Measurements for future outdoor cellular systems at 28 GHz and 38 GHz were conducted in urban microcellular environments in New York City and Austin, Texas, respectively. Measurements in both line-of-sight and non-line-of-sight scenarios used multiple combinations of steerable transmit and receive antennas (e.g. 24.5 dBi horn antennas with 10.9° half power beamwidths at 28 GHz, 25 dBi horn antennas with 7.8°half power beamwidths at 38 GHz, and 13.3 dBi horn antennas with 24.7° half power beamwidths at 38 GHz) at different transmit antenna heights. Based on the measured data, we present path loss models suitable for the development of fifth generation (5G) standards that show the distance dependency of received power. In this paper, path loss is expressed in easy-to-use formulas as the sum of a distant dependent path loss factor, a floating intercept, and a shadowing factor that minimizes the mean square error fit to the empirical data. The new models are compared with previous models that were limited to using a close-in free space reference distance. Here, we illustrate the differences of the two modeling approaches, and show that a floating intercept model reduces the shadow factors by several dB and offers smaller path loss exponents while simultaneously providing a better fit to the empirical data. The upshot of these new path loss models is that coverage is actually better than first suggested by work in ,  and .