JOINT-HYPOTHESIS TESTING PANEL DATA

[Guy Salgado]

Hi,

I am currently using stata12 and would like to make a joint-hypothesis test on the following regression attached as an image. I have a panel of two years and I am using fixed effects as my estimator.

The joint test is that beta1=0 and beta2=1, in other words the first explanatory variable in the regression (TOTAL_income) equals zero and that the second explanatory variable (AV_village_consumption) equals one.

I hope I was clear enough and would really like to ask for someone's help on this.

thank you

[Clyde Schechter]

For the convenience of those who might respond to you, the best way to show us output is to put it into a code block in the forum's editor. (Select the Advanced Editor by clicking underlined A button, Then click on the # button, which will bring up the delimiters of a code block. Copy your Stata material directly from the do-file, or results window, or log, and paste it between the two delimiters. The result will display in a way that is large enough to read, properly aligned, and also can be copied directly into Stata for those of us who might want to work with it. A posted screen shot is, perhaps, better than nothing, but is small and difficult to read, and impossible to do further work with.

So, I don't actually know what the names of the two variables involved are, but I will just call them var1 and var2. The code you want is

test var1 = 0
test var2 = 1, accum

The first command will just test if the first coefficient is significantly different from zero. The second one, because of the -accum- option, will give you the joint test you are looking for. If you prefer to suppress the output for the first command, you can add the -notest- option to that one.

[Guy Salgado]

Thank you very much Clyde.

Next time ill make sure I present the output as you advised.

kind regards

[Richard Williams]

A little trick I just discovered while reading the latest edition of Long and Freese (available from the Stata Bookstore):

test (var1 = 0) (var2 = 1)

also works. That seems like a good alternative for most of the times I have used accum.