non-linear regression

I'm a newbie to STATA and am trying to estimate a parameter using the 'nl' command. Here is the syntax:

. nl (c1 = (c2^{a}-c2^{a}*y1^{a}/x1^{a}+y1^{a})^(1/{a}))

The dataset has c1,c2,x1 & y1 given and I want to estimate a. However, I am getting the following error: "starting values invalid or some RHS variables have missing values"

Here is the actual equation that I want to fit: c1^a=c2^a - (c2^a*y1^a)/x1^a + y1^a

I realize that a=0 solves the equation and so tried setting a=0.1 etc., but it still gives the same error.

Any suggestions?