If it were not for Post hoc ergo propter hoc, we would need to attribute predictive capabilities of a global pandemic to the phenomenon of yield curve inversion.
You miss (amongst others) division of labour and specialisation / non-uniform skills. In your example the GDP might not be a good measure because every participant can stand on either side of each transaction, which makes it kind of a zero sum game. It is well known that economic interchange is not a zero sum game (citation needed, could not come up with a precise one). Extrapolating this, you would need to be able to stand on each side of all interactions that lead to the global GDP.
Your example even breaks down when you assume that the bread-baking neighbour is baking bread not only for you but for others, too. In that case economies of scale come into play which favour specialisation towards one baking all the bread and others (maybe investing into better lawnmowers and their handling skills) mowing the lawns.
I don't get how (re)investing should lower the profits? I mean they are buying assets and not burning the cash in the backyard. Are there instant depreciations on investments, lowering the profit?