dg/run_output.txt at main · marek-chadim/dg · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
Dynamic Entry/Exit Game: Equilibrium and Estimation
======================================================================

Q4-5: Equilibrium Computation
  Init π       : V̄(3,0)=8.6810, μ̄(3,0)=8.6320
  Init 0       : V̄(3,0)=8.6810, μ̄(3,0)=8.6320
  Init 1       : V̄(3,0)=8.6810, μ̄(3,0)=8.6320
  Init U[0,10] : V̄(3,0)=8.6810, μ̄(3,0)=8.6320
  Init U[-5,15]: V̄(3,0)=8.6810, μ̄(3,0)=8.6320
  Result: UNIQUE

Q6: Equilibrium at (N,x)=(3,0)
  μ̄(3,0) = 8.631981
  γ̄(3,0) = 7.024259
  V̄(3,0) = 8.681050
  V(3,0,-2) = 8.631981 (firm stays)

Q7: Market Simulation (10,000 periods)
  Average firms: 3.4740

Q8: Entry Tax Counterfactual
  With 5-unit tax: 3.3625 firms
  Change: -0.1115 firms (-3.21%)

Q9: BBL-like Estimation (forward simulation of Λ, Λ^E)
  Step 1: Nonparametric CCP estimation from simulated data
    Simulating 10000 periods from equilibrium...
    Estimated 15 incumbent CCPs and 15 entry CCPs from data
    Sample CCP comparison (estimated vs true):
      d(N=3,x=0): est=0.9497, true=0.9478
      d(N=3,x=-5): est=0.5147, true=0.5157
      d(N=3,x=5): est=0.9990, true=1.0000
      e(N=3,x=0): est=0.8206, true=0.8173
      e(N=3,x=-5): est=0.9990, true=0.9980
      e(N=3,x=5): est=0.9290, true=0.9261
  Step 2: Forward simulation (Λ, Λ^E) and minimum distance search
  Using 50 simulations, horizon=1000
    Start rng-1   : obj=0.051855, μ= 6.059, σ_μ=2.799, γ= 5.040, σ_γ=3.440, iters=26
    Start rng-2   : obj=0.051855, μ= 6.059, σ_μ=2.799, γ= 5.040, σ_γ=3.440, iters=23
    Start rng-3   : obj=0.051855, μ= 6.059, σ_μ=2.799, γ= 5.040, σ_γ=3.440, iters=30
    Settings: n_sim=50, horizon=1000
    Objective breakdown: total=0.051855, inc_sum=0.041749, ent_sum=0.010106
  Using 1000 simulations, horizon=1000
    Start rng-1   : obj=0.002690, μ= 4.928, σ_μ=2.359, γ= 5.055, σ_γ=2.185, iters=28
    Start rng-2   : obj=0.002690, μ= 4.928, σ_μ=2.359, γ= 5.055, σ_γ=2.185, iters=26
    Start rng-3   : obj=0.002690, μ= 4.928, σ_μ=2.359, γ= 5.054, σ_γ=2.186, iters=26
    Settings: n_sim=1000, horizon=1000
    Objective breakdown: total=0.002690, inc_sum=0.002342, ent_sum=0.000348
  Bootstrap using mode 'hp' with res n_sim=1000
Running 100 BBL bootstraps in parallel (n_jobs=-1)...
  Completed 100 bootstrap replications
  Bootstrap SE:         μ= 0.390, σ_μ=0.193, γ= 0.481, σ_γ=0.320
  Original θ:           μ= 4.928, σ_μ=2.359, γ= 5.055, σ_γ=2.185
  Bootstrap bias:       μ= 0.189, σ_μ=-0.048, γ= 0.003, σ_γ=0.129
  Bias-corrected θ:     μ= 4.740, σ_μ=2.407, γ= 5.052, σ_γ=2.056
  Percentile 95% CI:    μ=[ 4.360,  5.850], σ_μ=[1.973, 2.706], γ=[ 4.069,  5.947], σ_γ=[1.709, 3.018]
  Bias-corrected 95% CI: μ=[ 4.007,  5.496], σ_μ=[2.012, 2.745], γ=[ 4.163,  6.041], σ_γ=[1.352, 2.662]

======================================================================