Downvoting correct answers because you do not like the question is not ever appropriate. If you believe the question is not off-topic, vote to close or move the question. That's what that option is for. If others agree with you, the question will be closed or moved. Otherwise you are downvoting good answers to questions that others think are fine.
Despite the opinion ubiquitous gives that this practice is the consensus on other forums, I have not seen it anywhere but here. The fact that we can find people who say they do it elsewhere doesn't mean it's the consensus and it certainly doesn't mean it's the right thing to do.
- Downvoting an answer sends a signal that the answer, not the question, is bad. The purpose of downvotes is to say (to the OP) that the answer is bad in some way. Standard practice that I have seen on StackExchange is to upvote helpful answers, leave answers that are right but not the best alone and downvote only when an answer is wrong. If you downvote a correct answer, the OP and others reading the question will likely think the answer is wrong and actually be harmed my the question, rather than helped.
- The line between "on-topic" and "off-topic" questions is blurry. That seems to be especially true in the econ stack exchange. Most questions here are either requesting help on homework (whether at the undergraduate or Ph.D. level) or are people who are not economists asking big picture questions. Either one could easily be viewed as on or off topic. If there is insufficient agreement to close the question, then penalizing people who think the question is valid and try to help is counterproductive.
- Downvoting answers will do little or nothing to improve the quality of new questions. Off-topic questions pretty much universally come from people who don't spend much time in the forum. Dinging people who answer questions you consider off topic will not affect the people who are the source of these questions.
- Penalizing good answers harms the community. The purpose of StackExchange is to help people. For this we need a community of experts answering questions. Trying to coerce experts into not answering certain types of questions by giving negative feedback to good answers encourages these experts not to participate on other questions. This behavior makes StackExchange a less productive place, to say nothing of being rude.
- Downvoting answers breaks the logic of moved questions. The most blatantly off-topic questions often are on topic elsewhere. If the question does get moved, as it should, then a downvote on a correct answer becomes especially inappropriate.
- Downvoting correct answers causes confusion. I came here because the OP (denesp) caused some significant confusion by downvoting an answer that is both correct and helpful to a whole class of problems. Because downvoting is an inappropriate response to that, there was confusion about the reason for the downvote--I figured that because the poster was continuing to have problems, he had either voted the answer down or someone who thought my answer was not explicit enough had done so. Downvotes mean something. When they are used for something else, confusion ensues.
When you downvote an answer, StackExchange suggests that you leave a comment in case the answer can be improved. This is clear evidence that a downvote is meant to indicate that the answer is bad, not that the question is off-topic. That's how it should be used.
I've seen people who do all sorts of negative things, like downvoting good answers so their answer looks better, appealing to their (non-proveable) credentials, and posting off-topic comments. The fact that these behaviors happen does not mean they are right nor helpful. Before doing anything on StackExchange (upvoting, downvoting, or answering), ask yourself if what you are doing will help someone learn more about economics. If the answer is no, don't do it.
============== Additional Context ==================
I posted this meta answer because of a problem this notion just caused. Someone posted a mid-level undergraduate econ question, and demonstrated that effort had been put in but they were stumped. In my opinion this is not off-topic but I respect those who think it is. I provided an outline on how to solve this type of question and some hints that, if followed, would have solved the problem and others of that form. The poster commented that they were glad for my help and learned a lot but my answer was wrong (apparently they made an algebra mistake while following my directions). My answer also got a downvote. To me this means either the poster downvoted because they thought I was the one who made a mistake or someone who didn't know how to solve the problem downvoted because they thought it was wrong based on the comment. I left a more explicit solution and told the poster not to say things are incorrect unless they know that they are incorrect.
In fact, denesp had felt that the question was too introductory and therefore downvoted my answer, causing the confusion. This type of behavior is not unknown among new stack exchange members, but for a high repuation expert to do it is strange. To have a meta discussion coming to an apparent consensus that this is desirable behavior is bizarre. I see that this meta question was initiated by denesp, who has also taken it to be a consensus view and is practicing it. I hold no personal grudge--I'd be happy to go get a beer with you IRL--but I do want to make it clear that this idea was introduced, promoted, and practiced here by you and does not represent normal expert behavior from what I have seen elsewhere on StackExchange. If StackExchange wanted a "penalize an expert for answering a question you think is bad" button, we would have one.