Sussex assigned me to a mentoring circle, and our homework from the first meeting was to ask people we know who have recently completed PhDs for what they wish they knew when they started, to share with the group. Here is the really excellent advice I got from a couple of friends, both with recently(ish) finished economics PhDs and now with great jobs in applied policy research. Further tips gratefully received!
My views are far from standard, but here's two ideas;
First, be McKinsey about it, never forget about what the deliverable is. Project manage yourself. Get 2-3 finished PhDs from the library and work out exactly what you need to do over the next 3-4 years (choose book style or three papers), get a really good feel for the what the end product looks like. Even if your three ideas are only slowly becoming specific enough you could work out the headings and sub-headings of the output, perhaps get the ball park literature review down in final format so something is already delivered.
Second, be clear on what you want from the PhD. I'd argue it's mostly a signal. But if you can produce useful published work go for it. If not compromise and aim for one stellar output from 3 papers and then a satisficing strategy can reduce what could be enormous opportunity cost. Remember the job route or post-doc route need quite different attributes. Balance delivering the minimum output with earning, teaching, network building (work or academic), and publishing (not necessary but wonderful. A PhD only needs to be potentially publishable with more work/ a post-Doc; it's intended as research training.And from A:
1) Use BibTex or another bibliography program to keep track of all of your papers as you download them. I did do this, and it has been a life saver. I also started a spreadsheet of all of the papers which I read, but I started that too late. It is really useful to note down things like context of study, data used, main method, key finding etc. I absolutely hate doing literature reviews and answering the question "where does your research fit into the literature", especially after I've assured myself that what I'm doing is worthwhile.
2) Read the abstracts in the top 3 or 4 journals as they come out. And also JDE, EDCC, WBER and a few other field journals, just to get an idea of what people are doing.
3) Don't lose sight of the bigger picture. You may not have this problem at Sussex, but I found X was SOOOOO obsessed with empirical identification of effects and fetishizing causality that I stopped thinking about the bigger picture and started focussing on why my paper was bad and couldn't be identified. Obviously there has to be some balance between the interesting question and the rigorous empirical identification, and in some cases these things are complementary but don't let yourself get pulled too far down that rabbit hole.
4) Keep perspective and stay confident in your research. Obviously this is easier said than done, but so much of academic work is smoke and mirrors. People bullshit a lot and inflate how important their work. This can lead people to put down other people's work, which I think is totally unconscionable. Anyways, I don't know exactly how I would have done things differently on this point, but I never really got over the confidence thing (you can ask A about how many times he had to pick me up). I think one thing you have to do is develop a thick skin; if people are critiquing your paper, it isn't because they don't think you are a good researcher. Also, never, ever read Econ Job Market rumors. Being accepted to a PhD means you are good enough to do good quality research that people will be interested in. It may not be published in the AER or QJE and you may not get a faculty position and Harvard, but you should remember (and I should have remembered) that this isn't actually the point.
5) Present your work early and often. This relates to the previous point. I didn't present my work soon enough and you really do get a kick out it. One of the big challenges I found was that I always felt my work was worthless and stupid, but when you present to other people they almost always see the interesting and good parts of it. This helped me a lot to stay motivated and think of new ways to approach my papers.
6) Do not, whatever you do, lose contact with your supervisor. I had long spells where I just disappeared. As with any bad thing, I was fully aware of what I was doing, but couldn't muster the willpower to break out of it. Most often your supervisor doesn't care that you haven't done what you were going to do, but they will help you get back on track. I fell into the trap of not having done enough, so thinking "If I just spend one more week on this, it will be good enough to take to my supervisor", and then postponing my meetings. Don't do this, because every time you postpone, you build up the pressure to create or do or present something to your supervisor that is even better than what you currently have. Owning up to not having done anything and getting yourself back on track is WAY WAY WAY better than letting things slide.
7) Check out the sites that give advice for doing PhDs, presenting and writing academic work. I really liked John Cochrane's advice, but there are many others. I found them really helpful in preparing my slides and papers. Also, if you find articles that you like, copy their model for presenting your research. I did this on a few occasions. It works.
8) If you are doing data work, spend time figuring out the best way to store files, proper etiquette (that isn't the right word) for writing do-files. It will save you lots of time.
9) Get involved with research projects with other people, either as an RA or as a co-author. I didn't do this enough and I really thrive off of working with other people. Being a RA also helps to get you familiar with data-sets and opens up questions based on other people's work that you might find interesting. I think A benefited a huge amount from being able to work on the Y team; not only did he have a great social group, but he also really got to know the data and develop his own ideas about what to do.
10) Find yourself a supervisor who you will work well with. Some supervisors are hands-off, some are harsh, some are supportive, some are really anal and organized. I think this is probably a key decision (not sure why it is at the bottom of the page) but it is really important.
11) Write down, not up. Paul Klemperer gives this advice to people. I think it is really helpful. Basically write down your ideas and models and empirical findings so they are on paper and you've expressed them (or tried to express them). But don't write up into a paper until you have the argument and outline ready to go.
12) Keep making sure you love to do research. It is amazing to be funded to do research, so make sure you enjoy it. I loved all the other stuff around doing a DPhil: teaching, traveling, running a survey. But in retrospect, I don't think I loved the solitude and focussed effort that goes into polishing and re-polishing and being exactly right about something. Part of me thinks I should have quit early on, but I don't think I had the balls to do it. I don't know that I would have been happier, but there you have it.