It's a valid approach just for general cases simply because most LEDs happen to have a forward voltage drop near that value.
Here is a table take from wikipedia on the voltage-drop of different LEDs:
You can see that for some reds, oranges, yellows, and some greens, 2V is close to the value. I suppose if you were making a circuit with ultraviolet LEDs you could assume 3.5, or 4V. Now if you actually know the forward voltage drop on the LEDs you are using, it would make a lot more sense to use that value.
Going off of what echad said, the constant voltage drop model is the simplest one, and speeds up analysis. In reality, voltage drop on diodes have an exponential relationship.
Also, there are several different models for analyzing circuits that contain diodes.
Taken from a textbook I use at school, Microelectronic Circuits 6th Ed, by Sedra and Smith:
Graphical Analysis of the Exponential Model, using a load line
Constant Voltage Drop Model
Now this is for plain silicon diodes, but the same math holds true for all diodes, just the parameters are slightly different and the drop for LEDs comes out different based on how they are manufactured.