Should I avoid local journals?

I'm only going on my own experience here but I believe:

Location of the journal should only be taken into consideration after some other, more important, factors.

1. You should choose a journal that is a good fit for the topic, length and style of your paper.
2. You should pick a journal with at least one editor who will know roughly what you are talking about.
3. You should aim for as good a quality journal as is reasonable for the quality of the paper (this will most help your career).
4. There may be certain journals you just prefer to avoid (obviously junk/scam journals, but also eg. many mathematicians avoid Elsevier).
5. You need to consider any publishing costs (eg. for open access journals) and any rules you must follow (eg. UK open access requirements).
6. Where your field allows, avoid sending multiple papers to the same couple of journals (which could suggest one or two editors happen to like you).

Once you have a short list that meets these requirements, you could reasonably choose a local journal in favour of a non-local one. I think the other requirements will avoid the suggested negative impacts, when your work is viewed as a whole.

A couple of reasons publishing in an X-land journal (like Math.Scand.) could make sense:

• You'd like to commemorate and emphasize the fact that you were a postdoc in country X;
• You're interested in getting another job in country X;
• Your work is of somewhat general mathematical interest and not narrowly focused within a certain mathematical field (like that of the Journal of Topic).

Generally, a good criterion may be whether your work is of interest to editorial board members of the X-land journal (and conversely whether their work interests you), or whether the same is more true of Journal of Topic.

One consideration is if you want to help such local journals. If publishing your paper in a local journal will help maintain or even increase the quality of mathematical publications in that country, and you would like to contribute to the flourishing of mathematics there, then that would be a perfectly valid reason to do so, even if it means passing up on a more prestigious journal. It all depends on what your goals are.