18
Policy process perceptions of senior Canadian federal civil servants: a view of the state and its environment Abstmct: Based on interviews with senior civil federal civil servants in Ottawa, perceptions of various actors’ influence on public policy are examined. It is hypoth- esized that perceived influence patterns will change as one moves across the major phases of the policy process. Although it is expected that actors within the state will be of primary importance throughout the process, it is also expected that the relative importance of actors outside the state will increase as one moves from forming a policy agenda and selecting policy towards policy implementation. This should be particu- larly true of the perceived influence of private companies. Despite this change in perceived levels of influence of state versus non-state actors, the structure of correlation patterns for perceived influence patterns will probably not change dramatically from one policy process phase to the next. Results are also presented pertaining to the actual interaction patterns of the civil servants who were interviewed, as well as the relation between interaction and perceived influence. It is expected that there will be positive relations between these two types of variables but that direct interaction will not consistently explain the majority of variation in perceived influence variables. Sommuire : A la suite d’entrevues avec des hauts fonctionnaires fedtraux i Ottawa, on examine ici les perceptions quant il’influence de divers intervenants sur les politiques gouvernementales. On prend pour hypothese que ces perceptions Cvolueront P mesure que l’on avance d’une grande ttape i I’autre dans le processus de definition des politiques. On s’attend 1 ce que les intervenants etatiques jouent un r81e primordial tout au long du processus, mais aussi P ce que le r61e relatif des intervenants exterieurs au gouvernement prenne de I’ampleur lorsqu’on passe de I’ttablissement d’une liste de politiques au choix des politiques elles-mimes et, enfin, P la mise en oeuvre de ces politiques. Ceci devrait s’appliquer tout particulierement i la maniere dont on percoit I’influence des sociCtCs privtes. MalgrC ce changement du niveau d’influence percu entre les intervenants gouvernementaux et les nongouvernementaux, la structure corrklative des influences percues ne changera probablement pas dc facon dramatique d’une etape i I’autre du processus d’tlaboration des politiques. Scott Bennett teaches in the department of political science, Carleton University. Margaret McPhail is astudent in that department. This research was in part funded by Carleton University and the Social Sciences and Humanities Research Council. The authors appreciate this support as well as the comments provided by several anonymous referees for this journal. CANADIAN PUBLIC ADMINISTRATION / ADMINISTRATION PUBI.IQUE DU CANADA VOLUME 35, NO. ? I (AUTUMN/AUTOMNE). PP. 299-316.

Policy process perceptions of senior Canadian federal civil servants: a view of the state and its environment

Embed Size (px)

Citation preview

Page 1: Policy process perceptions of senior Canadian federal civil servants: a view of the state and its environment

Policy process perceptions of senior Canadian federal civil servants: a view of the state and its environment

Abstmct: Based on interviews with senior civil federal civil servants in Ottawa, perceptions of various actors’ influence on public policy are examined. It is hypoth- esized that perceived influence patterns will change as one moves across the major phases of the policy process. Although i t is expected that actors within the state will be of primary importance throughout the process, i t is also expected that the relative importance of actors outside the state will increase as one moves from forming a policy agenda and selecting policy towards policy implementation. This should be particu- larly true of the perceived influence of private companies. Despite this change in perceived levels of influence of state versus non-state actors, the structure of correlation patterns for perceived influence patterns will probably not change dramatically from one policy process phase to the next.

Results are also presented pertaining to the actual interaction patterns of the civil servants who were interviewed, as well as the relation between interaction and perceived influence. It is expected that there will be positive relations between these two types of variables but that direct interaction will not consistently explain the majority of variation in perceived influence variables.

Sommuire : A la suite d’entrevues avec des hauts fonctionnaires fedtraux i Ottawa, on examine ici les perceptions quant il’influence de divers intervenants sur les politiques gouvernementales. On prend pour hypothese que ces perceptions Cvolueront P mesure que l’on avance d’une grande t tape i I’autre dans le processus de definition des politiques. On s’attend 1 ce que les intervenants etatiques jouent un r81e primordial tout au long du processus, mais aussi P ce que le r61e relatif des intervenants exterieurs au gouvernement prenne de I’ampleur lorsqu’on passe de I’ttablissement d’une liste de politiques au choix des politiques elles-mimes et, enfin, P la mise en oeuvre de ces politiques. Ceci devrait s’appliquer tout particulierement i la maniere dont on percoit I’influence des sociCtCs privtes. MalgrC ce changement du niveau d’influence percu entre les intervenants gouvernementaux et les nongouvernementaux, la structure corrklative des influences percues ne changera probablement pas dc facon dramatique d’une etape i I’autre d u processus d’tlaboration des politiques.

Scott Bennett teaches in the department of political science, Carleton University. Margaret McPhail is astudent in that department. This research was in part funded by Carleton University and the Social Sciences and Humanities Research Council. The authors appreciate this support as well as the comments provided by several anonymous referees for this journal.

CANADIAN PUBLIC ADMINISTRATION / ADMINISTRATION PUBI.IQUE DU CANADA VOLUME 35, NO. ?I (AUTUMN/AUTOMNE). PP. 299-316.

Page 2: Policy process perceptions of senior Canadian federal civil servants: a view of the state and its environment

SCOTT BENNETT, MARGARET McPHAIL

On prksente tgalement des rksultats concernant l’interaction effective des fonc- tionnaires interview&, ainsi que le rapport en t re cette interaction e t l’influence percue. On prkvoit des rapports positifs entre ces deux groupes de variables, mais aussi que l’interaction directe ne saurait expliquer de maniitre cohkrente la plupart des variations dans les variables de l’influence percue.

The primary question addressed in this study is how senior civil servants view the relative importance of different types of policy process participants at different points in that process. We also wish to see whether the relations between indicators of perceived importance are essentially similar from one phase of the policy process to another. In addition, to put our main results in context, we want to describe some of the characteristics of civil servant interaction to determine whether this has a predominant influence on their perceptions of the policy process.

Our study has some points of reference in existing literature, but we go beyond that literature in several significant ways. Earlier work has been concerned mainly with describing interaction patterns of senior civil servants and trying to link such interaction to the attitudes of civil servants toward the political dimension of policy-making.’ In addition, there have been examina- tions of how different types of policy-making may be associated with different kinds of interaction patterns between bureaucrats, other state actors, and societal actors.* However, there is little systematic empirical analysis of how the relative importance of different types of actors, state and otherwise, might change during the different stages of a generic policy process which applies to’many types of policy-making.s It is this issue that we will emphasize in our discussion, concentrating on that part of our data which shows civil servant perceptions of actors in the policy process. Obviously, there are other vantage points from which this process can be viewed, but the senior officials in our sample provide one valid and informed perspective on the patterns of influence throughout the policy process.

1 For examples see: Robert Presthus, Elite Accommodation in Canadian Politics (Toronto: Macrnillan, 1973), pp. 244-66 and 283-332; Joel D. Aberbach, Robert D. Putnam and Bert A. Rockman, Bureaucrats and Politicians in Western Democracies (Cambridge, Mass.: Harvard University Press, 1981), pp. 209-37; and Michael M. Atkinson and William D. Coleman, “Bureaucrats and Politicians in Canada,” Comparative Political Studies 18, no. 1 (April 1985), pp. 58-80. Presthus was concerned with a somewhat different set of orientations towards politics than the other authors cited. 2 Michael M. Atkinson and William D. Coleman, “Is There a Crisis in Business-Government Relations?” Canadian Journal of Administrative Science 4, no. 4 (1987), pp. 321-40. The idea of variations in influence patterns by substantive policy type is also an important part of the policy communities literature. See A. Paul Pross, Groups Politics and Public Policy (Toronto: Oxford University Press, 1986). 3 In a limited way, the work of Presthus does probe the question of when the primary influence of different actors is felt in the sense that Presthus studied the kinds of contributions expected from different types of policy actors (see in particular Elife Accomniodafion, pp. 224-26).

300 CANADIAN PUBLlC ADMINISTRATION

Page 3: Policy process perceptions of senior Canadian federal civil servants: a view of the state and its environment

POLICY PROCESS PERCEPTIONS OF SENIOR CIVIL SERVANTS

To the extent that we can be guided by past research, we might expect that there would be some kind of political-administration distinction in percep- tions of influence across the policy process. Such distinctions have been found in characterizing the interactions and attitudes of senior bu rea~c ra t s .~ We would also expect that contact and perceptions of importance may be positively related. However, senior bureaucrats are well able to assess the importance of policy actors apart from their immediate contact with them. Particularly in terms of Canada, we would expect to find that individual businesses play a large part in the extra-bureaucratic interactions of senior bureaucrat^.^ However, this may be more important with some types of policy making that others.6

Yet these indications in existing work do not relate to changes in the perceived importance of policy actors as one moves from agenda formation to policy selection and on to policy implementation. Our expectation is that there will be changes in the perception of senior bureaucrats with respect to policy process participants as one moves through policy process phases. Specifically, we expect that, when one looks at a varied set of departments, actors within the state will always be viewed as having the most impact at every stage of the policy process, but there will be a distinct tendency for non-state or societal actors to increase in perceived importance as one proceeds from agenda setting to policy implementation. Further, we expect that, in light of previous Canadian findings, individual businesses will generally be perceived to be the most important of all non-state actors in every phase of the policy process. Alternatively, we do not expect that there will be major changes in the correlations between perceptions of the importance of different types of policy actors as one moves through the policy process. In other words, the structure of perceptions will have essentially the same form throughout the policy process, even though levels of particular variables are changing. Quite

4 See Aberbach, Putnam and Rockman, Bureaucrats and Politicians, and Atkinson and Coleman, “Bureaucrats and Politicians.” In general, the perspective emerging from such works does not suggest that bureaucrats are uninvolved in policy and political matters. Both politicians and bureaucrats have a concern with such matters but may play different roles in their development. In addition, there are certainly some bureaucrats who are more connected to political concerns than others, and this may be linked to variations in tolerance for political aspects of policy development. 5 Atkinson and Coleman “Bureaucrats and Politicians,” pp. 66-67. The data from this study suggest that Canadian bureaucrats tended to have fewer contacts external to the state than comparable officials in other developed democracies. However, this is primarily true with respect to contact with interest groups. With respect to contact with private firms, Canadian officials exhibit a relatively high rate of contact. 6 Atkinson and Coleman, “Business-Government Relations,”p. 332. The results of this study indicate that officials involved in regulatory policy rely relatively heavily on contact with business interest groups, while officials involved in distributive policy rely relatively heavily on contact with firms. Officials involved in state management policy rely less on both firms and interest groups than other officials, but have somewhat more contact with interest groups than with firms.

301 ADMINISTMTION PUBLIQUE DU C.4N.4D.4

Page 4: Policy process perceptions of senior Canadian federal civil servants: a view of the state and its environment

SCOlT BENNETT, MARGARET McPHAIL

likely that structure will be a function of distinctions in civil service percep- tions of political actors versus administrative actors, senior versus junior positions in a hierarchy and “insiders“ versus “outsiders.” Regarding this last distinction, in some instances the outsiders might be actors outside the federal state, but in some instances, relevant boundaries will be one’s own department rather than the federal state apparatus.

Our term for the types of policy process we expect to find is the “selectively permeable” policy process. It is a process in which non-state actors can have a substantial influence but only after basic parameters have been set. To some extent this agrees with those who have written about the autonomy of the state, but those concerned with state autonomy have not concentrated on the changing nature of autonomy throughout the policy process.’ Our work also has resonance with what is sometimes termed the “managerialist” perspective in that it does emphasize the interactions and perceptions of state officials within networks of organizations.’

It is useful to observe, as Pross has noted, that major participants in a policy community do not necessarily seek to exercise influence in the policy process to make or reformulate specific policies as much as they use a community to dampen the influence of those not strongly tied to the cornm~nity.~ However, this still constitutes influence on the various stages of the policy process, albeit a somewhat diffuse form of influence. In fact, this is a perspective which is generally consistent with the selectively permeable policy process. Key ele- ments of sub-governments act to control agenda-setting and policy selection by limiting the involvement of many actors external to the state. This sort of limitation may not only apply to unorganized components of the public but also to more organized non-state actors at certain phases in policy-making. If this is the case, then we will probably find some indications of it in the policy process perceptions of senior officials.

The study In order to address the questions of interest, personal interviews were conducted with sixty-eight individuals who were directors, directors general, and assistant deputy ministers situated in Canada’s National Capital Region and working in regular government departments.I0 These interviews were

7 See, for example, Eric Nordlinger, On the Autonomy ofthe Democratic State (Cambridge, Mass: Hanard University Press, 1981). 8 For a general discussion of the managerialist perspective, see Robert R. Alford and Roger Friedland, Powers of Theoq: Capitalisni, fhe State and Democracy (New York, NY: Cambridge University Press, 1985), pp. 161-267. Note that we do not take the view that this sort of perspective must be studied entirely in terms of organizational units o f analysis. 9 Pross, Groups Politics, p. 107. 10 By regular departments, we mean all agencies defined as departments or ministries. They are the direct responsibility of a minister. We excluded PCO and PMO, but did not exclude Treasury Board Secretariat which, for organizational purposes, was viewed as part of the Department of Finance. The sampling frame was the Government of Canada Telephone Directoq:

302 CANADIAN PUBLIC ADMINISTRATION

Page 5: Policy process perceptions of senior Canadian federal civil servants: a view of the state and its environment

POLICY PROCESS PERCEPTIONS OF SENIOR CIVIL SERVANTS

conducted between October 1989 and June 1990. Calculatcd in a conservative manner, the response rate for this randomly selected sample was 62 per cent.” Deputy ministers were excluded from the sample because earlier pretesting had suggested that deputies often wcre either reluctant to provide frank responses on certain topics or had insufficient time in their position to acquire a stable view of the interactions involved in filling the position. Apart from that, deputies’ placement on the border between bureaucratic and political cultures has often been used as a basis for distinguishing them from those immediately beneath them in department hierarchies.’*

There were several sets of questions included in the interview instrument. The major sets pertained to topics such as what types of actors a respondent had interacted with, for official purposes, during the twelve months preceding the interview, the relative importance of such interactions in terms of the time they consumed, and the respondent’s perceptions of how influential different types of actors were at different stages of the respondent’s department’s policy process. Other questions focused on characteristics of the respondent and her or his position, and on the perceived importance of informal interactions in the policy process. When it was relevant to define interactions in a particular set of questions, they were defined broadly - to include any sort of contact or communication.

The variables most directly related to our model of the selectively permeable policy process, and which will be the centre of our analysis, relate to respondents’ perceptions of the importance of various actors at different stages in the departmental policy process and, to a lesser extent, the variables relating to the occurrence and importance of respondents’ own interactions. Most of these central variables are measured using true ranking for~nats.’~

National Capital Regzon (June 1988). Specific departments sampled and yielding completed interviews are Agriculture, Communications, Consumer and Corporate Affaiis, Defence, Employment and Immigration, Energy, Mines and Resources, Environment, External Affairs, Finance (including Treasury Board Secretariat), Fisheries and Oceans, Health and Welfare, Indian and Northern Affairs, Industry, Science and Technology, Justice, Labour, Public Works, Revenue Canada, Secretary of State, Solicitor General, Supply and Services, Transport, and Veterans’ Affairs. 11 There are alternative, marginally different ways to compute response rates. Had we excluded selected cases where contact was not possible because of what was probably a vacant or abolished position, our response rate would be slightly higher. 12 For recent comments on the nature of the deputy minister position, see Gordon Osbaldeston, “Job Description of DMs, ‘Policy OpLio~(” (January 1988), p. 35, and Timothy Plumptre, “New Perspectives on the Role of the Deputy Minister,” CANADIAN PUBLIC ADMINISTKATION 30, no. 3 (Fall 1987). pp. 376-98. IS Whenever w e use rankings, the respondent is given a list of twenty-two types of actors who might be involved in the policy process or in the respondent’s work interactions. The respondent is asked to choose the top five and rank them in terms of either the importance of the policy process or in terms of the time they absorb in the respondent’s interactions. Any item not explicitly ranked in the top five is given a residual ranking of six. Experience with true ranking methods suggests that asking respondents to rank all twenty-two actors would have

303 ADMINISTRATION PUBLIQUE DU CANADA

Page 6: Policy process perceptions of senior Canadian federal civil servants: a view of the state and its environment

SCO'IT BENNE'IT, MARGARET McPHAlL

Our use of true ranking measures represent a departure from the fixed category ordinal measures used in many earlier studies. Our own experience with pretesting and an examination of other studies suggested that this was a realistic improvement. Under ideal circumstances, some of the data could best be gathered through a diary process. However, it would be difficult to secure the cooperation of busy executives in such an endeavour. Our own method provides an efficient way of measuring the relative importance of actors and interactions without pressing the limits of ranking procedures, without assuming that frequency is directly linked to total time consumed by interac- tion, and in a manner which engages the respondent's recall more naturally than many other formats.I4

The results presented here are based on relatively simple analytic tech- niques, ranging from basic univariate summaries to simple correlations. l5

These techniques provide appropriate tests of our main hypotheses in a clear and accessible manner. Additional tests with more complex techniques have been used with these data and will be pursued further. However, more sophisticated techniques such as factor or principal components analysis can be problematic when a large number of observed variables are being exam- ined in terms of a relatively small sample. In fact, this is a potentially important consideration in analysis of any small sample.16 In our presentation of results, the sample size is sixty-eight unless noted otherwise.

been overly burdensome, often irrelevant and may well have negatively influenced the flow of interviews and quality of data.

The statistical impact of this measurement approach is that it increases the probability of obtaining skewed distributions, and the implications of this for statistical techniques used in analysis were thoroughly explored and turned out to be minimal. In this context, one should note that even in the realm of fixed category ordinal variables, one often observes non- normalities in response distributions in terms of skewness, kurtosis, or both. 14 There are conceptual problems with the type ofmeasurement strategy which has dominated earlier studies ofbureaucratic interaction and related variables. At the most general level, many studies in this tradition concentrate on the relative frequency of interaction without consider- ing the overall amount of time devoted to particular interactions. Thus, there is an unexplored distinction, since the frequency of a particular type of interaction may not correlate very precisely with the amount of time absorbed by that type of interaction over a long period. In addition, the spec.ific fixed response categories in some earlier attempts to measure frequency of interaction appear to be problematic. 15 All of the correlations reported are Pearson product moment correlations. Even though most of the variables are ordinal, careful consideration of the overall analytic strategy' indicated that Pearson correlations were the most useful foundation for analysis. A variety of non- parametric alternatives, as well as alternatives based on transformed variables, were examined. There was almost no difference between these alternatives and straight Pearson correlations in terms of strength or significance. If anything, results from the Pearson correlations are slightly more conservative than the alternatives. 16 See James Stevens, Applied Multivariate Slatisticsfor the Social Sciences (London: Lawrence Erlbaum Associates, 1986). Stevens indicates that a conservative standard for the construction of reliable factors is that there should be at least four to five times as many observations as there are observed variables. Some elite samples in the social sciences would not meet this standard

304 CANADIAN PUBLIC ADMINISTRATION

Page 7: Policy process perceptions of senior Canadian federal civil servants: a view of the state and its environment

POLICY PROCESS PERCEPTIONS OF SENIOR CIVIL SERVANTS

Results Perceived importance of various actors throughout the policy process

Our major hypothesis was that actors within the public sector would always be the most important at every stage of the policy process. However, as one proceeds from the earlier to the later stages of the poIicy process, the relative importance of public sector contacts is expected to decrease compared to contacts with businesses and interest groups. We expect companies will be consistently more important than other non-state actors. The stages of the process that were explicitly dealt with were agenda-setting, policy selection, and policy implementation. The following table shows the most pertinent results, with low numbers representing high aggregate ranked influence and high numbers representing less aggregate ranked influence.

Table 1: Civil Servant Rankings of Perceived Imfiortance of Dgerent Actors at Three Stages i n the Policy Process

Aggregate Aggregate ranking at ranking at Aggregate

agenda- policy ranking at

Type of actor setting selection implementation stage stage stage

Own minister Other ministers Own minister's staff Other ministers' staff Own DM Other DMs Own ADM Other ADMs Own directors Other directors Junior staff Legislators in committee Legislators outside committee Committee staff Other federal staff

1 6

11 14 3 8 2

13 4 8 8

15 17 18 16

1 6

10 16 2

14 3

13 4 9

11 19 18 21 17

1 11 15 17 4

16 3

19 6

10 9

20 18 20 14

Other domestic governments 7 8 7

Private companies 5 5 2 Business interest groups 10 7 5 Non-business interest groups 10 12 8 Consultants 15 20 17 Other contacts 17 22 12

Foreign governments 12 15 13

in all aspects of analysis. I t is problematic for some parts of our study. This does not mean that factor and principal components analyses are inappropriate for this type of study, but i t does mean that they should be applied with caution when the variable to case ratio is very high.

305 ADMINISTRATION PUBLlQUE DU CANADA

Page 8: Policy process perceptions of senior Canadian federal civil servants: a view of the state and its environment

S C O l T BENNElT, MARGARET McPHAlL

Note that there are ties in some of the rank sequences. Contact types that are tied are given the same ranking number.

Our expectations are essentially confirmed by the results provided in the figures in Table 1. At every stage, a department’s own minister and senior civil servants are ranked in high influence positions. Other cabinet ministers and other levels of Canadian government have moderate influence at most stages. Other types of public sector contacts, particularly legislative ones, tend to be of minimal importance. This confirms the general dominance of public sector contacts in the policy process, but, as one would expect, some types of such contacts are much more important than others. In general, non-public sector contacts tend to move up in importance in the implementation stage, but private companies are by far the most important of such contacts at every stage. This is also consistent with a major aspect of our main hypothesis.

We will now examine the correlations between pairs of variables at each stage of the process. We turn first to variables for the agenda-setting stage, presenting only correlations which are significant at the .01 level or better (Table 2).

Table 2: Relations between Civil Seruant Perceptions of Actor Imj9ortance at the Agenda- setting Stage of the Policy Process

Strength of Variables in Relation Correlntion

Own minister - other minister Own minister - own DM Own minister - own ADMs

3 4 37 -37 -.45 Own minister - directors in own dept.

Own minister - sub-director staff in own dept. Other ministers - own ADMs Other minister - directors in own dept. Own DM - directors in other depts. Own DM - subdirector staff in own dept. Own DM - foreign governments Own DM - private businesses Own DM - business interest groups Own ADMs - directors in own dept. Own ADMs - sub-director staff in own dept. Directors in own dept. - sub-director staff in own department Directors in own dept. - other federal officials

-.48 - 3 5 - 3 4 - 3 5 -.43 -.32 -36 -30

.46 3 1 .56 .29

Directors in other depu. - foreign governments

Other federal officials - other contacts not specified elsewhere Foreign governments - private businesses

.33 3 1 .55 .54

Subdirector staff in own dept. - other contacts not specified elsewhere

Foreign governments - business interest groups Private businesses - private consultants

.4 1

.4 1 Business interest groups - private consultants .53

306 CANADIAN PUBLIC ADMINISTRATION

Page 9: Policy process perceptions of senior Canadian federal civil servants: a view of the state and its environment

POLICY PROCESS PERCEPTIONS OF SENIOR CIVIL SERVANTS

These correlations indicate some basic patterns exist in the perceptions of senior civil servants concerning agenda-setting in their departments. Clearly, there is a direct or positive relation between perceptions of one’s own minister’s impact and perceptions of the impact of other ministers. Similarly, the perceptions of ministerial influence on the agenda are directly linked to perceptions of one’s own deputy minister’s influence. However, below the deputy level, the correlations between perceptions of ministerial impact and other departmental staff are negative. Thus, for example, someone who sees her or his minister as having a high impact on agenda formation would view ADMs And directors as having a low impact and uice versa. There are also negative relations between perceptions of other ministers’ impacts and the impact of departmental staff below the deputy level. In a rudimentary sense, this suggests that there is a political versus regular civil service dimension to our respondents’ perceptions.

There is also an internal versus external dimension to the correlations, in that perceptions of departmental staff influence are negatively related to perceptions of agenda-setting influence by staff in other departments, busi- nesses, business interest groups, and foreign governments. On the other hand, the relations between perceptions of many of these external actors are positive, as are the relations between perceptions of different types of departmental actors. At this stage of the policy process, it would appear that the internal versus external distinction in perceptions has both a department versus non-department element and a state versus non-state element.

We now move on to the policy selection stage of the policy formation process (Table 3).

Table 3: Relations between Civil Seruant Perceptions of Actor Importance at the Policy Selection Stage of the Policy Process

Strength of Variables in Relation Correlation

Own minister - own DM Own minister - directors in own dept. Own ministers - directors in other depts. Other ministers - directors in own dept. Other ministers - legislators outside committee Own minister’s staff - other minister’s staff Own DM - own ADMs Own DM - subdirector staff in own dept. Own DM - private businesses Own ADMs - directors in own dept. Own Directors - subdirector staff in own dept. Sub-Director staff in own dept. - private consultants Other federal officials - other Canadian governments Foreign governments - private businesses Private businesses - private consultants

.31 -.30 -.30 -.33

.32

.38

.29 -.28 - 3 1 .39 .46 .3 1 .42 .45 .40

Business interest groups - private consultants .37

30’7 ADMINISTRATION PUBLIQUE DU CANADA

Page 10: Policy process perceptions of senior Canadian federal civil servants: a view of the state and its environment

S C O m BENNETT. MARGARET McPHAIL

There are major similarities between correlations at this stage and those already discussed pertaining to the agenda-setting stage of policy formation. Again, we see evidence of a political-regular civil service distinction in that the relationship between perceived policy selection importance of ministers and people at the director level is a negative one. The impact of one’s own deputy, probably seen primarily as a member of the political team, is inversely related to the impact of staff below the director. Alternatively, the perceived policy selection influence of one’s own minister and one’s own DM are positively related. Also, the relation between perceived policy selection influence of one’s own minister’s staff and the staff of other ministers is positive. There is also a positive link between perceived impact of other ministers and the impact of legislators acting in caucus and constituency roles. Relations between perceptions of impact of one’s own ADM, directors, andjunior staff tend to be positive, suggesting that people at this level are evaluated as being similar in their impact on policy selection. The only type of contact who is perceived to bridge the political and regular civil service domains in terms of having some positive correlation in perceived influence with each is the ADM. Apart from being evaluated similarly to lower level officials, there is also a positive correlation between perceptions of the impact of one’s ADM on policy selection and perceptions of the impact of one’s DM.

There is also evidence of an internal-external distinction in evaluating influence on policy selection. This is primarily manifested in a negative relation between the perceived importance of one’s own DM and the perceived importance of businesses. It is also manifested in a number of positive correlations between perceptions of different types of actors outside the federal government: businesses and foreign government, businesses and consultants, business interest groups and consultants, other types of federal agencies and other levels of Canadian government. The only civil service officials perceived as being similar to these external actors in impact are directors in one’s own department. There is a positive correlation between perceptions of their impact on policy choice and perceptions of the impact of consultants. Overall, it appears as though this external-internal dimension also has department versus outside department as well as federal state versus outside federal state elements. However, these elements have a somewhat different mix than was the case in the agenda formation stage of the process.

Finally, we turn to correlations between pairs of variables reflecting the perceived influence of different types of actors on policy implementation (Table 4). With some variation, the correlations are similar to what was presented for earlier stages of the policy formation process. There are generally strong positive relations between indicators of perceived influence for the political apex of the department. Correlations between indicators of perceived civil service impact below the political level also are positively related. There are some linkages between perceptions of different types of senior officials in other departments. In the same manner, correlations among

308 CANADIAN PUBLIC ADMINISTRATION

Page 11: Policy process perceptions of senior Canadian federal civil servants: a view of the state and its environment

POLICY PROCESS PERCEPTIONS OF SENIOR CIVIL SERVANTS

perceptions of actors outside of the federal government, in its environment, tend to be positive.

Table 4: Relations between Civil Servant Perceptions of Actor Importance at the Implemen- tation Stage of the Policy Process

Strength of Variables in Relation Association

Own minister - other minister Own minister - other ministers’ staff Own minister - own DM Own minister - sub-director staff in own dept. Own minister’s staff - other ministers’ staff Own DM - own ADMs Own DM - private businesses Own DM - business interest groups Other DMs - ADMs in other depts. Other DMs - directors in other depts. Own ADMs - directors in own depts. Own ADMs - sub-director staff in own dept. Own ADMs - private businesses Directors in own dept. - sub-director staff in own dept Foreign governments - private businesses Private businesses - business interest groups Private businesses - private consultants Business interest groups - private consultants

3 7 .33 3 2

-.30 .5 1 .57

-.36 -.30 .59 3 9 .44 .29

-.40 .67 .29 .32 3 2 32

Turning to correlations between blocks of variables representing different types of actors, the correlation between the political apex and the federal state environment is weak or negative, as partially exhibited in correlations between one’s own DM and private businesses or business interest groups. Interest- ingly, this sort of correlation also extends a bit further down the ranks of the departments, as there is also a negative correlation between ADMs in one’s own department and private businesses. A similar significant correlation does not occur for ADMs in the earIier phases of the policy process. There is only one negative correlation between a political apex indicator and an indicator for the civil service core of the department, and that is between perceptions of one’s own minister and perceptions of sub-director staff in one’s own department. These sorts of correlations were more common in the earlier phases of the policy formation process. Apparently, the perceived divergence in influence of the political apex and civil service core narrows somewhat as one proceeds to implementation. This is probably due to the fact that all of these variables are moving down in relative importance compared to indicators of non-state influence.

In summarizing this part of the results, the initial hypotheses about correlations at each stage of the policy formation process were, by and large, supported. There tended to be positive correlations within blocks of related

309 ADMINISTRATION PUBLIQUE DU CANADA

Page 12: Policy process perceptions of senior Canadian federal civil servants: a view of the state and its environment

SCOTT BENNElT, MARGARET McPHAIL

variables and some negative correlations between blocks. Certainly, the correlations were not exactly the same in comparing one stage to another, but the general patterns were similar.” Probably the most striking things to note are that there were not more negative correlations between variables in different blocks and that perceptions of the impact of legislators acting in committee, caucus, or constituency roles had so little relation with other perceptions at any stage. Apart from one positive correlation between the perceived policy selection impact of other ministers and the perceived policy selection impact of legislators acting in caucus or constituency roles, there were no significant correlations involving perceptions of legislators’ impact on policy.

An overview of respondent interactions Although we are primarily concerned with respondents’ perceptions of other actors as opposed to their own interaction patterns, a summary of their typical interactions will better characterize the type of people included in our study and part of the basis for their perceptions. We asked respondents whether they had interacted with certain types of other actors in the course of performing their official work during a twelve-month period (Table 5) . Members of our sample are commonly involved in a wide range of interac- tions.I8 Percentages of respondents involved in a certain type of contact don’t drop below 40 per cent until we reach the three least common types of contact - contact with parliamentary committee staff, parliamentary commit- tees, and cabinet ministers other than one’s own. We also see that the eight highest percentages are all associated with public sector type contacts. Note that not all of the items in the top eight represent contacts within a respond- ent’s own department. So we are not simply seeing a trivial effect due to obvious physical or organizational proximity. This clearly, and not surpris- ingly, indicates the state or inward-oriented focus of many of the interactions in which civil service executives are involved.

17 By way of background, the basic similarity of the correlational structures is revealed in principal components analysis. Approximately the same number of strong components are extracted from each stage of the policy process - eight in the agenda stage, nine in policy selection and seven in implentation. The substantive meaning of these components is quite similar across phases, and there are always two components at each stage which are much stronger than the others. Although generally informative, the large number of reasonably strong components underlines our earlier point about caution in using factor analysis or principal components analysis with small elite samples and large numbers of observed variables. It also shows that simple correlations have a certain interpretive economy which may exceed that of factor or principal components analysis when there are a large number of acceptable factors or components. At least, this may be true unless one is already testing highly structured hypotheses and can impose some constraints on the factor solution. 18 Supplemental analysis showed that the average number of types of interactions for members of our sample is 9.75, and everyone was involved in at least four types of interactions.

310 CANADIAN PUBLIC ADMINISTRATION

Page 13: Policy process perceptions of senior Canadian federal civil servants: a view of the state and its environment

POLICY PROCESS PERCEPTIONS OF SENIOR CIVIL SERVANTS

Table 5: Percentage of Senior Ofjcials Having Certain Types of Contacts

Type of Contact Percentage with Type of Contact

Interacts with directors in other departments 100.0 95.6 95.6 94.1 85.3 83.8 83.1 80.9 79.4

Interacts with directors in own department Interacts with ADMs in own department Interacts with civil servants below director in own department Interacts with civil servants below director in other departments Interacts with staff of own cabinet minister Interacts with other governments in Canada or elsewhere Interacts with own deputy minister Interacts with executives of private companies Interacts with interest groups Interacts with ADMs of other departments Interacts with private consultants Interacts with deputy ministers from other departments Interacts with cabinet members’ staff from other departments Interacts with minister of own department Interacts with other kinds of federal officials Interacts with MPs/senators on constituency or caucus matters Interacts with parliamentary committee staff Interacts with MPs/senators on committee matters Interacts with cabinet ministers from other departments

77.9 76.5 70.6 52.9 48.5 47.1 47.1 42.6 35.3 30.9 27.9

We also examined the correlations among the variables for which we have just presented the univariate results. Although these relations are not central to the themes developed in this paper, they do contain a few points of interest. First, there is a large number of significant correlations (forty-four of them at the .01 level), and nearly all of the correlations are positive. This means that involvement in one type of contact does not greatly diminish the probability ofbeing involved in other contacts, at least to a limited degree. A second point of interest is that many of the significant correlations are anchored in the political apex of departments. They tend to involve ministers or their staff. This probably means that those with a large number of contact types tend to also have contacts with the political apex. Finally, none of the significant correlations involve interactions with private businesses. In the context of other results, this probably means that people with many different types of contact patterns have about the same likelihood of interacting with private busines~es.’~

What happens when we look at these types of interactions in terms of ranked differences in time allocated to each type of interaction rather than in terms of simple presence or absence of some interaction? Respondents were asked

19 O n a related point, see Atkinson and Coleman, “Bureaucrats and Politicians,” p. 69. They indicate that both senior and junior officials have substantial contact with businesses and interest groups, but senior officials are somewhat more likely than junior to interact with chief executive officers of companies.

31 1 ADMINISTRATION PUBLIQUE DU CANADA

Page 14: Policy process perceptions of senior Canadian federal civil servants: a view of the state and its environment

SCOTT BENNETT, MARGARET McPHAIL

to use our standard ranking procedure to rank the top five contacts in terms of time involved in contact, with those below the fifth rank being placed in a sixth category.

We expect that interactions with public sector actors should generally have a higher ranked time allocation than interactions with private sector contacts. The results are shown in Table 6 in which different types of contacts are listed, starting with those that have the highest ranked time allocation for interaction and proceeding to those with lower ranked time allocation. Actors with the same average ranking are given the same tied rank in the table.

Table 6: Ranked Importance of DifJerent Types of Interactions in T e r m of Time Allocated to Such Interactions

Type of Contact Aggregate Rank

ADM in own department Director in own department Junior staff in own department DM in own department Minister in own department Directors in other departments Private businesses Staff of own minister Other levels of Canadian government Business interest groups Other types of federal officials Foreign governments Consultants ADMs in other departments Non-business interest groups MPs/senators on committee business DMs in other departments Other contacts not listed MPs/senators on caucus or constituency business Ministers in other departments Staff of parliamentary committees Staff of other ministers

1 2 3 4 5 6 7 8 9

10 11 11 12 13 14 15 16 17 18 19 19 20

Generally, the figures in Table 6 tend to confirm our expectations. The top five aggregate rankings all refer to types of contacts within a respondent’s own department, and these, of course, would all be considered public sector contacts within the federal state. The contact with the sixth highest aggregate ranking is “directors in other departments.” So the most frequent contacts of our sample are other people in the federal state. In particular, they are primarily involved in relationships within their own departments, with a secondary horizontal dimension ofcontact linking them to people with similar positions in other departments. This pattern is diluted somewhat as we move

312 CANADIAN PUBLIC ADMINISTRATION

Page 15: Policy process perceptions of senior Canadian federal civil servants: a view of the state and its environment

POLICY PROCESS PERCEPTIONS OF SENIOR CIVIL SERVANTS

into the less frequent types of contacts, with many other types of public sector contacts ranking behind time allocated to interaction with private companies and business interest groups. Certainly, contacts with Parliament and its staff do not absorb much time on the average, and contact with certain levels of officials in other departments is also relatively limited.

We also ran correlations among these ranking variables, but the results are not of direct salience to this paper. It is useful to note that the results were not inconsistent with the correlations among the presence/absence of interaction variables. However, there were fewer significant correlations and more of a balance between positive and negative relations. This is not unexpected in that variables reflecting ranking of time usage enforce a greater degree of selectiv- ity. Rankings by their very nature will mean that high rankings allowed for some interactions indicate an absorption of time which is then not available for other interactions.

The relation between contact and perception

For both methodological and theoretical reasons, it is important to consider whether the perception of influence variables examined earlier are largely a function of direct interaction with certain types of policy actors. The idea that the two might be related for theoretical reasons goes back at least to the work of Presthus.*O From a methodological perspective, we might also want to establish whether perceived importance was largely a function of whom respondents dealt with. Ideally, we would hope that the relationship is not a perfect one because we wanted civil servants to make judgments about the overall policy process in a manner that was not totally a function of their own direct contact patterns in work. Valid perceptions of policy importance can be formed from a great many sources of information apart from direct contact.

In order to address this question, w e ran correlations between each ranked perception variable and each ranked time of contact variable, and results are given in Table 7. Correlations with a double asterisk are significant at the .01 level using a two-tailed test, and a single asterisk indicates the .05 level of significance using a two-tailed test. Certain variables are not reported because they had insufficient variation, and the same is true for any correlation in the table designated with the word "none."'

20 Presthus, Elite Acconrnrodalion, pp. 92-94.

313 ADMINISTRATION PUBLIQUE DU CANADA

Page 16: Policy process perceptions of senior Canadian federal civil servants: a view of the state and its environment

SCOTT BENNETT, MARGARET McPHAIL

Table 7: Relations between Civil Seroant Policy Process Perceptions and Civil Sewant Time of Interaction Rankings

Correlation Correlation Correlation at at Agenda at Selection

Contact Type Phase Phase

Own minister .38** .46** Other minister .04 .02 Own minister's staff .59** .17 Own DM .32** .37** Other DMs .29* -.03 Own ADMs .15 .23

Own directors .38** .26* Other directors .31* .52** Own sub-director staff .39** .38**

Legislators in caucus or constituency .70** .15 Other federal officials .35** .45**

Foreign governments .79** .87** Private businesses .50** .34**

Other ADMs .06 -.08

Legislators in committee .27* .15

Other Canadian governments .76** .47**

Business interest groups .75** .51** Non-business interest groups .10 .37** Private consultants .19 .26* Other contacts -.01 (none)

Implementation Phase

.26*

.05 -.08

.22

.12

.23 -.06

.26*

.41**

.39** (none)

.67**

.34**

.69**

.76**

.31**

.41**

.10

.70** -.04

The table reveals predominantly positive relations between perception and interaction. We find a mixture of strengths in the correlations, ranging from extremely strong to very weak. In general, there is a relationship between perception and interaction, but there are certainly other elements that enter into our respondents' perceptions of the policy process apart from their direct interactions.

Reasonably strong correlations appear for a variety of types of actors. However, it would appear as though interaction and perception are particu- larly strongly linked with respect to actors outside the federal state. For example, this is true of other levels of Canadian government, foreign govern- ments, businesses, and business interest groups. On the margin of the federal state, there is also a strong link between interaction with and perception of legislators acting in caucus or constituency capacities at the beginning and end of the policy process. This probably reflects the fact that there are few institutional rules that structure interaction with these actors. If they are viewed as being important, such an assessment is more likely to arise from actual interaction rather than the residue of institutional rules.

Conclusion Our central hypothesis related to the idea of the selectively permeable policy process. We expected that there would be variations in the perceived influence

314 CANADIAN P u n m ADMINISTRATION

Page 17: Policy process perceptions of senior Canadian federal civil servants: a view of the state and its environment

POLICY PROCESS PERCEPTIONS OF SENIOR CIVIL SERVANTS

of actors across different phases of the policy process. In particular, it was expected that actors within the federal state would always be viewed as important throughout the policy process but that the relative importance of actors outside the state would increase as one moved through the phases of that process. Private businesses were expected to be consistently the most important non-state actors.

This central hypothesis was, for the most part, confirmed. It is fair to say that those who are near the apex of a respondent’s department are seen as having a major impact at all stage of the policy process. However, it is also true that actors outside the state, particularly private businesses, are seen as having an increased impact at the implementation stage of the process. The correlations among indicators of perceived influence at each stage tend to show that there are hierarchical, political versus administrative, and insider versus outsider distinctions in perceptions. The insider-outsider distinction has sometimes had both department versus non-department and federal state versus outside federal state elements. These patterns vary somewhat by phase, but are generally consistent across agenda-setting, policy choice, and implementa- tion. At implementation, the distinction between the political apex and the civil service is not as clear, but the distinction between the state and its environment is enhanced in correlational patterns.

We also found that our respondents were themselves often involved in a wide variety of interactions. Not unexpectedly, the interactions that con- sumed the greatest volume of time tended to be those within a respondent’s own department or with peers in other departments. However, private businesses had a fairly high ranking in terms of time of interactions, confirm- ing their relative importance as elements in the environment of the state.

When we examined the relations between rankings of time devoted to different civil service interactions and civil service perceptions of impact on policy process stages, we found that some strong positive relations existed. Yet in many instances it is clear that perceptions of impact are not predominately determined by one’s own direct interactions. Other factors of likely impor- tance are the respondent’s rank, type of department, position, and a variety of personal characteristics such as type of education. We have data on these variables, but a proper elaboration of their influence requires separate treatment in a more elaborate modelling exercise.

Thus, the main thrust of our theory is supported. The state is dominant throughout the policy process, but, in relative terms, recedes in importance as implementation of policy is reached. Environmental influences, particu- larly business influences, seem to become more important once the formal outlines of policy have been set. It is not difficult to appreciate that, in an era of increased emphasis on privatization, external interests may play an impor- tant role in determining the details of policy implementation, including delivery. Yet we also suspect that this is probably not a new phenomenon.

In closing, it is important to recall that the data do in fact represent the

315 ADMINISTRATION PUBLIQUE DU CANADA

Page 18: Policy process perceptions of senior Canadian federal civil servants: a view of the state and its environment

SCOTT BENNETT. MARGARET McPHAlL

perceptions of senior civil servants, and other types of actors might well hold alternative views. This is worth exploring in the future. It is also interesting to consider whether the various types of actors we have identified as being part of the state can be viewed as also representing environmental interests because of professional or ideological attachments. It is easily appreciated that senior officials may come to represent views that have a great deal to do with their professional attachment to civil society or that the political apex of the department will be involved in career-long civil society alliances that influence the general thrust of policy. It is our intention to consider such issues in additional research, but we do not expect that such considerations will totally dilute the impact of patterns within the state.

3 16 CANADIAN PUBLIC ADMINISTRATION