When you scale the recipe, are you using the same equipment profile? My guess would be that this is the case by the way you described your issue.
BeerSmith works by setting the total amount of sugars delivered to the fermentor (Brew House Efficiency or BHE) in your equipment profile. As you scale your recipe using the same equipment profile, having the same process losses for both batch sizes, the program configures the recipe based upon the brew house efficiency you have in your equipment profile and adjusts the amount of sugars extracted from the grain based upon this number. Doing so for a smaller batch means that the proportion of process losses (loss to trub and chiller, dead space in mash tun) becomes a greater percentage of the sugars which need to be extracted. BeerSmith handles this by increasing the sugars extracted during the mash, thus a higher mash efficiency.
In your case, your original recipe is set for 5.5 gallons at a 60% BHE, which means that the amount of sugars extracted during the mash is equivalent to 70% of the sugars available. if the target gravity at the end is 1.060, you have 5.5 gal at 60 points or 330 gravity (sugar) points into the fermentor. This is equal to 60% of the sugar points available in the grains, so your grain bill has the potential of 330/60% or 550 points of sugar. Your mash efficiency is 70%, so your mash delivers 70% * 550 = 385 sugar points and your process loss is 385 - 330 = 65 points of sugar.
When you scale the recipe to 2.75 gal with the same BHE, you are looking for 2.75 * 60 points = 165 sugar points. Your process loss is still at 65 points so your extracted sugars from the mash is 230 sugar points. The scaled grain bill will deliver 165 points / 60% or 275 points of sugar. Since you are extracting 230 points of this, your mash efficiency is 230/275 or about 84%.
Normally, I find that the mash efficiency in my different processes changes only slightly between different batch sizes and my BHE takes the hit (or increase). So when I set up my equipment profiles, I made one for 20 liter batches, one for 16 liter batches and one for 10 liter batches (the three most common sizes I brew) and each has a corresponding BHE which takes into account the effect that the process losses have on each of the batch sizes.