Most of the time, it's best to assume the summation is zero, but it really depends on your application. If a sum like this pops up somewhere for some reason, you should interpret it in context to see if it makes sense at all.
In some cases, for instance in finite calculus, it might be more useful to interpret it slightly differently. It is well known that
[tex]\int_a^b f(x) dx = -\!\!\!\int_b^a f(x) dx,[/tex]
and since integrals are sort of like infinite sums, perhaps sums should work similarly. If [tex]a < b < c[/tex], we have
[tex]\sum_{k = a}^b f(k) + \sum_{k = b + 1}^c f(k) = \sum_{k = a}^c f(k).[/tex]
It might be nice to extend this to cases other than [tex]a < b < c[/tex]. For example, we would have
[tex]\sum_{k = a}^{b - 1} f(k) + \sum_{k = b}^a f(k) = \sum_{k = a}^a f(k) = f(a),[/tex]
which would imply that
[tex]\sum_{k = b}^a = f(a) - \sum_{k = a}^{b - 1} = -\!\!\!\sum_{k = a + 1}^{b - 1}.[/tex]
We would then have
[tex]\sum_{k = a}^{a - 1} f(k) = 0,[/tex]
but
[tex]\sum_{k = a}^{a - 2} f(k) = -f(a - 1).[/tex]
Which interpretation you should choose is up to the application.