Hawkins has claimed that a substantial fraction of the long-term optical variability of many high-redshift QSOs is caused by microlensing. This is an interesting hypothesis and, as his papers show, there are many reasons to believe it. However, it doesn't hold up to a quantitative analysis.
The links below point to one file containing the complete text, all figures and tables etc.
As are most of my publications this is available in more than one form: