Over in Google Plusville, +Levi Kornelsenmentioned the absence of a magic item economy in 5e. This is a core design assumption on WotC's part, and they haven't been shy about it. In most settings I agree with this choice, but in his post he's specifically talking about Eberron. That setting gets to be the exception, because it internalizes the magic item economy of 3.x and renders it into industrial fantasy - both low-end magic items (tools to make life easier) and large-scale magic items (lightning rail!) are essential to Eberron's feel. Large-scale items are created at the speed and inflationary whims of Plot, but low-end stuff needs support, and the definition of "low-end" scales by level. The ideas I'm presenting here are only for people who want to introduce that approach to money and the economy.
Levi was good enough to present five basic actions that he felt were critical acceptance criteria to anything that wanted to call itself a magic item economy.
Buying magic items from an NPC
Selling magic items to an NPC
This one is already covered in the DMG, though I'll review its handling of the task
Commissioning an item for an NPC to make - that is, outsourcing the Crafting action
Disenchanting a magic item (and presumably getting something that has a use)
Salvaging something useful from abandoned workshops
In 3.x and 4e, the game's default assumption is that you can turn an appropriately large pile of cash into the magic item you want; in this way you might look at your choices on magic items as part of your "build." Oh, sure, you still get magic items other than your optimal choices from adventures, and maybe some of those are things you want to use. Likewise, not all DMs support a magic item market. I was usually pretty uncomfortable with it, and tried to present it as something more than a straightforward transaction.
3.x and 4e did not - in any particular rule or guideline that I recall, anyway - deal with the merchant's profit margin. Maybe the DM includes one, and maybe the DM assumes that the list price in the DMG (3.x) or PH (4e) is intended to be the merchant's asking price. In 3.x, there's still room for the list price to include a profit margin, because that's how the magic item crafting rules work. In 4e, not so much: it costs exactly as much money to make something as the list price, plus time.
For 5e, the "list price" is explicitly a function of rarity, which is to say power level. The treasure tables are arranged so that rarity also signifies "number of these that exist in the world," but a one-off item that has not yet been duplicated is not intrinsically legendary or unique in a rules-terminology sense. Since economies don't make sense if production cost equals purchase cost, I've got to build in a surcharge. Magic items of uncommon rarity and higher are "big-ticket" in the overall economy we see in the PH and DMG - luxury goods by any definition of the word. Conveniently, it's easy to figure out 10%, 20%, 50%, 100%, and 150% surcharges on 500, 5k, 50k, and 500k prices.
Digression: On p. 135, the DMG presents cost ranges that are a little complicated to resolve against the fixed numbers presented in other rules, and suggests that a merchant might require a service from the PCs as part of the transaction. The PCs of my 4e game remember with some chagrin the magic item they received in exchange for cash and a service - the service turned out to be a quest that took them twenty-something sessions to complete, and the PC who received the item had to quit the game to move to Canada within two sessions of receiving the item (they got the item up-front, and were bound by contract to the quest).
I'd like to turn this into a full Downtime Action, in the same way that Crafting and Selling magic items are Downtime Actions. This needs some guidance to the DM as to what kind of surcharge would be appropriate, and maybe some way for the player to influence the outcome too. Dungeon World and Apocalypse World influence my thinking here, as I find myself phrasing this as a PBTA-style move.
When you spend some time searching a marketplace for a particular magic item, you can find a source for it, but they almost never have or admit to having it on hand. Choose two:
Your source gets it to you fast (1d10 hours for common, days for uncommon or rare, weeks for very rare or legendary)
Your source gets it to you cheap (default surcharge of 50%)
Your source gets you exactly the thing you want, with no chance of flaws or discovering that it's actually a cursed version of the item
If your source gets it to you slowly, add 2d10 to your roll. If you source is expensive, the default surcharge is 100%. If your source is not so reliable, things get... interesting. I have all kinds of minor-to-major drawbacks I am happy to suggest, from taking an extra d12 damage whenever you take damage of a particular type to suffering disadvantage on all rolls with (skill) to... well, you can read the Scroll Mishap and Potion Miscibility tables on p. 140 just as well as I can. I've never liked silly or humiliating drawbacks; I think thorny tradeoffs are a lot more compelling (and I can get all the humiliation I need some time that I'm not playing a game of heroic adventure).
I realize that most PBTA moves would have gone something more like "10+, choose two; 7-9, choose 1; 6-, screw you and the horse you traded for this cursed item." 5e downtime actions aren't usually dice-driven, unless they're d100-based, and I felt that this was a hard enough choice as-is. The thing that probably makes a lot of players uncomfortable here is leaving it entirely up to the DM whether "unreliable" means "drawback" or "outright cursed," but Dungeon World gets away with that kind of trust for the GM, so let's not assume that D&D DMs are less trustworthy.
The other complicated thing here, the part I'm not sure how to resolve, is at what point the PC pays various portions of the price, and is there room for identifying the item and renegotiating the deal if you're not happy with a flawed item? Payment should probably be half up front, half on delivery. No refunds, and the seller had better have some serious goons backing him up, or skip town after basically every sale. If they can enforce the "no refunds" thing, then okay.
Does the seller allow the PC to cast identify (or have someone else do it)? For a full-scale deconstruction of fantasy economics and the identify spell, Multiplexer on Critical Hits is doing some astoundingly creative work - even if what she's writing isn't the feel you want in your campaign, it's worth a read. Anyway, my concern is that identify unravels the interesting tradeoffs of the whole thing. On the other hand, it only protects you from cursed items; if the item you've purchased has an odd drawback, you've still paid half up front, so you either pay the rest of it to get something that is probably still okay, if imperfect - or you recognize a sunk-cost fallacy for what it is. Maybe you start over with different priorities, and maybe you don't.
Finally, the player-side predictability of getting exactly what you want in exchange for either more time or more money might not be ideal for some games. It does take mystery out of the process, though for Eberron that's working-as-intended. Still, if that's a dealbreaker for you, add a fourth choice to the list above, but still only let the player pick two: Your source gets it for you without breaking any laws or pissing anyone off. Congratulations, now you have some thorny choices (and maybe crafting it yourself starts to sound a lot better).
When you arrange the commissioning of a particular magic item, you can find someone with the ability to make it, as long as the Minimum Level to craft the item (DMG, p. 129) isn't more than two levels higher than your character level. Choose two:
Your source has the means to craft it faster than 25 gp per day (default 100 gp per day).
Not that it's particularly important how they're accomplishing this, but in Eberron it's completely plausible that a magic item, location, or NPC-class feature might be just as good of an answer as assuming that the NPC hired on a team to rush the work.
Your source is willing to make it for a reasonable profit margin (default surcharge of 50%).
Your source gets you exactly the thing you want, with no chance of flaws or discovering that it's actually a cursed version of the item.
Because, well, the only significant difference between the PC crafting it and an NPC crafting it is who spends the time, right? There's nothing wrong with tacking on a fourth choice here also - maybe your source is untrustworthy and reports your commission to the worst possible person (a crime boss, the Holy Office of the Inquisition, your bookie). Maybe for both buying and commissioning, you tack on more hard choices as the rarity of the item increases, or as other complicating factors present themselves. If the last person you commissioned to make something for you ended up in a gutter because you couldn't reach a satisfactory deal, you're definitely working with more desperate and untrustworthy people from there on out. If you've done a lot of favors for the right sort of people, maybe one of the options gets mitigated. Actions within the narrative take precedence over the guidelines that game rules represent.
Again, this is an examination of rules in print, not a proposal from me, other than one little bit at the end.
To sell a magic item of very rare or lower rarity, a character spends a random number of days (larger die value for higher rarity) shopping it around, and rolls an Intelligence (Investigation) check against DC 20. On a failure, you don't find a buyer and you spend a full 10 days. The good news is that you can offer around multiple items simultaneously, and spend time based on the largest time result.
The rarity of the item also modifies the d100 roll that determines the result. The table is set up to make sure that crafting magic items is not a way to make money. Only a d100 result of 91-100 results in a profit to the selling PC. Thanks to the modifiers from rarity, it's 20% likely for common items, 10% likely for uncommon, and impossible for rare or very rare items. Also, you're almost certainly breaking some sort of law or moral code, since the text describes the buyer as shady, and it's a no-questions-asked deal. The more likely results - 40% likely for very rare items - include getting 10% of the list price as the best offer.
As a result, the DMG's rules for selling items are for games where the trade in magic items is rare or highly restricted, and where the overwhelming majority of magic items come not from PC crafting, but from loot. In the setting that gave us the Artificer class, these rules are overly punitive. (Artificers shouldn't get to make permanent magic items for free, though, the way they do in the Unearthed Arcana article that presented them.)
I like that the DMG's rules for selling magic items include a cost in time, variable price point outcomes, and the possible complications of less-than-licit buyers. I think the solution is to make the "undocumented features" that I've proposed above a widespread and recognized part of the magic item market - so buyers get used to paying a premium when they buy an item that carries the seller's personal maker's mark. That gives them a door they can come kick down if the item turns out to be cursed or otherwise defective. A 20-50% markup should be a reasonable starting point for a PC's sales negotiation, if they made it themselves. Let's assume that a magic item's maker's mark is prohibitively difficult to fake, or doing so carries repercussions so severe that no one does it and lives to tell the tale.
I don't remember disenchanting magic items to any benefit as a part of 2e or 3.x, though if I've forgotten something there, I'm sure the internet will be quick to admonish me. In 4e, demonstrating some of the influence of World of Warcraft, ritual casters can disenchant magic items of their level or lower (4e magic items have a level, 1-30), turning the item into 20% of its list price in residuum, which is a powder that is a universal reagent for all rituals, including the Enchant Magic Item ritual. The 20% number isn't a coincidence - that's the same return you get in gold for selling a magic item. 4e's economy is transparent, but awfully shallow.
I recently wrote about a work-in-progress to deepen 5e's crafting rules, but I haven't yet given any consideration to how I'd handle disenchanting there. In brief, that system supplements the existing crafting rules with optional components. I see two different options that I like:
Disenchanting generates 1d3 components, taken from that item's formula.
There's a lot of loss here, but potentially less than the Selling Magic Items action. It avoids giving you a simple alternate currency.
Disenchanting generates a component that doesn't come from any other source. Each rarity generates its own quality of [component]. So maybe 1d3+1 ounces of [rarity] dust. Some formulas and spells call for this dust; it just happens that destroying magic items is the source.
I'm specifically avoiding presenting that as "100gp of arcane dust," because fixed values take people away from thinking of an economy - it feels less like an item in your inventory and more like a mere number on a page.
Even more than other items in my crafting system, I like the idea that you use this substance to accelerate the creation of new magic items. Maybe it is a universal accelerant.
The substance you gain from disenchanting a cursed item is itself cursed, but can probably be used for some really interesting, underhanded stuff as long as you're prepared to remove its curse on you after use.
The second option is a lot more meaningful for anyone who doesn't eventually adopt my alternate crafting rules, once they're complete. Ahem.
I would probably present disenchanting as another downtime action. I don't know what the compelling choices of a PBTA-style disenchanting move would be, but giving very rare and legendary items a chance to draw unwanted Outer Planar or divine attention has a certain appeal. (Especially for the DM. You jerks just pulped this setting's equivalent of the Shroud of Turin so you could get something with bigger pluses. Good luck with that!)
Scavenging a Workshop
This is really taking me back to my days as an MMO game designer. My advice here is simple: once I eventually release my alternate magic item crafting rules in full, use some of the treasure tables in that work to generate treasure that the PCs gain from scavenging a magical workshop.
The DM rolls on the treasure table(s) 1d4+1 times. This treasure table is weighted to always generate treasure - "no treasure" is a very rare or impossible outcome.
As a group, the players roll Intelligence (Investigation), Wisdom (Perception), and stat + relevant tool, each against DC 20. For each success, the DM rolls on the treasure table one additional time.
Variation comes from having apprentice, journeyman, master, and grandmaster workshops as unique treasure tables. You can also treat a single large foundry as two adjoining workshops, if a larger volume of treasure feels right to you.
I've tried to avoid supply and demand in creating these rules, because I find that that's a bottomless pit of complexity for an infinitesimally small return in verisimilitude - and also I am not enough of an armchair economist to make it remotely credible. WotC plainly made the same decision for 5e as a whole.
The selling rules are by far the most dangerous thing here, since there's no 3.x-style XP loss to discourage an endless profit loop. Once you can craft magic items and sell them for a profit - even just a net profit over multiple sales - a character that is strictly motivated by his bank account stops adventuring and sticks with making common and uncommon items, and that's obviously bad for getting any fun to happen at the table. Stop playing that character and introduce someone who has some Ideals and Bonds, will you? (Or the magical Mafia thinks you have a real nice workshop, wouldn't it be a shame if something happened to it...)