1Changes in Version 1.2-8 2 3 o Some examples from the AER book lead to small numeric differences across different 4 platforms. Hence, these have been excluded now from tests/Ch-*.R with an IGNORE_RDIFF 5 comment for CRAN. 6 7 o The examples on the manual pages for the books CameronTrivedi1998, Greene2003, 8 WinkelmannBoes2009 are now excluded from testing to reduce the computational demands 9 on CRAN. The corresponding analyses are mostly available on the manual pages for the 10 underlying data sets, though. 11 12 13Changes in Version 1.2-8 14 15 o Starting from survival >= 3.1-6 the survival provides (or corrected) some standard 16 S3 methods for "survreg" objects: fitted(), nobs(), weights(), vcov(). Previously, 17 these were registered by AER" in order to work for "tobit" objects. For now, 18 AER registers the methods for "tobit" objects. In the future, the methods will 19 be dropped by AER altogether and inherited from survival (when AER depends on 20 survival >= 3.1-6). 21 22 23Changes in Version 1.2-7 24 25 o Diagnostic tests ivdiag() for _weighted_ instrumental variables regression 26 was incorrect as weights were erroneously ignored (detected and tested by 27 Jonathan Siverskog). 28 29 o vcov() method for "survreg" objects in "survival" currently (version 2.44-1.1) 30 provides no row/column names. Hence, a new vcov() method for "tobit" objects 31 was added in "AER" mimicking the naming conventions from survival:::summary.survreg. 32 Analogously, a bread() method was added for "tobit" objects that calls 33 the vcov() method internally. 34 35 o linearHypothesis() method for tobit() objects no longer assumes that a scale 36 parameter was estimated as part of the model. This enables the somewhat 37 unusual but possible case of summary(tobit(..., dist = "exponential")). 38 39 o In examples/demos/tests with calls to set.seed(...) the RNG version is now fixed 40 to suppressWarnings(RNGversion("3.5.0")) to keep results exactly reproducible 41 after fixing RNG problems in R 3.6.0. 42 43 o New errata item in vignette("AER", package = "AER"): The formula interface for 44 pgmm() has changed. As of "plm" version 1.7-0, the function dynformula() is 45 deprecated. Examples, tests, and demos have been adapted accordingly. 46 47 o In tests/Ch-Intro.R some quantreg computations are now wrapped into try() 48 because the summary method yields non-finite standard errors on some platforms. 49 50 o In ?CameronTrivedi1998 one version of the negbin-negbin hurdle model is now 51 wrapped into try() because evaluating the log-likelihood at the given start 52 values becomes too unstable on some platforms. 53 54 55Changes in Version 1.2-6 56 57 o Added update() method for "ivreg" objects that correctly handles the 58 two-part right-hand side of the formula (suggested by Matthieu Stigler). 59 60 o Use pdata.frame() instead of plm.data() as preparation for modeling 61 with "plm" and certain "systemfit" functions (requires systemfit 62 1.1-20). 63 64 o The model.matrix() method for "ivreg" objects erroneously dropped 65 the dimension of single-column projected regressor matrices. This 66 propagated to the estfun() method and hence lead to problems with 67 sandwich covariances (reported by Justus Winkelmann). 68 69 o In a bug fix the survival package changed the internal structure of 70 survreg()$y starting from survival 2.42-7. This leads to incorrect 71 summary() output for "tobit" objects which has been worked around 72 now. Thanks to Terry Therneau for pointing out the problem and suggesting 73 a fix. 74 75 76Changes in Version 1.2-5 77 78 o Support for aliased coefficients in ivreg() (suggested and tested by 79 Liviu Andronic). 80 81 o New data sets GoldSilver, MotorCycles2 and MSCISwitzerland, taken from 82 Franses, van Dijk, Opschoor (2014): "Time Series Models for Business and 83 Economic Forecasting", 2nd ed. For replication of the corresponding 84 examples, several packages were added to the list of 'suggested' packages 85 (including fGarch, forecast, longmemo, rugarch, vars). 86 87 o Small improvements in DESCRIPTION/Imports and NAMESPACE for R CMD check. 88 89 90Changes in Version 1.2-4 91 92 o Reference output updated for recent versions of R. 93 94 95Changes in Version 1.2-3 96 97 o Package "splines" is loaded explicitly if needed (rather than assuming 98 that it is loaded along with "survival"). 99 100 o Some URLs in the manual pages had been outdated and are updated (or omitted) 101 now. 102 103 104Changes in Version 1.2-2 105 106 o Another bug fix in the new summary(ivreg_object, diagnostics = TRUE). If 107 sandwich standard errors (or other vcov) were used, the chi-squared form 108 rather than the F form of the diagnostic Wald tests was computed and hence 109 the p-values were incorrect. 110 111 o If there is more than one endogenous variable, summary(ivreg_object, 112 diagnostics = TRUE) now reports separate tests of weak instruments for 113 each endogenous variable. 114 115 116Changes in Version 1.2-1 117 118 o Bug fix in the new summary(ivreg_object, diagnostics = TRUE). If there 119 is more than one endogenous variable, the degrees of freedom (and hence 120 the associated p-values) for the Sargan test were too large. 121 122 o The examples employing "rgl" for 3d visualizations (e.g., for the SIC33 123 data) are not tested anymore in R CMD check (as "rgl" currently has some 124 problems on CRAN's checks for OS X). 125 126 127Changes in Version 1.2-0 128 129 o The summary() method for ivreg() now has a diagnostics=FALSE argument. 130 If set to TRUE, three diagnostic tests are performed: an F test of 131 the first stage regression for weak instruments, a Wu-Hausman test 132 for endogeneity, and a Sargan test of overidentifying restrictions 133 (only if there are more instruments than regressors). 134 135 o Added new data set EquationCitations provided by Fawcett & Higginson 136 (2012, PNAS). 137 138 o Changes in Depends/Imports/Suggests due to new CRAN check requirements. 139 In particular, the "Formula" package is now only imported (but not 140 loaded into the search path). 141 142 143Changes in Version 1.1-9 144 145 o Recompressed data sets in package to reduce file storage requirements. 146 147 o ivreg() failed when used without instruments. Now fixed. 148 149 o The summary() for ivreg() displayed the degrees of freedom of the 150 overall Wald test incorrectly (although the p-value was computed 151 correctly). 152 153 o Some technical changes for new R 2.14.0, e.g., adding Authors@R in 154 DESCRIPTION, recompressing data, etc. 155 156 157Changes in Version 1.1-8 158 159 o The hat values for instrumental variables regressions are now 160 computed in the hatvalues() method and not within ivreg.fit() 161 to save computation time for large data sets. 162 163 o Added nobs() method for "survreg" objects (and thus "tobit" objects). 164 Modified "ivreg" objects so that default nobs() methods works. 165 166 o Labeling in coeftest() method for "multinom" objects with 167 binary responses has been fixed. 168 169 o Example 21.4 in ?Greene2003 now employs the scaled regressor 170 fincome/10000. 171 172 173Changes in Version 1.1-7 174 175 o Adapted some example in ?Greene2003 in order to work both with 176 the old and new "dynlm" package. dynlm() now provides convenient 177 support for linear time trends via dynlm(y ~ trend(y)) etc. 178 179 180Changes in Version 1.1-6 181 182 o Adapted code/examples/tests to new car version which has deprecated 183 linear.hypothesis() in favor of linearHypothesis(). 184 185 186Changes in Version 1.1-5 187 188 o CPS1985 now has 534 observations (not 533 as in prior releases), 189 the original observation 1 had been omitted inadvertently. See 190 also the errata in vignette("AER", package = "AER"). 191 192 o Data and examples for Winkelmann and Boes (2009), 193 "Analysis of Microdata" (2nd ed.) have been added. For details and 194 extensive (but not quite complete) replication code see 195 help("WinkelmannBoes2009") as well as help("GSS7402") and 196 help("GSOEP9402"). 197 198 o As announced in the changes for version 1.1-0 of the "AER" package, 199 the variable "earnings" has now been removed from the 200 PSID1976 (aka Mroz) data. In 1.1-0 it was renamed to "wage" 201 to avoid confusion with other data sets. 202 203 o The coeftest() method for "polr" objects used to return wrong 204 standard errors (and hence wrong z tests) for the last intercept. 205 This was caused by an inconsistency between the summary() and 206 vcov() methods for "polr" objects which has been improved in 207 recent versions of the "MASS" package. The correct results are 208 computed by coeftest() for "polr" objects computed with MASS 209 version >= 7.3-6. See also the errata in vignette("AER", 210 package = "AER") 211 212 o The paper describing the various versions of the Grunfeld data 213 has been accepted for publication in the German Economic Review. 214 An updated version of the manuscript and associated replication 215 files -- mostly based on data("Grunfeld", package = "AER") -- 216 is available from http://statmath.wu.ac.at/~zeileis/grunfeld/. 217 218 o Added lrtest() method for "fitdistr" objects with intelligible 219 model name (instead of the usual formula for formula-based models). 220 221 222Changes in Version 1.1-4 223 224 o ivreg() now uses the "Formula" package (>= 0.2-0) for processing 225 of its model formulas. However, this only affects the internal 226 computations, the user interface has remained unchanged. 227 228 o Numerous spelling improvements in the documentation (thanks to 229 the new aspell() functionality in base R). 230 231 232Changes in Version 1.1-3 233 234 o Added PSID7682 data set which contains the full Cornwell & Rupert 235 (1988) panel data for the years 1976-1982. This should be used 236 for estimation of the Hausman-Taylor model in Exercise 6 from 237 Chapter 3 (instead of PSID1982 which does not provide panel data 238 but only the cross-section for 1982). See the errata and the 239 manual page for more details. 240 241 o Fixed overall Wald test in summary() for "tobit" models 242 with intercept only. 243 244 245Changes in Version 1.1-2 246 247 o New errata item in vignette("AER", package = "AER"): The comment 248 regarding the output from the Johansen test (p. 169) is in error. 249 The null hypothesis of no cointegration is not rejected at the 250 10% level (only at 15% level). 251 252 o Enhancements of the CigarettesSW examples from Stock & Watson. 253 254 o Fixed overall Wald test in summary() for "tobit" models 255 without intercept. 256 257 o Improved "rgl" code in the SIC33 example. 258 259 o The variable "gender" in the Parade2005 data set was 260 wrong for observation 70. It is now "male" (not "female"). 261 262 263Changes in Version 1.1-1 264 265 o A new improved version of the "plm" package is available 266 from CRAN (version 1.1-1). This fixes a bug in the summary() 267 of "plm" objects, see the vignette/errata for details. 268 Furthermore, there is now a vcovHC() method for "panelmodel" 269 objects: It gives equivalent results to pvcovHC() but 270 is now the recommended user interface and hence used 271 in the examples of some manual pages (see e.g. ?Fatalities). 272 273 274Changes in Version 1.1-0 275 276 o Some variable names in the PSID1976 (aka Mroz) data 277 have been renamed: "earnings" is now called "wage" 278 (to avoid confusion with other data sets), the previous 279 variable "wage" is renamed as "repwage" (reported wage). 280 Currently, "earnings" is kept; it will be removed in future 281 releases. 282 283 o Documentation of the Grunfeld data has been enhanced and 284 updated. Much more details are available in a recent 285 technical report: Kleiber and Zeileis (2008), "The 286 Grunfeld Data at 50", available from http://epub.wu-wien.ac.at/. 287 288 o Multinomial logit examples using Yves Croissant's "mlogit" 289 package have been added for the TravelMode and BankWages 290 data sets. 291 292 o Vignette/errata updated. 293 294 295Changes in Version 1.0-1 296 297 o Small changes for R 2.8.0. 298 299 300Changes in Version 1.0-0 301 302 o official version accompanying the release of the 303 book (contains all code from the book in demos 304 and tests) 305 306 o See the new 307 vignette("AER", package = "AER") 308 for an overview of the package and a list of errata. 309 310 311Changes in Version 0.9-0 312 313 o release of the version used for compiling the final 314 version of the book for Springer 315 316 317Changes in Version 0.2-0 318 319 o first CRAN release of the AER package 320