hurdle.IV.sim: Simulate a dataset

Description Usage Arguments Value

View source: R/hurdleIVsim.R

Description

Simulates a dataset from either Cragg or Lognormal distributions. Allows you to specify the data generating process, though it still can't handle more than one endogenous regression

Usage

1
2
3
4
5
hurdle.IV.sim(formula = F, pi = c(1, -1, 3), gamma = c(-0.2, 0.8, 0.07),
  beta = c(0.05, 0.06, 0.02), endog_reg = list(), exog_mean = 1,
  exog_sd = 1, z_mean = 3, z_sd = 1, endog_sd = 5, y_sd = 2,
  rho = 0.2, tau0 = 0.3, tau1 = 0.1, n = 10000, options = list(silent
  = F, cragg_errors = 4), type = "lognormal")

Arguments

formula

For now, must be False. In the future, will allow you to leave out exogenous variables you use in first stage regressions from second stage regressions.

pi

a vector of the coefficients on your first stage regression. Should be in the order: intercept, exogenous variables, instruments. Defaults to (1,-1,3).

gamma

a vector of the coefficients on your second stage probit regression. Should be in the order: intercept, exogenous variables, endogenous variable. Defaults to (-.2,.8,.07)

beta

a vector of the coefficients on your second stage linear regression. Should be in the order: intercept, exogenous variables, endogenous variable. Defaults to (.05,.06,.02)

endog_reg

for now, leave as an empty list. Will create one endogenous regression which includes the instrument and all exogenous variables

exog_mean

the mean(s) of your exogenous variable(s). Defaults to 1

exog_sd

the standard deviation(s) of your exogenous variable(s). Defaults to 1.

z_mean

the mean(s) of your instrument(s). Defaults to 3.

z_sd

the standard deviation(s) of your instrument(s). Defaults to 1.

y_sd

the standard deviation of your second stage linear regression. Defaults to 2.

rho

the covariance term between second stage linear and probit regressions. Defaults to 0.2.

tau0

the covariance term between the second stage probit and the first stage regression. Defaults to 0.3.

tau1

the covariance term between the second stage linear regression and the first stage regression. Defaults to 0.1.

n

the number of observations. Defaults to 10,000.

options

Options are: silent = F (will print histograms and percent censored), cragg_errors = 1,2,3,4 cragg error type 4 will generate x2 and y0 first, then generate y1 from a conditional distribution on the first two and on itself being positive. See documentation for cragg_errs for more detail.

type

defaults to lognormal. Enter "cragg" for cragg model.

Value

returns a simulated dataset


jackiemauro/hurdleIV documentation built on April 2, 2018, 8:28 p.m.