dlsem (2.4.6)

* added the linearly decreasing lag shape
* changed the name of the lag shape constructors: 'ecq' (endpoint-constrained quadratic), 'qd' (quadratic decreasing), 'ld' (linearly decreasing), 'gam' (gamma) 
* the method vcovHAC() was replaced by lmHAC()
* added the method drawSample()

dlsem (2.4.5)

* added the KPSS test
* the KPSS is set as the default unit root test in dlsem() if the sample size is less than 100
* removed the Choi's method for combining multiple group p-values of the unit root test. The Demetrescu's method is the one in use.


dlsem (2.4.4)

* fixed an error in the computation of the standard error of cumulative lag shapes
* added the method compareModels() for class 'dlsem' in place of methods AIC(), AICc() and BIC()


dlsem (2.4.3)

* fixed an error in the methods causalEff() and lagPlot() when the lag shape is not specified
* fixed an error in unirootTest() when the group factor is not specified
* automated lag order selection was implemented in unirootTest()
* the default value of argument 'hac' in dlsem() is set to TRUE
* added the method lagShapes()
* added the method autoCode()
* added the method residualPlot()
* added the method AICc() for class 'dlsem'
* improved the heuristic procedure for the adaptation of lag shapes


dlsem (2.4.2)

* added a deseasonalization procedure into dlsem()
* the imputation procedure into dlsem() is now performed before differentiation and takes into
  account of the relevant autoregressive terms
* changed the algorithm for the adaptation of gamma lag shapes:
  now a grid search on several pairs of delta and lambda is performed, ignoring the constraints
  on the gestation lag and on the lag width, and considering a number of time lags equal to
  two thirds of the sample size
* added the option 'gammaOptim' into dlsem() to speed up the adaptation of gamma lag shapes


dlsem (2.4.1)

* fixed the bug in the visualization of the cumulative coefficient in lagPlot()
* the constructor gamm.lag() replaces gamma.lag() and is made visible to the user


dlsem (2.4)

* fixed the bug in lagPlot() for gamma lag shapes
* fixed the bug in causalEff() for multiple starting variables
* methods lagPlot() and causalEff() now return NULL instead of an error in case of inexisting path
* lag shape transformations are visible to the user
* added the minimum gestation lag option in dlsem()
* added the gestation and the lead lags in the summary
* added the 'npar' method for class 'dlsem'
* added the vcovHAC() method for class 'lm' and 'dlsem'
* added the confint() method for class 'dlsem'
* added the HAC estimate of standard errors in method dlsem()