You have Maximum Render Quality enabled. With MRQ, an export can take about 6 times longer than it would without. Your processor is pretty dated (read: slow) in terms of the current processor line, so I wouldn't be surprised if your MPEG2 exports were approximately 1:1 without MRQ enabled.
So, the math: if a real-time, 1:1 MPEG2 export was going to take 2.75 hours (the length of your sequence) without MRQ, enabling MRQ would take 2.75 x 6 = 16.5 hours. Your 16-17 hour range is just about right. That's probably why using Encore to transcode is faster and lower quality.
If you don't have MRQ checked... well, I can't think of another reason.
If I understand it correctly, your time-line is 2:45 (hh:mm) long and takes around 17 hours to export. Correct?
That is pretty good for a Q9450 with 8 GB RAM. Those systems are on average around 15 - 20 times slower than a fast system with MPEG2-DVD export, so it would take a fast system around 1 hour. Now that looks like a perfectly acceptable figure. You have to keep in mind that with hardware MPE you can only compare that with CS4 export times with MRQ (Maximum Render Quality) turned on, and my guess is that you are comparing it to CS4 times WITHOUT MRQ.
Have a look at the PPBM5 Benchmark and notice where the Q9450 systems are. They are all pretty close to the bottom of the list.
Seeing those results, one can argue your statement.
so it's not a particularly slow machine.
Of course much depends on your particular time-line, but if I take my benchmark results (with a lowly i7-920) versus another Q9450 system, I guess that my export times with your time-line would be around 1:06.
Maybe not the reaction you were looking for, but I can't change statistical results.
@ Colin: The OP never said he has MRQ enabled. Where do you get that from? He said he had hardware MPE enabled. And with his timeline, setting MRQ on would not influence his export times, since there is nothing in his single track time-line to profit from MRQ and it is all handled by the MPE engine. At least in this case.