There is no simple, single number. It depends on a number of factors: which benchmark you're using for the comparison, whether the benchmark is multithreaded (and the P4 has HT enabled), and so on. Platform differences further complicate things: you can't just pop a Pentium-M in the socket for a P4, so issues like chipset and memory controller differences, memory speeds, etc factor in. And there are two generations of Pentium M -- the older chips with 1MB of cache and the newer "Dothan" chips with 2MB (the older chips run at lower clockspeeds, but the two overlap at the 1.5GHz mark).
That said, if you dig around you can find a few comparisons that try to minimize the confounding variables. One of the better ones is at
CPUID but note that it is somewhat dated now: it's only looking at the older 1MB cache PM. However, they do run some benches with the P-M, P4, and some other chips at the same clockspeed. In general, for synthetic benches (raw CPU speed), a P-M would seem to be almost twice as fast as a P4 at the same clock. So you could say a 1.5Ghz PM is roughly equivalent to a 2.8 or maybe 3.0GHz P4. However, once you look at "real world" benches and take platform issues into account (P4s have HT, run with faster memory, usually have faster HDs and better video processors, etc) the Pentium M doesn't have quite the advantage. Nevertheless, it's probably safe to say a Dothan can run with a P4 that is clocked a full GHz higher (unless you're talking about games, where the much better GPUs available to desktops vs laptops can make a huge difference).
