The problem isn't so much about the rating for the luminaire (in 
particular the bulb holder itself - plastic or ceramic) so much as it is 
about the maximum temperature the lamp can tolerate within the confines 
of said luminaire.
 A tungsten filament lamp typically converts more than 95% of the input 
energy to 'waste heat' (infra-red and convected/conducted 'waste heat') 
and is designed to cope with the relatively high temperatures produced 
(circa 200 deg C). The luminaire's rating is based on these temperatures 
(60W for a plastic bulb holder, 100 to 150 W for a ceramic one - 100W 
rating being the limit due to other materials used by the luminaire and 
its level of ventillation).
 CFLs and LEDs both use electronic ballast circuits built into the lamp 
base which have much lower temperature limits (silicon devices max out at 
150 deg C and high temperature electrolytic caps at 105 or so deg C).
 Since the silicon devices aren't perfect lossless switching devices, 
they generate additional heat which requires their immediate environment 
to be a few tens of degrees lower than their maximum ratings in order for 
the heat to be conducted/convected away.
 Whilst good quality electrolytic capacitors don't normally generate much 
waste heat in a properly designed circuit, these too need to be operated 
in an environment several degrees lower than their maximum rating (the 
lower the better since this extends their lifetime hours rating 
significantly beyond the typical ten to twenty thousand hours at max 
rated temperature).
 I haven't seen any 'environmental temperature' ratings mentioned as part 
of the specification of these CFL and LEDs. No doubt the manufacturers do 
have such limits in the 'spec sheet', they just don't bother 'troubling 
the end consumer' with such pesky details. CFLs have a definite 'optimum 
temperature range' for the actual fluorescent tubes themselves (50 to 60 
deg - significantly higher than the 30 to 40 deg for conventional linear 
tubes).
 Allowing for the required temperature gradients between the electronic 
components used by the lamp ballast circuitry confined to the lamp base 
and thermal contact to the luminaire acting as a 'heatsink', I'd be very 
much surprised to see an environmental temperature limit much above 70 
deg for the lamp base of an LED lamp, considerably lower than the 150 to 
200 deg for a tungsten filament lamp.
 Rather conveniently, the 11W or so LED/CFL equivalent of a "60W" 
tungsten lamp offers enough temperature reduction in most cases 
sufficient to allow the use of such "60W Equivalents" in a 60W rated 
luminaire without undue overheating risk. Unfortunately, this is often an 
unknown parameter as far as most luminaires are concerned (no temperature 
versus wattage figures being readily available - it's down to the end 
user to take test measurements if they want to determine this all too 
critical factor in regard of the effect on LED lamp life).
 I'm facing a similar conundrum with an external "60W rated" porch 
luminare I recently fitted over our front door as a stop gap to tide me 
over the period between now and the appearance of 200Lm per watt security 
lamps finally appearing in the stores (the lousy 90Lm per watt lamps are 
about to be obsoleted any time soon - yet another 12 to 24 months later 
than Cree and Philips were promising would appear some 18 to 24 months 
later nearly 24 months back with their 300Lm per watt laboratory samples).
 I'm trialling a 20W CFL in the new porch light which replaces an 
original porch light I fitted some 30 years back but which failed and was 
sidelined by a 500W halogen security light over twenty years ago. I 
fitted the security light quite high up to give better coverage and less 
light spill. Unfortunately, whenever I need to change a burnt out lamp or 
even replace the whole unit (I'm on my third one this time round), I need 
to set up a three piece ladder to gain access and the novelty of 'working 
at height' has worn rather thin.
 I'm determined that the next security light will be a 200lm/W 25W LED 
that will last me out for the remainder of my days, hence the 'temporary 
porch light' to silence the XYL's moaning about trying to park up on our 
drive in darkness.
 The CFL is, as you could predict, rather slow in ramping up to full 
brightness at this time of the year (about 5 minutes run up time). 
Nevertheless, it does seem to reach full brightness.
 Since it's only twenty watts, the protracted run up period can be 
sidelined by leaving it switched on until bedtime. The fact that it seems 
to be surviving 6 to 8 hour sessions ok suggests that I might be ok with 
a modern 15W LED, especially if it's a 200 Lm/w type rather than an 81 to 
90 Lm/w type typically on today's shop shelves (even less 'waste heat' to 
be dissipated - the main benefit rather than just the higher lamp 
efficacy).
 The *real* benefit of the later 200 and 300 Lm/W LEDs isn't the 
reduction of lighting energy costs compared to the 67Lm/W of CFL and 80 
to 90 Lm/W of today's LED, it's the reduction of waste heat that has to 
be handled by our existing luminaires.
 The simplest way to demonstrate this benefit is to assume the 
replacement of a "60W 810 Lm" 10W 810 Lm LED (in reality, a 75W 240v lamp 
- 60W is the American 120v 750 hour rated 810 Lm standard) with a 2000Lm 
10W LED. For the same input wattage you get over a doubling in light 
output meaning even less of the input energy showing up as non-productive 
waste heat in the luminaire.
 The former 10W 810 Lm is emitting about 1.5W as useful light[1] leaving 
8.5 watts to heat the lamp and the luminaire, just doubling up the lamp's 
efficacy means 3 watts goes to useful radiation in the case of a 10W 1620 
Lm LED lamp, leaving only 7 watts as heat for the same 10 watt input 
power. If all you need is an 810Lm 5W lamp then the waste heat drops from 
8.5 watts down to 3.5 watts.
 The next generation of LED lamps will make a lot of existing light 
fittings (luminaires), that aren't suited to the current generation of 
LEDs, entirely adequate to the new task in hand, neatly avoiding any need 
to upgrade them for use with the currently available LED lamps.
[1] This figure is just a best 'guestimate' I've used to provide example 
figures to demonstrate the effect of improved luminous efficacy on 
luminaire (and hence lamp) temperature. 
-- 
Johnny B Good