The two main power quality issues with SMPSs is the fact that they are generally direct rectified (the first components are the diodes), and usually incorporate a filter between the mains and the rectifiers.
The effort and expense put into the filters is purely done to stop the high frequencies generated by the supply reaching the network. If such high frequencies were to reach the network they could radiate and cause radio frequency interference which is not permissible by EMC regulations.
These high frequencies are usually not high power and are therefore not directly responsible for any power quality issues. The filters do assist the SMPS by reducing any differential mode transients that may occur on the input (due to the design, common mode transients do not affect the power supply unless they exceed the breakdown voltage of the transformer), but this reduction is not much and such filters must never be viewed as protection.
A further problem with such filters is the 'leakage' current they introduce between Live and Ground. This is as a result of the Live being a different potential to the Neutral with respect to Ground. The bottom capacitors, as shown in the above circuit, have little current flowing in them as Ground and Neutral are effectively tied together.
The Live, wobbling about at 230VAC, causes a leakage of 73ľA/nF. Now this may not sound a lot but the typical capacitors used will total about 4 to 5nF on each input. This will result in approximately 340ľA leakage current per device. As there are typically 3 devices on the desk, the total leakage will be close to 1mA. It can now be seen it does not take many desks to create a substantial leakage current.
What the filters do not do is 'correct' the current curve of such power supplies, or any power supply for that matter where rectifiers are succeeded by 'smoothing capacitors'. The main difference between SMPS and conventional transformer fed power supplies is the current curve is 'softened' by the transformer.