Now I should be clear here: dealing with costs in non-monetary ways doesn’t make those costs go away. Someone, somehow has to bear the costs, regardless of if the state pays in grain or coin or tax remission or simply makes someone do it for free (in the latter case, the forced laborer is bearing the costs). In all of these cases, labor still has to be taken out of the civilian economy and it has to be subsisted while it does something military in purpose, be that soldiering itself or providing for military capital. Just because something isn’t paid for in money does not make it ‘free.’

However, it is also the case that the cash revenues of many states are both really complex and often quite limited. The thing to understand is that these are generally traditional polities with tax regimes that are also customary and traditional, which is to say that the ruler often has very limited latitude to simply change the system without triggering intense resistance. As a result, rulers often focus on developing revenues in the areas where they do have substantial latitude, even if those areas are smaller parts of the overall economy (remember: most of the economy is in farming).

[...]

A lot of tax systems, when one looks closely at them, have these sorts of quirks. Roman taxes were, for instance, divided into two categories: tributum (a property tax based on land) and the vectigalia, which covered a wide variety of state revenues from things like renting state owned land or state monopolies (as on silver mining). Rates of tributum outside of Italy (where the tax wasn’t collected after the 160s, since the whole point of having an empire is to make someone else pay taxes) were often set by truly ancient tradition, with the Romans generally preferring (for reasons of local stability) to preserve whatever taxes existed before they conquered a region, merely redirecting them to the Roman treasury (the aerarium Saturni). But that too might mean that while Roman revenues could be vast, they could also be remarkably inflexible as changing tax rates on a region was a breach of tradition which could provoke instability (and was ‘being a bad emperor’ to boot!). The workaround of all of this was the emperor’s private purse: property of successive emperors becoming a parallel form of revenue called the fiscus (the word for a household’s private money supply, literally a box of cash in the house), which at least notionally could be a bit more flexible.

In short, these state revenues tend to be messy, complicated and idiosyncratic, the product of generational layers of both innovation and stubborn tradition. But even as an economy grows, state revenues may stay stubbornly static.

[...]

When the state shifts an expense downward to individuals or communities, we say that the cost is devolved on to them. Devolution is thus a strategy for shifting costs off of the state balance sheet and given the above discussion, you may already be able to see the value: for a polity that has a lot of economic activity happening which (because of low administrative capacity, sticky traditions or a lack of coinage-based economics) it cannot effectively tax, devolution provides a means of shifting military costs directly onto those economics actors.

In historically-inspired or fantasy worldbuilding, this is a strategy that is often both neglected and unintentionally evoked. It is neglected in that it is rarely explicitly placed as part of the system: no one says, “oh, the town guards have to buy their own equipment” and generally the town guards never look as motley as they ought if that were the case. On the other hand, the basic nature of the ‘adventuring party’ involves a lot of devolved costs: the state needs monster hunters, but it expects those hunters to equip and supply themselves and often doesn’t do much to pay them (though part of this is ‘payment in loot,’ discussed below). But cost devolution was very common and worked on both smaller and larger scales.

[...]

That said, the recruitment principle matters a fair bit here. You can compel farmers to reach into their own resources a little bit, but if you want them to really dig deep for a war effort, they need to motivated by something beyond compulsion. Systems that devolve heavy infantry service – which demands a considerable investment in armor – are generally entitlement-principle recruitment systems. We see this with the hoplite armies of ancient Greece, the citizen-militia armies of the Roman Republic and also the heavy infantry militias of many medieval towns: what gets these men to work harder in order to afford to be able to shell out for that expensive equipment is the fact that their status in the community and their political position in the community are connected to it. Polities that are unwilling to devolve any political power to the commons are going to struggle to get the commons to buy expensive equipment or be highly motivated on the battlefield.

[...]

Taking loot, meanwhile, was an expected part of nearly all pre-modern warfare and so the promise of loot was a regular inducement for service. What I want to note here is that the promise of loot was almost never sufficient inducement: it was very rare for armies to serve only for loot. Instead, promises of loot were layered on top of other recruitment principles: loot and pay, loot and social status, loot and a role in the community. And that should make sense for two reasons. First, loot is never guaranteed, it requires winning, which generally only one side is going to do. Indemnities – which unlike loot, flow entirely to the state, rather than at least partially to individual soldiers – require winning the war and imposing a peace and again, only one side is generally in a position to impose indemnities (and often neither side is!).

[...]

This is something, I will note, that RPG-economies (both table top and computer) get quite wrong. The problem is three-fold on the one hand, these games invariably underestimate the cost of simply subsisting even a small adventuring party. Food and basic clothing consume quite a lot of resources in a pre-modern context, but that would be irritating to players and so it is often ignored or the cost reduced massively. Second, the loot gained is often over-valued, with itemization systems that fail to take into account that an old, busted hauberk pulled off of a corpse is not going to command the same market value as a shiny new one, freshly crafted to order.

But most importantly, these economies fall apart because they assume an insane amount of fighting and an absurd ‘win rate.’ Recall that, for an aged hoplite, having been in three battles was quite a respectable number even in a very violent period in ancient Greece. By contrast, your typical Dungeons and Dragons adventuring party has been in three battles before they unlock their subclass features at level 3. Moreover, most of the combatants on the losing side of a battle typically flee. In a battle between two armies of 10,000 men, we might expect the winning army to have lost around 500 men (5%) and the losing army to have lost perhaps 1,500 (15%), so that is 9,500 survivors splitting the loot of 1,500 fallen (2,000 even if they’re willing to rob dead comrades). So while your D&D party or Mount and Blade II: Bannerlord company sustains itself by splitting the loot of dozens of foes for every party member, in an actual army, you’re lucky to get your ~1/6ths share of a fallen foe.4 Loot is still a factor, but one cannot expect to run an army on it, long-term.

That said, loot distribution can have interesting distorting effects even if it isn’t enough to relieve the whole burden of running an army. Loot is a high-variance sort of thing: many soldiers get none, but some soldiers, if they are lucky to be on the right campaign, might get a great deal, potentially enough to alter their social position and status. Again, this simply cannot happen to everyone in a society, but it can happen to select individuals. Some of Alexander’s soldiers did get rich off of his conquests and certainly some Romans did too, although it is worth noting that in most societies, the structure of power channels looted wealth upwards: most of it ends up in the hands of the elite (as was certainly the case for both of those examples). Often this was institutionalized, with the proceeds of conquest being distributed in shares based on rank, with higher ranks getting a larger slice of the pie.