What are the potential problems of matching?

It is my pleasure to answer!

Thanks!

Even though there have been statistical methods to match several confounders at a time, scholars found that the more variables included in matching, the larger the deviance between a randomized sample and a matched sample would be. Therefore, we must carefully design the matching strategy and select the matching variables. We should also pay attention to the overmatching or post-treatment bias. Overmatching refers to a match for a mediator affected by the treatment. If the mediator itself is stratified, an obscured relation between the treatment and the outcomes would be induced and thus causing statistical bias.
Matching may also increase sampling bias in decreasing the external validity because the characteristics of matched data become more similar to the exposure cases than the general population.