The parallel formula is not valid here. It applies only to Ls away from each other. When you consider wider trace as a parallel of several narrow traces, mutual inductance comes into play. To a good approximation, the inductance can be considered as independent of width as far as I remember ( or google it)