Changes in Version 0.3-6 o Correctly expand '.' in dynlm(y ~ ., data = d) (suggested by Julien Mc Donald-Guimond). Changes in Version 0.3-5 o Preserve class attribute of "na.action" (based on fix suggested by Matthieu Stigler). Changes in Version 0.3-4 o In previous versions the weights argument did not work when used with "ts" series. Changes in Version 0.3-3 o Streamlined Depends/Imports/Suggests. Now "dynlm" only depends on zoo but imports from stats, car, and lmtest. Packages datasets, strucchange, sandwich, and TSA are suggested (for use in examples). o Improved evaluation of pre-computed formulas (suggested by Vaidotas Zemlys). o Added tests/ with output from examples for R CMD check. Changes in Version 0.3-2 o Bug fix so that weights/offset can actually be used. Changes in Version 0.3-1 o Fixed bug in computation of R-squared values and Wald statistic for two-stage least squares. Changes in Version 0.3-0 o Added new formula functions trend() and harmon() that allow convenient specification of linear or cyclical patterns. Changes in Version 0.2-3 o Fixed bug when lagged variables are used in instrumental variables regression. Changes in Version 0.2-2 o Added CITATION file. Changes in Version 0.2-1 o Enhanced documentation ?dynlm. Changes in Version 0.2-0 o Added support for instrumental variables regression (two-stage least squares) via formulas like dynlm(y ~ x1 + x2 | z1 + z2 + z3, data = mydata) where z1, z2, z3 are the instruments. o Enabled specification of multiple lags via formulas like dynlm(y ~ L(x, 0:4), data = mydata) where y is regressed on x and lags 1 through 4 of x. o Fixed bug in time properties when there were leading NAs in the data.