It is necessary to scale the magnitude of detail (or approximation) wavelet coefficients using wavedec?

I'm using wavedec toolbox to calculate the Wavelets coefficients in order to analyze the Energy composition of Power Systems signals. When I observed the coefficient magnitude of different detail coefficient I have noticed that this value is 1% or even less than the original signal's magnitude. I also noticed that the coefficient's magnitude depends on the vector length of the input signal. So, a doubt asalt me. It is necessary to perform some normalization of the coefficient magnitude once wavedec is applied?