Industrial Organization PDF

What is industrial organization Chapter 1 For the purpose of industrial organization, we’ll consider the de nition of industrial as “any large-scale business activity”. Industrial organization is concerned with the workings of market and industries, in particular the way rms compete with each other. What distinguishes industrial organization from microeconomics is the emphasis on rms’ strategies that are characteristic of market interaction (pricing policies, price competition, product positioning, advertising etc.), and is primarily concerned with imperfect competition and oligopoly. So another de nition could be the economics of imperfect competition The goal of Industrial Organization is to answer four broad questions: “Market power is the ability to charge prices above marginal cost, earning more than normal pro ts” 1. Do rms have market power Assuming that costs are proportional to output, a good approximation of market power can be obtained from data on prices, output and pro t rates A study shows that American economy’s extent of market power is very low; is consistent with ndings from the Chicago school of economics: “as long as there’s free entry into each industry, the extent of market power is never signi cant” There are a large number of industries were rms hold little or no market power, but there are some industries where market power exists to a signi cant extent 2. How do rms acquire and maintain market power? Market power = higher pro ts. It’s an important part of a rm’s value-maximization strategy To acquire market power Legal protection: it is possible to be legally protected from competition so that high prices can be set without new competitors entering the market Firm strategy: like Samsung that provided signi cantly lower prices than Apple and being rst to use Android operating system To maintain market power Establish a reputation that discourages new entries Provide a service that is extremely valuable (ex. Apple’s ecosystem of devices that are interconnected) 3. What are the implications of market power? Policymakers, if we assume their goal to be pursuing the collective good, should see the e ect of high prices as a negative outcome (the transfer of an extra dollar from consumer to rm) It also implies an ine cient allocation of resources: allocative ine ciency Firms with market power have less incentive to be cost e cient: productive ine ciency When market power is arti cially maintained by government intervention may result in rent seeking. Unproductive resources spent by rms in an attempt to in uence policymakers 4. What role is there for public policy with regard to market power? In the context of Industrial Organization, the primary role of public policy is to avoid the negative consequences of market power. Public policies in this area can be divided into 1 fi fi fi fi fi ffi fi fi fi fi fi fi fi fi fi fi ffi fl fi ffi fi ffi fi ff fi Economic regulation: refers to the case when a rm retains monopoly or near-monopoly power and its actions are directly under a regulator’s oversight. It can also be the case of industries with some market imperfections (asymmetry of information, experience goods) in which government may limit the rms’ actions Antitrust policy (competition policy): the idea is to prevent rms from taking actions that increase market power in a detrimental way, or abuse market power Industrial policy: the goal is to strengthen the position of speci c rms of industries, for example against foreign rms 2 fi fi fi fi fi fi Consumers Chapter 2 The demand function tells a lot about consumers’ tastes and preferences, so it’s useful for the rm to shape it’s behavior and production plans - We consume two goods A and B, of quantity qA and qB - The utility function U(qA, qB) expresses how much we like various quantities of the two goods - If we prefer (qA’, qB’) to (qA, qB), that are two di erent combinations, then U(qA’, qB’) > U(qA, qB) - If two combinations are equally good, U(qA’, qB’) = U(qA, qB) We can represent preferences with lines connecting combinations of qA and qB holding the same level of utility U(qA, qB)= constant. These lines are indi erence curves Indi erence curves are downward sloping because we need more of a good to compensate for less of the other (no satiation principle). The further from the origin, the higher the utility If two goods are closer substitutes, the indi erence curve will be atter The budget set is given by pAqA + pB qB ≤ y, that we use to see what choices a consumer can a ord. Total income is y. The y p budget set is the area below the budget line qA = − B qB, pA pA whose position depend on income and prices. If income increases, it shifts to the right b’, if pB increases, the line rotates inward b’’ By putting together tastes (indi erence curves) and consumption possibilities (budget line), we nd what the consumer should do, if we assume that the goal is to maximize utility. The optimal choice corresponds to the higher indi erence curve’s common point with the budget line The demand of good A depends on - The price of good A - The price of good B - Income - Consumer’s tastes expressed through indi erence curves - External conditions and addictions may matter too The demand curve gives the quantity demanded of a given good as a function of its price and of other factors. It can be derived from consumer preferences The inverse demand function tells us what the price would be if the quantity demanded is qi 3 fl ff fi ff ff ff ff ff ff ff fi A change in the price leads to a movement along the demand curve. A change in other factors leads to a shift in the demand curve itself The quantity demanded of a good depends on The prices of substitutes and complements Population and income Advertising It falls as prices rise Consumer surplus is generate for consumers by market transactions. It’s the di erence between the price and the willingness to pay (the maximum price at which I would still want to buy). Consumer surplus is an important measure of the e ciency of an economic system From this example about the slices of pizza, we can predict the number of slices bought (the number of slices such that the willingness to pay is greater or equal to the price) In both graphs the purple area is the consumer’ surplus DEMAND ELASTICITY It tells us how sensitive demand is to price changes. It’s the ratio between the percentage change in Δq q % Δquantit y quantity and the percentage change in price, for a small change in price ϵ ≡ = Δp % Δprice p When we refer to demand elasticity, we are talking of price elasticity of demand. Elasticity is negative, and it’s independent of units | ϵ | > 1 have an elastic demand so the quantity demanded is very sensitive to price 0 < | ϵ | < 1 elasticity is small, demand is inelastic, so it’s relatively insensitive to prices The percentage in revenue (R=p x q(p)) introduced by a small change in price is d( pxq( p)) dq( p) q( p) p dq( p) = q( p) + p = q( p)[ + ] = q( p)(1 + ϵ( p)) dp dp q( p) q( p) dp Revenue falls if ϵ < − 1 Revenue rises if ϵ > − 1 Revenue is unchanged if ϵ = − 1 4 ffi ff Cross-price elasticity tells how sensitive is demand for a product to prices of competing products dqA qA ϵAB = dpB pB - If ϵAB >0 the goods are substitutes - If ϵAB 1. 1. Demand for luxuries is more sensitive to price changes than demand for necessities 2. Demand for speci c products more sensitive to price changes than demand for a whole category 3. Long-run demand more sensitive to price changes than short-run demand 5 fi The production side of the market Chapter 3 Firms are organization, made of people, but for the purpose of Industrial organization we focus o the rm as a “blackbox”, mainly a process of transforming inputs into outputs. The “box” in economics is what corresponds mathematically to the rm’s production function The rm’s production function tells how much output q can be produced for a given set of inputs xi q = f (x1, x2,... , xn). Of course there are rms more e cient than others at transforming inputs into outputs, it also depends on the quality of inputs (ex. Skilled or unskilled workers) For the purpose of our analysis we consider: Two inputs: L labor and K capital Prices of the inputs W wages, and R as rent Firm is price-taker in the inputs market To understand what combinations of input lead to a given level of output we use the isoquant curve Extreme cases: - A production process operating on xed proportions is f (x1, x2) = min[x1, x2], inputs are perfect complements, leading to a right-angle isoquant curve - When the two inputs are interchangeable, inputs are perfect substitutes, leading to a straight isoquant curve - The intermediate case, where inputs are neither perfect complements nor perfect substitutes shows convex isoquant curves (due to diminishing marginal returns) The isoquant curves in the last gure are derived from a particular production function called Cobb-Douglas production function q = K α L β. For a given output level, a decrease in L requires to use more K to compensate (diminishing marginal return). By observing the three gures we can say that the closer complements two inputs are, the more convex the isoquants are PRODUCTIVITY A good measure of performance is TFP, total factor productivity We assume a. Market price is given (the same for all rms b. Each rm has a Cobb-Douglas function with same α and β but a di erent multiplying coe cient qi = ωi Kiα Liβ The coe cient ωi will be the TFP of each rm. To estimate TFP lnωi = lnqi − αlnKi − βlnLi 6 fi fi fi ffi fi fi fi fi fi fi fi ffi ff ffi A pro t maximizing rm chooses input levels such that the ratio between input costs and total revenue is equal to the corresponding coe cient in rK wL the function α= and β = pq pq COST MINIMIZATION Firms seek the lowest cost with a certain level of output which is the w tangency point between the isoquant and a line whose slope is − r which is the isocost line. By rearranging the cost function C = w × L + r × K we get the isocost line C w K = − L that are all the combinations of L and K that give the same cost C r r COST FUNCTION C(q) It shows the least cost of inputs the rm needs to pay in order to produce output q Fixed cost FC doesn’t depend on output level Variable cost VC would be Xero if output levels were zero Total cost TC is FC+VC, also AC x q Average cost AC also known as unit cost is TC/output level or C(q)/q Marginal cost MC is the cost of one additional unit. The total cost of producing q +1 units minus the cost of dC producing q units dq Marginal cost is the appropriate concept to use to decide how much to produce, while average cost to decide whether to produce or not When p < AC implies that p x q < AC x q that means revenues (p x q) are less than total cost (AC x q) the rm shouldn’t produce at all PRICING Often rms face a tradeo between price and quantity. To sell more, they must charge less The level of output should be chosen do that the value of the marginal revenue is equal to the marginal cost R(q) = pq = P(q)q revenue d R(q) d(P(q) × q) MR(q) = = = P(q) + P′(q) × q = p + P′(q) × q < p dq dq π(q) = R(q) − C(q) the optimal output level is found by maximizing pro t dπ(q) = MR(q) − MC(q) = 0 dq 7 fi fi fi fi ff fi ffi fi d R(q) dC(q) MR(q) = and MC(q) = dq dq The pro t maximizing output level is such that marginal revenue equals marginal cost MC With the elasticity rule p = , the rm should operate at a 1 1+ ϵ p − MC 1 positive margin m ≡ = which means, the lower the p −ϵ price elasticity of demand in absolute value, the higher the price-cost margin should be set p − MC Often, instead of price margin, we use price markup: k ≡ MC 1 which in terms of elasticity is k ≡ it is always optimal for −ϵ − 1 the seller to increase the price as long as | ϵ | < 1 DO FIRMS MAXIMIZE PROFIT? Although management and ownership are normally separated, there are reasons to believe that deviations from pro t maximization cannot be too large. The precise meaning of “not too large” remains an unresolved empirical question The horizontal boundaries of the rm are largely determined by cost considerations. The vertical boundaries result from the balance between investment incentives (speci c assets) and performance incentives 8 fi fi fi fi fi Competition, equilibrium and e ciency Chapter 4 Perfect competition: -No entry barriers (free entry) -Perfect information: rms and consumers know about prices and product characteristics -Atomistic rms: many competitors, unable to a ect the market price -Homogeneous products: competitors produce exactly the same products We study perfect competition as a benchmark to study the e ects of competitive forces; very few markets are perfectly competitive Each rm faces a at demand curve: can sell how much it desires at the market price, or nothing at an higher price A rm’s total revenue is R(q) = pq, and since p is given, the rm’s marginal revenue is p, hence the pro t maximization is MC(q) = p Industry supply curve is the sum of all the rms’ supply S( p) = q1 + q2 +... + qn The point in which demand and supply curve intersect is the market equilibrium The law of demand and supply says that whenever the price is below or above the equilibrium price, there are forces pushing it back to the equilibrium point Comparative statics is the study of what happens to equilibrium if an exogenous factor changes, shifting supply or demand In the short run the number of rms is xed In the long run rms can enter or exit the market. In the long-run equilibrium, rms produce at the minimum average cost and make zero pro ts (if the price is above the average cost, we expect new rms to enter the market in the long run, shifting to the right the supply curve, driving down the price) GAINS FROM TRADE In a voluntary transaction, both parties become better o. In an economy, value is also created from trade, not only by production Welfare is measured by Consumers' surplus: the di erence between the demand curve (the consumers’ willingness to pay) and the market price. Buyers pay less than their willingness to pay Producers’ surplus: the di erence between the price and the supply curve (the price at which producers are willing to sell). Sellers sell for more than marginal cost THE FUNDAMENTAL THEOREM In a competitive market, the equilibrium levels of output and price correspond to the maximum total surplus (total gains from trade (surplus) are maximized) All trades at which willingness to pay is higher than marginal cost take place Only the best rms survive competition 9 fi fi fi fi fi fi fi fl fi ff ff fi fi fi fi ff ffi ff ff fi fi If q ≠ q * then there is allocative ine ciency (C area) If S ≠ SL then there is productive ine ciency (D area) Real world observations are at odd with the model of perfect competition - Pro t rates are persistent in the long- run - Productivity di ers across rms - Simultaneous entry and exit - Entrants are of much smaller size than the industry average size MONOPOLISTIC COMPETITION If we drop the homogeneity assumption Large number of rms Downward sloping demand function ( rm is price maker, given the other rms’ price) Free entry and exit In equilibrium MR=MC Zero pro ts Firms do not produce at the minimum of their average cost 10 ffi ffi fi fi ff fi fi fi fi Price discrimination Chapter 6 The goal of price discrimination is selling at higher price to consumers whose willingness to pay is higher; sell at lower price to consumers whose willingness to pay is lower Perfect price discrimination: each customer is charged a di erent price; exactly his willingness to pay Discriminatory pricing policies are common in markets like customer markets where the number of customers is relatively small and the seller has considerable information on the buyers With respect to normal pricing - The seller gains - The low-price buyer often gains - The high-price buyer often loses Overall welfare (consumers’ + producers’ surplus) increases, but distribution of welfare changes drastically, now there is zero consumers’ surplus Di culties Di cult to know the willingness to pay of customers Arbitrage: resale, grey markets Legal limits Fairness: perceived as unfair to charge di erent prices to di erent customers TYPES OF PRICE DISCRIMINATION - By indicators: di erent segments can be easily identi ed - Higher price in less elastic markets - By self-selection: rms cannot identify the di erent segments - Versioning: designing product lines that appeal to di erent consumers (like normal and business class tickets) UB(NR) − pNR ≥ UN (R) − pR and UT (R) − pR ≥ UT (NR) − pNR to make sure that every traveller buys the ticket aimed at his type (incentive compatibility constraint) - If the inexpensive version is too attractive to the high-price customer, they get a positive surplus buying it - Need to reduce the attractiveness of this version to the high-price consumer enough to make the incentive compatibility constraint to hold - Non-linear pricing: unit price varies with quantity purchased - Quantity discount - Two-part tari : xed entry fee (F) to extract consumer surplus F=CS(p=MC), per-unit price (p) to maximize consumers’ surplus P=MC 11 ffi ffi ff ff fi ff fi ff ff ff fi ff - Under FULL INFORMATION (the rm has signals to distinguish consumers). The rm designs a menu of contracts but the consumer is constrained to buy that designed for him - Under ASYMMETRIC INFORMATION (the rm has no observable signs to distinguish consumers), the rm o ers a menu of contracts but the consumer can buy the package he prefers, without constraint. The pro t maximizing o er must satisfy the incentive compatibility constraint and the participation constraint UL( pL, FL ) ≥ 0 and UH ( pH , FH ) ≥ 0, to reduce the gain that H-type consumer gets if pretends to be L- type, we increase the price aimed at L-type, so less L-type buy, but reduces the gain from cheating of the H-type L-type consumers pay an higher pen unit price, but a lower xed fee, and are left with zero surplus H-type consumer pay lower per unit price but higher xed fee, but are left with positive surplus - Bundling: more products are on sale in a “bundle” - Pure bundling: products are NOT on sale separately - Mixed bundling: products are on sale in a bundle and also separately AUCTIONS Ascending price/English auction: highest bid wins: rise the current highest bid as long as it is equal to your value Descending price/dutch auction: lowest one wins First-price auction: sealed bid submitted, highest bid wins Second-price auction: sealed bid, the winner pays the amount bid by the second-highest bidder: best strategy, bid your highest value The winner is the person with the highest value, and their expected payment equals the expectation of the second-highest value winner’s curse: the winner’s bid exceeds the common-value item’s value, so the winner ends up paying too much Multi-unit auctions: multiple, identical units at auction (simultaneous ascending auction) 12 fi ff fi fi fi fi ff fi fi Oligopoly Chapter 8 When there are few competitors in a market we call it oligopoly or duopoly if there are only two Unlike monopoly or perfect competition, where one needn’t to worry about the reactions of competitors, in an oligopoly, an important characteristic is the strategic interdependence among competitors. An action by rm 1 will likely in uence rm 2’s pro ts. So a rm’s decision process should take into account how the other rm is expected to react The demand received by each rm depends on the price it sets, and also on the prices set by the rivals (if there are few) BERTRAND MODEL Players: two rms produce identical products Identical and constant marginal cost MC Strategies: rms set prices simultaneously Total demand Q=D(p) where p is the lowest price D( pi ) i f pi < pj For each rm i Di( pi, pj ) = αD( pi ) i f pi = pj where α ∈ (0,1) normally for simplicity equal 1/2 0 i f pi > pj Continuous-variable strategies Suppose that rms can set any value of p from 0 to +in nity including non integer values p ∈ ℝ+ Payo is now not represented by a payo matrix but by πi( pi, pj ) Firm i’s best response is pi*(pj) that gives for each price rm j sets, rm i’s optimal price Best response mapping: values of p* i ( pj ) such that πi( pi, pj ) ≤ πi( p* i , pj ) for all pi Nash equilibrium: values ( pî , pĵ ) such that πi( pi, pĵ ) ≤ πi( pî , pĵ ) for all pi πi( pî , pj ) ≤ πi( pî , pĵ ) for all pj Which is equivalent to pî ∈ p* i ( pĵ ) and pĵ ∈ p* j ( pî ) Nash equilibrium p1=p2=MC Under price competition with homogeneous product and constant, symmetrical cost (Bertrand competition), rms price at the level of marginal cost The Bertrand trap: The Bertrand competition is unattractive for sellers, because the presence of even one competitor can drive down prices tremendously with respect to the absence of competitors (monopoly) 13 ff fi fi fi fi fi fi fl fi fi ff fi fi fi fi fi fi To avoid the Bertrand trap 1. Product di erentiation: the model assumes that both rms sell the same product. If products are di erentiated, then duopoly price competition doesn’t not necessarily drive down to marginal cost as Bertrand model predicts 2. Dynamic competition: the model assumes that rms compete in one period only, but considering more periods implies that the rival with retaliate in the next period by undercutting your prices, possibly initiating a price war Asymmetric costs 3.Asymmetric costs: the assumption of the model is that the rms have the same marginal cost. If one of the rms has a lower marginal cost, then it is no longer true that both rms earn zero pro ts. The Nash equilibrium is p2=MC2 and p1=MC2-ϵ. Firm 1 just undercuts Firm 2’s marginal cost and gets all of the market demand 4. Capacity constraints: if the rm doesn’t have enough capacity to satisfy the whole market, then it is useless to undercut the rival to obtain the whole market demand. (One assumption Capacity constraint of the Bertrand model is that rms do not have capacity constraint). Suppose p2>p1 and that D(p1)>k1 (k1 is the max. capacity of rm 1 sales will be given by k1 if sets price lower than the rival, and rm 2 will be D(p2)-k1. If both rms are capacity constrained and k2>k1, both rms should set prices such that total demand equals industry capacity p1=p2=P(k1+k2), that is the intersection of the inverse demand curve P(Q) with the total capacity curve k2+k1, If total industry capacity is low in relation to market demand, then equilibrium prices are greater than marginal cost. If industry capacity is very high, then an equilibrium with positive margins may turn into the Bertrand trap 14 ff fi fi fi ff fi fi fi fi fi fi fi fi fi COURNOT MODEL Players: two rms produce identical products Constant marginal cost MC=c Strategies: - Firms set output simultaneously - Market price is a function of total output p=P(q1+q2) where P(Q) is the inverse demand curve Pro ts: for rm i equal to πi = P(q1 + q2)qi − c(qi ) Total output: Q = q1 + q2 Demand curve: P(Q) = a − bQ Cost function: C(q) = cq Firm 1’s pro t function: π1 = Pq1 − C(q1) = (a − b(q1 + q2))q1 − cq1 Δπ1 First order condition for pro t maximization =0 Δq1 −bq1 + a − b(q1 + q2) − c = 0 a − c q2 Firm 1’s best response q* 1 (q2 ) = − 2b 2 An equilibrium is de ned by a system of equations q1̂ = q* (q̂ ) and q2̂ = q* 1 2 (q̂ ) 2 1 Since the two rms are identical the equilibrium will be symmetrical q1̂ = q2̂ = q̂ a − c q̂ a−c q̂ = − = 2b 2 3b Under output competition (Cournot), equilibrium output is greater than monopoly output and lower than perfect competition output. Likewise, the duopoly price is lower than the monopoly price and greater than the price under perfect competition a−c a + nc When there are n rms in a Cournot oligopoly, the solution is q̂ = and p̂ = (n + 1)b n+1 Which shows than as n → ∞, p → c, that is as the number of competitors grows large, the equilibrium price under Cournot competition converges to the competitive price level, which is also the equilibrium under Bertrand competition If players make more than one strategic choice, one needs to model the sequence of moves Each player’s choices may be made at the same time or in a sequence Players make short-term moves given their long-term choices If capacity and output can be adjusted easily, then the Bertrand model describes duopoly competition better. If output and capacity are di cult to adjust, then the Cournot model describes duopoly better 15 fi fi fi fi fi fi fi fi ffi Collusion and price wars Chapter 9 Competition implies externality, rms maximize their own pro ts, not joint pro ts Collusion: concerted practices in which individual choices are meant to maximize joint pro ts, which are then shared (at the expense of consumers) - Cartel agreements are a particular institutional form of collusion - Secret agreements; because illegal - Tacit agreements: attained for historical reasons or as a natural conclusion Types of collusive behaviors include: price increase, reduced supply, territory restrictions and setting levels of service quality STABILITY OF COLLUSIVE AGREEMENTS - Homogeneous product duopoly ( rms set simultaneous prices, with constant marginal cost and no capacity constraints) - If one shot game, Bertrand equilibrium (p = MC) - If inde nitely repeated game, the strategy is a list of actions at each node, dependent on the history of the game - A grim trigger strategy is one of the simplest 1 M - In the rst period, each rm sets the monopoly price pM and share monopoly pro ts equally π 2 - Set p = pM if both rms charged p = pM in the past ( rms respected collusive agreement) - Set p = MC otherwise - The claim is that it may exist a grim trigger strategy equilibrium where p = pM at each t If both rms stick to the equilibrium strategy, Firm 1’s expected discounted payo (NPV) is 1 M 1 1 1 1 V= π + δ π M + δ2 π M +... = π M where delta is the discount factor, that is the 2 2 2 2 1−δ value of 1$ one period into the future compared to its value now If I deviate and set p = MC, my rival in the future will do the same, so I will do the same over and over 1 δ V′ = π M + πP = πM + 0 1−δ 1−δ 1 Conditions for an equilibrium in which both follow a grim trigger strategy V ≥ V′ or simply δ ≥ 2 If the discount factor is su ciently high, then there exists a Nash equilibrium of the repeated game where rms set monopoly price in every period under the threat that if there is a deviation, both rms will revert forever to setting the marginal cost 16 fi fi fi fi fi fi ffi fi fi fi fi fi ff fi fi fi DISCOUNT FACTOR Normally 0< δ < 1, delta is lower than zero, an explanation would be the opportunity cost of time, 1$ 1 can be used to gain (1+r) in the next period, where r is the interest rate. So δ = 1+r Determinants of collusion are How many times rms choose the price in a year (how long the rm is bene ting from the deviation). f is the frequency of interaction How is the market evolving, g is the industry growth rate Is the market interaction and the bene ts from collusion going to continue, h is the hazard rate (probability of industry or rm end) (1 + g)(1 − h) δ= collusion is more likely when r is low, f is high, g is high, h is low 1 + rf Predictions from the model: - Stable high price in case of collusion - Endless punishment of deviations: stable low price after the break of the collusive outcome PRICE WARS - Demand uctuates - Unobservable demand shocks ( rms can observe only their own price and demand, not market’s) - Low rm demand implies guessing problem - Low market demand or price cutting by rival Suppose rm a decides not to punish rm b assuming its low demand is resulted from a market downturn, not from rm b cheating. It cannot be an equilibrium, as rm b could be better o by cutting prices and blaming market conditions for rm a low demand But also, in nitely punishing rm b will revert into an inde nite price war even if no cheating on the agreement occurred - An intermediate strategy: punishment for T periods (large enough to discourage price undercutting the rival). It is an equilibrium with colluding phases alternating with price war phases (as empirical observation suggests) - Price wars are a necessary evil for collusion to take place, as the incentives for cheating would be too great for the collusive agreement to be stable without price wars DEMAND FLUCTUATIONS - Observable demand uctuations (demand shocks are interdependent across periods, the future looks the same in terms of demand shock, regardless of the current state of demand) - Incentive to undercut is greater during periods of high demand - Collusive equilibrium may require prices to be lower during periods of high demand - Unobservable demand shock: price low, demand low - Observable demand shock: price low, demand high - Prices move in a counter-cyclical way, price wars take place during booms 17 fi fl fi fi fi fi fl fi fi fi fi fi fi fi fi fi fi ff FACTORS FACILITATING / HAMPERING COLLUSION Symmetry between rms; a weaker rm may decrease prices due to its condition, or due to having a lower discount factor (implies that the future may be less relevant, bigger incentive to cheat) Market structure: collusion is more likely in concentrated industries than in fragmented ones. Same for similar rms (if di erent rms, the more e cient has incentive to deviate if pro ts are shared, the less e cient rm has incentive to deviate if pro ts and production go to the more e cient) Easier to establish a collusive agreement (less communication) Easier to maintain a collusive agreement (lower incentive to deviate) Multi-market contact: if all markets are identical, multi-market contact makes no di erence Firm a has an advantage in market a and charges C In market a price equal to the rival’s cost p (c̄) − ϵ → entire market C In market b price equal to its own cost p (c̄) → no market Taken in isolation, the prices charged in the collusive equilibrium are not stable, while taken together the threat of reverting to the competitive equilibrium may be su cient to sustain collusive equilibrium Information sharing: Perfect information: condition for perfect competition Natural presumption: the more natural the market is, the more competitive it is 18 ffi fi fi fi ff fi fi ffi fi ffi fi ffi ff Market structure Chapter 10 The determinants of market structure are Technology Market size Mode of competition Market power HOW TO MEASURE MARKET CONCENTRATION AND MARKET POWER In a symmetric oligopoly model, the number of rms n is a good indicator of industry structure. The lower n the more concentrated the industry is. 1/n is a good measure of concentration, ranging from 0 to 1 Considering rms with di erent market shares, we need an alternative to counting the number of rms. We measure the coe cient Cm that is the sum of the market shares of the largest m rms in the market m ∑ Cm = si ranges from 0 to 1, or from 0 to 100 if market share is in percentage i=1 Another index is the Her ndahl-Hirschman or HHI which is the sum of squared values of market shares of all rms in the market N (si )2 ranges from 0 to 1, or from 0 to 10.000 if market share is in percentage ∑ HHI(or H ) = i=1 It provides a better measure of market concentration but is harder to compute Measuring market power p − MC - Before now, by measuring the markup × 100 ( ne if all rms have the same cost) MC - Lerner index, is de ned as the weighted average of each rm’s margin (weights given by market n p − MCi ∑ share) L ≡ si p i=1 Cournot competition Symmetric rms (concentration = n of rms). Markets are assumed identical once taken into account di erences in size - Cost function Ci = F + cqi - Demand q = (a − P)S where S is the market size (n + 1) 2 a−c - Equilibrium pro t level Π(n) = S −F 19 fi ff fi fi fi fi fi fi ff ffi fi fi fi fi fi fi [ ] S S - Free entry equilibrium Π(n) = 0 → n = (a − c) −1 n ̂ = (a − c) − 1 where [x] F F denotes the highest integer lower than x. A free entry equilibrium is characterized by a set of active rms such that: - No active rm wishes to leave the market Π(n)̂ ≥ 0 - No inactive rm wishes to enter the market Π(n ̂ + 1) ≤ 0 - n ̂ is the equilibrium value of n, decreasing in F, increasing in a and S The intuition is that higher S leads to higher n, but higher n leads to lower p, which in turn limits the increase in n (due to increased price competition, the equilibrium number of active rms varies less than proportionally with respect to market size) Market size is an inverse function of xed F and variable c costs. For high values of n,̂ the relationship between S and n ̂ is almost quadratic. To double the number of rms, market size increases 4 times ROLE OF TECHNOLOGY F is the crudest measure of increasing returns to scale. The larger the degree of increasing returns to scale, the more concentrated the market. The relationship is illustrated by F - Minimum e cient scale: quantity such that AC is minimal, or close enough MES = → MES c′ − c and F move together (AC=c’) F AC q +c F - Scale economies: ρ coe cient of scale economies ρ ≡ = = 1 + → ρ and F MC c cq move together. It ρ>1, economies of scales, if ρ0, it pays E - Demand is given by P = a - bQ - Cost for rm 1: marginal cost c1 - Cost for rm 2: marginal cost c2 and entry cost E 25 fi fi fi fi fi ff fi fi 1. Look at stage 2 rst (backward induction) Taking rm 1’s choice as given, rm 2 maximizes by π2 = (a − b(q1 + q2)) − c2)q2 a − c2 1 Solving for q2 by the rst order condition we get q* 2 (q1 ) = − q1 2b 2 2. Stage 1 now Firm 1 anticipates rm 2’s best reply by incorporating q2*(q1) into its pro ts 1 π1 = (a − b(q1 + q* (q )) − c1)q1 = 2 1 (a − c2 − 2c1 − bq1)q1 2 a + c2 − 2c1 a − 3c2 + 2c1 Optimal quantity for rm 1 is q* 1 = and for rm 2 q* (q* 2 1 ) = 2b 4b 3. Equilibrium Quantities q* 1 > q* 2 1 1 Pro ts π* 1 = (a + c2 − 2c1)2 π* 2 = (a − 3c2 + 2c1)2 − E π* 1 > π* 2 8b 16b E may be too large and the entry doesn’t occur at all 4. Revised stage 1 (entry deterrence in capacity setting games) 1 Entry by rm 2 occurs if (a − c 2 − bq1) ≥ E 4b In order to deter entry, rm 1 myst set 11 so that the above equality doesn’t hold 1 Firm 1 entry deterring capacity level is q1D = (a − c 2 − 2 bE) b ENTRY DETERRENCE When E is very low, deterring entry is costly (too large quantity) and the incumbent would be better o by accommodating entry When xed costs are very high, rm 1 is better o by setting monopoly output (blockaded entry) When xed costs are intermediary, rm 1 can choose if to deter (entry deterrence) Quantity choices are irreversible for the incumbent, otherwise, it could revise them if entry occurs Knowing this, entrant may decide to enter even if incumbent choose the entry deterring quantity PRODUCT PROLIFERATION An industry like cereals has a low economies of scale, and low levels of technology. It should be easy to enter into the industry, from a technological perspective. But there were few entries over the years, even thought pro ts were signi cant by the incumbents. The number of brands sold by the incumbents also increased signi cantly, and keeps increasing - Incumbent rms lled-in the product space to eliminate opportunities for pro table entry (product proliferation strategy to deter entry without lowering prices) - This strategy applies not only to density in the product space, also to density in geographical space 26 fi fi fi fi fi fi fi fi fi fi fi fi fi fi fi fi fi ff fi fi fi fi ff FORECLOSURE AND LONG TERM CONTRACTS Naked exclusion: signing up a su cient number of buyers, the incumbent e ectively makes entry unpro table. Suppose entry is pro table only if entrants manages to attract a fraction of the market given by α (0 uAl ) ∀ l ≠ B - The problems arise with aggregate demand data, since consumers have di erent tastes and their own sets of coe cients to estimate COMPETITION WITH DIFFERENTIATED PRODUCTS Vertical product di erentiation: consumers agree on what is better (unanimously prefer more of a given characteristic) Horizontal product di erentiation: di erent consumers have di erent ideal points (di erent buyers’ preferences have di erent values, some think a feature is a good thing, some that it’s bad) VERTICAL DIFFERENTIATION - The simplest version is a duopoly where rms o er products that only di er on the quality dimension - Di erent consumers have di erent valuations for higher quality, but all agree that more of it is better - Similar to Bertrand model. (Firms simultaneously set prices, consumers decide from which rm to buy, then rms produce and supply the amount demanded) vj with (j = 1, 2) is the amount of relevant characteristics that rm j’s product has. (We refer to it as j’s quality). Assume 2 v2 > v1 ( rm 2’s product is better) Let b be the value in $ that a consumer attributes to a product quality (varies from consumer to consumer) b ∈ [bL, bH ] uniformly distributed ph is price set by rm h utility is uj(b) = bvj − pj I. Not buying the car yields a net utility u0 = 0 (b < b1) 31 ff fi ffi ff fi fi ff ff ff ff fi ff fi ff fi fi ff ff fi ff fi ff ff ff fi II. Buying a car from rm 1, which yields a net utility u1 = bv1 − p1 (b > b1 and b < b2) III. Buying a car from rm 2, which yields a net utility u2 = bv2 − p2 (b > b2 U0 is not a function of b, while the others are increasing in b. Consumers who value the relevant characteristic more, are willing to pay more for the product in question The rate at which net utility changes with b di ers across products. The increase is particularly signi cant for the higher-quality product ( rm 2’s) The di erence from Bertrand is that a small change in price doesn’t imply a big change in demand, and it is not su cient for a rm to price lower than the rival to capture all market demand - Assume total number of consumers to be N=1 (ex. 1 mln) - Assume bL=0 and bH=1 and that all consumers want to buy ( p2 − p1) - Demand levels q1 and q2 = 1 − q1 = b2 = (v2 − v1) - Solving the rst order condition to maximize πi = ( pi − c)qi we get 1 2 - Equilibrium prices p1̂ = (bH − 2bL )(v2 − v1) and p2̂ = (2bH − bL )(v2 − v1) 3 3 1 4 - Equilibrium pro ts π1̂ = (bH − 2bL )2(v2 − v1) and π2̂ = (2bH − bL )2(v2 − v1) 9 9 PRODUCT POSITIONING Assume product quality is chosen by rms, v2 is given, and rm 1’s chooses v1 Firm 1’s pro ts decrease as v1 increases due to two e ects 1. Direct e ect: an increase in v1 corresponds to the change in π1 that happens if prices remain constant at their initial equilibrium level (positive) 2. Strategic e ect: the e ect of the equilibrium price adjustments following an increase in v1 (negative) For the lower quality rm to increase its quality levels means to bring v1 closer to v2.The closer v1 to v2 the more competitive pricing is HORIZONTAL DIFFERENTIATION Consumers are distributed along line segment (sellers o er products which di er to some characteristic and buyer di er among themselves as how they value such characteristic) -> hotelling model ( rms sell the same product but consumers do not endure the same transportation costs to purchase the product, so rms can price di erently. The rms are located at some di erent point along a straight line) 32 fi ff ff ffi ff fi fi ff fi fi fi ff fi fi ff fi fi fi ff fi ff ff fi ff ff fi Horizontal axis measures each consumer’s address (their preferred product characteristic) A consumer of type x has a valuation v (the utility, same for all consumers) and must pay a total cost pj + t | x − aj | where pj is rm j’s price, aj is rm j’s address (a1=0, a2=1) and x-aj is the distance between consumer’s preference and the rm’s o ering. t is consumer’s transportation cost (the drop in utility resulting from one unit of “distance” Not buying from any rm u0=0 Buying from rm j (j = 1, 2) which yields utility v − pj − t | x − aj | Consumers with low value of x (relative preference for rm 1) buy from rm 1. For these consumers - v − p1 − t | x − a1 | > 0 - v − p1 − t | x − a1 | > v − p2 − t | x − a2 | Consumers with high value of x (relative preference for rm 2) buy from rm 2. Consumers with intermediate value of x do not purchase at all The di erence from Bertrand is that a small change in price doesn’t imply a big change in demand, and it is not su cient for a rm to price lower than the rival to capture all market demand Suppose v is su ciently high that a consumer always buys A consumer with address x is indi erent between the rms if t x + p1 = t(1 − x) + p2 - Consumers to the left of x buy from rm 1, and those to the right from rm 2 (consumers are uniformly distributed along the segment [0,1] ( rm 1’s demand is given by x, rm 2’s demand by 1-x) (2 2t ) 1 pj − pi 1 pj − pi q = - i 2 + and rm i’s pro t function is πi = q ( i ip − c) = + ( pi − ci ) 2t 1 - Solving the rst order condition to maximize pro t, rm i’s best response is pi = (ci + t + pj ) 2 - Suppose that ci = cj hence pi = pj which means p=c+t (this suggests that, the greater the degree of product di erentiation, the greater market power) the Bertrand model corresponds to a particular case of the Hotelling model in which t = 0 (no transportation cost) all consumer go to the lower price setting seller - If price competition is very intense, then rms tend to locate far apart (high degree of di erentiation) if price competition is not very intense, then rms tend to locate close to each other (low degree of di erentiation) - Product di erentiation is a solution to the Bertrand trap: price competition, but pricing above marginal cost 33 ff ff ffi ff ff fi fi fi ffi fi fi fi fi ff ff fi ff fi fi fi fi fi fi fi fi fi fi fi fi fi fi ADVERTISING AND BRANDING Now we consider di erentiation by di erent perceptions of consumers as e ect of advertising or branding Economists classify goods according to two categories Search goods: has features that consumers can evaluate before purchasing Experience goods: whose feature can only be evaluated after consumption (like services) Advertising can be classi ed from this distinction as - Informative advertising: describes the product’s existence, its characteristics, selling items etc. - Persuasive advertising: designed to change consumers’ perception about the product’s value Advertising is the mean by which brands are created (a bundle of physical features and brand images). A strong brand is vehicle of expensiveness, higher quality, top product features Implicit contract: a strong brand creates perception of higher quality, so consumers exec yo pay a higher price Umbrella branding: or brand extension, selling products a η A pro t maximizing monopolist chooses = where a/R is the advertising to sales revenue ratio R −ϵ dq a and epsilon is the rm’s price elasticity of demand and η ≡ that is the demand elasticity with da q respect to advertising expenditures (measure how much quantity demand increases (%) when advertising expenditures are increased by 1%) ADVERTISING AND PRICE COMPETITION In a context in which two products sold are physically di erent but consumers do not know which product the di erence, compete in a Bertrand model, with prices equal to marginal cost. In this context, informative advertising softens price competition. The moment consumers perceive the di erent product as di erent, the rms do not have any reason to undercut the rival (advertising product features) In a context in which two products are homogeneous but consumers do not know at which price products are sold, informative advertising lead sto solutions closer to Bertrand (advertising prices) PUBLIC POLICY Advertising leads to new areas public policy, like consumer protection. - Changes the rules of product di erentiation - Information increases, improved allocative e ciency CONSUMER BEHAVIOR De facto di erentiation can result also from imperfect information or consumer behavior Search cost: if consumers have to spend to be informed about the price of a product, then rms can more easily set higher prices without losing customers (price dispersion, with heterogeneous consumers, some will buy at higher price at closer location, others will travel to nd lower prices) Shrouded attributes: sellers can engage in obfuscation practices, making selling terms unclear. Or shrouding attributes or features of a product or of the sale contract until the purchase takes place Switching cost: the combination of monetary and non monetary Barries to change from one brand to another Harvesting e ect: rms can increase prices without losing many of their current customers Investment e ect: when competing for new customers, the expectation of future rents from captive customers leads rms to price more aggressively 34 ff fi ff ff ff ff fi fi fi fi ff fi ff ff ffi ff ff fi ff fi

Industrial Organization PDF

Document Details

Tags

Related

Summary

Full Transcript

Upgrade to continue