The Costs of Environmental Regulation in a Concentrated Industry

Transcription

1 The Costs of Environmental Regulation in a Concentrated Industry Stephen P. Ryan MIT Department of Economics

2 Research Motivation Question: How do we measure the costs of a regulation in an oligopolistic industry? Motivation: EPA is required to estimate costs of regulation, usually relying on an engineering estimate of compliance costs Problem: This static analysis ignores the influence of the regulation on entry and investment behavior, and its subsequent dynamic effect on market power Goal: measure the welfare effects of an environmental regulation on a concentrated industry, accounting for the dynamics of market structure Research Motivation Application Strategy Major Contributions Preview Results Literature Review

3 Research Motivation Question: How do we measure the costs of a regulation in an oligopolistic industry? Motivation: EPA is required to estimate costs of regulation, usually relying on an engineering estimate of compliance costs Problem: This static analysis ignores the influence of the regulation on entry and investment behavior, and its subsequent dynamic effect on market power Goal: measure the welfare effects of an environmental regulation on a concentrated industry, accounting for the dynamics of market structure Research Motivation Application Strategy Major Contributions Preview Results Literature Review

4 Research Motivation Question: How do we measure the costs of a regulation in an oligopolistic industry? Motivation: EPA is required to estimate costs of regulation, usually relying on an engineering estimate of compliance costs Problem: This static analysis ignores the influence of the regulation on entry and investment behavior, and its subsequent dynamic effect on market power Goal: measure the welfare effects of an environmental regulation on a concentrated industry, accounting for the dynamics of market structure Research Motivation Application Strategy Major Contributions Preview Results Literature Review

5 Research Motivation Question: How do we measure the costs of a regulation in an oligopolistic industry? Motivation: EPA is required to estimate costs of regulation, usually relying on an engineering estimate of compliance costs Problem: This static analysis ignores the influence of the regulation on entry and investment behavior, and its subsequent dynamic effect on market power Goal: measure the welfare effects of an environmental regulation on a concentrated industry, accounting for the dynamics of market structure Research Motivation Application Strategy Major Contributions Preview Results Literature Review

6 Application Industry: US Portland cement industry Large fixed costs: $200M+ facilities Long horizons: 100 year operational lifetimes Large adjustment costs lumpy investment Geographically concentrated Second largest industrial emitter of SO 2 and CO 2, major source of NO x and particulates Regulation: 1990 Amendments to Clean Air Act Major revision to principal environmental regulation Requires permits for operation and instituted a new class of emissions New plants have additional fixed costs of entry Surprise legislation EPA took 12 years to promulgate parts of legislation pertaining to variable costs Research Motivation Application Strategy Major Contributions Preview Results Literature Review

7 Application Industry: US Portland cement industry Large fixed costs: $200M+ facilities Long horizons: 100 year operational lifetimes Large adjustment costs lumpy investment Geographically concentrated Second largest industrial emitter of SO 2 and CO 2, major source of NO x and particulates Regulation: 1990 Amendments to Clean Air Act Major revision to principal environmental regulation Requires permits for operation and instituted a new class of emissions New plants have additional fixed costs of entry Surprise legislation EPA took 12 years to promulgate parts of legislation pertaining to variable costs Research Motivation Application Strategy Major Contributions Preview Results Literature Review

8 Research Strategy Three-pronged approach: 1. Exploiting natural experiment of surprise legislation, estimate reduced form policy functions that let the data tell what effects the Amendments had on investment, entry, and exit behavior 2. Recover the cost structure of the industry before and after the Amendments by mapping the observed policy functions into an underlying dynamic model tailored to Portland cement industry 3. Simulate counterfactual distributions of consumer and producer surplus to evaluate welfare changes due to Amendments Research Motivation Application Strategy Major Contributions Preview Results Literature Review

9 Research Strategy Three-pronged approach: 1. Exploiting natural experiment of surprise legislation, estimate reduced form policy functions that let the data tell what effects the Amendments had on investment, entry, and exit behavior 2. Recover the cost structure of the industry before and after the Amendments by mapping the observed policy functions into an underlying dynamic model tailored to Portland cement industry 3. Simulate counterfactual distributions of consumer and producer surplus to evaluate welfare changes due to Amendments Research Motivation Application Strategy Major Contributions Preview Results Literature Review

10 Research Strategy Three-pronged approach: 1. Exploiting natural experiment of surprise legislation, estimate reduced form policy functions that let the data tell what effects the Amendments had on investment, entry, and exit behavior 2. Recover the cost structure of the industry before and after the Amendments by mapping the observed policy functions into an underlying dynamic model tailored to Portland cement industry 3. Simulate counterfactual distributions of consumer and producer surplus to evaluate welfare changes due to Amendments Research Motivation Application Strategy Major Contributions Preview Results Literature Review

11 Major Contributions Extend the analysis of regulation in a concentrated industry to a strategic, dynamic setting Pose an underlying dynamic model that allows for flexible investment choices, adjustment costs, endogenous entry capacity Estimate a dynamic model of oligopoly with lumpy adjustment, recovering full cost structure of industry, including distribution of sunk entry costs and fixed and variable costs of investment Provide a cost estimate of the Clean Air Act in this industry Research Motivation Application Strategy Major Contributions Preview Results Literature Review

12 Major Contributions Extend the analysis of regulation in a concentrated industry to a strategic, dynamic setting Pose an underlying dynamic model that allows for flexible investment choices, adjustment costs, endogenous entry capacity Estimate a dynamic model of oligopoly with lumpy adjustment, recovering full cost structure of industry, including distribution of sunk entry costs and fixed and variable costs of investment Provide a cost estimate of the Clean Air Act in this industry Research Motivation Application Strategy Major Contributions Preview Results Literature Review

13 Major Contributions Extend the analysis of regulation in a concentrated industry to a strategic, dynamic setting Pose an underlying dynamic model that allows for flexible investment choices, adjustment costs, endogenous entry capacity Estimate a dynamic model of oligopoly with lumpy adjustment, recovering full cost structure of industry, including distribution of sunk entry costs and fixed and variable costs of investment Provide a cost estimate of the Clean Air Act in this industry Research Motivation Application Strategy Major Contributions Preview Results Literature Review

14 Major Contributions Extend the analysis of regulation in a concentrated industry to a strategic, dynamic setting Pose an underlying dynamic model that allows for flexible investment choices, adjustment costs, endogenous entry capacity Estimate a dynamic model of oligopoly with lumpy adjustment, recovering full cost structure of industry, including distribution of sunk entry costs and fixed and variable costs of investment Provide a cost estimate of the Clean Air Act in this industry Research Motivation Application Strategy Major Contributions Preview Results Literature Review

15 Preview of Results Amendments roughly doubled sunk costs of entry, to $35M, increasing market power Consumer welfare decreased 25% due to lower entry and increased market power ( $1.2B) Amendments led to higher investment by incumbents, but lower overall market capacity Static analysis would ignore the benefits of increased market power on incumbent firms, obtaining welfare effect of wrong sign Research Motivation Application Strategy Major Contributions Preview Results Literature Review

16 Preview of Results Amendments roughly doubled sunk costs of entry, to $35M, increasing market power Consumer welfare decreased 25% due to lower entry and increased market power ( $1.2B) Amendments led to higher investment by incumbents, but lower overall market capacity Static analysis would ignore the benefits of increased market power on incumbent firms, obtaining welfare effect of wrong sign Research Motivation Application Strategy Major Contributions Preview Results Literature Review

17 Preview of Results Amendments roughly doubled sunk costs of entry, to $35M, increasing market power Consumer welfare decreased 25% due to lower entry and increased market power ( $1.2B) Amendments led to higher investment by incumbents, but lower overall market capacity Static analysis would ignore the benefits of increased market power on incumbent firms, obtaining welfare effect of wrong sign Research Motivation Application Strategy Major Contributions Preview Results Literature Review

18 Preview of Results Amendments roughly doubled sunk costs of entry, to $35M, increasing market power Consumer welfare decreased 25% due to lower entry and increased market power ( $1.2B) Amendments led to higher investment by incumbents, but lower overall market capacity Static analysis would ignore the benefits of increased market power on incumbent firms, obtaining welfare effect of wrong sign Research Motivation Application Strategy Major Contributions Preview Results Literature Review

19 Related Literature : Ericson and Pakes (1995), Fershtman and Pakes (2000), Gowrisankaran and Town (1997), Besanko and Doraszelski (2004), Doraszelski and Satterthwaite (2004) Econometric technique: Hotz and Miller (1993), Aguirregabiria and Mira (2002), Pakes, Ostrovsky, and Berry (2003), Pesendorfer and Schmidt-Dengler (2003), Jofre-Benet and Pesendorfer (2003), Bajari, Benkard, and Levin (2004) Empirical application: Benkard (2004) Empirical dynamic effects of regulation: Mansur (2003) Research Motivation Application Strategy Major Contributions Preview Results Literature Review

20 Ash Grove Cement Plant in Utah Plant Photograph Data Sources Industry Statistics Summary Statistics

21 Data Sources USGS Minerals Yearbook Market-level data for prices and quantities 27 markets covering United States Source of market definitions 517 observations Instruments on energy prices, labor inputs from Dept. Energy and County Business Patterns Portland Cement Association Plant Information Survey (PIS) Every plant in United States Kiln-level data on capacity and production 2233 observations Plant Photograph Data Sources Industry Statistics Summary Statistics

22 Industry Trends Capacity Year Production Imports Consumption Price Plants Per Kiln ,242 3,035 70, ,054 2,514 66, ,475 2,231 59, ,884 2,960 65, ,488 6,016 76, ,665 8,939 78, ,473 11,201 82, ,940 12,753 84, ,733 14,124 83, ,025 12,697 82, ,954 10,344 80, ,755 6,548 71, ,585 4,582 76, ,807 5,532 79, ,948 9,074 86, ,906 10,969 86, ,266 11,565 90, ,582 14,523 96, ,931 19, , Plant Photograph Data Sources Industry Statistics Summary Statistics

23 Summary Statistics Plant Data Standard Variable Minimum Mean Maximum Deviation Demand Data MARKETQ 186 2, ,262 1, PRICE PLANTS WAGE COAL ELECTRICITY POPULATION 689,584 10,224,352 33,145,121 7,416,485 GAS Production Data QUANTITY CAPACITY Plant Photograph Data Sources Industry Statistics Summary Statistics Investment INVESTMENT ,

24 of Cement Industry Fully dynamic model of oligopoly based on Ericson and Pakes (1995) Markets defined by state vector of firm capacities (s) Focus on interactions in independent regional markets, delineated by geography Discrete time, infinite horizon Long-lived incumbents and short-lived potential entrants Firms maximize profits through production, investment, entry, and exit Equilibrium is Markov-perfect Nash equilibrium (MPNE) Product Market Per-Period Payoffs Entry and Exit Value Functions Equilibrium

25 Product Market Homogeneous goods market with CED ln Q = α 0 + α 1 ln P Firms choose quantities, have a privately-known shock to marginal costs (ɛ i ), and face increasing costs once they exceed a capacity bound: c i (q i ) = q i (MC + ɛ i ) 1 ( qi s i ν ) [ ( ) ] 2 qi CAPCOST ν s i Product Market Per-Period Payoffs Entry and Exit Value Functions Equilibrium

26 Per-Period Payoff Functions Investment Costs Firms have non-convex investment costs: I (x) = 1(x > 0) ( ADJPOS + INVPOS x + INVPOS2 x 2) + 1(x < 0) ( ADJNEG + INVNEG x + INVNEG2 x 2) Firms make optimal investment decisions conditional on state, s Incur non-convex adjustment costs in investment, x Deterministic investment: new capacity = old capacity + adjustment Product Market Per-Period Payoffs Entry and Exit Value Functions Equilibrium

27 Entry and Exit Incumbent firms who exit obtain product market profits and a scrap value, SCRAP Firms make optimal exit decisions conditional on strategies of competitors: Φ(s) New entrants pay a privately-known setup fee, SUNK i, and investment costs, and become active next period In equilibrium firms follow cutoff rules for entry: Θ(s, SUNK i ) Uncertainty in configuration of new period derives from private information in entry decision, mixed strategies over exit and investment Product Market Per-Period Payoffs Entry and Exit Value Functions Equilibrium

28 Value Functions Potential entrant: Vi e [ (s, SUNK i ) = max SUNKi I (xi e ) + βe(v (s ) s) ] x e i Incumbent firm s value depends on its exit choice: V i (s, ɛ i ) = max [π i(s) + Φ(s) SCRAP {x i,φ(s)} +(1 Φ(s)) ( I (x i ) + βe(v (s ) s) )] Product Market Per-Period Payoffs Entry and Exit Value Functions Equilibrium

29 Equilibrium Definition Markov perfect Nash equilibrium (MPNE) obtains when: V (σ i s, σ i ) V (σ i s, σ i ) for optimal strategy σ i against any alternative σ i. In equilibrium, every firm follows policies which maximize its expected discounted stream of revenues given the strategies of its competitors Equilibrium restriction is basis for empirical estimator Product Market Per-Period Payoffs Entry and Exit Value Functions Equilibrium

30 Linking the to the Data I follow Bajari, Benkard, and Levin (2004) and estimate the parameters of the model in two stages First stage: Flexibly describe what firms will do at every state vector Second stage: Impose restrictions of MPNE to find parameters that best explain why firms follow these equilibrium policies Intuition: Observed policies are result of firms solving a dynamic programming problem, so I find parameters which best rationalize these policies as optimal given the underlying model Estimator Intuition Two Types Production Parameters Entry and Exit Policies (S, s) Investment Policy Simulated Investment Dynamic Estimator Constructing the Estimator Sunk Entry Costs

31 Two Types of Parameters Static Parameters Demand curve Production costs Dynamic Parameters Fixed and variable costs of adjustment Scrap value of exiting Distribution of sunk entry costs entry and exit policies as probits, as implied by theoretical model Characterize investment policy as (S, s) rule, a flexible method of handling lumpy capacity adjustment Estimator Intuition Two Types Production Parameters Entry and Exit Policies (S, s) Investment Policy Simulated Investment Dynamic Estimator Constructing the Estimator Sunk Entry Costs

32 Production Parameters Estimating Equation (Production Parameters) π i = P(Q) + q i P (Q) MC + ɛ i q i ( ) [ qi 2 CAPCOST 1 > ν s i Firm sets q i such that MR = MC s i ( )] qi ν = 0. s i Solve system of equations at market level in each period Recover production parameters: baseline cost (MC), capacity cost (CAPCOST), capacity binding level (ν) Estimator Intuition Two Types Production Parameters Entry and Exit Policies (S, s) Investment Policy Simulated Investment Dynamic Estimator Constructing the Estimator Sunk Entry Costs

33 Entry and Exit Estimating Equation (Entry and Exit Policy Functions) Pr(Entry s) = Φ(γ 0 + γ 1 SUMCAP + γ 2 AFTER1990) Pr(Exit s) = Φ(γ 0 + γ 1 SUMCAP + γ 2 AFTER1990 +γ 3 CAPACITY + γ 4 ɛ) Estimate entry and exit policies for all states using a probit Probits capture mixed strategy, cutoff rule equilibrium behavior as implied by theoretical model Estimator Intuition Two Types Production Parameters Entry and Exit Policies (S, s) Investment Policy Simulated Investment Dynamic Estimator Constructing the Estimator Sunk Entry Costs

34 Lumpy Investment Policy Estimating Equation (Investment Policy) TARGET it = γ 4s 1 (CAPACITY it ) + γ 5s 2 (SUMCAP it ) + γ 6 ɛ it + u d it BAND it = TARGET it ± exp ( γ 7s 3 (CAPACITY it ) +γ 8s 4 (SUMCAP it ) + γ 9 ɛ it + u b it Firms adjust to TARGET when CAPACITY falls outside either the upper or lower BAND s i (x) are flexible functions of x ) Estimator Intuition Two Types Production Parameters Entry and Exit Policies (S, s) Investment Policy Simulated Investment Dynamic Estimator Constructing the Estimator Sunk Entry Costs

35 Simulated Path of (S, s) Rule Estimator Intuition Two Types Production Parameters Entry and Exit Policies (S, s) Investment Policy Simulated Investment Dynamic Estimator Constructing the Estimator Sunk Entry Costs

36 Dynamic Parameter Estimation Policy functions provide complete description of how a firm will act for any state vector Demand curve and production parameters tell how much profit firm derives from product market at each state Sufficient information to simulate industry evolution and assign revenues to each path Intuition of dynamic estimator is that expected payoffs from following observed policies must dominate payoffs generated by alternative policies Estimator Intuition Two Types Production Parameters Entry and Exit Policies (S, s) Investment Policy Simulated Investment Dynamic Estimator Constructing the Estimator Sunk Entry Costs

37 Linear Basis of Estimator Can expand V i ( ) for incumbent as: V i ( ) = u i (s t ) + E { βu i (s t+1 ) + β 2 u i (s t+2 ) +... } Unknowns in per-period payoffs enter linearly: u i (s) = π i (s) + Φ SCRAP + 1(x > 0)(ADJPOS + INVPOS x + INVPOS2 x 2 ) + 1(x < 0)(ADJNEG + INVNEG x + INVNEG2 x 2 ) = α ζ i (s ) Therefore, rewrite V i ( ) as: [ E β t { α ζ i (s ) }] [ ] = E β t ζ i (s ) α = W i (s) α t=0 t=0 Estimator Intuition Two Types Production Parameters Entry and Exit Policies (S, s) Investment Policy Simulated Investment Dynamic Estimator Constructing the Estimator Sunk Entry Costs

38 Equilibrium Structure of Estimator MPNE requires: W i (s; σ i, σ i ) α W i (s; σ i, σ i ) α Alternative policies constructed by adding noise to optimal policies Draw n k different alternative policies, compute their values, and find difference against optimal policy payoff: g j (α) = [ W j (s; σ i, σ i ) W j (s; σ i, σ i ) ] α Estimator finds parameters that rationalize observed policies as optimal by minimizing profitable deviations: 1 min α n k n k 1(g j (α) < 0)g j (α) 2 j=1 Estimator Intuition Two Types Production Parameters Entry and Exit Policies (S, s) Investment Policy Simulated Investment Dynamic Estimator Constructing the Estimator Sunk Entry Costs

39 Distribution of Entry Costs Entry probit describes when firm enters at a given state Previous estimation sufficient to compute dollar value if firm did enter The only way a firm does not enter is if it gets a draw larger than that value Estimator matches probability of entry to probability draw of sunk entry costs value of entering: Pr(Entry s) = Pr(SUNK i V e (s)) = G(V e (s); µ, σ 2 ) G( ) is the CDF of sunk entry costs Estimator Intuition Two Types Production Parameters Entry and Exit Policies (S, s) Investment Policy Simulated Investment Dynamic Estimator Constructing the Estimator Sunk Entry Costs

40 Demand Estimates Table: log q = log p + Intercept + Region Variable Coefficient (Std. Err.) logp (0.303) Intercept (1.251) Instruments are cost-side shifters Market fixed effects for heterogeneity in level of demand First-stage F statistic = 30, fail to reject orthogonality of instruments Robust against specifications with market-specific time trends, imports in border states Demand Parameters Production Parameters Implied Markups and Profits Entry and Exit Policies Investment and Exit Parameters Distribution of Entry Costs

41 Production Parameters 95% Confidence Parameter Mean Median Interval CAPCOST ( 10 7 ) [1.105, 3.782] BINDING LEVEL (ν) [1.806, 2.016] MARGINAL COST [30.761, ] CAPCOST DUMMY ( 10 7 ) [-3.056, 0.642] BINDING DUMMY [-0.131, 0.180] MC DUMMY [-2.23, 4.36] No statistically significant differences in production costs Binding capacity utilization level roughly 87% Very expensive to produce beyond this Demand Parameters Production Parameters Implied Markups and Profits Entry and Exit Policies Investment and Exit Parameters Distribution of Entry Costs

42 Implied Markups and Profits Standard Variable Min Mean Max Deviation Price Revenues 9,819 41, ,071 19,389 Shock (ɛ) -3, , Profit -5,041 13,327 60,470 9,372 Margin Implied prices, shocks, revenues, and profits for every firm at estimated demand and production parameters Margin is implied profit margin calculated as profits divided by revenues Demand Parameters Production Parameters Implied Markups and Profits Entry and Exit Policies Investment and Exit Parameters Distribution of Entry Costs

43 Entry and Exit Probit Results Standard Parameter Coefficient Error Exit Policy Constant CAP ɛ SUMCAP Late Dummy Entry Policy Constant SUMCAP Late Dummy Sample size for exit policy function = 2233; sample size for entry policy function = 414. Both entry and exit less likely after Amendments As operation costs have not changed, must reflect investment or entry cost shifts Demand Parameters Production Parameters Implied Markups and Profits Entry and Exit Policies Investment and Exit Parameters Distribution of Entry Costs

44 Investment and Exit Parameters Standard 95% Confidence Parameter Median Deviation Interval Early Period ADJPOS 30, [30,491, 30,963] INVPOS [125, 131] INVPOS [0.018, 0.021] ADJNEG 22, [21,754, 23,562] INVNEG -1, [-1,279, -925] INVNEG [28.428, ] SCRAP 84, [82,640, 84,109] Late Period ADJPOS 27, [27,562, 27,663] INVPOS [69.36, 71.67] INVPOS E-5 [0.015, 0.015] ADJNEG 22, [20,062, 22,996] INVNEG -1, [-1,645, -1,291] INVNEG [49.38, 57.25] SCRAP 54, [54,423, 55,749] Demand Parameters Production Parameters Implied Markups and Profits Entry and Exit Policies Investment and Exit Parameters Distribution of Entry Costs Fixed adjustment costs are economically significant Investment costs have decreased after Amendments Implies there was a shift in entry costs

45 Distribution of Sunk Costs of Entry Mean Standard 95% Confidence Parameter (000 $) Deviation Interval Before Amendments 120,976 11,603 [93,321, 132,865] After Amendments 162,470 7,728 [145,133, 173,115] 34% increase in mean of sunk entry cost distribution Mean sunk cost draw if entered in early period: $17M Mean sunk cost draw if entered in later period: $35M Firms recover their initial outlays within 20 years Demand Parameters Production Parameters Implied Markups and Profits Entry and Exit Policies Investment and Exit Parameters Distribution of Entry Costs

46 Policy Experiment I compare the distribution of consumer surplus and producer profits before and after Amendments Exploit timing in regulation to identify Amendments shifting sunk cost of entry Benchmark against social planner s solution I find consumer surplus is lowered due to reduced entry rates Incumbent firms are better off under the new regulations due to increased barriers to entry Investment per firm higher under Amendments, total market capacity lower Description Simulation Results

47 Simulation Results Post 1990 Amendments Social Planner New Market Producer profit 56, , , Consumer welfare 308, , ,279, Average Capacity Active Firms 2,035 1,950 3,773 Average Market Capacity 5,855 5,969 15,094 Average Market Quantity 5,158 5,299 13,041 Average Market Price Average Number Active Firms Incumbent Market Producer profit 316, , , Consumer welfare 477, , ,656, Profits of firm 1 717, , , Average Capacity Active Firms 3,337 3,324 4, Average Market Capacity 6,867 8,169 13,409 Average Market Quantity 6,069 7,225 11,794 Average Market Price Average Number Active Firms Description Simulation Results Static estimate of costs during this period: $0 EPA estimates of future sunk investments $5-10M

48 Simulation Results Post 1990 Amendments Social Planner New Market Producer profit 56, , , Consumer welfare 308, , ,279, Average Capacity Active Firms 2,035 1,950 3,773 Average Market Capacity 5,855 5,969 15,094 Average Market Quantity 5,158 5,299 13,041 Average Market Price Average Number Active Firms Incumbent Market Producer profit 316, , , Consumer welfare 477, , ,656, Profits of firm 1 717, , , Average Capacity Active Firms 3,337 3,324 4, Average Market Capacity 6,867 8,169 13,409 Average Market Quantity 6,069 7,225 11,794 Average Market Price Average Number Active Firms Description Simulation Results Static estimate of costs during this period: $0 EPA estimates of future sunk investments $5-10M

49 Simulation Results Post 1990 Amendments Social Planner New Market Producer profit 56, , , Consumer welfare 308, , ,279, Average Capacity Active Firms 2,035 1,950 3,773 Average Market Capacity 5,855 5,969 15,094 Average Market Quantity 5,158 5,299 13,041 Average Market Price Average Number Active Firms Incumbent Market Producer profit 316, , , Consumer welfare 477, , ,656, Profits of firm 1 717, , , Average Capacity Active Firms 3,337 3,324 4, Average Market Capacity 6,867 8,169 13,409 Average Market Quantity 6,069 7,225 11,794 Average Market Price Average Number Active Firms Description Simulation Results Static estimate of costs during this period: $0 EPA estimates of future sunk investments $5-10M

50 The 1990 Amendments to the Clear Air Act changed the cost structure of the industry, primarily the sunk cost of entry Consumer surplus is decreased due to lower entry rates, resulting in lower quantities and higher prices Producer surplus is higher for incumbent firms Industry has socially-inefficient low capacity Static cost analysis misses almost all of the relevant economic costs in this industry, underpredicting costs of regulation and obtaining welfare estimates of wrong sign on producers

51 Cost effectiveness: can EPA achieve same reductions in output using different policy? Efficiency: does the cost of the Amendments exceed the benefits of cleaner air Long-term goal: link model to pollution data to examine strategic and distributional effects of a trading permits market

52 Cost effectiveness: can EPA achieve same reductions in output using different policy? Efficiency: does the cost of the Amendments exceed the benefits of cleaner air Long-term goal: link model to pollution data to examine strategic and distributional effects of a trading permits market

53 Cost effectiveness: can EPA achieve same reductions in output using different policy? Efficiency: does the cost of the Amendments exceed the benefits of cleaner air Long-term goal: link model to pollution data to examine strategic and distributional effects of a trading permits market

54 Entry and Exit Cross-Tabulations Entry After 1990 False True Row Count Column Count Exit After 1990 False True Row Count 0 1, Column Count 1, ,233 Entry and Exit Upper number within cell is count Lower number within cell is proportion to overall