Centered and Uncentered R-square and RMSE

[ANU THAKUR]

Dear Statalist

I am working on the topic diversification strategy and firm performance. And my literature suggests that there is a causal relationship between these two variables and suggested by my model as well.
Firstly I ran ivreg2 command only after that ivregress 2sls command, results were same by using both the commands. And then reg3 command to apply both the equations simultaneously.

reg3 (dep1.var_endog1.var_indeplist1.var) (dep2.var_endog2.var_indeplist2.va), 2sls inst (x_y_z_f_e_r)

There are some confusions in my mind like, how to interpret centered and uncentered R-square, RMSElikewise negative value of R-square.
For example: Is 0.0724 value of “R-sq” correct with 0.4389877 RMSE?

I already read this link: http://www.stata.com/support/faqs/st...least-squares/ but I am unclear with the meanings in a layman language.
Please correct me if I am wrong I think one should report centered R-sq. and in case of artificial regression (when constant is not present) uncentered R-sq should report. But what is artificial regression in that case.
And is there any cut-off value for RMSE?

Thanks in advance...
With Regards...
Anu Thakur

[Mark Schaffer]

You can find a discussion of centered vs. uncentered R-sq in Jeff Wooldridge's textbook, Introductory Econometrics: A Modern Approach. In the latest edition (2016), it comes up on page 214.