Joe Chen 1 1 Introduction 1.1 The Beginning Joe Bain and Edward Mason (the Harvard tradition): Empirical in nature (loose theories) The Structure Conduct Performance (SCP) paradigm: Structure (market structure): entry conditions, market concentration, number and size of firms, the degree of product differentiation, cost structure, the degree of vertical integration,...; Conduct (firm behavior): price, R&D expenditures, investment, advertisement, capacity utilization, collusion...; Performance: efficiency, price-cost margins, the variety of products, innovative rate, profits,...; The SCP paradigm was the brain child of the Harvard school of thought and popularized during 1940-60 with its empirical work involving the identification of correlations between industry structure and performance. The standard SCP paradigm asserts that there is a direct relationship between the degree of market concentration and the degree of competition among firms. This hypothesis will be supported if positive relationship between market concentration (measured by concentration ratio) and performance (measured by profits). Results: stylized facts: correlations, descriptive statistics, stories. Anyway, these are not causal relationship. New IO: During 1980-90 game theory took center stage with emphasis on strategic decision making and Nash equilibrium concept. Theoretical development: game theory, firm and market dynamics, and asymmetric information. New tools are used to re-examine many existing stories from the SCP paradigm.
Joe Chen 2 After 1990, empirical industrial organization with the use of economic theory and econometrics lead to complex empirical modeling of technological changes, merger analysis, entry-exit and identification of market power. New empirical approaches: the empirical renaissance in industrial economics (Bresnahan and Schmalensee, 1987): The importance of (economic) cost information; The estimation of causal relationship. 1.2 History of the field of I.O. 1. (a) 1930 s-1950 s: Case studies i. Detailed analyses of specific industries; heavy in institutional detail ii. Induction: sought to go from the specific to the general iii. Absence of a theory which seeks to cover a class of industries (b) 1950 s-1970 s: Introduction of econometric studies i. Basic unit is the industry ii. Mostly cross-sectional reduced form analyses iii. General inter-industry hypotheses but no formal theory iv. The hope was to show that certain general statement can be made across all manufacturing industries. This was unsuccessful. (c) 1970 s-1980 s: Theory i. Deduction: Developing general theory which has implications for a class of industries ii. Coincident with revival of game theory iii. Theory may be too general A. It may be trying to explain too many industries at once and thereby explaining none of them B. More institutional detail than is typically modelled may be vital to understanding behavior.
Joe Chen 3 (d) 1980 s present: Empirical renaissance i. Industry time-series studies that use formal theory to build empirical models. (Unfortunately, these theories have not usually been adequately tested as to their empirical validity.) ii. Much more disaggregated data iii. Case studies with econometrics iv. Recognizes the idiosyncratic institutional features v. But where is this empirical literature heading? Will it always be a collection of narrow case studies or will empirical regularities emerge that are applicable to a broad class of industries? 1.3 The Departure of IO from the General Equilibrium Theory The First Welfare Theorem: Competitive (Walrasian) equilibrium implies Pareto optimality. Problems : Price taking behavior: market power; Externality: network externality; Private goods: dispersion of new technology; Perfect information: product quality. Partial Equilibrium Framework: Look at markets and industries (theory of second best). What is a market? Correlation of product prices? Bottom-up? Theoretical vs. empirical work
Joe Chen 4 References [1] Bresnahan, T. F. and R. Schmalensee, "The Empirical Renaissance in Industrial Economics: An Overview", Journal of Industrial Economics, 35 (1987), 371-8.
Joe Chen 5 2 Monopoly Pricing Basic assumption: goods are given (no product selection problem), qualities are known (no experience goods), same price for the same commodity (no price discrimination). 2.1 A single product monopoly The monopoly solves the following problem: max p pd(p) C(D(p)). FOC implies the inverse elasticity rule: p m C 0 p m =1/ε, where:ε = D 0 (P m )P m /D(P m ). Lerner index; p m >C 0, dead-weight loss; Rule-of-sum (constant mark-up) pricing ; non-profit maximization; ε>1, notethat:p m = εc 0 /(ε 1); p m is a nondecreasing function of marginal cost. Consider a more general setup. Let C2 0 (q) >C0 1 (q), forallq>0,and(pm i,qm i ) maximizes profit when the cost function is C i (q i ) so that: p m 1 qm 1 C 1(q1 m) pm 2 qm 2 C 1(q2 m). p m 2 qm 2 C 2(q2 m) pm 1 qm 1 C 2(q1 m) This implies: C 2 (q1 m ) C 1 (q1 m ) [C 2 (q2 m ) C 1 (q2 m )] 0; or, Hence, q m 1 qm 2 ;or,pm 1 pm 2. Z q m 1 q m 2 [C 0 2(q) C 0 1(q)]dq 0. How to tax the monopoly in order to induce social optimal?
Joe Chen 6 Intuition: positive or negative tax (subsidy)? Consider the following problem: max p pd(p + t) C(D(p + t)). FOC implies: (p C 0 )D(p + t)+d(p + t) =0; or, (p + t C 0 )D 0 (p + t)+d(p + t) td 0 (p + t) =0. Social optimal requires: p + t = C 0 ; hence, t = D(p c )/D 0 (p c ) < 0. 2.2 Monopoly with multiple products Now, consider a monopoly with n products. Let p =(p 1,...,p n ). The monopoly s problem is: FOC implies: D i + p i D i (p) P i max p 1,...,p n + X j6=i nx i=1 p j D j (p) P i = p i D i (p) C(D 1 (p),...,d n (p)). nx j=1 C(D 1 (p),...,d n (p)) D j (p), for all i. q j p i Independent demands, separable costs: if D i (p) =D i (p i ),andifc(d 1 (p),...,d n (p)) = X n j=1 C i(d i (p)), then: Same as before: the maximization of each product is equivalent to the maximization of the aggregate profit. Dependent demands, separable costs: C(D 1 (p),...,d n (p)) = X n C i(d i (p)) j=1 FOC becomes: D i + p i D i (p) P i Ci 0 D i (p) = X (p j C p j) 0 D j(p) ; i P i j6=i
Joe Chen 7 and, p i Ci 0 = 1 X (p j Cj 0)D jε ij ; p i ε ii R i ε ii j6=i where: ε ij = ( D j (p)/ p i )(p i /D j (p)) for all i, j, andr i = p i D i (p). Substitutes: D j (p)/ p i > 0. This implies ε ij < 0. If the monopoly were decentralized, there will be de facto competition. Complements: D j (p)/ p i < 0. This implies ε ij > 0. It is likely that in some market k, p k Ck 0. An example: (Good will). Consider the following problem for a monopoly producer of a single good: max p 1 D 1 (p 1 ) C 1 (D 1 (p 1 )) + δ [p 2 D 2 (p 2,p 1 ) C 2 (D 2 (p 2, p 1 ))]. p 1,p 2 Let e D 2 = δd 2 and e C 2 = δc 2. Then it is the same problem when demands are dependent and costs are separable. If there exists good will: D 2 (p 2,p 1 )/ p 1 < 0; and, it is the complement goods case. Independent demands, dependent costs: D i (p) =D i (p i ). Let s consider an example: max p 1 D 1 (p 1 ) C 1 (D 1 (p 1 )) + δ [p 2 D 2 (p 2 ) C 2 (D 2 (p 2 ),D 1 (p 1 ))] p 1,p 2 Learning by doing: C 2 (q 2,q 1 )/ q 1 < 0; The FOC for p 1 : (p 1 C 0 1)D 0 1(p 1 )+D 1 (p 1 ) δ C 2 q 1 D 0 1(p 1 )=0; hence, p 1 C1 0 = 1 + δ C 2. p 1 ε 1 p 1 q {z 1 } <0 2.3 Durable good monopoly Durability of the product creates competition for the monopoly.
Joe Chen 8 Let s start with an example. Consider a market for an infinitely durable good. The good costs nothing for the monopoly firminthemarkettoproduce. Thereare7 consumers with valuations (discounted value of the flow of services) of the good, v i, being 1,2,..., 7. What s the monopoly price for an once-and-for-all offer? p m Demand 0 7 1 7 2 6.. 7 1 8 0 Now, consider the multiperiod case. Suppose the monopoly charge p m =4in the first period. At the second period, it faces a residual demand of: p m 0 3 1 3 2 2 3 1 4 0 Demand So for consumers such that: v i 4 <δ(v i 2), or,v i < (4 2δ)/(1 δ), waiting is worthwhile (when δ =0.95, v i < 42). To find an equilibrium, one has to find a sequence of prices and consumer expectations such that: the prices are optimal given the consumer s expectation, and the expectations are rational given the firm s prices. Let s do this for a two-period case. Let D(p) =1 p in each period, so that the inverse demand function each period is P (q) =1 q. Say the monopoly supply q 1 in the first
Joe Chen 9 period. In the second period, it then solves: FOC gives us: max q 2 [(1 q 1 ) q 2 ]) {z } q 2 p 2 =residual inverse demand q 2 =(1 q 1 )/2. Given this result, let s move back to the first period. Note that consumers are willing to pay: p 1 =1 q 1 + δp a 2. Rational expectation requires: p a 2 = p 2,andp 2 =(1 q 1 ) q 2 =(1 q 1 )/2. Thus, The monopoly solves: Hence: p 1 =1 q 1 + δp a 2 =1 q 1 + δp 2 =(1+δ/2)(1 q 1 ). max (1 + δ/2)(1 q 1 )q 1 + δ( 1 q 1 ) 2. q 1 2 q 1 = 2 4+δ p 1 = (2+δ)2 2(4+δ) Note that: p 1 >p 2 = (2+δ) 2(4+δ). In general, the equilibrium price sequence is a decreasing one. What if the monopoly leases out the infinitely durable good instead of selling it? In each period t, the monopolist maximizes p t D(P t ),andtherfore,p 1 = p 2 = 1/2, andq 1 =1/2 and q 2 =0(no depreciation). The profit of leasing is then: 1/4+δ 1/4 =(1+δ) /4.. Coase conjecture: when price adjustments become more and more frequent, the monopoly s profit converges to zero. All trades take place almost instantaneously at prices close to marginal cost. (See the Appendix of Chapter 1 for a proof if you are interested).
Joe Chen 10 The monopoly would better off if it could lease the product or it could ex ante commit itself to no-haggle. Leasing has its own problems. In general, without a commitment device, no-haggle is not credible (not subgame perfect). Some commitment devices are: Arbitrators: third-party arbitration; Money back guarantee; Reputation; Destroy factory; Decreasing returns to scale; In formation asymmetry: the consumers do not know if the cost is high or low; New customers arrive each period; Planned obsolescence. Bottom line: one should not take it for granted that a durable good monopoly would charge the competitive price and makes no profit. 2.4 How bad is monopoly? In general, monopoly is not good : Dead-weight loss: p m >C 0 ; X-inefficiency (Leibenstein, 1966): No competition, the managers of monopoly firms are more likely to engage in slacking. This inflates the cost (theoretical foundation is shaky); Rent seeking (wasteful expenses): A group of firms engage in acquisition and maintenance of the monopoly franchise: Rent dissipation: the total expenditure firms are willing to pay equals the total amount of the rent;
Joe Chen 11 Socially wasteful dissipation: this expenditure has no socially valuable by-products (lobbying). Possible other distortions as well. Well, maybe we are too pessimistic. Some mitigating factors: For X-inefficiency: Yardstick or tournament, Align the manager s incentive to that of the monopoly shareholders : stock option for CEOs; Increasing returns to scale: fixed-cost vs. dead-weight loss; R&D: patent protection.