VARIABILITY IN CONSTRUCTION

A number of commentators have argued that variability is the main explanatory variable in the organisation of the construction industry. In a comparative study of the construction and mass production sectors, Stinchcombe (1959) argued that the distinctive "craft organisation" of production in construction was different from the "bureaucratic" organisation of production in mass production because of the greater "variability" of production process in the former. This variability, he argued was due to the seasonality of construction work, fluctuations in the volume of work, and variety in the product mix. He then deployed data to show that the more seasonal the work, the less bureaucratic was the organisation. Building on her earlier work (1984), Campinos-Dubernet (1988) also deployed "variabilité" as an explanatory variable in her cross national analysis of the organisation of the structural trades in the construction industries of France, (west) Germany, Italy, and the UK. Variability in production processes was here defined as a result of "l'instabilité de leur lieu de production", which would appear to be a similar usage to that of Stinchcombe. She reported that general contractors in France and Germany deployed a strategy of the "acceptation de la variabilité des processus" (internalisation of variability) by directly employing the labour and plant required for structural works, while their counterparts in Italy and the UK deployed a strategy of the "extériorisation de la variabilité des processus" (externalisation of variability) by sub-contracting for such inputs. The use of variability by Stinchcombe and Campinos-Dubernet appears to fit within the definition of variability made earlier. Both develop their analysis in contrast to the organisation of the mass production sector - Stinchcombe's focus is a critique of bureaucracy, while Campinos-Dubernet's is on taylorism. Stinchcombe puts the emphasis upon seasonality in the construction industry, and Campinos-Dubernet prefers to emphasise the site-specific nature of the production process, but both can be seen as applications of the idea of variations within a predictable range of possibilities. However, Eccles (1981b) has criticised Stinchombe's work, and argued that "subcontracting in the construction industry is a response to uncertainty arising from complexity" - it is complexity rather than variability which explains the development of subcontracting in the construction industry, where complexity is defined in terms of the number of functions performed by the structure. In a study drawing on US Census data, he shows that firms which subcontract (i.e. externalise the production process) more tend to specialise in technologically more complex projects, in larger projects, and to operate in more extensive market. Variability, defined as seasonality, was found not to correlate with the extent of subcontracting. Eccles' work has been supported by Üsdiken and his colleagues' (1988) findings that extent of subcontracting was correlated with project technology. This debate raises a number of complex conceptual and empirical issues which cannot be addressed here. In particular, there is a lack of clarity in the definition of terms such as variability and complexity in the work of these three researchers, and the empirical testing of the relationships is dubious. Eccles' ranking of the complexity of building types is open to contention; Stinchcombe only attempts to measure one component of his own definition of variability, while Campinos-Dubernet offers no empirical measure of variability. The position maintained here is that Galbraith's contention about the nature of uncertainty remains central. Uncertainty can defined as consisting of two distinct components, complexity and dynamism as defined above, where it is not possible to define the full range of options in advance. Rationality is inevitably bounded, and decisions tend to be made on a sequential adaptive basis. Actors which contract to take the consequences of uncertainty and the pattern of rewards and liabilities flowing therefrom can be considered to be taking a risk. Under certain conditions uncertainty can be commuted into a weaker condition defined as variability where the full range of options can be specified in advance, while the final choice remains contingent. Under certain other conditions, actors can insure their actuarial risk, thereby hedging their liabilities. In both cases, the level of risk taken by actors is reduced. In the cross-national comparison of phenomena linked to uncertainty/variability, a number points can be made. Firstly, there is evidence that behind the apparent similarity in the definition of the terms lie profound cultural differences. Hofstede (1984) notes that on a measure of "uncertainty avoidance", French managers score much more highly than the Americans and British. In other words, French managers might be expected to try much harder to avoid conditions where they face uncertainty than the British. This may be summarised in the aphorism that, for British managers, coping with uncertainty is the central management task, while, for the French, uncertainty is what you cannot manage. Secondly, there is the possibility that firms in different countries face differing levels of system-derived uncertainty. From a comparative study of Britain and France (Winch and Campagnac 1995), it would appear that the differences identified by Campinos-Dubernet are strongly linked to the ability of French contractors to choose the structural technology themselves, while such choices are the responsibility of the principal designer in Britain. In other words, French principal contractors may face less uncertainty simply because they have the power to choose the structural technology. By opting overwhelmingly for in situ concrete French contractors can raise transaction frequencies and move down the experience curve more rapidly than their British counterparts. They can, therefore, transact hierarchically for the construction of structural elements, while the British favour collaborative trilateral governance due to lower transaction frequencies. Thirdly, the strategic orientation of contractors is also correlated with extent of subcontracting. Campinos-Dubernet concludes that her findings show how contractor's strategies in the construction sector reflect the more general differences between the industrial strategies favoured by firms within the French and German economies, and financial strategies favoured by firms within the British and Italian economies - that there is a national effect in the strategic orientation of principal contractors. She does not elaborate the linkages between national system and principal contractor strategic orientation, and there can be considerable variation in the strategic orientation of principal contractors within a particular economy, as Üsdiken and his colleagues (1988) show for Turkey. However, the link between the national financial system and firm behaviour is now well established (Whitely 1992) and warrants further exploration in the context of construction.