Is the local density approximation exact for short wavelength fluctuations?