As Betwix pointed out, the phase is always relative. For instance phase of output of an amplifier is expressed relative to the input.
To define a phase in degrees, one has to have a repetitive signal, which then has also a frequency of repetition. As our time goes only forward, the output is always behind (later) than the input, as no circuit can have a negative time delay.
Because the amplifier's delay time is usually roughly similar for frequencies in same magnitude (is not exactly constant - but mostly doesn't change abruptly) , the phase delay in degrees tend to increase pretty consistently with frequency -- as the cycle time is reduced with higher frequencies.
That causes the phase shift to increase, and if the amplifier feedback is on low frequencies properly negative, on higher frequencies it tends to turn, until it is even positive, leading to instability - if the gain is still sufficient on that high frequency to matter. (Usually the amplifier is designed so, that the gain rolls to below one before the phase has twisted 180 degrees)