Even in the ideal case, where your central planners are competent and moral indefinitely, centralization is a swingy strategy, for countries or companies. Every investment is ultimately a bet, that you have correctly predicted the future needs of your citizens or customers. Centralized control, of states through an autocrat or central committee, or of markets by monopolies or megacorps, ultimately amounts to placing fewer, larger bets. When the bets are good, it is very very good; when the bets are bad, it is ruinous.
Decentralized decision making is ultimately conservative, risk adverse: it is never great -- a lot of people will be wrong, so things will never go as good as they might. But it is seldom terrible: even when context shifts and the underlying conditions everyone was betting on change, the wide diversity of bets means that at least some people won't have been completely wrong, maybe for stupid reasons, but it can prevent a complete catastrophe.
(The difficulty, of course, lies in socializing the losses and gains, so that you can benefit from decentralized decision making without it ultimately reverting to a de facto central committee through successive rounds of winnowing those who guess wrong, until the small number of people who had so far guessed right can begin guessing wrong at scale.)