How agents matter

1、chapter 7: hawkins and jacobychapter 7how agents matterdarren hawkins and wade jacoby* the authors are from brigham young university. they thank david lake, daniel nielson, michael tierney, andrew cortell, lisa martin, rachel cichowski, karen alter, jay goodliffe, mark pollack, kelly patterson, scot

2、t cooper, jon pevehouse, rachel epstein, and many others who commented during a series of presentations. camille jackson and anna sanders provided invaluable research assistance.in spite of the growing sophistication of the principal-agent (pa) literature, it still contains a remarkably thin view of

3、 agent behavior. that is, pa theorists have made surprisingly few direct claims about agents. almost twenty years after it was written, williamsons (1985, 30) pithy formulation - that agents are “self-interest seeking with guile” - remains the classic statement, and most current formulations do not

4、go far beyond it. for example, bergman, müller, and strøm (2000, 257) note that “delegation is often problematic. agents may have different interests from their principalsand/or the principal may be unable to observe the agents actions on his behalf”. mainly, the field has focused on what

5、principals can do to control such agents. these controls - including detailed rules, screening and selection, monitoring and reporting requirements, institutional checks, and sanctions, as detailed in the introduction - give us an indirect picture of agents as seen through the eyes of principals. wh

6、ile the indirect picture reinforces williamsons original notion of potentially troublesome agents, it also suggests that principals have many tools to control these agents.scholars have paid less attention to the strategies that agents use to try to circumvent these controls. agents often do more th

7、an just attempt to hide their information and their actions, as discussed in the introduction. in fact, as we discuss below, some agent strategies are not very hidden at all. other strategies are indeed hidden, but agents use different methods to cover their tracks. though scholars have made great e

8、fforts to articulate and describe a range of principal control strategies, as summarized in the introduction, a parallel effort needs to be made to understand agent strategies. moreover, a focus on principal control mechanisms privileges the ways in which principals design the contract governing age

9、nt behavior, essentially directing attention to moments of institutional creation. what happens between the creation moment and subsequent outcomes can depend on agent behavior and strategies. of course principals can later recontract, yet such recontracting is often quite difficult due to collectiv

10、e action problems among principals (nielson and tierney 2003a). if scholars are to successfully analyze the interaction between principals and agents, they need to understand agents in greater detail. other chapters in this volume identify agent characteristics as important explanatory factors - het

11、erogeneity of preferences for thompson, professional versus political staffing for cortell and peterson, and the particular nature of international courts for alter - while we focus on agent strategies.our central point is that principal preferences and control mechanisms alone cannot fully explain

12、which agents principals end up hiring or how those agents act once hired. more specifically, independent agent strategies can influence a principals decision to delegate and the agents level of autonomy. theorists who see agents as simply trying to hide information and action are likely to miss impo

13、rtant strategic interactions that alter pa outcomes. our arguments are relevant not only to the pa approach but also to broad theoretical debates about international institutions. much of the institutionalist literature is focused on theorizing state preferences and design of ios (abbott and snidal

14、2000; koremenos, lipson, and snidal 2001). we focus theoretical attention on ios as strategic actors with agency. ios matter not only because states have designed rules to resolve problems, but because those ios are themselves independent actors that interact strategically with states and others. wh

15、ile few might disagree with this contention, scholars have not yet - with a few exceptions (barnett and finnemore 1999 and 2004) - theorized the strategies of those actors and how they influence international outcomes. we share with barnett and finnemore (2004) a concern with taking ios seriously as

16、 agents. unlike those authors we focus not on the social knowledge that endows ios with authority (and that takes them beyond a principal-agent approach) but rather on the particular strategies that ios pursue in their relationship with states.most of us know intuitively that agent strategies differ

17、, and popular culture constantly reminds us. forrest gump is the perfect agent because he always does exactly what he is told with total commitment (and surprising competence). when his drill sergeant asks forrest what he thinks, forrest shouts, “whatever you tell me to think, sir!” to the sergeant,

18、 forrest is a “!%! genius.” to us, forrest is an agent who has no strategies that would trouble his principal. but forrest is unusual. much more common are what we might call the george castanza agent. like his seinfeld namesake, this agent is shiftless, marginally competent, and always on the take.

19、 screening and selection having already failed, close monitoring of this agent is a must. principals do get some work out of this kind of agent - otherwise they would terminate the contract - but its always a close call on whether the costs outweigh the benefits. pa theorists hardly expect to find m

20、any forrest gump agents - though like the sergeant, they know how to appreciate them when they see them - but they often describe george costanza agents, with whom the principal is never satisfied, but also not quite ready to abandon. more neglected in pa analysis is the kind of agent exemplified by

21、 the man with no name, made famous by clint eastwood. in high plains drifter, the man with no name is hired by a towns leading citizens to protect them against the outlaws who terrorize them. in a fistful of dollars, he is hired by each of two warring families to help in their fight against the othe

22、r. whether splitting principals (the town fathers) or playing them off against one another (the two families), this agent is a nightmare: he hides the way his preferences diverge from those of his principals, he waits for moments of maximum principal vulnerability to clarify contract terms, he embra

23、ces all the autonomy granted by the principals, and then uses his power to take more. self interested with guile, indeed. the point of the analogy is not to supplant negative stereotypes of io officials as feckless and incompetent with equally cartoonish pictures of them as ruthless and deceptive. t

24、he point is that agent strategies vary greatly and are likely to have some influence on outcomes. scholars may know this intuitively, but consideration of that variation has played little role in either pa or international institutions theory to date. agent strategies are likely to influence both pr

25、incipal delegation decisions and agent autonomy, and we adopt the introductions definition of both concepts. we argue that agent strategies can entice potential principals into delegating authority and then often increase the agents own autonomy once that authority has been delegated. while a whole

26、range of agent strategies are worthy of examination, we focus on four: interpreting principal mandates and other rules prior to delegation, reinterpreting those rules once states have delegated, expanding permeability (to non-principal third parties) and buffering (creating barriers to principal mon

27、itoring of agents). our point is not that principals are too dumb to anticipate these strategies and to devise counter-strategies. rather, given that all controls impose costs on the principals who use them and that as a result principals may not employ control mechanisms vigorously, we wish to iden

28、tify the strategies that agents use to exploit these difficulties in order to to secure delegated authority and to increase autonomy.for every endogenous aspect of the contract designed by principals, we stress an exogenous complement that can result from agent strategies. all principals propose a m

29、andate and a set of rules for agent behavior. as we show below, however, potential agents do not always wait patiently for principal delegation but seek to convince principals through principal-friendly interpretations that they will be excellent agents. on the other hand, once principals delegate m

30、ore authority, agents can openly attempt to reinterpret these mandates and rules in ways that increase their autonomy. similarly, principal monitoring through agent communication is a normal, endogenous part of most agency contracts. yet agents can develop strategies to buffer such monitoring, somet

31、imes by creating fairly elaborate organizational structures to raise the monitoring costs. finally, principals and agents exist in a broader political context, and principals may well allow agents to be permeable to certain third parties that can influence agent decision-making (gould 2003). since p

32、ermeability is a potential aid to principals - through the well-known mechanism of “fire alarms” (mccubbins and schwartz 1984), it is often endogenous to the agent contract. but agent strategies can also increase permeability in ways that let them serve third parties rather than principals. the pape

33、r has three sections plus a conclusion. first, we specify the scope conditions under which agent strategies will matter most. in the second section, we analyze the ways in which agent strategies may affect both delegation and autonomy. finally, we offer a case study of the european convention on hum

34、an rights. we show that the european human rights commission and court first helped convince states to delegate to them and then used three distinct strategies to gain greater autonomy. <a>when do strategies matter? agency costs and pool size as scope conditionsfor agent strategies to matter,

35、agents need leverage. in particular, the costs of creating new agents must be high compared to the costs of delegating to existing agents, and the pool of existing agents must be limited (see table 1). on the high costs of creating new agents, see keohane 1984 and weber 1994. this argument underpins

36、 all of the subsequent analysis and sets scope conditions on the arguments to come. where the cost of creating new agents is low or where the agent pool is large, agent characteristics and strategies are less likely to matter.table 1 near hereprincipals incur two types of costs in creating new agent

37、s: contracting costs and uncertainty costs. contracting costs include the time and resources required to negotiate with other potential principals and to set up new agents and new control mechanisms. although states delegating to existing agents must also pay contracting costs, we assume these are t

38、ypically lower than for new agents. principals that create new agents must negotiate fundamental agent characteristics, decision rules, funding methods, broad competencies, and decision-making structures. such features are already established for existing agents so that contracting costs are limited

39、 to negotiations over the task at hand. additionally, principals often minimize these costs by writing rules at the moment of agent creation about how new delegation should proceed (gruber 2000).uncertainty costs multiply these contracting costs for new agents. when creating new agents, principals a

40、re uncertain about whether those agents will operate in practice as they do on paper, whether the control mechanisms will work, whether other principals might have hidden agendas in agent creation, and whether and how the prospective agent will benefit the principal. with existing agents, principals

41、 have more information about the agents preferences and abilities, as well as the nature of the political interaction among principals or between principals and agents. as a result, principals have higher confidence in the predicted outcomes and - provided their preferences are sufficiently aligned

42、- are more likely to delegate due to the lower risks. where the cost of creating new agents is relatively high, a limited pool of existing agents further increases the importance of agent characteristics and strategies. where agent pools are large, screening and selection can work well as a control

43、mechanism. as pools diminish in size, however, screening and selection become increasingly irrelevant, and the characteristics of existing agents loom increasingly large. in both international and domestic politics, the number of available institutional agents is generally quite small; only a few bu

44、reaucracies or ios with the needed expertise are available for any given problem, and principals often lean on existing agents to take on new tasks. small pool size can also adversely affect other control mechanisms. sanctions, for example, are less effective when agents know that principals have fe

45、w other options. endless u.s. delays in u.n. budget payments would undoubtedly have been more effective in bringing reform if the united states could credibly threaten to use its money to employ other agents. the u.n. monopoly in so many issue areas makes its existing characteristics both more impor

46、tant and more difficult to change. moreover, agent strategies are likely to matter more in politics than in economics because the typical agent output is public policy, which is usually a monopoly good and rarely priced. we thank david lake for helping us make this point. this makes it difficult or

47、impossible for principals to compare alternative providers and measure efficiency.our argument so far suggests that limited agent pools can make agent characteristics and strategies more important to pa outcomes. this position echoes an early sympathetic critique of pa theories, which observed that

48、political bureaucracies are likely to be more difficult to control than economic agents and that political contexts work differently than economic ones (moe 1984). few have followed up this insight, despite the fact that a fair number of empirical studies have now shown that many political agents ha

49、ve more autonomy and slack than standard pa theories would suggest (eisner and meier 1990; hill and weissert 1995; krause 1996; rourke 1976; wood 1988). our approach also resonates with sociological approaches to organizational behavior, which have long emphasized the important role that organizatio

50、nal structure can have on political outcomes (perrow 1986; pierson 1996; dimaggio and powell 1991a; barnett and finnemore 1999).<a>how agent strategies influence delegation and autonomyscholars in both pa and international institutionalist traditions have focused on principal preferences to ex

51、plain delegation and control mechanisms to explain autonomy. invoking pa theory, pollack (2003a) argues that states delegate to ios to reduce transaction costs and gain credibility. abbott and snidal (2000) utilize institutionalist theory to reach a similar conclusion that states delegate to ios to

52、reduce transaction costs, strengthen credible commitments, and resolve problems of incomplete contracting. although based on a different approach, moravcsik (2000, 1997) also points to the importance of government (not state) preferences when he argues that governments delegate to ios to gain domest

53、ic policy lock-in. with respect to autonomy, pollack (2003a) echoes mainstream principal-agent explanations by arguing that state control mechanisms determine the scope of io autonomy. nielson and tierney (2003a) agree, but add that unresolved collective action problems among principals can provide

54、agents greater autonomy. without at all denying that principal preferences and control mechanisms are important, in this section we explain what happens when we reverse the causal arrows to treat agent strategies as independent influences on principal delegation and agent autonomy. we cover four suc

55、h strategies: agent interpretation of potential mandates prior to delegation, agent reinterpretation of their mandates once delegation has occurred, agent efforts to increase their permeability to third-parties, and agent efforts to buffer principal monitoring. these agent strategies can expand thei

56、r autonomy and/or persuade principals to delegate more authority to them. these strategies also can be used to circumvent principal control mechanisms. the strategies vary in interesting ways. interpretation and reinterpretation address ex ante principal controls (rules, screening and selection), wh

57、ile increasing permeability and buffering seek to circumvent ex post efforts (monitoring, sanctions) (strøm 2000). interpretation and reinterpretation are done openly, while buffering is generally hidden, with expanding permeability often a mix of overt and covert action. principals do not alwa

58、ys rebuff agent strategies for a variety of reasons, including a lack information on the extent to which their control structures have been challenged, the judgment that it is too costly to redesign or rebuild control structures, difficulties in reaching agreement within collective principals, or ev

59、en the conviction that the unexpected agent behavior actually suits their interests.<b>agents interpret and reinterpret rulesalthough the use of particular rules is a control mechanism in its own right, all of the control mechanisms developed in pa theory are based on rules, in a broader sense. monitoring and reporting requirements require rules - usually formalized and written down - about how much information agents must report and when and how to report it. institutional checks and balances rely on rules about which