psfm: psfm

In sfa: Stochastic Frontier Analysis

psfm

Description

Function to implement various panel data stochastic frontier estimators

Usage

psfm(formula, model_name = c("TRE_Z", "GTRE_Z", "TRE",
                    "GTRE", "TFE", "FD", "GTRE_SEQ1", "GTRE_SEQ2"), data,
                    maxit.bobyqa = 100, maxit.psoptim = 10, maxit.optim =
                    10, REPORT = 1, trace = 3, pgtol = 0, individual,
                    halton_num = NULL, start_val = FALSE, gamma = FALSE,
                    PSopt = FALSE, optHessian, inefdec= TRUE, Method = "L-BFGS-B",
                    verbose = FALSE,rand.gtre = NULL, rand.psoptim = NULL)

Arguments

formula	a symbolic description for the model to be estimated
model_name	model name for the estimation
data	a pdata.frame
maxit.bobyqa	Maximum number of iterations for the bobyqa optimization routine
maxit.psoptim	Maximum number of iterations for the psoptim optimization routine
maxit.optim	Maximum number of iterations for the optim optimization routine
REPORT	reporting parameter
trace	trace
pgtol	pgtol
individual	individual unit in the regression model
halton_num	number of Halton draws to use in SML models
start_val	starting value (optional)
gamma	gamma
PSopt	use psoptim optimization routine (T or F)
optHessian	Logical. Should a numerically differentiated Hessian matrix be returned while using the optim routine? (for optim routine)
inefdec	Production or cost function
Method	The method to be used for optim. See 'Details' within optim.
verbose	Logical. Print optimization progress messages? Default is FALSE.
rand.psoptim	Integer. Seed for replication of psoptim. Default to NULL.
rand.gtre	Integer. Seed for replication of the gtre model. Default to NULL.

Details

The generalized true random effects model (GTRE, 4-component model) and true random effects models (TRE) are both estimated by simulated maximum likelihood based on the paper by the Fillipini and Greene (2016, JPA). The TRE_Z and GTRE_Z allow for modeling the u-component of the GTRE and TRE with determinants of inefficiency. The first-difference estimator (FD) of Wang and Ho (2010, JoE) as well as the True Fixed Effect model estimated by within-maximum likelihood of Chen, Schmidt and Wang (2014, JoE) are also available.

Value

An object of class "sfareg" containing components that vary by model. All models return:

out	A matrix with parameter estimates, standard errors, and t-values.
opt	A list containing the optimization results from the final optimization procedure (not returned for GTRE_SEQ1 and GTRE_SEQ2).
total_time	The total computation time for model estimation.
start_v	The starting values used in the optimization (not returned for GTRE_SEQ1 and GTRE_SEQ2).
model_name	The name of the panel stochastic frontier model estimated.
formula	The formula used in the model specification.
coefficients	A vector of estimated parameters.
std.errors	A vector of standard errors for the estimated parameters (NA if optHessian = FALSE).
t.values	A vector of t-values for the estimated parameters (NA if optHessian = FALSE).
call	The matched call.
data	The data used in estimation.

Additional model-specific components:

For GTRE and GTRE_Z models:

H	Predicted time-invariant technical efficiency for each individual.

For GTRE, GTRE_Z, TRE and TRE_Z models:

U	Predicted time-varying technical efficiency for each observation.

For TFE model:

r_hat_m	Estimated individual-specific random effects.
exp_u_hat	Predicted technical efficiency.

For FD model:

u_hat	Predicted technical efficiency in levels.
h_hat	Estimated z heterogeneity function values.
exp_u_hat	Predicted technical efficiency.

For GTRE_SEQ1 and GTRE_SEQ2 models:

other_parms

A matrix of additional parameters (lambda, sigma, beta_0 for SEQ1; sigma_u, sigma_v, sigma_h, sigma_r, lambda, sigma for SEQ2).

Note

Standard errors require optHessian set to TRUE

Note

The GTRE_SEQ1 and GTRE_SEQ2 models use sequential estimation methods and do not return optimization objects or starting values. All panel models require the individual argument to identify panel units.

Author(s)

David Bernstein

References

Fillipini and Greene (2016, JPA); Wang and Ho (2010, JoE); Chen, Schmidt and Wang (2014, JoE)

See Also

see also

Examples

library(sfa)     

data_trial <- data_gen_p(t=10,N=100, rand = 100, 
                         sig_u = 1,  sig_v = 0.3, 
                         sig_r = .2, sig_h = .4, 
                         cons = 0.5, beta1 = 0.5,
                         beta2 = 0.5)

max_tre_z   <-  psfm(formula    = y_tre_z ~ x1 +x2| z_gtre, 
                     model_name = "TRE",                    ## "TRE_Z" also works
                     data       = data_trial,
                     individual = "name",
                     PSopt      = TRUE)

sfa documentation built on Jan. 22, 2026, 1:08 a.m.