mfx.hlp

是一个经济学管理应用软件 很难找的 但是经济学学生又必须用到

{smcl}
{* 07mar2005}{...}
{cmd:help mfx}{right:dialog:  {bf:{dialog mfx:mfx}}}
{hline}

{title:Title}

{p2colset 5 16 18 2}{...}
{p2col :{hi:[R] mfx} {hline 2}}Obtain marginal effects or elasticities after
estimation{p_end}
{p2colreset}{...}


{title:Syntax}

{p 8 12 2}
{cmd:mfx}
[{opt c:ompute}]
{ifin}
[{cmd:,} 
{it:options}]

{p 8 12 2}
{cmd:mfx}
{opt r:eplay}
[{cmd:,}
{opt l:evel(#)}]

{synoptset 27 tabbed}{...}
{synopthdr}
{synoptline}
{syntab :Model}
{synopt :{opt pred:ict(predict_option)}}calculate marginal effects (elasticities)
for {it:predict_option}{p_end}
{synopt :{opth var:list(varlist)}}calculate marginal effects (elasticities) for
{it:varlist}{p_end}
{synopt :{opt dydx}}calculate marginal effects; the default{p_end}
{synopt :{opt eyex}}calculate elasticities in the form of d(lny)/d(lnx){p_end}
{synopt :{opt dyex}}calculate elasticities in the form of d(y)/d(lnx){p_end}
{synopt :{opt eydx}}calculate elasticities in the form of d(lny)/d(x){p_end}
{synopt :{opt nod:iscrete}}treat dummy (indicator) variables as
continuous{p_end}
{synopt :{opt nos:e}}do not calculate standard errors{p_end}

{syntab :Model 2}
{synopt :{cmd:at(}{it:{help mfx##atlist:atlist}}{cmd:)}}estimate marginal effects (elasticities) at these values{p_end}
{synopt :{opt noe:sample}}do not restrict calculation of means, medians, and
mean offsets to the estimation sample{p_end}
{synopt :{opt now:ght}}ignore weights when calculating means, medians, and mean
offsets{p_end}

{syntab :Adv. model}
{synopt :{opt nonl:inear}}do not use the linear method{p_end}
{synopt :{opt force}}calculate marginal effects and standard errors in cases
where it would otherwise refuse to do so{p_end}

{syntab :Reporting}
{synopt :{opt l:evel(#)}}set confidence level; default is
{cmd:level(95)}{p_end}
{synopt :{cmdab:diag:nostics(beta)}}report suitability of marginal-effect
calculation{p_end}
{synopt :{cmdab:diag:nostics(vce)}}report suitability of standard-error
calculation{p_end}
{synopt :{cmdab:diag:nostics(all)}}report all diagnostic information{p_end}
{synopt :{opt tr:acelvl(#)}}report increasing levels of detail during
calculations{p_end}
{synoptline}
{p2colreset}{...}
{p 4 6 2}{marker atlist}
where {it:atlist} is {it:{help numlist}} or {it:matname} or

{pin2}
[{cmd:mean}{c |}{cmd:median}{c |}{cmd:zero}] [{varname} {cmd:=} {it:#} [{cmd:,} {it:varname} {cmd:=} {it:#}] [...]]

{pin}
where {opt mean} is the default.


{title:Description}

{pstd}
{opt mfx}
numerically calculates the marginal effects or the elasticities and their
standard errors after estimation.  Exactly what {opt mfx} can calculate is
determined by the previous estimation command and the
{opt predict(predict_option)} option.  The values at which the marginal
effects or elasticities are to be evaluated is determined by the
{opt at(atlist)} option.  By default, {opt mfx} calculates the marginal
effects or elasticities at the means of the independent variables using the
default prediction option associated with the previous estimation command.

{pstd}
The term partial effects is used in some disciplines, rather than marginal
effects, for what is computed by {cmd:mfx}.

{pstd}
{opt mfx replay} replays the results of the previous {opt mfx} computation.


{title:Options}

{dlgtab:Model}

{phang}
{opt predict(predict_option)} specifies the function
    (that is, the form of y) for which to calculate the marginal effects or
    elasticities.  The default is to use the default {opt predict} option of
    the preceding estimation command.  To see which {opt predict} options are
    available, see {opt help} for the particular estimation command.

{phang}
{opth varlist(varlist)} specifies the variables
    for which to calculate marginal effects (elasticities).
    The default is all variables.

{phang}{cmd:dydx} specifies that marginal effects be calculated. It is
the default.

{phang}
{opt eyex} specifies that elasticities be calculated in the form
    of d(lny)/d(lnx)

{phang}
{opt dyex} specifies that elasticities be calculated in the form
    of d(y)/d(lnx)

{phang}
{opt eydx} specifies that elasticities be calculated in the form
    of d(lny)/d(x)

{phang}
{opt nodiscrete} treats dummy variables as continuous.  A dummy variable is
    one that takes on the value 0 or 1 in the estimation sample.  If
    {opt nodiscrete} is not specified, the marginal effect of a dummy variable
    is calculated as the discrete change in y as the dummy variable changes
    from 0 to 1.  This option is irrelevant to the computation of the
    elasticities because all the dummy variables are treated as continuous
    when computing elasticities.

{phang}
{opt nose} specifies that standard errors of the marginal effects
    (elasticities) not be computed.

{dlgtab:Model 2}

{phang}
{opt at(atlist)} specifies the values at which the
    marginal effects (elasticities) are to be estimated.  The default is to
    evaluate at the means of the independent variables.

{pmore}
    {opth at(numlist)} specifies that the marginal effects (elasticities)
    be evaluated at the {it:numlist}.  For instance,

{p 12 16 2}{cmd:. probit foreign mpg weight price}{p_end}
{p 12 16 2}{cmd:. mfx, at(21 3000 6000)}

{pmore}
    The order of the values in the {it:numlist} is the same as the variables
    in the preceding estimation command, from left to right, without
    repetition.  For instance,

{p 12 16 2}{cmd:. sureg (price disp weight) (mpg for disp) }{p_end}
{p 12 16 2}{cmd:. mfx, predict(xb) at(200 3000 0.5)}

{pmore}
    {opth at(matname)} specifies the points in a matrix
    format.  The ordering of the variables is the same as that of
    {it:numlist}.  For instance,

{p 12 16 2}{cmd:. probit foreign mpg weight price}{p_end}
{p 12 16 2}{cmd:. mat A = (21, 3000, 6000)}{p_end}
{p 12 16 2}{cmd:. mfx, at(A)}

{pmore}
    {cmd:at(} {opt mean} | {opt median} | {opt zero} [ {it:varname}
    {cmd:=} {it:#} [{cmd:,} {it:varname} {cmd:=} {it:#} [{it:...}]]] {cmd:)}
    specifies that the marginal effects (elasticities) be
    evaluated at means, at medians of the independent variables, or at zeros.
    It also allows users to specify particular values for one or more
    independent variables, assuming that the rest are means, medians, or
    zeros.

{p 12 16 2}{cmd:. probit foreign mpg weight price}{p_end}
{p 12 16 2}{cmd:. mfx, at(mean mpg=30)}

{pmore}
    {cmd:at(}{varname} {cmd:=} {it:#} [{cmd:,} {it:varname} {cmd:=} {it:#} ]
    [...]{cmd:)} specifies that the marginal effects or the elasticities be
    particular values for one or more independent variables, assuming that the
    rest are means.

{p 12 16 2}{cmd:. probit foreign mpg weight price}{p_end}
{p 12 16 2}{cmd:. mfx, at(mpg=30)}

{phang}
{opt noesample} affects {opt at(atlist)}, any offsets used in the
    preceding estimation, and the determination of dummy variables.  It
    specifies that the whole dataset be considered instead of only those
    marked in the {cmd:e(sample)} defined by the previous estimation command.

{phang}
{opt nowght} only affects {opt at(atlist)} and offsets.
    It specifies that weights be ignored when calculating the means or
    medians for the {it:atlist} and when calculating the means for any
    offsets.

{dlgtab:Adv. model}

{phang}
{opt nonlinear} specifies that y, the function to be calculated for the
    marginal effects or the elasticities, does not meet the linear-form
    restriction.  By default, {opt mfx} assumes that y meets the linear-form
    restriction, unless one or more dependent variables are shared by multiple
    equations.  For instance, predictions after

{phang3}
{cmd:. heckman mpg price, sel(foreign=rep78)}

{pmore}
    meet the linear-form restriction, but those after

{phang3}
{cmd:. heckman mpg price, sel(foreign=rep78 price)}

{pmore}
    do not.  If y meets the linear-form restriction, specifying
    {opt nonlinear} should produce the same results as not specifying it.
    However, the nonlinear method is generally more time consuming.  Most
    likely, you do not need to specify {opt nonlinear} after an official Stata
    command.  For user-written commands, if you are not sure whether y is of
    linear form, specifying {opt nonlinear} is a safe choice.

{phang}
{opt force} specifies that marginal effects and their 
    standard errors be calculated in cases where it would otherwise refuse to
    do so.  Such cases arise, for instance, when the marginal effect is a
    function of a random quantity other than the coefficients of the model
    (e.g., a residual).  If you specify this option, there is no guarantee
    that the resulting marginal effects and standard errors are correct.

{dlgtab:Reporting}

{phang}
{opt level(#)} specifies the confidence level, as a percentage,
    for confidence intervals.  The default is {cmd:level(95)} or as set by
    {helpb set level}.

{phang} {opt diagnostics(diaglist)} asks {opt mfx} to display various
    diagnostic information.

{pmore}
    {cmd:diagnostics(beta)} shows the information used to determine
    whether the prediction option is suitable for computing
    marginal effects.

{pmore}
    {cmd:diagnostics(vce)} shows the information used to determine
    whether the prediction option is suitable for computing
    the standard errors of the marginal effects.

{pmore}
    {cmd:diagnostics(all)} shows all the above diagnostic information.

{phang}
{opt tracelvl(#)}
    shows increasing levels of detail during calculations.
    {it:#} may be 1, 2, 3, or 4.
    Level 1 shows the marginal effects and standard errors as they are computed,
    and which method, either linear or nonlinear, was used.
    Level 2 shows, in addition, the components of the matrix of partial
    derivatives needed for each standard error as they are computed.
    Level 3 shows counts of iterations in obtaining a suitable finite
    difference for each numerical derivative.  
    Level 4 shows the values of these finite differences.


{title:Examples}

{phang}{cmd:. sysuse auto, clear}{p_end}
{phang}{cmd:. logit foreign mpg price}{p_end}
{phang}{cmd:. mfx, predict(p)}{p_end}
{phang}{cmd:. mfx, predict(p) at(mpg = 20, price = 6000)}{p_end}
{phang}{cmd:. mfx, predict(p) at(20 6000)}{p_end}

{phang}{cmd:. mlogit rep78 mpg displ, nolog}{p_end}
{phang}{cmd:. mfx, predict(p outcome(2))}{p_end}
{phang}{cmd:. mfx, predict(p outcome(2)) at(20 400) }{p_end}
{phang}{cmd:. mfx, predict(p outcome(2)) varlist(mpg)}{p_end}

{phang}{cmd:. heckman mpg weight length, sel(foreign = length displ) nolog}{p_end}
{phang}{cmd:. mfx, predict(xb)}{p_end}
{phang}{cmd:. mfx, predict(xbsel)}{p_end}
{phang}{cmd:. mfx, predict(yexpected) varlist(length) }{p_end}

{phang}{cmd:. regress mpg length weight}{p_end}
{phang}{cmd:. mfx, eyex}{p_end}
{phang}{cmd:. mfx replay, level(90)}{p_end}


{title:Also see}

{psee}
Manual:  {bf:[R] mfx}

{psee}
Online:  {helpb predict}, {helpb _predict}, {helpb probit},
{helpb truncreg}
{p_end}