Speculation at places like Goldman Sachs isn't just about having better future-predicting power, though. It's also about sitting on buckets of money such that you have a lot of risk tolerance. Then people pay you to take their risk, and you win on the average (in terms of profit) and they also win on the average (in terms of reduced risk).
Starbucks, being as big as they are, may simply have felt that they were big and well-informed enough that the risk of self-insurance made more sense for them... a decision which is easy enough to snipe at ex post facto, but may still have made more sense.
In value-at-risk terms, Starbucks' hedging requires a negative amount of capital: they need more cash on hand if they refuse to hedge. With a given capital structure and no coffee hedging, there's a possible coffee price X that would wipe them out. With hedging, they can place an offestting bet that coffee will rise; then they're indifferent to that price fluctuation. That's one less way for them to unexpectedly lose money, so their results in the aggregate would be less aggregate.
Plus, the buckets of money thing is a canard; Goldman has to earn a percentage on their capital, just like anyone else. If they have 10X as much equity as Starbucks, they need to earn 10X as much net profit to get the same return. And ceteris paribus, that means taking 10X as much risk.
Starbucks, being as big as they are, may simply have felt that they were big and well-informed enough that the risk of self-insurance made more sense for them... a decision which is easy enough to snipe at ex post facto, but may still have made more sense.