Going back to the original post in this thread...
They pull their own weight under some circumstances. The former slightly more often than the latter. The higher the planet quality, always and hereafter: PQ, and population, though potentially only to a point, and the less the current overall planetary +economy, hereafter: econ, bonus, the more likely they are going to be to be buoyant. (If someone wants to take the time to make a formula for when to build each "+econ improvement with maintaince", be my guest. The three factors you'd be integrating, I believe, would be PQ, planetary +econ bonus, and the
actual tax revenue. The tricky part is you have to figure out the intricacies of the actual tax revenue formula. Meaning you have to track out how morale influences tax revenue.) If you end up being afforded the freedom to buy embassies early on, it is probably worth building them purely on their economic merit on major--high PQ, high population--worlds. When to get rid of them, if you aren't worried about influence, is another matter entirely.
On to the second part, the revenue
soft-cap is applied after the +econ bonus is factored in. After the
final revenue from a planet exceeds double the PQ there is a minute penalty to income; however, once you reach 4xPQ, the penalty to further income is very noticable.
Everything can potentially reach a point of critical mass, provided your overall +econ is high enough. A point where your income from every planet is already a decent bit over 4xPQ, and every percentage you raise taxes amounts to just a measily 1bc or maybe even nothing. At this point, it could actually net you more money to destroy +econ improvements that have upkeep. It could also net you more money to destroy morale improvements with upkeep; even if you have to lower the tax rate!
If this helps or not, I don't know. If you get to this point in the game, good money says you've won. (The pun wasn't intentional, I swear.) It's not often you get your economy to its apex and find the AI still in a position to be a challenge. (Although, I believe it is possible to get to this point with individual worlds, with economic capitols, much earlier on, if you are looking to skim a BC or two off the top for the duration. Be wary of the loss of econ anomalies or elections, though.)
What I want to know is how the AI manages to have worlds produce 8x their PQ.
To clarify something from one of the replies that may be misleading in the context of the original quandary, the only slider that will effect your
revenue is the tax slider. The amount of money
generated per turn is unrelated to government
expense or the mil/soc/tech sliders--they affect spending rather than revenue, which only affects
net gain, which still has no bearing on whether that %econ mod is pulling its own weight. (Except indirectly in that if you run into the red your morale will suffer; and, morale is a component in actual tax revenue.) Increasing social spending does not increase your base tax pool, only your ability to produce econ modifying projects...
You know, Truman was onto something when he said all he wanted was a one-armed economist.
But on the other hand, we should really be more concerned with net gain than increased revenue. So... now we must decide if, even if it nets no increase in revenue and is, in fact, a drain, which should be true most of the time, the neural net is worth keeping because of the research boon. Keep in mind only 1/3 of the 25% bonus, or 8.33%, it adds is actually "free." The other 2/3s of the bonus are simply added to your basic pool, which you pay for. Yet, this does mean you could conceivably set your research slider lower and decrease government spending a tiny amount to achieve the same Gross National Product--or, perhaps, Gross Imperial Product--for that slight drop in government expenditure. This would be "benefiscal" only when 8.33% of your base research pool exceeds 5bc/month.
One more point to keep in mind, though. 2/3 of the boon from research anomalies is bonus, free, production. So if you have developed several research starbases, it would probably be more cost effective to destroy your research labs and simply increase the research slider and government expenditure.
There are also, conceivably, situations in which the fusion and antimatter power plants would be worth scrapping. That would be if your base production is capped and your bonus production is significantly in excess of 3xPQ (the soft-cap) and would remain so without them; this, AND your revenue would still be over 4xPQ without their economy bonus. It seems unlikely to me, though, that you would fail to find more pragmatic manufactoring improvements to remove first. I've never looked very closely though.
If you are running at 100% expenditure and still in the green, neural nets and power plants are always worth it.
Ask a silly question; get a silly answer.