X1,...Xn ~ Gamma(lambda,theta)

lambda = shape parameter

theta = scale parameter

we assume that a<lambda*n

I need to prove that:

T=[(gf(n*lambda))/(gf(n*lambda-a))]*[sum(xi)]^-a

is an unbiased estimator of the function T(theta) = theta^a

I have no clue how to start it, any clue will be appreciated. I can type it again as a picture if you fail to understand the question.

thanks