Scholarly attention in the early 1980s was focused on the need to more precisely categorize types of policy instruments in order to better analyze the reasons for their use (Salamon 1981; Tupper and Doern 1981; Trebilcock and Hartle 1982; Bressers and Honigh 1986; Bressers and Klok 1988). Careful examination and systematic classification of implementation instruments and instrument choices, it was argued, would not only lead to insights into the factors driving the policy process and the characterization of long-term patterns of public policy-making, as Lasswell had hoped, but would also allow practitioners to more readily draw lessons from the experiences of others with the use of particular techniques in specific circumstances and hence improve policy designs and outcomes (Mayntz 1983; Linder and Peters 1984; Woodside 1986).

During this period studies in Europe and North America shed a great deal of light on the construction and establishment of regulatory and other political and administrative agencies and enterprises; traditional financial inducements, and the ‘command-and-control’ measures adopted by administrative agencies, during this period (Tupper and Doern 1981; Hood 1986; Howlett 1991; Vedung 1997; Landry et al. 1998). And this new emphasis upon the systematic study of policy instruments quickly generated a sizable academic literature and resulted in immediate application in the design of many new policy initiatives in emerging areas such as pollution prevention and professional regulation (Trebilcock 1983; Hippes 1988). Significant subjects such as the reasons behind shifts in patterns of instrument choices associated with the waves of privatization and deregulation which characterized the period also received attention in this period (Howlett and Ramesh 1993).

Soon the field of instrument studies had advanced enough that Salamon (1989) could argue that the ‘tools approach’ had become a major approach to policy studies in its own right, bringing a unique perspective to the policy sciences with its focus on policy outputs. At this point he framed two important research questions to be addressed in future analyses of the tools of government action: ‘What consequences does the choice of tool of government action have for the effectiveness and operation of a government program?’ and ‘What factors influence the choice of program tools?’ (265). These questions were taken up by the ‘tools approach’ and the policy design literature in the 1990s.

Assessing and answering Salamon’s questions required scholars interested in policy design to engage in a lengthy process of social scientific analysis and model-building related to the study of implementation tools. These efforts expanded the number of preliminary questions which needed to be answered before Salamon’s queries could be addressed (Salamon 1981; Timmermans et al. 1998; Hood 2007) to include:

1. What potential tools does any government have?
2. How can these be classified?
3. How have these been chosen in the past?
4. Is there a pattern for this use?
5. If so, how can we explain this (or these) pattern(s)?
6. Can we improve on past patterns of use?

In order to answer these questions, policy scientists pursuing the tools approach followed, not necessarily as systematically as might be hoped, a five-stage research and analytical model-building strategy; one quite typical of the social sciences (see Figure 1 below). Each of the stages in this process is set out and described below.

 Figure 1: Analytical steps in social science model-building (modified from McKelvey, Bill, 1982)