Transportation costs and economic geography

  • We have updated our Community Code of Conduct. Please read through the new rules for the forum that are an integral part of Paradox Interactive’s User Agreement.

Colon

Major
2 Badges
Feb 19, 2000
666
88
Visit site
  • Victoria 2
  • 500k Club
I was thinking that given that infrastructure will require pops to function, transportation costs and distance may be a thing in the game and with it a chance to better simulate the historical location patterns of various industries. In Victoria 2 an attempt was made to incorporate this element by giving a throughput bonus to a factory when an input was within the same state. It falls short for several reasons. Firstly, throughput efficiency is a production amplifier that maintains the ratio of inputs to output. Given that the most of a factory's costs consists of inputs purchases (wages coming from what's left over), throughput efficiency tends have fairly little influence on the odds a factory is profitable or not: it mostly amplifies profits and losses. For that reason a factory that's located in a state that produces one or more inputs doesn't have much of a greater chance of surviving than a factory that doesn't have inputs in the same state. Its profits will be bigger but so will its losses. The AI also never seems to consider the presence of inputs in its decisions so the locations of industries remain random. Second, the size of the bonuses seem unconnected to the inherent features of the input goods. Most factories have a max 25% throughput bonus and this bonus is subdivided between the various inputs in a rather haphazard way. Third, there's no consideration of distances beyond the state and no consideration of distanes to (sales) markets.

Urban Economics by O'Sullivan has a very interesting discussion of where firms are located, some which I think is very relevant in this regard and which I'll try to pass along. To start, ignore everything else for a moment, such as quality of infrastructure, local availability of workers and their productivity, the quality of local governance etc. In that case, the location of an industrial firm is decided by procurement costs (costs of transporting inputs to the factory) and distribution costs (costs of transporting output to markets). Where the sum of both is lowest is where a factory will be located. Either of these costs are in turn determined by monetary weight x distance, with monetary weight being the product of weight and transport cost rate. For instance, a sawmill will be closely located close to a forest because during the process of sawing them into lumber, a lot of weight is shed from the input logs, resulting in a higher (monetary) weight for wood than for lumber. A location near the forests results in the lowest total transportation costs. Generally, the more bulky, perishable, fragile, or hazardous a good is, the higher its monetary weight. A canner producing canned fruit has to deal with perishable fruit as an input while canned fruit isn't as perishable and cheaper to transport, so it will be located near a fruit farm. Assembled cars are more bulky than rolls of wire and sheets of metal so a car assembly plant will be located nearer to markets. In essence, there's a tug of war between procurement costs and distribution costs in determining a factory's location.
The same principle applies when transhipment points such as ports come into play, where goods are transfered from one transport mode to another. The sawmill may instead be located in the port, where it can combine procurement from several forests, shipped by train, and from where it can export to foreign markets by ship. Or a factory like a chemical plant or a textile mill could instead be located in a port near the ultimate market, from where it procures cotton or oil from several foreign sources. Wherever the sum transport costs are lowest, taking into account the monetary weight of the inputs and outputs.

I was thinking it might be possible to include something like this in Victoria 3, since the math isn't complicated, you just have to assign a monetary weight for a each good and multiply it by transportation distance. I'm guessing you can't include and track single cargo shipments without your computer exploding though, so maybe an approximation would be needed for that. What I think would be the big payoff is that if you'd include this you may get industries cluster or be spread out in a natural manner, and even shift in a natural manner. In the book O'Sullivan included the example of steel plants shifting from coal rich areas to iron rich areas as less coal content was required with technological change (that would perhaps tie in nicely with production methods).
If it isn't feasible to feature monetary weights, then I'd suggest you'd at least give input efficiency bonus to the presence of production of input goods in the same state, instead of throughput bonuses, and take into account that being near inputs shouldn't be as important for every type of industry. Steel plants and lumber mills should be having bigger total bonuses than bakeries or ammunition factories (hazardous goods). That should result in the former clustering near forests and coal/iron deposits and the latter being more scattered around.

Much the discussion in the book re this matter, with historical examples, can also be read in this paper: https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&ved=2ahUKEwj156647vTxAhXPO-wKHZkpDOcQFjABegQICBAD&url=http://www2.lawrence.edu/fast/finklerm/OSullivan%20chapter%20on%20Firm%20Location.pdf&usg=AOvVaw1H6fp5OuGhPQM4-BulLihY
 
Last edited:
  • 11Like
  • 11Love
  • 4
  • 3
Reactions:
We don't know exactly how detailed the transport model is. We know that industries impose an infrastructure usage cost for market access, but it's not clear if you can forgo some of that cost if producers and consumers are co-located. Also not clear if industries have to pay for transport to the market, or if an industry might pay more if it's further from most of the consumers in the local market. We don't know the costs for inter-market transfers of goods.

So a lot of open questions! Hopefully a transport DD is coming soon :)
 
  • 1Like
Reactions:
I was thinking that given that infrastructure will require pops to function, transportation costs and distance may be a thing in the game and with it a chance to better simulate the historical location patterns of various industries. In Victoria an attempt was made to incorporate this element by giving a throughput bonus to a factory when an input was within the same state. It falls short for several reasons. Firstly, throughput efficiency is a production amplifier that maintains the ratio of inputs to output. Given that the most of a factory's costs consists of inputs purchases (wages coming from what's left over), throughput efficiency tends have fairly little influence on the odds a factory is profitable or not: it mostly amplifies profits and losses. For that reason a factory that's located in a state that produces one or more inputs doesn't have much of a greater chance of surviving than a factory that doesn't have inputs in the same state. Its profits will be bigger but so will its losses. The AI also never seems to consider the presence of inputs in its decisions so the locations of industries remain random. Second, the size of the bonuses seem unconnected to the inherent features of the input goods. Most factories have a max 25% throughput bonus and this bonus is subdivided between the various inputs in a rather haphazard way. Third, there's no consideration of distances beyond the state and no consideration of distanes to (sales) markets.

Urban Economics by O'Sullivan has a very interesting discussion of where firms are located, some which I think is very relevant in this regard and which I'll try to pass along. To start, ignore everything else for a moment, such as quality of infrastructure, local availability of workers and their productivity, the quality of local governance etc. In that case, the location of an industrial firm is decided by procurement costs (costs of transporting inputs to the factory) and distribution costs (costs of transporting output to markets). Where the sum of both is lowest is where a factory will be located. Either of these costs are in turn determined by monetary weight x distance, with monetary weight being the product of weight and transport cost rate. For instance, a sawmill will be closely located close to a forest because during the process of sawing them into lumber, a lot of weight is shed from the input logs, resulting in a higher (monetary) weight for wood than for lumber. A location near the forests provides in the lowest total transportation costs. Generally, the more bulky, perishable, fragile, or hazardous a good is, the higher its monetary weight. A canner producing canned fruit has to deal with perishable fruit as an input while canned fruit isn't as perishable and cheaper to transport, so it will be located near a fruit farm. Assembled cars are more bulky than rolls of wire and sheets of metal so a car assembly factory will be located nearer to markets. In essence, there's a tug of war between procurement costs and distribution costs in determining a factory's location.
The same principle applies when transhipment points such as ports come into play, where goods are transfered from one transport mode to another. The sawmill may instead be located in the port, where it can combine procurement from several forests, shipped by train, and from where it can export to foreign markets by ship. Or a factory like a chemical plant or a textile mill could instead be located in a port near the ultimate market, from where it procures cotton or oil from several foreign sources. Wherever the sum transport costs are lowest, taking into account the monetary weight of the inputs and outputs.

I was thinking it might be possible to include something like this in Victoria 3, since the math isn't complicated, you just have to assign a monetary weight for a each good and multiply it by transportation distance. I'm guessing you can't include and track single cargo shipments without your computer exploding though, so maybe an approximation would be needed for that. What I think would be the big benefit is that if you'd include this you may get industries cluster or be spread out in a natural manner, and even shift in a natural manner. In the book O'Sullivan included the example of steel plants shifting from coal rich areas to iron rich areas as less coal content was required with technological change (that would perhaps tie in nicely with production methods).
If it isn't feasible to feature monetary weights, then I'd suggest you'd at least give input efficiency bonus to the presence of production of input goods in the same state, instead of throughput bonuses, and take into account that being near inputs shouldn't be as important for every type of industry. Steel plants and lumber mills should be having bigger total bonuses than bakeries or ammunition factories (hazardous goods). That should result in the former clustering near forests and coal/iron deposits and the latter being more scattered around.

Much the discussion in the book re this matter, with historical examples, can also be read in this paper: https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&ved=2ahUKEwj156647vTxAhXPO-wKHZkpDOcQFjABegQICBAD&url=http://www2.lawrence.edu/fast/finklerm/OSullivan%20chapter%20on%20Firm%20Location.pdf&usg=AOvVaw1H6fp5OuGhPQM4-BulLihY

I love this idea. If this is at all possible, then I hope they do it. I don't understand how you could do an accurate economic simulation without modeling transportation costs in some reasonably realistic way.
 
Thanks, great write up and suggestion! If I read it correctly though, shouldnt spoilage also be modeled as well (not just the monetary value, but destruction)? So certain goods cannot travel far enough without certain tech?
 
The period of Vic saw an enormous revolution in international shipping costs, and this had a prime importance in the industrial revolution because it allowed european nations to import foodstuffs from the new world beyond specialty goods like sugar. Notably, mass grain imports from the US became a thing.

All that notwithstanding, I have no idea how to make this shipping revolution an interesting game mechanic.
 
Thanks, great write up and suggestion! If I read it correctly though, shouldnt spoilage also be modeled as well (not just the monetary value, but destruction)? So certain goods cannot travel far enough without certain tech?

That would be an outcome of how perishable a good and is part of monetary weight. Monetary weight includes basically anything that makes it more desirable to keep transportation distances down. It could indeed change with new technology. Meat would have a high monetary weight at the start of the game, but would drop significantly with refrigeration (which was a major factor in Argentina's economic boom in the late 19th century), making it less vital for plants that use meat as an input to be located near ranches.

Excellent idea. Though judging by the standard of complexity of all other PDX's games I dont think they plan to go that deep.

If there aren't transportation costs or it isn't possible to add monetary weights somehow, they should still rework the bonuses factories receive when input goods are produced in the same state. In Victoria 2 fabric factories can have the same 25% throughput bonus as steel mills, even though textile mills historically often were nowhere near cotton fields (eg: Manchester), which suggest being close to sources of input wasn't as important to such factories as it was for steel mills I think that varying the max bonus with the type of factory and replacing throughput efficiency bonuses with input bonuses would do a lot more to encourage natural location patterns. Some you'd typically find in the same states as mines and plantations, others would be more dispersed.
In that vein, there should also be an output efficiency bonus for having customers in the same state. Explosives factories could receive such a bonus for being in the same state as mines for instance.
 
Last edited:
  • 2Like
Reactions:
@Colon those are all good ideas, they make industrial chains make more sense. Not to mention, all developed areas are targets for conquest. If you build up an arms industry closer to your mines, which happen to be next to the border with your nemesis, well, there is another layer of strategic depth right there. This is then kind of thing that separates trully challenging strategy games from the ones that only appear to be so on the surface (like Stellaris).

I just wonder if any dev team would want to go that far, on account of all in-game calculations and it's effect on performance. Stellaris is plagued by those things, the pops system is destined to never really be too good since it will always be either too bland or to taxing on the average machine.

If it was up to me to come up with a logistic system for your own industry, which is "simpler", I would make a "train cart pool" working in parallel to the "steam convoys" we had in the previous games. The size of your economy dictates how much of each you MUST have, under the penalty of severe maluses if you don't have enough convoys and trains (which would now have to become a comodity too so that it can be bought or manufactured). Now, so that they are not just a double-of-the-same type of thing, I would make it so that they work differently somehow. Haven't really thought of it. Maybe the required convoys pool is calculated on your flow of imports, and the required trains pool is calculated on a basis of your industry, something like that.
 
Very interesting read.
Even though it probably isn't related to "monetary weight" itself, I guess it's worth adding here that transportation costs for goods (at least using railways) have been confirmed.
I don't think they've explained what exactly that'll do yet, but it's there, a building with the "railway transportation" production method will need to buy tickets produced by railways.
1626983509184.png

Infrastructure (which railways also produce) is a tough one, it doesn't seem like it costs anything to use it, but it passively improves market access for goods and pops, we'll have to wait until it's better explained as well.
As for ports and markets, and how/if they influence transportation costs over longer distances (which I very much hope they do), is yet another thing yet to be explained in detail (I'm sensing a pattern here).
 
@Colon

I just wonder if any dev team would want to go that far, on account of all in-game calculations and it's effect on performance. Stellaris is plagued by those things, the pops system is destined to never really be too good since it will always be either too bland or to taxing on the average machine.
I also proposed to rework the efficiency bonuses factories had in Vic2, people seem to be missing that. :) I suspect it wouldn't be any more CPU intensive
If you let the size and the total sum of these bonuses depend on the characteristics of inputs goods (in such a way that for instance a sawmill gets much bigger bonuses for being near woods than a textile mill for being near cotton fields). And if you add bonuses for having custumers in the same state instead of just suppliers (like for dynamite factories and mines) you'd already go a long way towards recreating the tug of war between procurement costs and distribution costs and encouraging historical location patterns.
 
  • 1
Reactions:
All that notwithstanding, I have no idea how to make this shipping revolution an interesting game mechanic.
If the merchant marine is modeled, you could invest in it to lower the costs of importing goods (and in wartime, have it sunk by commerce raiders or submarines). This would lower prices for your pops to meet their needs, but also for factories to buy their inputs. So, it would be a necessary step toward building factories that aren’t connected to their suppliers by railroad.
 
  • 1Like
Reactions:
I was thinking that given that infrastructure will require pops to function, transportation costs and distance may be a thing in the game and with it a chance to better simulate the historical location patterns of various industries
I was actually thinking about something similar and came up will a possible algorithm to calculate the transport costs. You can check it out here https://forum.paradoxplaza.com/forum/threads/transport-cost-calculation-proposal.1489217/ ;)