Hubble Telescope


Interstellar Trade:
A Primer

February 20, 2001

If you ever wondered why Rigellian green-fuming brandy costs so much when you're not even buying a top brand, now you can find out ...


Blue Band I confess to being a sucker for a type of science fiction that is not always the most literary: space SF. At bottom this is surely the tourist impulse. I want to look out the viewport at the Pleiades and the Orion Nebula, or land on a habitable moon and see a giant ringed planet hanging in the sky above forests of unearthly trees. If doing so takes zapping the odd space pirate, I can live with that, too. Exploring cyberspace or mindspace just doesn't have quite the same thrill.

Space SF is a bit out of critical fashion (though it still sells like hotcakes, no small virtue to a writer). It certainly has its share of corny elements, as I admit in more detail here. Moreover, nearly all of it is really fantasy, every bit as much as elves & dragons. If you're going to tour the galaxy and come back to bore your friends while they're still alive, you have to use faster-than-light drive (FTL), i.e. effectively magic. 

I like space SF anyway, and would like to do it right. This means, among other things, making up a universe that hangs together so that it sounds believable even if it isn't.   This article started as notes to myself on the (pseudo-) economics of interstellar trade.  If I want to have Free Traders, space pirates, interstellar battle fleets, and all that cool stuff - and I do, even if I change a few names for good form - it's handy to know (i.e., make up) the underlying economic background.   Details like how much a University survey ship (read battlecruiser) costs, how big it is, how long it takes to get from Seychelle to Tilmunsen, and so forth. 

Fair warning: If you're just looking for the Cool Stuff, skip down to the end of the next page.  That's where I give some specs for interstellar battlecruisers and hyper-dreadnoughts.  Getting there, though, is largely a matter of estimating trade costs and volumes - about as exciting as any other cost estimates, real or imaginary.  But the people who sign building contracts for starships are going to know it, and you should know it too.


From jetliners to starships ...

How to figure out the cost of a starship ticket, or of shipping cargo aboard one?  I start with the premise that starships are, in a sense, the jetliners of a future century - they go fast, carry lots of people or cargo, and embody the most advanced technology that the society can crank out on a mass production basis. Starships go a whole lot faster and farther, of course, but the builders have a more advanced technology. Present-day spacecraft certainly won't do for an analogy:  No one is selling spaceflight tickets except maybe the Russians, for $10 million a pop.  Not really a viable commercial service.  So we have to assume that an interstellar civilization can build and use starships more or less the way we use jumbo jets. 

As a first approximation we'll say that things balance out:  In future centuries the overall cost of a starship is equivalent (in proportion to size and cargo capacity) to the costs of jets for us. But we can't merely assume that a starship ticket goes for the same price as an airline ticket. Even with FTL, starships probably take a lot more than a few hours to get where they're going.  They complete fewer trips in a year, so each ticket or cargo waybill has to cover a larger share of the ship's expenses.

The annual cost of running a jet (as opposed to the sticker price to buy one), is not something that gets reported regularly.  I guesstimated it by working backwards from the price of a ticket, since ultimately ticket revenue (plus airfreight) must be enough to keep airliners flying.

Suppose that the average cost of a seat on a transcontinental jetliner is $500. (A good deal less, to be sure, if you're willing to reserve a seat for a 3 AM red eye on a Tuesday next month, but a lot more if you want to fly business class tomorrow at 9 AM.)

The seating capacity for a midsize widebody jet is maybe 250 passengers; with 200 seats filled, revenue earned by the flight is $100,000. Takeoff fuel load is about 200,000 lbs or 30,000 gallons, so fuel cost is close to half the total operating cost. (Also landing fees, taxes, etc., which we'll ignore.) Say then that the basic operating cost of the flight, less fuel, is $50,000. This pays for plane, crew, maintenance, etc. ... and of course the airline's profit margin. (From an economics point of view this is just another cost: persuading someone to invest in an airline.)

What with ground turnaround time, preflighting, etc. - and of course the usual delays - say that the jet can make two transcontinental flights (one way) per day, 300 days/year. The annual cost of operating this jet is therefore some $30 million, less fuel and incidentals like landing fees and taxes. The purchase price is about $100 million, so one year's regular use of the jet costs about 1/3 of the purchase price. A guesstimate, of course, but it should be a valid ballpark figure.

As a cargo plane, this jet could carry about 50 tons, so the purchase price is roughly $2 million per ton of cargo capacity.  A starship, however, operates entirely in space (shuttles providing surface-to-orbit service).  It has no landing gear or heavy wing spars, and probably accelerates at well below 1 g, so it can have a lighter hull structure in proportion to cargo and fuel. We'll say that the starship carries twice as much cargo per deadweight ton of hull as a jet can carry.  Thus, it costs the equivalent of about $1 million per ton of cargo capacity. (We'll simply use equivalent present-day dollars for starship costs, instead of credits or whatever. We know what a dollar buys.) Assuming that the annual operating cost of the starship, relative to building cost, is similar to that of a jet, it therefore costs $300,000 per ton of cargo capacity per year to run a commercial starship.

(Again we are cheerfully ignoring details taxes, station-docking fees, and even the cost of all-important fuel.  We'll just assume that starship fuel is cheaper than JP-4.)

We may suppose that the designed service lifetime of a merchant starship is 30 years, so the initial purchase price is only about a tenth of overall lifetime service cost. (Closer to a fifth in practice due to interest on the purchase loan; you have to pay the shipyard up front for the ship's ability to haul cargo years from now.)  Ships may well last longer; we'll say more on that below.

This leaves one key remaining variable: how many cargoes can a starship carry in a year's service? That is, assuming full cargo turnover at each port of call, how many one-way runs can she make? This will of course vary depending on route, but let's make a quite conservative assumption - at least by SF standards, for ships with some kind of FTL.  Say that the one way trip between two nearby systems - from departure planet orbit to FTL jump point to arrival planet orbit - takes 3 months. (Thus, travel times are comparable to the age of sail.) Add another month for servicing and maintenance in parking orbit, plus selling the cargo and buying cargo for the next run. Four months then, departure-to-departure.

So ... This starship can turn over three cargoes a year, and therefore must earn $100,000 per ton of cargo: that is, selling price at destination minus purchase price at origin. The minimum selling price of imported goods is therefore $100,000/ton, or $100/kg. That's if you can get them for free on the origin planet! In practice, obviously, only goods of high intrinsic value will be shipped. Suppose that, on average, half the retail price of interstellar imports goes to shipping costs, so that imported goods typically cost $200/kg. For us English-unit people, call it $100/lb, or six dollars per ounce.

For the consumer market, this obviously means luxury goods only, with very high value per weight - things like jewelry, spices, fine liquor, designer-label clothing. Even luxury goods won't fly if they weigh very much; the future equivalent of an small imported sports car would cost $100 K just to import that sucker. But very high value-to-weight industrial goods can be shipped, e.g. the equivalent of microchips. Bulk goods obviously won't move in interstellar trade - not at $100,000/ton.

Now, suppose that the per capita GPP - Gross Planetary Product - of a colony world is $100,000 (about three times that of the US today.) If two percent of income goes to imported luxuries and high-value capital goods, that's $2000 per capita, of which half, or $1000, pays for shipping costs.  For a colony planet of ten million people, the total shipping cost of imported goods thus comes to $10 billion.

Working backwards, this means that 100,000 tons of interstellar cargo arrives annually. The planet will have to export a similar quantity of goods; otherwise it runs a trade deficit. Also, import prices will go up if ships can't load a cargo to sell at the next destination; a starship with empty cargo pods costs nearly as much to run as one with a full cargo load.

Take the simplest trade model, two planets trading with each other. Each year, 100,000 tons of cargo moves in each direction, or 200,000 tons total volume. If the average cargo starship carries 1000 tons - much less than sea cargo ships, but much more than cargo planes - this requires 200 annual loadings. (And an equal number of unloadings!) Since each ship can make three one-way legs per year, she will account for three loadings, so these two planets' combined merchant fleets should be about 65-70 ships. If each ship carries more cargo, fewer are needed. For 5000 tons, about the load of a typical freight train, 13-14 ships will be sufficient. In general, however, this trade economy will not favor many big ships; they have trouble finding enough cargo to fill their pods.

All of this makes for a viable if modest-scale interstellar economy. A trade network of a dozen colony worlds would support a few dozen to a few hundred cargo starships, depending on the average size of the ships.

We can also say something about passenger traffic. Jetliners can carry about 4 or 5 passengers per ton of equivalent cargo capacity. But interstellar passengers can't live in a coach seat for several months, so let's say that each passenger berth equates to a ton of cargo capacity. This is for berth space as well as apportioned galley/diner space, as well as food, etc., and of course the passenger plus baggage. Cost of a starship ticket to a nearby system: $100,000. Obviously there won't be a big tourist traffic. Passenger ships (or passenger pods carried by general-purpose cargo starships) will carry a few rich tourists, but mostly business or official travelers.

BUT ... alas, there is one kind of large-scale passenger traffic that is absolutely necessary for this trade network of colonies to exist at all:  Colonization. This must also cost at least $100,000 per colonist - more likely several times as much. The equivalent of tractors, etc. have to be shipped along with the colonists. Moreover, because a planet just being colonized is unlikely to have anything to export, the colonizing ships return empty, so the applicable cost is round-trip, not one-way.   Altogether it might add up to a million dollars per colonist.

This could be a fatal problem. Not many millionaires are going to want to emigrate to the rugged life of a new colony world.  Sending 10,000 people to a olony would cost some $10 billion; even if the planet has a known potential export, this is an awful lot of capital for private industry to raise for a fairly speculative venture. Not even governments are likely to spend this kind of money to establish colonies, at least not on a consistant basis.  Research bases on habitable planets might gradually develop into colonies, but the whole process would be painfully slow and always one budget cut away from extinction.

So the price schedule offered here seems to make interstellar colonization unlikely in the first place. (Non-FTL colonization by "slowboats" is even worse, because then you have to pay the whole cost of the ship for a single one-way voyage.)  In trying to run some numbers for interstellar trade, instead we've killed the whole damn thing.  I hate it when this happens!


Happily, there is substantial room in the model to bring costs down, either by making trade starships cheaper or making them faster. Since FTL is magic to start with, we may as well also wave the wand over operating costs and/or travel time.  Let's do both. 

We'll cut the annual service cost of starships to $100,000 per ton of cargo capacity, a third of the original estimate.  This perhaps is not unreasonable.  Space is a harsh environment, but harsh in a fairly steady way.  Unlike jets, starships don't have to ramp up every ounce of takeoff thrust, undergo constant pressure changes, get thrown around by turbulence, or thump down onto a concrete runway at 150 mph every few hours.  Even the cargo pods don't have baggage handlers tossing suitcases around inside them several times each day.

Since the starships are built for long-haul reliability, we'll say that building cost is reduced only by half, not two-thirds.  At $500,000 per ton of capacity, for a 30-year expected service life it now accounts for a sixth of total lifetime service cost rather than a tenth.   With interest payments this may be closer to a third; we've cut operating cost - for maintenance, crew, fuel and such - by a factor of about 3.5 or so. 

We'll also assume that a starship can make a one-way voyage and turn around for next departure in 35 days instead of 120 days.  Thus she can deliver ten cargoes in a year's service.  Given the reduced yearly operating cost, the actual transport cost per ton or per passenger drops from our original $100,000 to a measly $10,000.

Colonization transport cost is still higher, because of supplies and provisions and no return cargo.  But now it is perhaps $100,000/colonist or less, in reach of the middle class. (Relatives will probably have to kick in; why are they so eager to get rid of you?) Even in poor countries, a village, church, or mosque might scrape up the money to send off a young couple to the colonies. Large-scale interstellar colonization is now economically viable -  assuming the technology exists and that habitable planets are out there to colonize. (What's a minor assumption or two among friends?)

We now have an order-of-magnitude basis for estimating the conditions under which colonization can develop. If the cost of starships is comparable,  size for size, to that of jet planes, interstellar travel times must be measured in days or a couple of weeks, not months. (We could have reduced transport cost by the same factor of ten by cutting turnaround time to 12 days, without reducing annual costs.)  For longer travel times, starships must be significantly cheaper to operate than jet planes.

Given a starship technology that makes colonization viable in the first place, subsequent interstellar trade will be more robust, because the minimum cost of imported goods is lower. Under the revised $10,000/ton shipping cost model, the threshhold cost of imported goods is now around $10/lb - say, $25 a fifth for a cheap brand of that Rigellian green-fuming rotgut.  A sports car will cost some $10,000 to import (plus the factory price of the car). Bulk cargoes still won't figure in interstellar trade, though; the shipping cost of oil, for example, would be about $1500/barrel.

As with any freight rates, these will vary. Higher-value merchandise will support higher shipping charges, justifying transshipment or multi-leg runs to more distant customers. A long-term fixed contract, allowing the shipowner to depend on cargoes, will get much better rate than a rush shipment. The cargo equivalent of "standby" will also get a better rate; if a ship is making the run anyway, better to fill her cargo pods.

If imports still account for just two percent of GPP, as in the earlier estimate, the volume of goods moving in interstellar commerce would increase tenfold (but shipping capacity only about threefold, since each starship delivers three times as many cargoes each year). However, with shipping costs ten times lower and a much wider range of goods now worth importing, the import-export sector can expand in total value of goods shipped as well.  Apply an inverse square-root rule and say that reducing shipping costs by a factor of ten will increase spending on imported goods by a factor of three.  Six percent of GPP now goes to imports - undoubtedly a high-end figure, but not wholly out of reach for a mature trading zone in which worlds have developed their own exports. A colony of 10 million people thus has an annual import - and export - volume of about 3 million tons per year.

Each trade starship now can pick up and deliver 10 cargoes per year, so the ships needed to carry this volume have a net cargo capacity of 300,000 tons. For a trade network of a dozen colonies, the combined merchant marine has a capacity of some 3.6 million tons. Most ships will still be fairly small (though still much bigger than jumbo jets) in order to fill their cargo pods easily, but the heaviest-traffic routes will support big ships.

Say, then, that the trade network's merchant fleet is something like this:

  75 ships of  20,000 tons capacity each   =  1,500,000 tons
300 ships of     5000 tons capacity each   =  1,500,000 tons
400 ships of     1500 tons capacity each   =     600,000 tons

TOTAL: 775 trade starships, capacity   =   3,600,000 tons

If there is no ansible or FTL radio (and in my SF universe there isn't), some of the small freighters will sacrifice cargo capacity for speed.  They will serve as a sort of interstellar FedEx, delivering small parcels, mail, and paying-through-the-nose VIP passengers, if not overnight then at least as fast as technology permits.

Passenger traffic is probably only a small fraction of cargo volume. (Unless there is still colonization, i.e. emigration from Earth.) Someone makes a profit by consigning freight, but passengers are pure expense, either to the passenger or whoever pays her expense account. Perhaps they are one percent of total volume - that still comes to an impressive 360,000 passengers each year. A few routes may support scheduled passenger service (probably using fairly small ships), but most passengers may ride in pods - not unlike railroad sleeping cars - carried by freighters, or simply in spare crew quarters.

The full-load mass and physical size of all these ships depends heavily on assumptions about fuel mass ratio, fuel bulk, etc. But a typical proportional mass breakdown might be

Deadweight  = 1
Cargo load   = 2
Fuel load      = 3

Total mass   = 6     (i.e., three times cargo capacity)

Big freighters thus have a full-load mass of some 60,000 tons; the largest individual ships in service might be up to twice as big, or 120,000 tons. For our revised estimated building cost of $500,000 per ton of cargo capacity, the cost per deadweight ton - i.e., the ship itself; structure, engines, etc. - is a million dollars, and the annual cost in service is $200,000.  We haven't so much made starship hulls cheaper as allowed them to carry more cargo in proportion to their own structural mass.

The building cost of these ships is not cheap. At $500,000 per ton of cargo capacity, the largest giant freighter costs $20 billion to build - but she has the cargo capacity of two hundred 747 jets, and by herself accounts for over one percent of the whole fleet's cargo capacity. An average small freighter costs $750 million, and has seven times the cargo capacity of a 747.

Given the 30-year service life of ships, the combined yards of the trade network turn out about 25 ships each year. In fact, the hull structures might well last much longer than 30 years, but equipment wears out and has to be replaced. Probably ships go back into the yards for overhaul each decade or so, but eventually the cost of stripping everything out and replacing it exceeds the value of the ship. Depending on overhaul costs, however, yards may well live more on rebuilding than new construction, with some ships remaining in service for many decades. Others may be retained as the equivalent of naval hulks or the old passenger equipment that railroads use as work trains; every big commercial space station will have a bunch of these old ships around its outskirts.

If modular design is taken to its limit, however,"ships" may have no permanent existence, but be assembled out of modules and pods for each run, much as railroads make up rolling stock into trains. In this case a ship's identity probably attaches to a service than to physical structures, just as the Santa Fe Chief was identified by timetable and reputation, not a particular set of locomotives and cars.

Go to Page 2

Return to Homepage


Blue Line