3. 
part by governments, although to some extent large concerns do provide such infra 
structure investment (for example, power stations, etc.) themselves. 
This tentative picture of the proglem can now be related to the historical 
development of the capital coefficient as it appears to most economists nowadays: 
The impression is that the capital-output ratio for society as a whole has remained 
roughly the same over the decades. From the comparison of scale one might have 
expected it to decline since large firms have tended to replace small ones with 
the advance of technology. Insofar as it did not, this might be partly due to the 
above mentioned factor (increasing infra-structure, etc.). There is, however, 
another explanation: The comparison of firms in a cross section and the develop 
ment over time is not the same thing. Over time, the real wage increases, and with 
it the value of a given output in terms of labour input decreases. This will counter 
act the cheapening of capital equipment due to scale effects; in other words, the 
differential profit margin of the large firm will be reduced as the large scale 
methods spread, and the value of output will fall relatively to a given capital 
(2) 
equipment with given labour cost. The impression is that a race between the 
cheapening of equipment and the cheapening of output is continuously on and neither 
is getting too far ahead of the other. 
So much for the interpretation of data. What about the theoretical discussion 
of Chapter III? My analysis there shows under what conditions an increase in 
capital coefficient will be profitable and this analysis may be applied also outside 
the context of the problem of economies of scale. The conditions are these: The 
proportionate reduction in cost divided by the proportionate increase in capital- 
coefficient must not be smaller than the profit margin from which we start (that is, 
the profit margin obtained with alternative or usual methods), otherwise the profit 
rate will be lower than at the outset (that is, with alternative or usual methods). 
^ Simon Kuznets, Capital in the American Economy, Princeton 1961. Table 6, p. 80. 
(2) 
Capital is always measured at cost-value, inflated with a convenient price 
index to give a measure of reproduction cost.