Ramsey RESET test

[Daniel Gomez]

Hi,

Is it possible to use the Ramsey RESET test with only one regressor?:




Stata omits the variable fit2 because of collinearity.....


Thank you!

[Carlo Lazzaro]

Daniel:
yes, it is, but I do not think that my reply is what you're after:

. sysuse auto.dta
(1978 Automobile Data)

. reg price mpg

      Source |       SS           df       MS      Number of obs   =        74
-------------+----------------------------------   F(1, 72)        =     20.26
       Model |   139449474         1   139449474   Prob > F        =    0.0000
    Residual |   495615923        72  6883554.48   R-squared       =    0.2196
-------------+----------------------------------   Adj R-squared   =    0.2087
       Total |   635065396        73  8699525.97   Root MSE        =    2623.7

------------------------------------------------------------------------------
       price |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
         mpg |  -238.8943   53.07669    -4.50   0.000    -344.7008   -133.0879
       _cons |   11253.06   1170.813     9.61   0.000     8919.088    13587.03
------------------------------------------------------------------------------

. estat ovtest

Ramsey RESET test using powers of the fitted values of price
       Ho:  model has no omitted variables
                  F(3, 69) =      7.72
                  Prob > F =      0.0002

.

[Eric de Souza]

Please show your Stata output if you want a proper answer adapted to your case. And use the code delimiters before and after your output to make it readable.

[Daniel Gomez]

Yes, actually I am trying to implement the Ramsey RESET test for a gravity model with panel data.

Do you know how could I include the fixed values in the prediction for this Ramsey RESET test?:


cap egen imp=group(importer)
cap egen exp=group(exporter)
eststo: ppmlhdfe TRADE RTA ///
if exporter != importer, a(exp#year imp#year exp#imp) cluster(pairid) nolog
estimates store PPMLfesPAIR


*RESET test
predict fit, xb
generate fit2 = fit^2
ppmlhdfe TRADE RTA fit2 ///
if exporter != importer, a(exp#year imp#year exp#imp) cluster(pairid) nolog
test fit2 = 0
drop fit*


Thanks!

[Carlo Lazzaro]

Daniel:
I'm not familiar with gravity model, sorry.
See Joao Santos Silva 's post on this topic.

[Joao Santos Silva]

Dear Daniel Gomez,

The problem is that your only regressor is a dummy and the square of a dummy is the dummy itself. Essentially, the functional form of the model cannot be incorrect if the only regressor is a dummy, and that is why you cannot perform the RESET in that case.

You can do the RESET if you include the fixed effects in the predicted values, but in that case you need to worry about the potential incidental parameters problem.

Best wishes,

Joao

[Daniel Gomez]

Perfect! Thank you very much! 🙂

[Rob Baker]

Dear Joao Santos Silva ,

Like Daniel Gomez I would like to run a RESET test on a fixed effects model with two regressors (both dummies). I run the following:

ppmlhdfe tradevalueus ceta1 ceta4, a(imp#year exp#year imp#exp) cluster(imp#exp) nolog
predict fit, xb
gen fit2=fit^2
ppmlhdfe tradevalueus ceta1 ceta4 fit2, a(imp#year exp#year imp#exp) cluster(imp#exp) nolog
test fit2=0

After reading the above I understand why this doesn't work, but how do I include the fixed effects in the predicted values? Which commands should I employ?

Thanks,

Rob

[Joao Santos Silva]

Dear Rob Baker,

In your case it should work because you have two dummies and therefore the reset is equivalent to a test for the exclusion of their product.

Best wishes,

Joao

[Rob Baker]

Dear Joao Santos Silva ,

Thank you again for your help and patience.

When I run the second model 'fit2' is omitted because of collinearity (pictured below). Do you have any suggestions as to how I could solve this issue?

Running a RESET test on this kind of model (dummies for trade creation & trade diversion with importer year, exporter year & dyadic fixed effects) has been done before here - https://www.sciencedirect.com/scienc...43951X14000285

I'm just not sure how! Frustratingly, this is the final part of my thesis.

Thanks,

Rob

[Joao Santos Silva]

Dear Rob Baker,

Then I guess that the product of the two dummies is always zero, and therefore you do not need to run the test

Best wishes,

Joao