INDUSTRY STRUCTURE AND INCOME OPPORTUNITIES 153
by a constant percentage each time accumulated experience doubles.^29 How
much costs fall depends on the economics of specific businesses or activities.
Rules of thumb suggest that indirect costs tend to fall by around 30 percent
for each doubling, direct manufacturing labour by 6–8 percent and raw mate-
rial costs seldom by very much at all. On that basis, if they start at the same
time and each step takes a similar amount of time, a member of a team of ten
who concentrates on one or two steps will always have ten times the expe-
rience in a particular step than an individual doing all the steps by himself,
so his unit costs should be between (1 – 0.6^10 = 54%) and (1 – 0.08^10 = 43%)
of the non-specialist’s. In other words, the specialist in this instance will be
between 1.9 and 2.3 times as productive – the very estimate Smith made
almost 200 years before experience curve effects were properly measured!
Despite the excellence of Smith’s observation, it only goes a little way to
explaining firm size. Division of labour into discrete tasks increases produc-
tivity and (assuming the product can be sold) enables enterprise growth. It
therefore defines the minimum size required for efficient production.^30 It does
not, however, explain why a firm might be able to build a sustainably profit-
able business above this minimum size. There is no obvious reason why there
would be more job descriptions in a Toyota factory employing 10,000 peo-
ple than in a Rolls Royce factory employing 1,000 people – probably fewer,
since Toyota’s scale might mean that it would be more economic to use robots
for painting and welding than at Rolls Royce. There are five job descrip-
tions in a MacDonald’s franchise, three of them largely interchangeable types
of cook. MacDonald’s scale cannot be attributed to the productivity benefits
of further division of labour; rather it involves multiplying labour within job
descriptions as more franchises are opened up. Some of the world’s largest
employers have very few job descriptions (notable examples are the military
and the Catholic Church). In Nicias’ mine-slave rental business, claimed by
Xenophon to number a thousand men, it is likely they were all doing the
same basic job of digging (Xen. Vect. 23–4). The ancient industries discussed by
Bresson, which employed large numbers of labourers in single establishments,
making such items as shields, knives and beds, might have had longer produc-
tion chains than, say, potteries, but not by much.^31 If division of labour were
the only factor, Lysias’ shield factory would have employed one gang of six to
eight slaves, not ten or more.^32 Xenophon (Cyr. 8.2.5) famously describes how
shoemaking labour was divided among several specialists in urbanized areas,
but as Thompson points out ‘four workers dividing the tasks can turn out
more shoes than four men who do the entire job individually, but it is hard to
see how eight teams of four could turn out proportionally more than a single
team of four.’^33 Once a firm has exhausted the benefits of division of labour
(in the case of the pin factory, Smith thought the optimum number of workers
was about ten), if it is to grow to a size larger than the minimum required for