There exists a fundamental tension between managerial efficacy and administrative control. Admittedly, both functions are equally critical for the efficient functioning of a modern state. The managerial approach is indeed based on the assumption of the pre-existence of a rule-based bureaucracy, an assumption that does not hold in most developing countries. The key question, however, is which of these functions ought to dominate at any given time. Their relevance in specific country contexts depends on a variety of factors linked to the stage of development of the state, including the quality of the bureaucracy, the strength of internal controls and adherence to the rule of law.
The 1992 reforms in Argentina prematurely ended ex-ante compliance control in a public administration lacking a culture of transparency and probity. Conversely, in Chile, the bureaucracy is highly respectful of the law (probably too rigidly), which makes it ripe for shifting to ex-post performance auditing. However, the deep-seated legalistic culture of control of the CGR privileges formal compliance with rules over modern audit techniques focusing on performance. The tragedy of most Latin American states is that they have been unable to achieve neither government restraint nor managerial autonomy, rendering performance-management a premature task. The critical issue, as Schick (1998) advises, is to ‘get the basics right first.’
Five main policy implications can also be drawn from this research. First, politics matter to explain the effectiveness of external auditing. It is thus necessary to take full account of the political context in which AAAs functions. Independent oversight of government finances is inherently political. It tackles embedded political interest and is itself embedded in a political context. Electoral incentives, the nature of the political system, the distribution of political power, and the degree of political competition have a determinant influence on the willingness of legislators to use external audit findings to enforce accountability on government. The failing of fiscal control in Argentina is a failure of politics. The case of Argentina also shows that the choice of audit model is a political decision.
Relations between governments and auditors-general are often tensed and conflictive. Independent-minded auditors generals must strike a delicate balance between a confrontational and an accommodating strategy to have a meaningful impact on public sector performance. At the same time, external audit can also be a dangerous endeavor in inauspicious political contexts, such as in Chile during the military dictatorship. Auditors must show both courage and prudence.
Second, to build effective AAAs, mobilizing political power is often more important than increasing technical capacity. It is important to address the underlying causes of poor performance, rather than focus on the symptoms. Strengthening technical capacity per se does not automatically improve the effectiveness of AAAs, nor has it prevented them from being captured. External auditing inserts itself in the politics of the budget and the logics of executive-legislative relations. The degree to which legislatures use audit findings to hold the government to account is likely to depend on the configuration of political power, the degree of political contestation and the extent of electoral competition.
Legislatures are more likely to develop technical capacity to oversee government finances if the opposition is sufficiently strong and cohesive. In situations of fused government, where executive and legislative majorities coincide, legislatures have fewer incentives to oversee government and develop oversight capacities, either directly through public accounts committees or legislative budget offices, or indirectly through the AAAs. In Argentina, legislators do not necessarily lack technical capacity, but rather political incentives to exercise the oversight powers they have and invest in the legislature’s oversight functions. Low re-election rates, intra-party dynamics, and the