It reflects a modern concern to establish a more systematic understanding of the process of creative destruction

[8]

Abramovitz (1989), pp. 12-13, pp. 30-31 and p. 90.

• 
  [9]
  

If s_i is the share of industry i, in a constant price weighted measure of industry output, then in a small interval of time, ?s_i ? s_i (g_i ? g), g_i being the growth of the volume of output for industry i, and g being the growth rate of the aggregate. The contribution, which i makes to a constant average growth rate, s_ig_i , changes as ?(s_ig_i ) = s_ig_i (g_i ? g) + s_i ?g_i . Obviously, a fast growing sector need not have much effect on the aggregate growth rate since, typically, fast growth corresponds to the early life of the industry where its share is also small. See Kuznets (1971, chap. 8). Further analysis along these lines needs to take into account the disappearance of some activities and the introduction of new activities into the ensemble defining g. In turn, this raises interesting problems in comparing lists of activities that are qualitatively different between two dates.

• 
  [10]
  

Glenday (1938) applied Burns’ method to long production series for eight uk industries and found consistent evidence of retardation.

• 
  [11]
  

For some interesting commentary see Baumol et al. (1989, chap. 3).

• 
  [12]
  

For a valuable discussion of the competing alternatives, see Antonelli et al (1992), Stoneman (1983) and Mahajan and Peterson (1985). The pages of the journal Technological Forecasting and Social Change are a rich source of empirical studies using different growth curves. One of the classic early references based on the logistic curve is Fisher and Pry (1971).

• 
  [13]
  

In previous papers, Metcalfe (1981) and Metcalfe and Cameron (1988), I have explored similar ideas using the logistic process. The logistic is rather clumsy as a basis for treating time dependent shifts of the kind explored here.

• 
  [14]
  

Cf. Windsor (1932), Prescott (1922) and Peabody (1924) for early discussion and application. The logistic curve is symmetric around an inflection point, which corresponds to half the carrying capacity. I understand that the Gompertz curve was first proposed for actuarial purposes in 1825!

• 
  [15]
  

The industry is closed to foreign trade.

• 
  [16]
  

A point emphasised by Kuznets and Burns. See Metcalfe and Gibbons (1987) for such a treatment in the logistic case.

• 
  [17]
  

Cf. Metcalfe, Foster and Ramlogan, 2002, for some development of a more general multi-sector evolutionary model along Fabricant/Young lines.

• 
  [18]
  

Cf. Metcalfe (1998).