[Hang Long]

After executing the irf table command, Stata reports a table of impulse responses, but for many years, there has been no effective method to convert the output table into a .dta file, matrix, or export it as an .xlsx file. Furthermore, the r() function does not save all the data after the command execution. Currently, the only option is to manually copy and paste, which is cumbersome. It is hoped that a new feature can be developed to enable the output table displayed on the screen to be converted into other formats after the execution of irf table.

[Dirk Enzmann]

dragon hhh : For very good reasons you are asked to use your full real name in the Stata Forum, see the FAQ (#6) as well as here (#3).

[Tobias Renkin]

Native read/write support for .parquet files.

[Noah Case]

Perhaps a simple request, but I wish the Stata command line supported line continuation. Sometimes I'd like to debug my code by running it interactively, and for various reason I do not like to use the do file editor. I would like to be able to copy past code such as

generate double variable = ///
    expression

into the command line without an issue.

[Rich Goldstein]

re: #64 - since you can just put your cursor on the line you want and click on "Execute (do) line", I'm not sure why what you are requesting is needed or even desirable - please clarify

[Jeff Wooldridge]

In recent versions of Stata, interactions are allowed to be specified as endogenous variables using ivregress. For example, the following is allowed:

ivregress 2sls y (i.w i.w#c.x1 = c.z c.z#c.x1 c.z#c.x2) x2 x3 ... xK, vce(robust)
margins, dydx(w)

In this way, I can easily estimate models with interactions among endogenous and exogenous variables, models with squares, interactions among endogenous variables, and so on. Unfortunately, xtivreg does not support factor notation or interactions. There is no argument against it, and it would simplify estimation a lot for more complicated panel data applications with heterogeneous treatment effects. Having to created the interactions is clunking and precludes using the margins command after estimation.

Stata is best when options are consistent across estimation methods that are connected. For example, both reg and xtreg allow factor notation among the explanatory variables. But margins after xtreg does not support the vce(uncond) option, and it should. This is another addition I'd like to see in Stata 20.

[Homa Taheri]

Hi,

I think it would be useful to add features for calculating both "mode" of observations in the same rows across multiple columns. Something like similar to what commands like "egen----rowtotal and egen----rowmax" do. It could be something like:

egen X_mode =rowmode(X1-X10)(X1-X10)

[Rich Goldstein]

there is a "rowmedian" function in the egen command - it was added, I believe, in version 11

[Homa Taheri]

there is a "rowmedian" function in the egen command - it was added, I believe, in version 11

Thank you, Rich. I just edited my post, and dropped the part about calculating median.

[Bruce Weaver]

Thank you, Rich. I just edited my post, and dropped the part about calculating median.

That's a relief! For the life of me, I could not figure out why Rich was talking about rowmedian in #68. But I can see now that he was responding to the original version of #67.