First, a declaration. I'm NOT a theorist. I'm an experimentalist (and proud to be one, damn it!) :) So one can say that my take on this can easily be inaccurate and based on superficial observations. However, having looked at it for many, many years, and talking to many theorists for quite a while, I think I have a view that isn't too far off for someone who isn't one.
This thought came up because I keep coming across students just starting out (some even still in high school) wanting to be theoretical physicists. Neglecting the fact that many of them have a mistaken idea of what "theoretical physics" is, I think that most (if not all) of these kids do not realize just how difficult it is to not only graduate with a PhD in physics, but also having the chance to actually be employed as a theorist.
Let's start from the most obvious: there are more experimentalists than there are theorists working in physics. Regardless of the field of study (outside of string/etc, I mean), experimentalists tend to outnumber theorists, often by a lot (see, for example, condensed matter physics and accelerator physics). So already the "phase space" for employment does not look very appealing to theorists.
Experiments and experimentalists tend to bring in more funding to a particular institutions. Now granted that in many of these funding, both theorists and experimentalists are involved. But even in such situation, the funding proposal tends to have more experimentalists than theorists. This is also one reason why there are more employment for experimentalists than theorists.
A project may get by without a theorist, even if it requires theoretical work. More often than not, an experimentalist can pick up the task that a theorist does, but it is more daunting for a theorist to do an experimentalist job. I'm not saying that this is true all the time, but in my experience, I've seen experimentalists do theory (especially in high energy physics), or use tools such as packaged software to perform theoretical simulations (especially in accelerator physics) without officially needing a theorist. Now, they may consult a theorist on site, but such tasks are often done by experimentalists without needing to employ another theorist to do such jobs. I haven't seen the reverse yet in my experience, i.e. group of theorists taking on jobs done by experimentalists, without needing to hire or have the presence of experimentalists. In fact, last time a theorist got close to my vacuum components, he ruined it by touching a clean part with his bare hands!!
Finally, the competition for the few positions in theoretical physics, be it in Academia or other institutions, is fierce! I do not envy the theorists at in this aspect. Because of the small number of positions available, even the good ones will have a tough time finding a job in their respected fields. In fact, if you did not come from a top-tier school, and your mentor isn't a "brand-name, world famous theorist", there's a very good chance that you will not get accepted to such a position in a good institution in your field. I think that the "pedigree" factor is a lot more prominent for theorists than for experimentalists, mainly because of such limited job opportunities. There are just too many outstanding candidates. What this means is that newly-minted PhDs from less well-known schools or supervisors seldom have a chance for employment as a theorist in their fields, leading to many to go into other fields or even outside of physics completely.
I'm sure there are many exception to what I've just described. But I believe that, on average, this is what is going on based on my years of observation. So, are you a theorist? Did I get it right, or was I just blowing smoke?