See this section of the official scene making guide:
Vornne said:
Target stock counts (value 2 of the stockpile scene prop) for top tier things like plate armor, great swords, long axes, siege crossbows, war bows, and armoured horses should be set at low values such as 1 or 2, then the next tier down of weapons and armor at about 5 - 10, the mid tier stuff about 10 - 20, then clothes and peasant tools something like 20 - 50 (values might need to be adjusted, but the relative structure should remain). This is so players don't have as many problems getting necessary tools and changing their clothing style when starting out, and when stocks are low there will be a steady reward for craftsmen to bring a load of resources to restock; mid tier items should have a medium target count so that some will generally be in stock when a fighter wants equipment, but to equip an army requires engineers to produce more stock; but so troops heavily armed with valuable weapons need to work with factions or serfs and engineers to get their desired equipment made, with a large reward only if the stock is actually being bought and used, not encouraging the crafters to just make great swords for the gold reward. The more the current stock count is below the target count set by the scene maker, the higher the crafting reward: 0 stock always gives the highest possible reward for that item - a combination of 100 for each skill level required to make the weapon and 20% of the item price, the same as the amount lost when selling - and stock counts in between 0 and the target give a proportional amount of the reward; the refunding of the default resource costs remains unaffected by stack count.
Specifically responding to incorrect assumptions: a stockpile target of 0 basically disables crafting rewards for it (since the current stock must always be greater than or equal to the target), only rewarding the base prices of the resources consumed (nothing to do with the crafted item's price). Target values of 1 and above always have the maximum crafting reward at 0 stock scaling back to 0 extra reward when the existing stock count equals the target; the crafting reward calculated by averaging the alternate skill levels necessary rounding down to integers and multiplying that by 100, then adding 20% of the crafted item's price, then multiplied by the difference between the target and current stock, and divided by the target amount.
For example, pw_buy_red_gambeson with 15 in value 2: starts the mission at 15 stock count, a tailor crafts one for only the base refund of 286 (linen cloth default price 220 + leather piece 66), then 5 soldiers buy the armor and the tailor makes another, getting 286 resource refund plus 77 crafting reward (requires 3 tailoring or 2 engineering skill, averaged rounding down is 2, multiplied by 100 for 200, plus 20% of the item price which is 92, then multiplied by 4 (15 - 11) and divided by 15: 292 * 4 / 15 = 77). The maximum return at 0 stock would be 578, a crafting reward of 292.
pw_buy_long_axe_a with 3 in value 2 and a price multiplier in value 1 of 140%, changing the stockpile price from 3120 to 4368: after mission start two footmen buy axes, then an engineer crafts one, the base resource refund amount being 671 (wood pole 221, iron bar 450), and the extra reward being 915 ((skill 5 * 100) + (price 4368 * 20%) * 2 / 3) for a total of 1586; then crafts another getting 1128 (crafting reward reduced to 457 since the count is nearer the target), and then if the engineer keeps crafting (without anyone buying to bring the stock below the target again) he will only get the base refund amount of 671 each time, the same as if he sold the resources to a default export stockpile (but more than if selling to a resource stockpile, since the tax is not subtracted). The maximum total return at 0 stock would be 2044, a crafting reward portion of 1373.
pw_buy_warhorse_steppe with a target of 1 in value 2: base refund of 2597 (saddle 459 + horse armor 2103 + wheat sheaf 35), with the extra reward of 1434 (3 * 100 + 5670 * 20%) only when there is no stock (since the target of 1 has only one value below, 0). Result: 4031.
As you can see, expensive items give more crafting reward as well as the base resources generally being worth more; so even though they mostly take longer to craft, people would be inclined to only produce those types for the money, causing a lack of basic tools and items accessible to lower classes; this is why I suggest generally making the target counts of more expensive powerful items lower, down to about 1 (0 only if you want to disable the crafting reward), so there is incentive to keep them stocked but not to flood the market with hundreds of items at the expense of stocking necessary tools.
The base refund amount might seem high in some cases, but otherwise crafters would have to make a loss compared to selling the resources at export stations. If item prices are dramatically adjusted overall, this might make crafting generally less or more attractive; I have noticed a tendency to increase the price of iron which requires an increase in the price of leather and the price of flax, which requires an increase in the price of food, which might cause a perceived need for an increase in the iron price, ad infinitum, all in the name of keeping the resource gathering processes comparably efficient; but then the crafting system becomes less and less profitable since the resources are worth much more than the final results. I would suggest that it might be better to streamline certain gathering or processing systems by moving them closer rather than increase prices dramatically - but only when done slowly and carefully, one thing at a time, to try avoid causing unstable "feedback loops" of players crying "buff this since you buffed that!", and trying to avoid for secondary effects that might not be desirable, such as flooding the markets with weapons rather than clothes by placing iron mines easily accessible to forges and full price export stations.