7. Application: Industrial Organization Using theory to identify the degree of market power in a given market. That is, we want to estimate PMC/P. If it is low or zero, the market is highly competitive. If it a above zero by a significant amount, then the market is not very competitive. The problem is, in most cases, we do not observe marginal cost. Does this mean that we should give up? Maybe not. From IO economics, we know that many types of oligopoly markets have equilibrium prices that can be written like this: PPMC= Where is the price elasticity of demand. The parameter measures how competitive the market is. If =0, it is perfectly competitive. If =1, it is a monopoly, and values of between 0 and 1 are the in-between cases. For example, in the Cournot model with n firms, =1. IO economists often want to estimate . If we knew P,MC and could estimate , then estimating would just be arithmetic. However, in many cases, we don't observe MC. Firms just don't keep cost information that way. So how do we estimate if we don't observe MC? Review of Oligopoly Models Cournot Model P=P(q1+q2++qn)=P(Q) i=qiP(Q)cqii=1,2,,n Equilibrium condition: qii=P(Q)+qiP(Q)c=0 or MRforfirmiP(q1+q2++qn)+qiP(q1+q2++qn)=Mcc. Example: two firms, P=100Q,c=10 : 1=q1[100q1q210] 2=q2[100q1q210],q11=1002q1q210=0,q22=100q12q210=0,90=2q1+q2, 90=q1+2q2, q1=q2=30, P=100230=40. Monopoly: M=Q[100Q10]=100QQ210Q=90QQ2,QM=902Q=0,QM=45, PM=55. Robert Porter, Tim Bresnahan and others developed a way of estimating without needing to know MC. We do need to be able to estimate , though. This is called the NEIO" approach. 1. P=0+1Q+2Z+3ZQ+1. This is the demand curve, written inversely. Note that if we estimate this using OLS, we will get biased estimates of the coefficients, in particular the terms involving Q:1,3. Hence, we would estimate this using some form of instrumental variables. We'll come back to this. 2. Assume that firms compete using quantities as strategies. Assume also that when firm i chooses qi, it conjectures" that other firms may adjust their outputs in response. Define dqidQ As the "conjectural variation" of an increase in qi. Using this conjectural variation idea, we can write the marginal revenue of an individual firm as: MRi=P(Q)+QP(Q),Q=j=1nqj. Now write the profit maximization MR=MCP(Q)+QP(Q)Technically,=Qqiiq1Q=MC.PPMC=PQP=, where is the market price elasticity of demand. PPMCisoftencalledtheLernerindex,soLernerIndex=,iswhatwewanttoknow. How can we estimate ? 1. Suppose we know c and can estimate . (PPc)=,noproblem. 2. However, usually we cannot observe c, even if we have access to a firm's P\&L statement. - The P\&L may contain many products and/or services aggregated together. The product or service we are interested in may be concealed within that aggregation. - The "variable" costs may contain allocations of fixed costs. - Assets may be carried on a firm's balance sheet at "historical" cost, less funky accounting-based depreciation. What do we do then? 1. Using the assumed demand curve, MR=P+PQ=P+(1+3Z)Q 2. We don't observe MC, but we can assume MC=0+1q+2w+2. So if the equilibrium is symmetric, P+(1+3Z)Q=0+1nQ+2w+2, or P=0+Q[n113Z]+2w+2=0+B1Q+B2ZQ+2w+2, where B1=n11.B2=3. 3. Estimate the demand curve from (1) and the equilibrium condition (2). From demand curve: ^0,^1,^3. From equilibrium condition: ^0,B^1,B^2. 4. Identifying ^ : B^2=3, We estimate ^3 from (1): ^=^3B^2. 7. Application: Industrial Organization Using theory to identify the degree of market power in a given market. That is, we want to estimate PMC/P. If it is low or zero, the market is highly competitive. If it a above zero by a significant amount, then the market is not very competitive. The problem is, in most cases, we do not observe marginal cost. Does this mean that we should give up? Maybe not. From IO economics, we know that many types of oligopoly markets have equilibrium prices that can be written like this: PPMC= Where is the price elasticity of demand. The parameter measures how competitive the market is. If =0, it is perfectly competitive. If =1, it is a monopoly, and values of between 0 and 1 are the in-between cases. For example, in the Cournot model with n firms, =1. IO economists often want to estimate . If we knew P,MC and could estimate , then estimating would just be arithmetic. However, in many cases, we don't observe MC. Firms just don't keep cost information that way. So how do we estimate if we don't observe MC? Review of Oligopoly Models Cournot Model P=P(q1+q2++qn)=P(Q) i=qiP(Q)cqii=1,2,,n Equilibrium condition: qii=P(Q)+qiP(Q)c=0 or MRforfirmiP(q1+q2++qn)+qiP(q1+q2++qn)=Mcc. Example: two firms, P=100Q,c=10 : 1=q1[100q1q210] 2=q2[100q1q210],q11=1002q1q210=0,q22=100q12q210=0,90=2q1+q2, 90=q1+2q2, q1=q2=30, P=100230=40. Monopoly: M=Q[100Q10]=100QQ210Q=90QQ2,QM=902Q=0,QM=45, PM=55. Robert Porter, Tim Bresnahan and others developed a way of estimating without needing to know MC. We do need to be able to estimate , though. This is called the NEIO" approach. 1. P=0+1Q+2Z+3ZQ+1. This is the demand curve, written inversely. Note that if we estimate this using OLS, we will get biased estimates of the coefficients, in particular the terms involving Q:1,3. Hence, we would estimate this using some form of instrumental variables. We'll come back to this. 2. Assume that firms compete using quantities as strategies. Assume also that when firm i chooses qi, it conjectures" that other firms may adjust their outputs in response. Define dqidQ As the "conjectural variation" of an increase in qi. Using this conjectural variation idea, we can write the marginal revenue of an individual firm as: MRi=P(Q)+QP(Q),Q=j=1nqj. Now write the profit maximization MR=MCP(Q)+QP(Q)Technically,=Qqiiq1Q=MC.PPMC=PQP=, where is the market price elasticity of demand. PPMCisoftencalledtheLernerindex,soLernerIndex=,iswhatwewanttoknow. How can we estimate ? 1. Suppose we know c and can estimate . (PPc)=,noproblem. 2. However, usually we cannot observe c, even if we have access to a firm's P\&L statement. - The P\&L may contain many products and/or services aggregated together. The product or service we are interested in may be concealed within that aggregation. - The "variable" costs may contain allocations of fixed costs. - Assets may be carried on a firm's balance sheet at "historical" cost, less funky accounting-based depreciation. What do we do then? 1. Using the assumed demand curve, MR=P+PQ=P+(1+3Z)Q 2. We don't observe MC, but we can assume MC=0+1q+2w+2. So if the equilibrium is symmetric, P+(1+3Z)Q=0+1nQ+2w+2, or P=0+Q[n113Z]+2w+2=0+B1Q+B2ZQ+2w+2, where B1=n11.B2=3. 3. Estimate the demand curve from (1) and the equilibrium condition (2). From demand curve: ^0,^1,^3. From equilibrium condition: ^0,B^1,B^2. 4. Identifying ^ : B^2=3, We estimate ^3 from (1): ^=^3B^2