:
 Producer extracts ALL consumer surplus for that individual – it will charge him/her up
until the point where he/she is indifferent to purchasing or not purchasing the good.
 Second degree:
 Attempts to provide incentives to force the consumer to reveal his willingness to pay. It
may do this by engaging in ‘block’ pricing – offering discounts for bulk purchases etc.
 Third degree:
 Based on the assumption that separable markets exist for the product; these markets can
be differentiated by price elasticities.
 Therefore, with different price elasticities, the producer can charge different prices in
each market; this can be seen with last minute vs early bookings in train fares for
example.
 All these are based on the assumption of NO ARBITRAGE; if consumers could
resell goods, then they would aim to undercut the firm and therefore reduce
price differentials.
REGULATION
Price
 Regulation of monopolies is desirable because
of the DWL. D=AR
 However, for natural monopolies that exhibit
IRS (as shown), operating at P = MC will
incur losses.
 Thus, there are two main solutions:

 First Best; MC pricing.

 Force firm to operate at P = MC, and subsidise it
for losses.
 Second Best; AC (full cost) pricing.
 Force firm to operate at P = AC.
 Not the most efficient point, but close to and there 1
P=AC
is no need for subsidies – the firm earns a normal AC
profit. 2 MC
 Can be achieved by using two tier pricing; P= MC MR
essentially second degree price discrim, Quantity
segmenting the markets into two groups
Y
depending on Q bought.
 The gains made by those who pay above AC
should cancel out the losses made by those who
pay less than AC.
OLIGOPOLY
 Price + Output decisions in an Oligopoly can be modelled in
two ways:
 Cournot Model:
 Model where all firms set QUANTITIES simultaneously
 Bertrand Model:
 Model where all firms set PRICES simultaneously.
 Both models assume the firms face constant (and equal) MC,
are identical in terms of production functions and sell
identical goods. Furthermore, we will be using a version
which focuses on a DUOPOLY – a market with only two
providers.
 For both models we will also consider the ‘leader’ case –
where a firm set their output or price first.
COURNOT MODEL
 Reaction Functions
 To derive the Cournot Model, consider an industry where two firms produce Q1 and Q2
respectively. Total industry output is therefore Q = (Q1 + Q2).
 However, the inverse demand function for either firm depends on this quantity:
P= p(Q) = p(Q1 + Q2)
 Firm 1 therefore assumes a given level of Q2, and vice versa for Firm 2, in order to find
the optimal level of Q1.
 Problem for Firm 1:
Π = P(Q1 + Q2).Q1 – TC (Q1)
 dΠ/dQ1 = P(Q1 + Q2) + [dP/dQ1].Q1 – dTC/dQ1 = 0
 MR1 = MC1

 An identical (but respective) version of this holds for Firm 2.

 These results lead us to REACTION FUNCTIONS, graphically shown on the next slide.

 Reaction Functions are the locus of points where the Isoprofit curves are at their
maximum – each isoprofit curve showing potential profit for a firm GIVEN A CERTAIN
LEVEL OF THE OTHER FIRM’S OUTPUT.
 Reaction Function for Firm 1: Q1 = R1(Q2)

 Reaction Function for Firm 2: Q2 = R2(Q1)

 REACTION FUNCTIONS + ISOPROFIT
CURVES

q2
q2

Isoprofit curves
Firm 1’s reaction function

Isoprofit curves

Firm 2’ s reaction function

q1 q1
COURNOT MODEL
 To graphically derive the optimal output level, simply overlay
the two Reaction Curves and find the point where they intersect
– this is COURNOT EQUILIBRIUM.
 It is a Nash Equilibrium – each firm is doing it’s best given
what it believes the other firm is doing.
 If both firms have the same technology, then Q1* = Q2*.

 Factoids:
 Cournot Equilibrium produces a HIGHER OUTPUT than a
monopoly model, but a LOWER OUTPUT than a perfectly
competitive model; with all the implications for the size of the DWL.
 As the number of firms in the industry rises, the output rises, up until
the limit of perfect competition – however, the DWL shrinks very
fast, suggesting that it only takes relatively few firms to simulate
perfect competition.
COURNOT MODEL W/ LEADERS
 Quantity Leadership – The Stackelberg Model
 Assume that Firm 1 produces first, followed by
q2
Firm 2; and let us work backwards.
 The problem for Firm 2 is EXACTLY THE
SAME as in the original model – it still has to
produce a reaction function that depends on Q1:
 Q2 = R2(Q1)
 However, the problem is slightly different for the
first firm:
Π = P(Q1 + Q2).Q1 – TC (Q1)
Π = P(Q1 + R2(Q1)).Q1 – TC (Q1)
 Because we are able to substitute in Firm 2’s q2Cournot Firm 2’s reaction function
q2Stackelberg
reaction function, we can solve the above problem
as we would for a normal firm, and achieve a
concrete unconditional optimal Quantity:
 Firm 1 produces Q1*
 Firm q1Cournot q1Stackelberg q1
2 Produces Q2* = R2(Q1*)
 Graphically this is shown by Firm 1 producing on
it’s lowest Isoprofit line tangential to Firm 2’s
Reaction Function – thus producing the most and
gaining the most profit;
 IT IS ADVANTAGEOUS TO GO FIRST IN
COURNOT.
COURNOT MODEL W/ COLLUSION
 If the two firms collude, they will behave as a single monopolist would in setting output:
Π = P(Q1 + Q2).(Q1+Q2) – TC1 (Q1) – TC2(Q2)
 Differentiating to find the profit maximising conditions:
 dΠ/dQ1 = P(Q1 + Q2) + [dP/dQ1].(Q1+Q2) – dTC1/dQ1 = 0
 dΠ/dQ2 = P(Q1 + Q2) + [dP/dQ1].(Q1+Q2) – dTC2/dQ2 = 0

 Hence:
 MC1 = MR = MC2; If one firm has a cost advantage it will simply have to PROVIDE MORE.
 There is, however, a temptation to cheat in a cartel, as the Marginal Profit from expanding
output (cheating) is given by
 dΠ/dQ1 = P(Q1 + Q2) + [dP/dQ1].(Q1) – dTC1/dQ1
 Which, when compared to the Marginal Profit under collusion:
 P(Q1 + Q2) + [dP/dQ1].(Q1+Q2) – dTC1/dQ1 =
 P(Q1 + Q2) + [dP/dQ1].(Q1) + [dP/dQ1].(Q2) – dTC1/dQ1
 {P(Q1 + Q2) + [dP/dQ1].(Q1) + [dP/dQ1].(Q2) – dTC1/dQ1} – {P(Q1 + Q2) + [dP/dQ1].(Q1) –
dTC1/dQ1} =
 [dP/dQ1].(Q2) > 0

 Therefore, Firm 1 faces a positive MR when it cheats (same holds for Firm 2):
 If either firm believes the other will stick to the deal, it has an incentive to cheat.
 If either firm believes the other will cheat, it has an incentive to cheat faster.
BERTRAND MODEL
 Much like the Cournot Model, except instead of setting OUTPUT,
the firms set PRICES.
 Other assumptions remain the same, however;
 The firms are profit maximisers.
 They have identical products.
 They make their decisions simultaneously.

 Instead of considering the inverse demand functions, consider the

normal demand functions:
 Firm 1: Q1 = D1(P1,P2)
 Firm 2: Q2 = D2(P1,P2)

 Then solve via normal methods:

Π = P1.[D1(P1,P2)] – TC[D1(P1,P2)]
 FOC (using the Chain rule for MC):
 dΠ/dP1 = D1 + P1.[dD1 / dP1] – {dTC/dQ1}.{dD1/dP1} = 0
BERTRAND MODEL P1 Firm 2’s Reaction Curve
 Therefore, we arrive at a conclusion that Firms
1 and 2 will have REACTION FUNCTIONS:
 P1 = R1(P2)
 P2 = R2(P1)
 The R.F are UPWARD SLOPING, as shown
to the side.
 However, it is important to note that the only
Nash Equilibrium in this game is if both
firms price at MC. P1*
 If a firm does not price at MC, then the other
firm can simply undercut and take all profits. Firm 1’s Reaction
Curve
The firm cannot operate at a loss either,
therefore only ONE sensible policy is left.
 However, this is paradoxical; the Bertrand
Model suggests that P=MC in an oligopoly, Nash Equilibrium
which is the same result for perfect
competition.
 This isn’t true.
P2* P2
 We need to alter the assumptions.
BERTRAND MODEL VS COURNOT
MODEL
 Price Leadership in the Bertrand Model is NOT
advantageous.
 This is because of the SLOPE of the Reaction
Functions; the leader will actually do worse (it will have a
higher price) than the follower if it acts aggressively.
 The precise mathematics of the sequential game are broadly
similar to the maths of the Stackelberg Cournot instance.
 In any instance, you want to induce the other firm to act
less aggressively:
 If the RF SLOPE UPWARDS:
 Act like a ‘puppy dog’ to make the other firm act like one too.
 If the RF SLOPE DOWNWARDS:
 Act like the ‘Top Dog’ to make the other firm act like a ‘puppy dog’.
BERTRAND PRICE LEADERSHIP

P1

•Sub – Optimal Nash

Equilibrium with price
P1 leadership
•Prices are higher than
P1* they are in the original
model and output is
lower.

Nash Equilibrium

P2* P2
P2=R2(P1)