Skip to contents

BLP Estimation Results

Source: R/results.R
BLPResults.Rd

Stores estimation results and provides post-estimation methods.

Public fields

problem

The originating BLPProblem

sigma

Estimated Cholesky root of random coefficient covariance

pi

Estimated demographics interaction matrix

rho

Estimated nesting parameters

beta

Estimated demand linear parameters

gamma

Estimated supply linear parameters

delta

Estimated mean utilities

xi

Demand-side structural error

omega

Supply-side structural error

objective

GMM objective value

gradient

GMM gradient at solution

hessian

GMM Hessian at solution

se

Standard errors for nonlinear parameters

parameter_covariances

Covariance matrix for nonlinear parameters

W

Final weighting matrix

optimization_converged

Whether optimization converged

optimization_iterations

Number of optimization iterations

optimization_evaluations

Number of function evaluations

fp_converged

Whether all fixed points converged

fp_iterations

Total fixed-point iterations

method

GMM method ("1s" or "2s")

se_type

Standard error type

Methods

Public methods

Method new()

Create results object

Usage

BLPResults$new(
  problem,
  params,
  sigma,
  pi,
  rho,
  beta,
  gamma,
  delta,
  xi,
  omega,
  objective,
  gradient,
  hessian,
  se,
  parameter_covariances,
  W,
  step_results,
  optimization_converged,
  optimization_iterations,
  optimization_evaluations,
  fp_converged,
  fp_iterations,
  method,
  se_type,
  beta_se = NULL,
  gamma_se = NULL
)

Method compute_elasticities()

Compute own-price elasticities for a specific market

Usage

BLPResults$compute_elasticities(market_id = NULL)

Arguments

market_id

Market identifier

Returns

J x J elasticity matrix

Method compute_diversion_ratios()

Compute diversion ratios for a specific market

Usage

BLPResults$compute_diversion_ratios(market_id = NULL)

Arguments

market_id

Market identifier

Returns

J x J diversion ratio matrix

Method compute_costs()

Extract marginal costs

Usage

BLPResults$compute_costs()

Returns

Named list with costs per market, or full vector

Method compute_markups()

Compute markups (p - c) / p

Usage

BLPResults$compute_markups()

Returns

Markup vector

Method compute_consumer_surplus()

Compute consumer surplus per market

Usage

BLPResults$compute_consumer_surplus()

Returns

Named numeric vector of CS per market

Method compute_hhi()

Compute HHI per market

Usage

BLPResults$compute_hhi()

Returns

Named numeric vector of HHI per market

Method compute_merger()

Simulate merger and compute new equilibrium prices

Usage

BLPResults$compute_merger(new_firm_ids, iteration = NULL, costs = NULL)

Arguments

new_firm_ids

Updated firm ownership (length N)

iteration

BLPIteration for price equilibrium

costs

Optional pre-computed costs; computed if NULL

Returns

List with new_prices, new_shares, new_costs, delta_cs

Method run_hansen_test()

Run Hansen's J-test for overidentifying restrictions

Usage

BLPResults$run_hansen_test()

Returns

List with statistic, df, p_value

Method run_wald_test()

Run Wald test for parameter restrictions

Usage

BLPResults$run_wald_test(R, r = NULL)

Arguments

R

Restriction matrix (q x p)

r

Restriction values (q x 1), default 0

Returns

List with statistic, df, p_value

Method compute_aggregate_elasticities()

Compute aggregate elasticities per market

Usage

BLPResults$compute_aggregate_elasticities(factor = 0.01, market_id = NULL)

Arguments

factor

Percentage price increase (default 0.01 = 1%)

market_id

Optional market identifier (NULL = all markets)

Returns

Named numeric vector of aggregate elasticities per market

Method compute_passthrough()

Compute cost-to-price passthrough matrix

Usage

BLPResults$compute_passthrough(market_id)

Arguments

market_id

Market identifier

Returns

J x J passthrough matrix dp/dc

Method compute_long_run_diversion_ratios()

Compute long-run diversion ratios

Usage

BLPResults$compute_long_run_diversion_ratios(market_id = NULL)

Arguments

market_id

Market identifier (NULL = all markets)

Returns

J x J diversion matrix (or list of matrices)

Method compute_shares()

Compute predicted market shares at estimated parameters

Usage

BLPResults$compute_shares()

Returns

Numeric vector of predicted shares (length N)

Method compute_profits()

Compute per-product profits (p - c) * s

Usage

BLPResults$compute_profits(costs = NULL)

Arguments

costs

Optional pre-computed costs (computed if NULL)

Returns

Numeric vector of profits (length N)

Method compute_optimal_instruments()

Compute optimal instruments (BLP 1999, Chamberlain 1987)

Usage

BLPResults$compute_optimal_instruments(method = "approximate")

Arguments

method

"approximate" (default) or "exact"

Returns

List with optimal_instruments (matrix) and to_problem() function

Method bootstrap()

Parametric bootstrap for inference on post-estimation quantities

Usage

BLPResults$bootstrap(draws = 100L, seed = NULL, ...)

Arguments

draws

Number of bootstrap draws

seed

Random seed

...

Additional arguments passed to solve()

Returns

List of BLPResults objects (one per draw)

Method importance_sampling()

Construct importance sampling weights

Usage

BLPResults$importance_sampling(n_draws = 500L, seed = NULL)

Arguments

n_draws

Number of importance sampling draws

seed

Random seed

Returns

List with new agent_data and to_problem() function

Method sigma_squared()

Extract sigma squared (Sigma %*% Sigma')

Usage

BLPResults$sigma_squared()

Returns

K2 x K2 covariance matrix

Method summary_table()

Get a summary table of all estimated parameters

Usage

BLPResults$summary_table()

Returns

Data frame with estimates and standard errors

Method print()

Print estimation results

Usage

BLPResults$print(...)

Method clone()

The objects of this class are cloneable with this method.

Usage

BLPResults$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.