Skip to contents

BLP Problem

Source: R/problem.R
BLPProblem.Rd

The main user-facing class for BLP demand estimation. Inherits from BLPEconomy and adds the solve() interface.

Super class

rblp::BLPEconomy -> BLPProblem

Methods

Public methods

Inherited methods

Method new()

Create a BLP Problem

Usage

BLPProblem$new(
  product_formulations,
  product_data,
  agent_formulation = NULL,
  agent_data = NULL,
  integration = NULL,
  rc_types = NULL,
  epsilon_scale = 1,
  costs_type = "linear",
  add_exogenous = TRUE
)

Arguments

product_formulations

List of BLPFormulation objects (1-3). First: linear demand (X1). Second: nonlinear demand (X2). Third: supply (X3).

product_data

Data frame with columns: market_ids, shares, prices, and optionally firm_ids, nesting_ids, demand_instruments*, supply_instruments*

agent_formulation

Optional BLPFormulation for demographics

agent_data

Optional data frame with columns: market_ids, weights, nodes*

integration

Optional BLPIntegration object

rc_types

Character vector of random coefficient types ("linear", "log", "logit")

epsilon_scale

Scaling for epsilon (default 1)

costs_type

"linear" or "log"

add_exogenous

Whether to add exogenous X1/X3 columns to instruments

Method solve()

Solve the BLP estimation problem

Usage

BLPProblem$solve(
  sigma = NULL,
  pi = NULL,
  rho = NULL,
  beta = NULL,
  gamma = NULL,
  sigma_bounds = NULL,
  pi_bounds = NULL,
  rho_bounds = NULL,
  method = "2s",
  optimization = NULL,
  iteration = NULL,
  fp_type = "safe_linear",
  W_type = "robust",
  se_type = "robust",
  initial_W = NULL,
  initial_update = FALSE,
  scale_objective = TRUE,
  center_moments = TRUE,
  delta_behavior = "first",
  micro_moments = NULL,
  error_behavior = "revert",
  error_punishment = 1,
  processes = 1L
)

Arguments

sigma

Initial K2 x K2 Cholesky root (0 = fixed, non-zero = free)

pi

Initial K2 x D demographics interaction (0 = fixed)

rho

Initial nesting parameters (0 = fixed)

beta

Demand linear coefficients (NA = concentrated out)

gamma

Supply linear coefficients (NA = concentrated out)

sigma_bounds

List of lower, upper bound matrices for sigma

pi_bounds

List of lower, upper bound matrices for pi

rho_bounds

List of lower, upper bound vectors for rho

method

'1s' or '2s' for one-step or two-step GMM

optimization

BLPOptimization object

iteration

BLPIteration object (for contraction mapping)

fp_type

Fixed point type: "safe_linear", "linear", "nonlinear"

W_type

Weighting matrix type: "robust", "clustered", "unadjusted"

se_type

Standard error type: "robust", "clustered", "unadjusted"

initial_W

Optional initial weighting matrix

scale_objective

Whether to scale objective by N

center_moments

Whether to center moment conditions

delta_behavior

How to initialize delta: "first", "logit", "last"

micro_moments

Optional list of MicroMoment objects

error_behavior

"revert", "punish", or "raise"

error_punishment

Punishment scale for objective on error

processes

Number of parallel processes (1 = no parallel)

Returns

A BLPResults object

Method print()

Print the problem

Usage

BLPProblem$print(...)

Method clone()

The objects of this class are cloneable with this method.

Usage

BLPProblem$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.