On the theme of economic reductionism / turning flesh and blood human beings into cold robot calculators: the central theme running through the PhD and the last project was the effect of distance on economic choices: what people buy, where they buy it from, what/where organisations buy, how they change buying/selling and location together as a single choice set. Which is more-or-less a definition of spatial economics.

One of the ideas that turned out to be very useful for thinking about it is value density. As I use the concept, this is just the ratio of an object's value over 'cost to shift it a unit of distance'. So if it's selling for five pounds and you need to ship it ten miles at fifty pence a mile, its full cost is going to £10 and its value density is ten.

In the usual use of value density, the focus is on weight and bulk. Weight hasn't really been a part of recent spatial economics, even if it's obviously continued to be essential for actual logistics planning. Weber's is the classic work that looks at optimal siting giving inputs, weights and distances - but it's past a hundred years old now and 'proper' spatial economists dismiss it as mere geometry. ('That literature plays no role in our discussion' is all it gets from Fujita/Krugman/Venables.)

But shifting the idea of value density to 'how much it costs to ship per unit of distance' alters this: weight/bulk are important only for how they change that cost. Value density as I'm defining it can also change if, for example, fuel costs go up - which is why the idea ended up being superbly useful. It gets more complex if one starts to factor in time issues (the main reason containerisation transformed the planet was due to its impact on time, not distance) - something that, up to now, I've studiously avoided - so in true quant style, since it's difficult, let's just pretend it doesn't matter for now...)

Value density is a concept that, as far as I can tell, only gets used in the logistics literature - a very applied setting where analysts support decisions about actual production networks. I first heard it mentioned by Steve Sorrell and thought, "wossat then?" Because of this, it doesn't seem to have been used as a 'how do spatial economies work' tool.

It's the 'value' part that makes value density interesting. Because the value of something is determined mostly by its economics, only very minimally by its physical weight/bulk/distance-cost, it can be changed by anything that can add or decrease that value. From a non-spatial point of view, that's trivially obvious: something's real cost drops if, for example, wages rise. But when these two are combined, value density means that changing something's real value also changes how far it can economically be moved. That puts it right at the centre of how spatial economies wire themselves.

So where Weber cited the weight of certain primary inputs as the reason they don't move very far, logistics folks know that's not it. This can be seen along whole production chains: primary inputs will often need to be close to the first stage of production. At each following stage, more labour is added - and hence more value. Extra value can be added anywhere that extra economies of scale can be squeezed out. This is possibly what determines the dropoff of UK domestic trade (see pics) I dug out of the transport data: low value-density cement, sand, gravel, clay at one end, manufactures at the other (though there are likely plenty of other factors at play, since it's just domestic movement).

This leads to what, for me, seems like a rather startling conclusion: it's the ratio of value to per-unit distance cost, not either alone, that determines the spatial reach of economic networks. This was a very useful idea for the last project as it meant I only needed to determine one number, not two. Again, if we're talking about value versus weight - not a shock, huh? It's when fuel costs or other per-unit distance costs change that it gets interesting. You can increase the spatial reach of your economy by decreasing fuel or time costs - or by increasing value. That can happen through, for example, a more educated workforce with higher wages, through new production methods, through Jacobs-style diversity externalities caused by diversifying clusters - whatever pushes value up.

But there's an exception. There's one input into the production process, one feedstock into the great grinding millstone of the economy, that doesn't quite work in the same way. Human beings. Superb quote:

Humans remain the containers for shipping complex uncodifiable information. The time costs of shipping these containers is on the rise because of congestion on the roads and in the airports while the financial costs of so doing are also rising due to increases in real wages of knowledge workers who are the human containers. (Leamer/Storper, the Economic Geography of the Internet Age, Journal of Intl Business Studies, 2001 p.648).

So humans can be value dense, just like anything else you shift on a truck and back into a factory. But, as production inputs, they have some idiosyncracies that set them apart from coal or crankshafts or hard drives. At the top of the list: with some exceptions (i.e. people who don't stay in one place) they all need to be within a few hours' radius of the place they input their labour. They also have a range of other functions - some of them not even economic! - that determine their spatial patterning. A whole range of essential inputs, for instance, are required to guarantee creation of the next generation of shipping containers for complex uncodifiable information.

Putting aside my flippancy, this is sort of car-crash fascinating to me. It is actually possible to see a quite generalisable theory of value density that just includes some of these tweaks to humans. After all, there's nothing novel about the approach: humans are one side of the most commonly-used production economics idea, the Cobb-Douglas, with capital on one side and labour the other. And we've mostly become blithe to the presence of 'human resource' teams in all large workplaces. But this is a little more specifically economising, isn't it? Not only labour, but an input with certain ratio of value to per-unit distance cost, just with the addition of some tweaks.

I think it works though. That's the marvelous thing. It's this fundamental difference - something explored in detail by Glaeser, for example - that is determining the shape of modern cities. Decreasing distance costs, either relatively through increasing value or otherwise, increases the value density of us lot.

What I'm still not clear on, and why I'm writing all this down: we know that theories change reality. I go on about that all the time. They also change how we see the world individually - obiously, really. Even if there was some clean way of separating out my analysis of how society's structured from my daily world-view, how would that make sense? Don't we try and learn about the world specifically to change how we see it, both as people and collectively?

So I find myself asking the Dr Malcolm question ('Yeah, yeah, but your scientists were so preoccupied with whether or not they could that they didn't stop to think if they should'). Not because, of course, it'll actually make me turn off the computer and walk away. At least not yet. But I do wonder - the kind of reductionism I've just laid out might produce genuine insights, but at what cost? I've been wibbling recently about the inseparable link between language networks/production, and this stuff is just another version of that. The cosmology of economics creates some kind of productive landscape - but is it one we want? Should I be comfortable in promoting the use of that language, justifying it with claims that simplifying produces useful views of reality?
---
New year's earnestness 6/17. If Eddie Izzard's having to do two marathons on his last day, I should be able to catch up on the blog target in eight weeks, right? Hmm.