[Splm-commits] r118 - pkg/man
noreply at r-forge.r-project.org
noreply at r-forge.r-project.org
Thu May 26 08:39:23 CEST 2011
Author: the_sculler
Date: 2011-05-26 08:39:23 +0200 (Thu, 26 May 2011)
New Revision: 118
Removed:
pkg/man/spfeml.Rd
pkg/man/spreml.Rd
Log:
Removed man files spreml.Rd spfeml.Rd as now functions are outside NAMESPACE
Deleted: pkg/man/spfeml.Rd
===================================================================
--- pkg/man/spfeml.Rd 2011-05-26 06:37:37 UTC (rev 117)
+++ pkg/man/spfeml.Rd 2011-05-26 06:39:23 UTC (rev 118)
@@ -1,140 +0,0 @@
-\name{spfeml}
-\alias{spfeml}
-\title{Spatial Panel Fixed Effects Models Estimation}
-\description{Maximum likelihood (ML) estimation of spatial panel
-fixed effects models}
-
-\usage{
-spfeml(formula, data = list(), index = NULL,listw,listw2 = NULL,
-model= c("lag","error", "sarar"), effects=c('pooled','spfe','tpfe','sptpfe'),
-method ="eigen", na.action = na.fail, quiet = TRUE, zero.policy = NULL, interval = NULL,
-tol.solve = 1e-10, control = list(), legacy = FALSE, llprof = NULL)
-}
-
-
-\arguments{
- \item{formula}{a description of the model to be fit }
- \item{data}{an object of class \code{\link{data.frame}} or \code{pdata.frame}. An optional data frame containing the variables
- in the model. When the obect is a \code{\link{data.frame}}, the first two columns may contain the indexes. See \code{index}}
- \item{index}{if not NULL (default), a character vector to identify the indexes among the columns of the \code{\link{data.frame}}}
- \item{listw}{an object of class \code{listw} created for example by \code{nb2listw}}
- \item{listw2}{only if \code{model} is \code{"sarar"}. an object of class \code{listw} created for example by \code{nb2listw}. if not given, set to the same spatial weights as the listw argument}
- \item{model}{one of \code{c("lag", "error", "sarar")} }
- \item{effects}{one of \code{"pooled"} (no spatial effects, i.e. OLS on the pooled model),
- \code{"spfe"} (only spatial fixed effects), \code{"tpfe"} (only time period fixed effects),
- \code{"sptpfe"} (both time period and spatial fixed effects)}
-
- \item{method}{\code{"eigen"} (default) - the Jacobian is computed as
- \eqn{\prod_{i=1}^N (1 - \rho \omega_i) } (with \eqn{\omega_i} the eigenvalues of the spatial
- weigths matrix) using \code{eigenw} - (see \code{lagsarlm} for details).
- Other available methods are \code{"spam"} or \code{"Matrix"} for strictly symmetric weights lists of styles "B" and "C",
- or made symmetric by similarity (Ord, 1975, Appendix C) if possible for styles "W" and "S",
- using code from the \pkg{spam} or \pkg{Matrix} packages to calculate the determinant; \code{"LU"}
- provides an alternative sparse matrix decomposition approach.
- Finally, \code{"Chebyshev"} and Monte Carlo \code{"MC"} provides approximations to the computation of the log-determinant.}
-\item{na.action}{see \code{lagsarlm} for details }
- \item{quiet}{default=\code{TRUE}; if \code{FALSE}, reports function values during optimization}
- \item{zero.policy}{see \code{lagsarlm} for details}
- \item{interval}{search interval for spatial parameters}
- \item{tol.solve}{tolerance for detecting linear dependence in the columns of matrices to be inverted (default= 1.0e-10)- (see \code{lagsarlm} for details)}
- \item{control}{a list of control parameters for the optimization - see \code{lagsarlm} for details}
- \item{legacy}{FALSE When TRUE and model is \code{"lag"} calculate goodness of fit measures}
- \item{llprof}{default NULL, can either be an integer, to divide the feasible ranges into a grid of points, or a two-column matrix of spatial coefficient values, at which to evaluate the likelihood function}
-}
-
-
-\details{
-The function deals with fixed effects spatial sarar, lag and error models, of the general form:
-
-\deqn{ y = \lambda (I_T \otimes W_N) y + (\iota_T \otimes \alpha) + X\beta + \epsilon }
-\deqn{ u = \rho (\iota_T \otimes W_N)u + \epsilon}
-
-where \eqn{\rho} is called the spatial autoregressive coefficient, and
-\eqn{\lambda} is the spatial autocorrelation coefficient.
-
-Both \eqn{\rho} and \eqn{\lambda} are found by \code{\link{optimize}},
-and the model parameters by a feasible GLS procedure.
-
-The asymptotic standard error of
-\eqn{\rho} in the error model is only computed when
-\code{method=eigen}, because the full matrix
-operations involved would be costly for
-dimensions associated with the choice of \code{"spam"} or \code{"Matrix"}.
-The same applies to the coefficient covariance matrix.
-
-The asymptotic standard error of
-\eqn{\rho} and \eqn{lambda} in the full model
-are calculated by a numerical Hessian.
-
-The model variables are first transformed depending on the
-typology of fixed effects. As an example, if time period
-fixed effects are considered, the demeaned form is obtained
-by subtracting the average for each cross-sectional unit computed over the time
-dimension.
-
-The transformed variables are then used in a concentrated likelihood
-approach to estimate the spatial parameters.
-
-Once the spatial parameters are obtained, the model parameters are
-then estimated by a feasible GLS procedure.
-
-Fixed effects can then be extracted using the function \code{\link{effects}}.
-}
-
-
-\value{
- An object of class \code{"splm"}.
- \item{coefficients}{coefficients estimate of the model parameters }
- \item{vcov}{the asymptotic variance covariance matrix of the estimated coefficients (when calculated)}
- \item{type}{'fixed effects spatial lag (or error) model'}
- \item{spat.coeff}{the spatial coefficients \eqn{\rho} or \eqn{\lambda} }
- \item{residuals}{the GLS residuals}
- \item{fitted.values}{difference between response variable and residuals}
- \item{sigma2}{GLS residuals variance}
- \item{model}{the matrix of the data used}
- \item{call}{the call used to create the object}
- \item{logLik}{the value of the log likelihood function at the optimum }
- \item{method}{the method used to calculate the Jacobian}
- \item{effects}{the typology of effects included in the model}
- \item{res.eff}{a list whose elements are the fixed effects (see \code{\link{effects}})}
-
-}
-
-\references{
- Elhorst, J.P. (2003)
-Specification and estimation of spatial panel data models,
- \emph{International Regional Science Review}, \bold{26}, pages 244--268.
-
- Elhorst, J.P. (2009)
-Spatial panel data models,
- \emph{In} Fischer, M.M. and Getis, A. (eds),
- \emph{Handbook of Applied Spatial Analysis} Springer, Berlin.
-
-Anselin, L. and Le Gallo, J. and Jayet, H. (2008)
-Spatial Panel Econometrics, \emph{In}
-Matyas, L. and Sevestre, P. (eds), \emph{The econometrics of Panel
-Data, Fundamentals and Recent Developments in Theory and Practice (3rd
-Edition)}, pages 624--660. Springer-Verlag, Berlin Heidelberg.
-}
-
-\author{ Gianfranco Piras \email{gpiras at mac.com}}
-
-\seealso{
- \code{\link{spreml}}, \code{\link{spgm}}, \code{\link{effects}}
-}
-\examples{
-data(Produc, package = "Ecdat")
-data(usaww)
-Produc <- Produc[Produc$year<1975, ]
-fm <- log(gsp) ~ log(pcap) + log(pc) + log(emp) + unemp
-lag <- spfeml(fm, data = Produc, listw = mat2listw(usaww), effects = "sptpfe", method = "eigen", quiet = TRUE)
-summary(lag)
-eff <- effects(lag)
-err <- spfeml(fm, data = Produc, listw = mat2listw(usaww), model = "error", effects = "tpfe", method = "eigen", quiet = TRUE)
-summary(err)
-eff <- effects(err)
-print(eff)
-write.effects.splm(eff)
-}
-
-\keyword{spatial}
\ No newline at end of file
Deleted: pkg/man/spreml.Rd
===================================================================
--- pkg/man/spreml.Rd 2011-05-26 06:37:37 UTC (rev 117)
+++ pkg/man/spreml.Rd 2011-05-26 06:39:23 UTC (rev 118)
@@ -1,112 +0,0 @@
-\name{spreml}
-\alias{spreml}
-\title{Spatial Panel Random Effects Model Estimation}
-\description{Maximum likelihood (ML) estimation of spatial random effects panel of the form:
-\deqn{
- y_it = X'_it \beta + u_it,
- i=1,\dots,N, t=1,\dots,T
-}
-with
-\deqn{
-u_t=\mu +\epsilon_t
-}
-and
-\deqn{
-\epsilon_t=\lambda W \epsilon_t+ \nu_t }
-where
-\deqn{\nu_t=\rho \nu_t-1 +e_t.
-}
-Depending on the restrictions on the vector of parameters one can
-differently combine error features giving rise to various nested
-specifications.
-}
-
-
-\usage{
-spreml(formula, data, index = NULL, w, w2=w, lag=FALSE,
- errors = c("semsrre","semsr","srre","semre",
- "re", "sr", "sem","ols", "sem2re"),
- pvar = FALSE, hess=FALSE, quiet=TRUE,
- initval = c("zeros", "estimate"),
- x.tol=1.5e-18, rel.tol=1e-15,
- ...)}
-
-\arguments{
- \item{formula}{a symbolic description of the model to be estimated}
- \item{data}{an object of class \code{\link{data.frame}} or \code{pdata.frame}. An optional data frame containing the variables
- in the model. When the obect is a \code{\link{data.frame}}, the first two columns may contain the indexes. See \code{index}}
- \item{index}{if not NULL (default), a character vector to identify the indexes among the columns of the \code{\link{data.frame}}}
- \item{w}{an object of class \code{listw} or a \code{matrix}}
- \item{w2}{an object of class \code{listw} or a \code{matrix}}
- \item{lag}{default=\code{FALSE}. If \code{TRUE}, a spatial lag of the dependent variable is added.}
- \item{errors}{one of \code{c("semsrre","semsr","srre","semre","re","sr","sem", "ols", "sem2re")}. See details.}
- \item{pvar}{if \code{TRUE} the \code{pvar} function is called}
- \item{hess}{if \code{TRUE} use numerical Hessian instead of GLS for the standard errors of the estimates}
- \item{quiet}{if \code{FALSE} report function and parameters values during optimization}
- \item{initval}{one of \code{c("zeros", "estimate")}, the initial values for the parameters. If \code{"zeros"} a vector of zeros is used. if \code{"estimate"} the initial
- values are retreived from the estimation of the nested specifications. Alternatively, a numeric vector can be specified.}
- \item{x.tol}{Tolerance. See \code{\link{nlminb}} for details. }
- \item{rel.tol}{Relative tolerance. See \code{\link{nlminb}} for details. }
- \item{...}{additional argument to pass over to other functions}
-}
-\details{
- The models are estimated by two-step Maximum Likelihood.
-
- The covariance structures allowed are:
- \code{"semsrre"} random effects, spatial and serial correlation in the idiosyncratic error
- term;
- \code{"semsr"} spatial and serial correlation in the idiosyncratic error
- term;
- \code{"srre"} individual random effects and serial correlation;
- \code{"semre"} individual random effects and spatial correlation;
- \code{"re"} individual random effects;
- \code{"sr"} serial correlation;
- \code{"sem"} spatial correlation.
- \code{"ols"} none (spherical residuals).
- \code{"sem2re"} spatial correlation in both individual random effects
- and idiosncratic errors (Kapoor, Kelejian and Prucha's specification).
-}
-\value{
- An object of class \code{"splm"}.
- \item{coefficients}{coefficients estimate of the model parameters }
- \item{arcoef}{the coefficient for the spatial lag on \code{y}}
- \item{errcomp}{the estimates of the error variance components}
- \item{vcov}{the asymptotic variance covariance matrix of the estimated coefficients}
- \item{vcov.arcoef}{the asymptotic variance of the
- estimated spatial lag parameter}
- \item{vcov.errcomp}{the asymptotic variance covariance matrix of the
- estimated error covariance parameters}
- \item{type}{'random effects ML'}
- \item{residuals}{the model residuals}
- \item{fitted.values}{the fitted values, calculated as \eqn{\hat{y}=X \hat{\beta}}}
- \item{sigma2}{GLS residuals variance}
- \item{model}{the matrix of the data used}
- \item{call}{the call used to create the object}
- \item{logLik}{the value of the log likelihood function at the optimum }
- \item{errors}{the value of the \code{errors} argument}
-}
-\references{
-Baltagi, B.H., Song, S.H., Jung B. and Koh, W. (2007)
-Testing panel data regression models with spatial and serial error correlation.
-\emph{Journal of Econometrics}, \bold{140}, 5-51.
-}
-
-\author{Giovanni Millo}
-
-\seealso{\code{\link{spgm}}, \code{\link{spfeml}}}
-
-\examples{
-data(Produc, package = "Ecdat")
-data(usaww)
-Produc <- Produc[Produc$year<1974, ]
-fm <- log(gsp) ~ log(pcap) + log(pc) + log(emp) + unemp
-## the two standard specifications:
-## random effects panel with spatial errors
-respaterr <- spreml(fm, data = Produc, w = usaww, errors="semre")
-summary(respaterr)
-## random effects panel with spatial lag
-\dontrun{respatlag <- spreml(fm, data = Produc, w = usaww, errors="re", lag=TRUE)}
-\dontrun{summary(respatlag)}
-}
-
-\keyword{spatial}
\ No newline at end of file
More information about the Splm-commits
mailing list