I don't know whether this is too obvious to be worth pointing out, but money is just token for goods of inherent value. If you give money to a charity, you aren't directly increasing net global wealth, you are just moving resources from one area to another.

Inherent value? Most models see value as subjective, which allows for gains from trade. Charity, in particular, can increase wealth if the donor and recipient are both more satisfied afterwards, as swapping an apple and an orange can increase wealth if the traders are both more satisfied afterwards.

It also needs to be noted that jobs have social value apart from what one does with one's salary. So, if you work as a heart surgeon, and you save a bunch of lives, then you're contributed to society, even if you don't give your money to charity.

Part of the argument for earning to give is that high-compensation fields are generally restricted entry; only so many people work on Wall Street, or as doctors, or so on, and so the more altruists seek to enter those fields, the more resources that those fields control will be directed towards altruistic purposes. This seems very dependent on the field- the number of working doctors will not be altered by a marginal student deciding to apply to medical school, but the number of FAI philosophers may be altered by a marginal student deciding whether or not to work as one.