institutionalizing public accountability - white paper

7/31/2019 Institutionalizing Public Accountability - white paper

1/82

Institutionalizing Public Accountability -

Why citizen-oriented, independent external public auditing

thrives in some countries and fails in others

Katharina Noussi, BA, MASDoctoral Candidate, Institute of Political Science, University of Vienna, Austria, [email protected];[email protected]

Paper prepared for delivery at the 26th Annual International Consortium on Gov-ernmental Financial Management (ICGFM) International Conference: Public Finance

Management (PFM) in the 21st Century - The PFM Architecture, Institutions, andTools to Meet the Challenges of the Modern Worlds, Sunday, April 29 Friday, May4, 2012; Miami, Florida USA; www.icgfm.org
mailto:[email protected]:[email protected]://www.icgfm.org/http://www.icgfm.org/mailto:[email protected]


2/82

Katharina Noussi 1

Contents: Page

1. Introduction 2

2. The concept SAIs as accountability arrangements 3

3. The problem ISSAI implementation 7

4. Hypotheses development 11

5. Exploring statistical data 22

6. What the heads of SAIs think 34

7. Conclusion 41

Appendix A: Content analysis of the ISSAIs 43

Appendix B: Dependent Variables measuring SAI power 45

Appendix C: Independent Variables 53

Appendix D: Regression tables 66

References 78

Support for participation at the conference was provided by the University of Vienna and the Austrian Political

Science Association (OEGPW). Special thanks go to all anonymous interview partners, as well to Sophia Gollwitzer,

Jack Mills, Paolo de Renzio and particularly the INTOSAI Development Initiative for the provision of data. I also thank

my thesis supervisor Otmar Hll, director of the Austrian Institute for International Affairs (oiip), for his continuous

encouragement and advice and Joachim Wehner, Stephen Krasner and Martin Eijnar Hansen for advice and helpful

comments during various stages of the research. All errors remain the authors exclusive responsibility.


3/82

Katharina Noussi 2

1. Introduction

Supreme audit institutions (SAIs) such as auditors general and courts of audit werecreated to conduct external audits of public administration. External auditing is notan end in itself; it is ideally a component of a regulatory system of public finance

and administration oversight that includes internal auditing, parliamentary re-views, public debate facilitated by the media, judicial prosecution, and other mech-anisms.

The declared objective of this conference is to learn about the PFM institutionsthat provide the foundation for meeting the demands for greater accountability,

transparency, and better management of public resources. Supreme audit institu-tions undeniably are among the institutions of interest here. But do they meet thedemands for greater accountability, transparency, and better management of pub-lic resources? If yes, to what extent? Can we generalize or do the experiences varyacross regions, institutional models or along other factors? The question of effec-

tive, independent public auditing is furthermore of high political relevance, if wethink about the European and US debt crises and the uprising in the Arab world,which was partly sparked by the public demand for more democracy, public ac-countability and corruption control.

The intention of this paper is to share parts of my doctoral research findings withthe community of public finance experts and scholars. My personal contributionsto the expertise provided by public finance professionals and scholars are insightson processes of political and institutional development. Understanding the politicalfactors which support or constrain SAIs helps to draw up context specific reformstrategies. At the same time the author hopes to receive valuable feedback and crit-

ical comments from the conference participants which will be included in the re-search and round up the dissertation project.

My research applies the mixed method approach ofpraxeology(Bourdieu, 1973 inHarrits, 2011) and thus integrates quantitative and qualitative methods with thepurpose to incorporate reflexivity into objective knowledge. This paper starts bydiscussing the concept of accountability in order to analyze whether supreme auditinstitutions actually understand themselves as accountability arrangements, and ifyes, with what intention. I conclude that SAIs do indeed thrive to be accountabilityarrangements with the main purpose of guaranteeing constitutional accountability.I then proceed to examine the extent to which they meet their aspirations. Unfor-

tunately, there still seems to be an implementation gap between the intended goalsas laid out in the International Standards of Supreme Audit Institutions (ISSAIs)and the recent UN General Assembly Resolution on the independence of SAIs andactual auditing practices in many countries. At the same time the findings also sug-gest a strong variation across countries with similar levels of democratization andsocio-economic development.

This leads me to the main research question of this paper, Why does independentexternal public auditing thrive in some countries and fail in others? I propose sev-en main hypotheses which explain the differences across SAIs. I tested these hy-potheses through multiple regression analysis using data mainly from the Open

Budget Survey (IBP, 2010) to develop my dependent variable measuring the pow-er of SAIs. I cross-checked the results on other dependent variables measuring thepower of SAIs, such as the PEFA reports, the Global Integrity reports, the OECD


4/82

Katharina Noussi 3

Budget Database and an anonymized dataset from the INTOSAI-Donor CooperationStocktaking Report 2010.

In a next step I compared the insights from the statistical analyses with insightsfrom the practice of political reforms based on 17 confidential expert interviews

with heads of supreme audit institutions (SAIs) from six INTOSAI regions. This es-say concludes that both the quantitative as well as the qualitative research suggestthat inter-elite relationships and (SAI) leadership conditions the success or failureof SAI reform initiatives. I close by proposing four strategies for effectively consol-idating SAI reforms.

2. The concept of public accountability

Accountability has become a very fashionable word and concept during the lastdecades. For policy makers accountability has become an easily used and seldom

reflected icon for good governance (Dubnick, 2002). It serves as a synonym formany other words such as responsibility, responsiveness, answerability, transpar-ency, controllability, fidelity or participation and thus the concept is not adequatefor discourse analysis.1 Accountability today stands as a standard term for anymechanism that makes powerful institutions responsive to their respective pub-lics. (Mulgan, 2000, 2003:8)

Equally, scholars of various academic disciplines such as public administration,law, political science, education and psychology use the concept widely, oftenwithout reflecting about the meaning of it. Accountability is used in different con-texts, with a wide variety of different understandings and nearly every author sets

out to produce his or her own definition of accountability.2 This tendency has beena strong impediment to systematic comparative, scholarly analysis. However, dur-ing the last few years the scholarly interest of overcoming this deficit has grownand there is currently a lively cross-disciplinary debate about the content and lim-its of accountability, as well as a first consensus among the main scholars in thefield that there are broadly speaking two main concepts of public accountability:accountability as a virtue and as a mechanism.3

In the former case, mostly dominant, but certainly not exclusively, in the Americanacademic and political discourse, accountability is used as a normative concept. Itoften serves as a set of standards for the evaluation of the behavior of public ac-

tors, being accountableis seen as a virtue, as a positive quality of organizations orofficials. In British, Australian, Canadian and continental European scholarly de-bates, accountability is more often used in a narrower, descriptive sense. The focusof this strand of accountability studies is on the way in which institutional ar-rangements operate and accountability is seen primarily as a social mechanism.4

1 Behn 2001:3-6; Dubnick 2007a, Mulgan 2000:555; all qtd. in Bovens (2007:453-454)2 Bovens, 2010: 946 quoting Behn 2001:3-6; Dubnick 2005; Mulgan 2000:555, Pollitt 2003:89; and

cf. Biela and Papadopoulos 2010:4 quoting Bostrom and Garsten 2008; Dowdle 2006b; Ebrahim

and Weisband 2007; Curtin, Mair et al. 20103 Dubnick and Frederickson, 2010, in Bovens, 20104 Bovens, 2010:947-8 based on Aucoin and Jarvis 2005; Bovens 2007b; Day and Klein 1987; Goodin

2003; Mulgan 2003; Philp 2009; Scott 2000


5/82

Katharina Noussi 4

While both concepts, accountability as a virtue and as a mechanism, are useful forthe study of democratic governance, each of them addresses different issues andimplies different sorts of standards, frameworks and analytical dimensions. Theinterest of this paper is to analyze accountability in the second sense, thus as aninstitutional arrangement or mechanism. Here accountability refers to a specific

set of social relations, of which the most concise definition is, the obligation to ex-plain and justify conduct.5 This basic definition implies a relationship between anactor, the accountor, and a forum, the account-holder or accountee6. On the basis ofthis definition Bovens (2007) has developed a more elaborate and now widelyused definition of accountability arrangements:

Accountability is a social relationship between an actor and a forum, in

which the actor has an obligation to explain and to justify his or her conduct,

the forum can pose questions and pass judgment, and the actor may face con-

sequences.7

Thus, a relationship qualifies as an accountability mechanism if it meets the follow-ing criteria:8

YES NO

Is there explanation and justification ofconduct by the actor?

Is there propaganda, or only the provi-sion of information or instructions tothe general public?

Is the explanation directed at a specificforum?

Is the explanation given at random?

Does the actor feel obliged to come

forward?

Is the actor at liberty to provide any

account whatsoever?Is there a possibility for debate andjudgment by the forum?

Is there only a monologue without en-gagement?

Is there the optional imposition of (in-formal) sanctions or rewards?

Are there no consequences?

THREE PURPOSES OF ACCOUNTABILITY

An altogether different question is the assessment of the adequacy of a particularaccountability arrangement9. Aucoing and Heintzman10 propose to engage in such

an evaluation along three main purposes of accountability:

5 Bovens, 2005:184; 2007:450 The Oxford Dictionary of English (revised second edition 2006) pro-vides a very similar definition: accountable > adjective 1 Required or expected to justify actionsor decisions; responsible: ministers are accountable to Parliament | parents cannot be held ac-countable for their childrens actions. 2 able to be explained or understood.DERIVATIVES ac-countability noun, accountably adverb.

6 Pollitt, 2003:89, qtd. in Bovens, 2007:4507 Bovens, 2007:450 quoting Day and Klein 1987:5; Romzek and Dubnick 1998:6; Lerner and

Tetlock 1999:255; McCandless 2001:22; Scott, 2000:40; Pollit 2003:89; Mulgan, 2003:7-14. Fora detailed discussion of all the elements in this definition see Bovens, 2007:450-452. This defini-

tion implies that the focus of accountability studies is more on ex post facto processes in govern-ance than on ex ante inputs. Most of the ex ante processes should then be studied separately forwhat they are, forms of deliberation, participation and control Bovens (2007:467)

8 Table based on Bovens (2007:452; 2005;185)


6/82

Katharina Noussi 5

1. Accountability serves as a democratic means to monitor and control gov-ernment conduct.

2. Accountability is important for preventing the development of concentra-tions of power.

3. The purpose of accountability is to enhance the learning capacity and effec-tiveness of public administration.

Behind each of these three arguments lies a different theoretical perspective on therationale for accountability and thus a separate perspective for the actual assess-ment of the adequacy of an accountability mechanism. From the democratic per-spective, public accountability is essential for citizens to control those holding pub-lic office.11 The second rationale for public accountability is to prevent corruptionand abuse of power. This constitutional perspective is derived from the liberal tra-dition of Locke, Montesquieu and the American Federalists. The main concern hereis to prevent the abuse of power through the organization of checks and balances

in the form of institutional countervailing powers. The third rationale for publicaccountability is to enhance government effectiveness through learning processes.Here the purpose of public accountability is to induce the executive branch tolearn. Accountability offers a regular mechanism to confront administrators withinformation about their own functioning and forces them to reflect on the success-es and failures of their past policy. Political scientists12 have long discussed theadvantages of a pluralist democracy compared to other political systems as it en-courages learning in the process of policy making.13

My first research question is, Are SAIs accountability arrangements? I decided toanswer this question through a structured qualitative content analysis14 ofInterna-

tional Standards of Supreme Audit Institutions (ISSAIs15). My hypothesis was that

9 When analysing the effects of various accountability arrangements, inadequacies can either takethe form of accountability deficits a lack of accountability arrangements or of accountabilityexcesses a dysfunctional accumulation of a range of accountability mechanisms. The former isoften cited in relation to various aspects of European Union governance (Arnull and Wincott,2000; Fisher:2004; Harlow, 2002; all quoted in Bovens, 2007:462). The latter is thematised byexecutive agencies and public managers particularly in countries, which have extensively im-plemented NPM reforms. (Anechiarico and Jacobs, 1996; Power, 1997; Behn,2001; Halachmi,2002; all quoted in Bovens, 2007:462)

10 Aucoing and Heintzman, 2000:45, qtd. in Bovens, 2007:46211

The rationale for vertical accountability through elections and other forms of citizen engagementreaches back to the tenets of Rousseau and Weber, and has been theoretically defined using theprincipal-agent model. see March and Olsen 1995:141-181 and Mulgan 2003, both qtd. inBovens, 2007

12 e.g. Easton 1965, Luhman 1966, Lindblom 1965; all qtd. in Bovens, 200713 See Bovens (2007:463; 2005:192ff; 2006:25-27) for a more detailed overview of the three ra-

tionales for public accountability. Next to these three rationales are two additional concerns ofaccountability, which Bovens (2007:464-465) accords only indirect importance. The first indi-rect importance is the assurance of legitimacy of governance, the second is public account givingafter incidences of tragedy, fiasco and failure.

14 See Bovens, 2007 and Mayring, 2000 for a description of the method used15 The ISSAIs present common standards as defined by INTOSAI and its governing bodies, as such

they are not legally binding but present important norms to INTOSAI members. (cf. ISSAI 100,

17); for more information see http://www.issai.org. However the Lima (ISSAI 1)and Mexico(ISSAI 10) Declarations are the founding principles of INTOSAI and thus constitute a very highlevel of commonly shared norms and principles, they have both also been recognized in theUnited Nations General Assembly Resolution A/RES/66 as important declarations which should
http://www.issai.org/http://www.issai.org/


7/82

Katharina Noussi 6

SAIs do qualify as accountability arrangements in all three perspectives. Table A.1.in Appendix A presents a summary of my research findings. I conclude that in prin-ciple SAIs understand themselves as arrangements guaranteeing all three types ofaccountability, namely democratic, constitutional and learning accountability. Myanalysis suggests that the main objective of SAIs as laid out in the ISSAIs, is to

guarantee constitutional accountability. The aspect of learning accountability isonly an additional rationale for their work. This aspect of their work has howevergained in importance with the rise of performance auditing. Finally, SAIs onlypartly understand themselves as guaranteeing democratic accountability. The val-ues of democracy are an essential part of their objectives but they do not primarilyunderstand themselves as service agents of parliament. The ISSAIs very clearlypoint out that SAIs do not intend to conform to the preferences of the legislaturebut instead they are independent actors which pursue a primarily professionalobjective.

To conclude, SAIs aspire to be accountability arrangements mainly in the sense of

providing for constitutional accountability. Their power to impose sanctions on theactor (the executive) is however limited in most cases. Some SAIs do have judicialpowers and can impose sanctions themselves, but most SAIs depend on other insti-tutions to follow up on their reports. Still, I argue that SAIs meet the criteria toqualify as accountability arrangements. According to ISSAI1, Section 11.2. SAIsshould be empowered to approach the authority which is responsible for taking thenecessary measures and require the accountable party to accept responsibility.

The understanding of SAIs as guarantors of constitutional accountability is alsoreflected in the United Nations General Assembly Resolution A/RES/66/20916 onPromoting the efficiency, accountability, effectiveness and transparency of public

administration by strengthening supreme audit institutions, adopted in December2011. This path-breaking resolution stresses that SAIs need to be independent andprotected against outside influence in order to accomplish their tasks objectively and

effectively. Equally, in the January 2012 issue of the INTOSAI Journal on Govern-ment Auditing17 Terence Nombembe, Chairman of the INTOSAI Governing Boardand Auditor General of South Africa, identified the main purpose of SAIs as beingan independent voice of reason, again highlighting the main objective of SAIs to bea powerful watchdog of government as defined by the constitutional purpose ofaccountability.

The General Assembly resolution, which was initiated under the leadership of

INTOSAI, and the standards on public auditing as set out in the ISSAIs reflects theincreased self-confidence of SAIs. It started with INTOSAIs strategic plan of 2005,whereby INTOSAI decided to complement its focus of mutual exchange facilitationamong SAIs with a strengthening of its external profile and communication. Thisincreased self-confidence of SAIs is also reflected in the endorsement of the MexicoDeclaration on SAI Independence in 200718, in the Johannesburg Accords and the

be applied by member states, although only in a manner consistent with their national institu-tional structures.

16http://daccess-dds-ny.un.org/doc/UNDOC/GEN/N11/471/36/PDF/N1147136.pdf?OpenElement(last accessed April 18, 2012)

17 http://www.intosaijournal.org/ (last accessed January 31, 2012)18 http://www.issai.org/media(626,1033)/ISSAI_10_E.pdf(last accessed on 16 Sept.2011)
http://daccess-dds-ny.un.org/doc/UNDOC/GEN/N11/471/36/PDF/N1147136.pdf?OpenElementhttp://daccess-dds-ny.un.org/doc/UNDOC/GEN/N11/471/36/PDF/N1147136.pdf?OpenElementhttp://www.intosaijournal.org/http://www.issai.org/media(626,1033)/ISSAI_10_E.pdfhttp://www.issai.org/media(626,1033)/ISSAI_10_E.pdfhttp://www.intosaijournal.org/http://daccess-dds-ny.un.org/doc/UNDOC/GEN/N11/471/36/PDF/N1147136.pdf?OpenElement


8/82

Katharina Noussi 7

South African Declaration on the ISSAIs, both of which were adopted during the XX.INCOSAI in November 2010 in South Africa19.

To summarize, I use accountability as a concept which describes a way in whichinstitutional arrangements operate. I have analyzed the extent to which SAIs un-

derstand themselves as accountability arrangements according to the ISSAIs andwhich purpose (democratic, constitutional, learning) they pursue when aspiring tobe accountability arrangements. I have come to the conclusion, that SAIs commit-ted themselves to serving public accountability in all three senses, whereby theirmain rationale is to be a powerful institution serving as a constitutional watchdog.

3. The problem ISSAI implementation

So far I have revealed that both the SAIs self understanding as agreed in the ISSAIsand the recent UNGA resolution representing the agreements by the heads of

states clearly recognizes the important role of SAIs as independent accountabilityarrangements. While there is thus official consensus on the purpose of SAIs asserving constitutional accountability, the question now is whether the accountabil-ity arrangement is also institutionalized, de facto as well as de jure. I define institu-tions along North (1990:3-4) as the rules of the game. They span formal laws, in-formal norms of behavior, and the shared beliefs that individuals hold about theworld. The same institution then produces different results depending on the con-text. (North et al. 2009:15) In order to investigate to what extent the formalagreement of SAIs as accountability arrangements have been institutionalized atcountry level, I conducted a literature review of previous qualitative and quantita-tive studies on SAIs and developed, constructed and compared rankings of SAIpower.

SAIs but also other oversight agencies (anti-corruption groups, the judiciary, om-budsmen, electoral commissions etc.) often fail to investigate abuses of power inthe public or private bodies over which they exercise jurisdiction for reasons ofboth elite capture and political bias.20 The budget process is particularly prone tocapture and rent seeking,21 which makes public finance control extremely difficultand explains why there is often a gap between the formal and the actual power ofSAIs. SAIs thus face a difficult challenge. They need to assert themselves againstpowerful groups within the state, which try to keep privileges and the status quounchanged. I thus differentiate between the de jure and the de facto powers of SAIs.De facto applies if a specific indicator does not only exist by law but is also applied.I call the de facto strength of SAIs their power. Basically, I consider the power ofSAIs as their capacity to conduct audits without systematic political bias or cap-ture. This is measured by their degree of independence, their equipment with ade-quate resources and power to obtain relevant information, their discretion to se-lect audits, their reporting procedures, their transparency and whether they areempowered to approach the organization responsible for effective follow-up, in

19 http://www.intosai.org/blueline/upload/jhbaccordsen.pdf(last accessed on 16 Sept.2011)20 see Goetz and Jenkins (2005) for a detailed discussion of the meanings of capture and bias.21

Economists often define rents as excess returns above normal levels in competitive markets. More specifically, rent is a returnin excess of the resource owners opportunity cost. (Tollison, 1982 cited athttp://en.wikipedia.org/wiki/Economic_rent#cite_note-4, accessed 11 April 2011).
http://www.intosai.org/blueline/upload/jhbaccordsen.pdfhttp://www.intosai.org/blueline/upload/jhbaccordsen.pdf


9/82

Katharina Noussi 8

sum, whether public organs are obliged to justify and explain their financial con-duct to the SAI.

Data specifically on SAIs has only recently been developed by a number of organi-zations and are mostly of the subjective type.22 I have developed four separate in-

dices measuring the de facto power of SAIs. In order to increase the country cover-age the rankings combine data for the years 2007-2011:

The ranking OBS2010 presents results from the Open Budget Survey2010, which is based on data up to 15 Sept. 2009 and covers 94 countries. 23

The GI08-10 ranking covers 70 countries and is based on the Global Integ-rity Reports 2008, 2009 and 2010 covering data from the periods June 2008up to June 2010.24

The PEFA index includes publicly available data for 61 Public Expenditureand Financial Accountability (PEFA) Reports for the period January 2008 toDecember 2011.25

The OECD ranking covers data for 97 countries for the period2007/2008.26

The INTOSAI Development Initiative (IDI) conducted a stocktaking report in2010. Unfortunately the IDI data is not available to the public, but IDI wasso kind to arrange a fully anonymized version of the dataset for me, forwhich I am very grateful. Although, it is not possible to combine the datawith other variables, and thus to test it on the same variables as the otherdatasets, it was possible for me to test some of my hypotheses on data con-tained in the IDI dataset.

Appendix B provides a description and discussion of the dependent variables in-cluding the questions that they are based on and their descriptive statistics (table

22 Haggard and Tiede (2011:676) explain that we can differentiate between two basic groups of ruleof law indicators: subjective and objective indicators. The subjective indicators are based onevaluations of experts, investors, or citizens. The purportedly objective indicators include dis-crete features of political institutions that are subject to measurement error or proxies designed tocapture features of the institutional and legal environment. The authors further elaborate thatthe difference between subjective and objective indicators has been an ongoing point of contro-versy. While some scholars (Glaeser et al, 2004; Kurtz and Schrank, 2007; qtd. in Haggard and

Tiede, 2011:676) argue against subjective measures because of the risk of bias, others havedemonstrated that subjective measures may capture interesting gaps between de jure and defacto institutions (Woodruff, 2006; Feld and Voigt, 2003; Rios-figueroa and Staton, 2008; qtd. inHaggard and Tiede, 2011:676). The strength of supreme audit institutions has mainly beenmeasured through subjective indicators, there are only a few institutional features (one suchexample is the question for the legal basis of the SAI, thus whether external auditing is en-shrined in the constitution, legal framework or based on internal rules of the government)which have been measured and could by themselves serve as indicators for SAI strength.

23 http://internationalbudget.org/what-we-do/open-budget-survey/24 37 countries were covered more than once during this period, for which the mean of all annual

scores was calculated.http://www.globalintegrity.org/report25There are basically two versions of this ranking, PEFA-26 is treated as an ordinal variable while

PEFA-26(i,ii,iii) is an average of the scores PI-26 (i), (ii) and (iii). Two countries have conduct-

ed two PEFA assessments during this period. Rwanda conducted PEFA assessments in 2008 and2010 and Kosovo conducted PEFA assessments in 2009 and 2011. For these two countries, themean of the scores for the two years were calculated.http://www.pefa.org

26 http://www.oecd.org/gov/budget/database
http://www.globalintegrity.org/reporthttp://www.globalintegrity.org/reporthttp://www.globalintegrity.org/reporthttp://www.pefa.org/http://www.oecd.org/gov/budget/databasehttp://www.oecd.org/gov/budget/databasehttp://www.pefa.org/http://www.globalintegrity.org/report


10/82

Katharina Noussi 9

B.1). After having examined the descriptive statistics and theoretical content ofthese five dependent variables, I decided that the OBS2010 variable is the mostappropriate variable for a measure of the power of SAIs27 for reasons of theoreticalcompatibility (guaranteed independence, transparent reporting, and effective fol-low-up) with the accountability concept and for statistical reasons such as number

of observations and statistical distribution28. When comparing the various SAIrankings a great variation across data sources becomes apparent. SAIs which areranked high in one index are rated as quite weak in another. This points out to theunreliability of the data sources, the problem of using different years, researchquestions and different methods for data collection. Despite this problem of crosssurvey consistency, all surveys point out that there is in general a great variation ofSAI power within regions and that very few SAIs are rated as very powerful. Someof the challenges of SAIs as presented in the OBS2010 survey (covering 94 coun-tries and using data as of September 2009) are:

- A lack of independence from the Executive: In 24 countries the executive

decides the SAI head. In 35 countries the executive also decides the budgetof the SAI and in 11 countries the SAI has limited freedom to decide what itwill audit.

- A lack of follow-up by the legislature: In 21 countries there is no follow-upat all and in 19 countries there is only minimal follow-up on the audit find-ings by the legislature.

- This leads to the fact that audit reports risk having no consequences. In 78countries the executive does not reveal steps taken to the public and in 72countries neither the SAI nor the legislature reports steps taken.

- In 34 countries the public is not informed about the annual accounts of itsgovernment as the final audited accounts are not completed within 24months after the end of the fiscal year or they are not released to the public.

- Furthermore, in 45 countries the public cannot give input to the SAIs workas the SAI does not maintain formal mechanisms with the public.

In summary, the 2010 Open Budget Survey (IBP, 2010:6) confirms these findingsof generally weak SAIs. Among the 94 countries assessed in the 2010 Survey, theaverage score for the strength of SAIs was just 49 of 100. Only 24 of the 94 coun-tries surveyed have strong SAIs while 32 countries have weak SAIs. Other studiesand data also reveal major weaknesses in government auditing in many countries.

For example, de Renzio (2009) analyzing PEFA29 data shows that while most coun-tries fare quite well in regard to the initial stages of the budget process, their per-

27 The OBS2010 covers a high number of countries and territories (n=94), the mean is 48.35 with a

median of 50, a minimum value of 0, a maximum value of 100 and quite a high standard devia-tion of 24.27. It is not strongly skewed and its small kurtosis points to a normal distribution.

28 The OBS2010 covers a high number of countries and territories (n=94), the mean is 48.35 with amedian of 50, a minimum value of 0, a maximum value of 100 and quite a high standard devia-tion of 24.27. It is not strongly skewed and its small kurtosis points to a normal distribution.

29 Public Expenditure and Financial Accountability (PEFA) Program was started in 2001 as a multi-donor partnership with the goals goal to strengthen recipient and donor ability to (i) assess the

condition of country public expenditure, procurement and financial accountability systems, and (ii)develop a practical sequence of reform and capacity-building actions.One important tool of thePEFA program is the development of PEFA assessment reports of country performances . Formore information visitwww.pefa.org (06 March 2011)
http://www.pefa.org/http://www.pefa.org/


11/82

Katharina Noussi 10

formance is rather weak when it comes to the final stages of the budget processincluding external scrutiny and audit.Andrews (2010) differentiates in his study ofpublic finance management (PFM) reforms in Africa between various performanceleagues. Thus, some countries are making more progress than others. Still, thereare some findings which are typical for nearly all African countries, one of which is

that budget preparation processes are comparatively stronger than budget execu-tion and oversight processes. 30

The IDI stocktaking report 2010, which was distributed to 204 SAIs31globally andeight INTOSAI Regions and Sub regions, achieving an overall response rate ofabove 90 percent, shows substantial variations among SAIs in the differentINTOSAI Regions and across World Development Indicator (WDI) classifications.The main results are summarized as follows:

While financial audit coverage is high among SAIs in AFROSAI-E andARABOSAI, the corresponding figures for SAIs in CAROSAI and CREFIAF are

much lower. Compliance audit coverage is lower than for financial auditing, with the ex-

ception of SAIs in the OLACEFS Region.

While 91 percent of the respondents have the mandate to carry out perfor-mance auditing, the data suggests that many SAIs are of the view that theydo not adequately meet demands, expectations and their own plans interms of performance auditing.

75 percent of the respondents, or 130 SAIs, indicated that their annual re-ports were issued to the legislature or other designated recipients in a time-ly fashion. Timeliness appears particularly high among SAIs in ASOSAI andEUROSAI where no SAIs have reported delays. (IDI, 2010:6-7)

This IDI stocktaking report also cautiously questions the current methods of capac-ity building as not showing clear results. While the success of capacity building ofSAIs according to the PEFA indicators seems to vary strongly across countries 32,the SAIs themselves report largely positive results. This positive effect is confirmedby case studies on the comprehensive follow up by peers on SAI capacity building,which also suggests that SAI support to other SAIs can have a crucial impact.(World Bank, 2010)

There is thus a discrepancy between positive and negative findings on the success

of SAI reforms, the reason for which might be due to the limited sample for thePEFA data used as well as the time aspect of reform. As pointed out in the INTOSAI

30 Andrews, 2010:131 This includes, in addition to the INTOSAI members, those SAIs that are members of an INTOSAI

Region but not INTOSAI, as well as SAIs that are not members of INTOSAI or any of the INTOSAIRegions.(IDI, 2010:6)

32 The overall analysis of PEFA Indicator 26 for the 14 LI and LMI countries that have been as-sessed twice or more in 2005-2010 shows improvements for five SAIs. The results are howevervaried, and for six countries the results have remained identical, while three countries have hadtheir scores reduced. Taking into consideration that the SAIs in all these countries have been re-

cipients of capacity development support during this period, this raises questions about the im-pact and sustainability of the support. These results are also in contrast to the perceived suc-cess of capacity development support as reported by both providers and recipient SAIs in thestocktaking, where around 92 percent of projects are categorized as successful. (IDI, 2010:7-8)


12/82

Katharina Noussi 11

publication on Good Practices for SAI Capacity Development, support needs to belong term and predictable. Thus it may take more than ten years according to theauthors to see any real impact of capacity development interventions. Apart fromthe time factor, other authors criticize the largely technical approach taken in re-forming budget institutions in poor countries.33 Positive experiences with reforms

would then result from political economy approaches, which emphasize a systemicapproach to reform. If the SAI is not independent, then the priority of reform mightnot be improving the internal governance of the SAI but rather strengthening thedemand for accountability. Andrews (2010:ii) concludes in his review of PFM re-forms in Africa that there needs to be a shift in focus from technicalities to creatingreform space, that reform needs to expand engagements across a larger set of ac-tors and that a stronger emphasis needs to be placed on analyzing the particularpolitical economy of reform, on the basis of which context-appropriate reforms canbe developed.

In sum, current scholarship, including the World Bank (2010) analysis of SAIs in

the Middle East and North Africa region, concludes that the political economy ofreform is crucial to its success. Ultimately the success of efforts to build the capacityof the audit body will depend on whether the parliament is prepared to utilize the

auditors report. (World Bank, 2010: 33) Appearances might thus be deceiving.SAIs might be in place in nearly all countries, and some are even granted with largeconstitutional powers. However, this power might only be in place de jure, while defacto they might fail to realize their aspired role as accountability arrangements.

However, there is currently a momentum of reform, which might change the situa-tion of SAIs in the next decades. Academics and PFM practitioners alike have re-cently started to pay closer attention on SAIs. INTOSAI has started its cooperation

with the donor community only in 2010. This has confirmed that there is a newunderstanding of the vital role of SAIs and that the donor community has acknowl-edged that it is not enough to include SAIs in regular PFM reform programs, whereconsequently they were often neglected by the implementing ministry of finance.To conclude, this literature review of empiric data of SAIs has revealed that there isa great variation across SAIs but that very few SAIs qualify to date as powerful ac-countability arrangements.

4. Hypothesis development

The next step of my research investigates what potential explanations exist to ex-plain the variation in the de facto power of SAIs across countries. Why does citizen-oriented independent external public auditing thrive in some countries and fail inothers? There exist numerous explanations and no consensus in the scholarly de-bate on the potential determinants for the effective institutionalization of publicaccountability. The main argument of this paper is that while the institutional ar-rangements for government auditing, the nature of the source of national income,the external influence on reform, the local demand for public accountability andthe technical and organizational capacities of SAIs matter, the political economy ofelite relations and the existence of plural, functional and problem-oriented leader-

33 such as Allen (2008, 2009); Santiso (2006, 2009); Pretorius & Pretorius (2008); Shah (2007);

Stapenhurst et al (2008)


13/82

Katharina Noussi 12

ship creating reform space ultimately conditions the development and consolida-tion of the de facto power of SAIs. I will now proceed to discuss these arguments inturn.

1. Institutional theoriesH1a: Although the institutional model and linked to that the colonial heritage

do have some influence (with court/Napoleonic institutions performing worst

and the monocratic/Westminster model performing best), institutional fea-

tures are not the main determinants for the variation in SAI power.

Several authors have conducted research on the effect that the institutional designof SAIs can have on their strengths per se as well as on national economic features.Existing qualitative research (f.i. Santiso, 2009) supports the hypothesis that theinstitutional model, although influential to some extent, is not a main explanatoryfactor for the variation of SAIs political strengths. Blume and Voigts quantitative

study (2011) suggests that the court model is associated with higher levels ofcorruption. However, the authors also believe that the reason behind is the relativelack of parliamentary and public involvement in the auditing process under thecourt model and not the institutional model per se.

H1b: Countries with plurality rule have stronger accountability systems than

countries with proportional representation.

Another strand of research suggests that the type of electoral system does have animpact on the functioning of fiscal institutions. Van Hagen (2006) supports the hy-pothesis that countries with plurality electoral rule have stronger accountabilityinstitutions than countries with proportional representation.

2. Theories on the source of national incomeH2a: The higher is the share of tax endowments in a countrys national in-

come, the more likely accountability systems including SAIs work properly.

The early development of supreme audit institutions was part of a long process ofeconomic and political institutional development in Europe and the USA. State the-ories (f.i. Tilly, 1992) explain how through the process of institutionalizing taxa-tion, wars were financed and states were formed. There is a broad scholarshipinvestigating the question whether the source of national income has an influenceon the institutions that a state develops and on the way these institutions function.For instance, the recent study by Broms (2011) compares the effect of tax systemsand incomes on the governance of African states. He suggests that it is not the taxlevel per se but the type of taxation which makes a difference.

H2b: The share of aid endowments in the national budget does not have any

effect on the de facto power of SAIs.

Apart from the effect of taxation, scholars also study the effect of aid and naturalresources on institutional development. The question whether aid plays a role in


14/82

Katharina Noussi 13

the development of accountability is currently a hotly debated issue. 34 There is noconsensus in the (particularly US-American based) academic debate on the role ofaid on macro-development and democracy. Furthermore, a generalization is diffi-cult as we have to differentiate between various aid modalities and priorities. Forthe purpose of this study, I suppose that the share of official development aid in the

national budget does not have any significant effect on the de facto power of SAIs.

H2c: The higher is the share of natural resource endowments in a countrys

national income, the less likely accountability systems including SAIs work

properly.

The literature exploring the so called resource curse is very rich but also incon-clusive. New insights (Haber and Menaldo, 2011) suggest that the supposed linkbetween natural resources and the large probability of an authoritarian regimemight be mistaken. Instead of natural resource endowments causing authoritarian-ism, authoritarian governments decide to invest in natural resource exports as thishighly specialized industry sector cannot easily be expropriated and used by po-tential opposition forces. Despite the question of what causes what, there seems tobe a strong correlation between large rents stemming from natural resource en-dowments and authoritarian regimes, and by definition also weak independentpublic auditing.

3. Diffusion theoriesH3a: The stronger is the regional influence by good performers, the more like-

ly a country implements similar reforms.

Theories on the diffusion of democracy suggest that external influence can have astrong effect on the probability for the creation and endurance of democracy. Thisstrand of literature further argues that the stronger the regional influence of goodperformers is, the more likely a country is to also adopt reforms of accountability.

H3b: The more influence the international community has in building state

capacity (i.e. in a post-conflict situation with international involvement) the

more likely it is that institutions of accountability can effectively be put in

place.

Under this rubric I also classify the concept of institutional inertia caused mainlyby path dependency35. Here it is assumed that some exogenous shock at different

critical junctures sets a state or institution on its development trajectory. The moreinfluence the international community has in building state capacity (i.e. in a post-conflict situation or after a total regime change such as South Africa, Rwanda, Libe-ria, Kosovo etc.) the more likely it is that institutions of accountability can effec-tively be put in place as there is less resistance from powerful elite groups as theyhave been substantially weakened through the conflict. (Andrews, 2010:3; Dia-mond, 2009)

34 Critics like Easterly (2006), Moyo (2009) and Collier (2010) claim that aid has helped little andinstead supported authoritarian regimes. On the other hand, Sachs (2005, 2008) and prominentpro-aid movements argue that aid has helped so little because it was too little.

35 Putnam, 1993; Acemoglu et al. 2001


15/82

Katharina Noussi 14

H3c: Countries comply to international norms only formally, and do not

transform these values into de facto institutions.

Finally, socialization or decoupling theory36 argues that countries choose insti-tutions to comply to international norms, even if those institutions do not meet the

countries needs.

4. Modernization theories37H4a: The higher is the level of socio-economic development of a country the

more likely is the country to become democratic and to stay democratic once

democracy has been established, thus the more likely institutions for account-

ability develop and work effectively.

H2b: The stronger is a societys demand for accountability, the more likely it

is that the ruling elites will make democratic concessions. Thus, the stronger is

a societys capacity to understand and request audit reports and to organizecampaigns, the more likely are SAI reforms and thus the stronger is a SAI inthe country.

Modernization theories and other structural accounts of democratization occupy alarge share of scholarly research on the development of democracy including pub-lic accountability.38 Basically, the assumption is that economic and social develop-ment of a country is both a prerequisite for the development and for the enduranceof democracy. In this perspective an educated middle class demands more voice insociety, which sets the conditions that ultimately lead to democratic institutions.The policy consequence would then be to support socio-economic development inpoor countries and build the capacity of the wider public to demand their rights. Thus, the stronger is societys demand for accountability, its capacity to under-stand and request audit reports and to organize campaigns, the more likely it isthat the SAI will gain in power.

5. Conventional institutional capacity theory (function follows form)H5a: The technical capacity of a SAI including staff number and profession-

alism as well as infrastructure has a positive effect on the development of de

facto SAI power.

According to institutional accounts of development, development is only possiblewith political institutionalization. The essence of the institutional capacity theoryis that development will only occur if states succeed to build effective autonomousinstitutions able to organize politics. Samuel P. Huntington famously wrote: Themost important political distinction among countries concerns not their form of gov-

ernment but their degree of government. (Huntington, 1968:1) The institutionalcapacity approach has emerged in the aid debate after the influential comparativestudy of democratic transitions by Linz and Stepan (1996), the World Develop-

36 Meyer et al. 1997, qtd. in Handler, 2010.37 I adopted the following three types of classifications from Krasner (2009a,b, 2011a,b) who differ-

entiates between modernization theory, institutional capacity theory and rational choice theoryas the three main paradigms in the current aid debate.

38 Lipset, 1959, Przeworski et al, 2000; Boix and Stokes, 2003; Sachs, 2005, 2008; Inglehart andWelzel, 2008


16/82

Katharina Noussi 15

ment Report of 1997, The State in a Changing World, and the insights by Fukuyama(2004). He explained that although it was necessary to limit the scope of statefunctions in poor countries as part of structural adjustment programs during the1990s, it was a mistake to also reduce state capacity. The mainstream of develop-ment policy today follows this institutional capacity approach and stresses the

need for building capacity of state institutions in order for them to become effec-tive and for development to occur. Thus, while modernization theory believes thateconomic development will lead to political development, institutional capacitytheory stresses that states need to build political institutions first in order to de-velop.

POLITICAL WILL

While modernization theory and institutional capacity theory assume that the pro-vision of enough external support will lead to development, I agree with North et al

(2009a,b) and others that there is no such teleology and that we need to put poli-tics at the heart of development processes. In my opinion political will to institu-tionalize accountability develops not automatically if there is the demand for pub-lic accountability (just think of the various outcomes of political revolts during theArab spring). The demand for public accountability (be they led by public outcry,donor conditionality or within the elite itself) is metde facto as well as de jure onlyif there is 1) the presence of specific inter-elite relationships which are conduciveto reform as proposed by rational choice institutionalism and 2) if there is (SAI)leadership that creates a momentum for reform as explained through organiza-tional theories.

6. Rational choice institutionalismThe same institution produces different results

depending on the context.39

H6a: The more the doorstep conditions of North, Wallis and Weingast

(2009a,b) are realized the more reforms of institutions of accountability can

succeed.

Rational choice institutionalism investigates the question why democratic institu-tions at time not work. De jure the institution might be in place, but the facto otherrules of the games apply. The argument that inter-elite relationships condition suc-cess or failure of reform was developed by North, Wallis and Weingast (2009a,b).Their book can be allocated among the new institutional economics literature,which studies the role of institutional designs on economic performance. However,their main focus is on studying how these institutions come about, and by doing sothey touch on a very topical debate in current development scholarship and prac-tice. Their work is part of a new approach to studying the integrated politicaleconomy of democracy and capitalism, institution-building, state-development and

39 North, 1990: 54


17/82

Katharina Noussi 16

authoritarianism: rational choice institutionalism.40 In recent years, the rationalchoice institutionalist literature has exploded, particularly in the US political sci-ence academia.41 This approach agrees with institutional capacity theory that insti-tutions are a prerequisite for development but criticizes its focus on institutionalfunctionality rather than possession. Rational choice sees economic growth and

effective governance as the result of decisions taken by key actors, who are self-interested.As individuals and groups act according to their available options, theresults of various political deals are always open and not linearly defined. Contraryto conventional institutional capacity theory, for rational choice scholars institu-tions can never be autonomous and a concentration of power is detrimental. In-stead, institutions must be embedded in the interests of all powerful elite groups tomake them stable. Consequently, contrary to the assumption of the conventionalpolicy approach of institutional capacity building, democratic institutions will notwork only because enough resources have been provided for organizational devel-opment, training and infrastructure. Instead, institutions will work if they are in

the interests of the elites because they have been put into a position, where theylose less by making democratic concessions than by holding on to power.

The main argument of rational choice institutionalists is that the feature of sus-tainable political orders and institutions with de facto power is that they must beself-enforcing and pareto-improving. They are the result of a conclusion of mutu-ally beneficial bargains as they solve commitment problems.

Democracy endures only if it is self-enforcing. It is not a contract because there areno third parties to enforce it. To survive, democracy must be an equilibrium at least

for those political forces which can overthrow it: given that others respect democra-

cy, each must prefer it over the feasible alternatives. (Przeworski 1991, qtd. in

Przeworski, 2006:300)This logic of a self-enforcing equilibrium contradicts the contractarian theoremthat democracies develop out of constitutions42. Rational choice scholars see insti-tutions as mechanisms that can make political bargains stable and enduring, in thesense thatform follows function and not the other way around43. Thus, there needsto be a political process which leads to bargains and commitments, which are theninstitutionalized.

North et al. (2009a,b) explain that the transition from authoritarian to democraticstructures must be consistent with the logic of the current social order.44 The au-

40 North, Wallis and Weingast (2009:269) have called this stream of research an integrated politi-

cal economy approach, Iversen (2006:617) has defined it as new structuralism and Krasner(2009a,b, 2011 a,b) has named itrational choice institutionalism.

41see f.i.: Acemoglu and Robinson, 2006; Fisman and Miguel, 2008; Collier, 2010; Carothers, 2002;

Grzymala-Busse and Jones Luong, 2006; Greif, 2006; Haber et al., 2003, 2006; Persson andTabellini, 2000, Spiller and Tommasi; 2007; Magaloni, 2008; Khan, 2005, 2006, Fisman, 2001;and Meisel and Ould Aoudia, 2008

42Constitutions are not necessary because democracy survives not because of exogenous rules but forendogenous reasons, they are self-enforcing. The rules that regulate the functioning of a

democratic system need not be immutable or even hard to change. When a society is sufficiently

wealthy, the incumbents in their own interest moderate their distributional zeal and tolerate fair

electoral chances. Democratic governments are moderate because they face a threat of rebellion.

Democratic rules must be thought of as endogenous. (Calvert, 1994, 1995, in Przeworski,2006:320-321)

43 As explained above, institutional capacity theory assumes thatfunction follows form44cf. Huntingtons distinction of countries according to their political order, 1965, 1968, see above


18/82

Katharina Noussi 17

thors differentiate between social orders with open and such with limited accessorders, meaning access to both political and economic activities. Social orders withlimited access, also labeled natural states, exist on a continuum ranging from frag-ile states, characterized by political instability and violence, at one extreme, to ma-ture natural states such as emerging marketsthat are close to satisfying the

doorstep conditions. The new contribution of their framework is the concept ofhow societies reduce or control the problem of violence. They first explain that thecontrol over violence is disbursed in many groups in natural states, which wouldmake them very fragile to civil wars. However, natural states solve the problem ofviolence through rent-creation, granting powerful individuals and groups privi-leges so that they have incentives to cooperate rather than fight. The resultingrents, limits on competition, and limited access to organizations hinder long-termeconomic development. In contrast, open access orders use competition, open ac-cess to organization and institutions to control violence and are characterized byrent-erosion and long-term economic growth. 45 The idea that Western institutions

cannot easily be transferred to other contexts is not new but the theory why thedynamics in developing countries at times resist the development and consolida-tion of democratic institutions or transform them to hollow institutional shells(namely to prevent violence) is new.

According to North et al. (2009a,b) in order to trigger a transition to an open ac-cess society, it is important to create preconditions that put the elites in a positionwhere it is in their own interest to transform personal and privileged deals amongthemselves into impersonal laws, so that all members of the elite groups are treat-ed the same way. In a second step it can then be possible that elites transform theirunique and personal rights to impersonal rights for all citizens.

The natural tendency of powerful groups faced with uncertainty and novel situa-tions, is to consolidate privileges, not to expand them to include more elites. The tran-

sition proper is the process by which elites open access within the dominant coalition,

secure that open access through institutional changes, and then begin to expand ac-

cess to citizenship rights to a wider share of the population. (North et al. 2009:190)

Thus, contrary to modernization theory, there is no direct link between public de-mand and the development of sustainable democratic institutions but the publicdemand is only met if the elites are put into a position where it is in their own in-terest to make these concessions and make the institutions work. North et al.(2009a,b) call the preconditions for open access for elites to develop the doorstepconditions. They are:

DC#1. Rule of law for elites. (The objective is to separate personal identityfrom privilege and thus to contribute to new ways for impersonal exchangeand trade.)

DC#2. Perpetually lived organisations in the public and private spheres.(These are sophisticated organizations that last and are not tied to one spe-

45 While I agree basically with this classification of political orders, and I also acknowledge the fact

that the authors mentioned in their book that no country has achieved full open access, I thinkthe authors could have said more about the political order in the so called open-access coun-

tries. While equality before the law and other features of open access orders are definitely moresecure in the rich industrialized OECD countries than in the rest of the world, here too, I think,inequalities persist and crises (such as the recent financial crisis and the debt crisis) reveal vul-nerabilities of the apparently stable political order.


19/82

Katharina Noussi 18

cific person, which increases impersonal exchange as one can count on thefirm or public entity to last and therefore one is more inclined to invest inor trade with this organization or to take on the regulations of this publicentity.)

DC#3. Consolidated control of the military (This is important to prevent thedisbursement of violence among various groups and get stability for eco-nomic and political development.)

To sum up, North, Wallis and Weingast (2009a,b) provide a conceptual frameworkfor understanding social orders. The authors show that the analysis of violence,institutions, organizations, and beliefs is central for an understanding of economicand political development processes. They base their insights on detailed historicanalysis, mainly of the USA, Great Britain and France. While this method allowsthem to paint a clear picture of the processes in concrete historic contexts, it is notclear to what extent the same processes can be observed today (and less even it isfeasible in the context of an ongoing climate crisis, financial crisis and food crisis).Furthermore, even in their historic analysis they leave out several aspects such asthe international political economy and the role of the non-elites, including the roleof gender relations, on development. Despite these limitations46, I consider theirframework a useful tool for the analysis of inter-elite relationships.

H6b: The power of SAIs in autocracies differs depending on the strategy that

the autocrat uses in order to stay in power.

Recent scholarship on authoritarianism agrees with rational choice institutional-

ism. This scholarship first of all observed that many autocracies have created fa-cade democratic institutions, which however leave the autocratic essence of theregimes unchanged.47 Why are some appearances deceiving? Interviewee 5 de-scribed the practice of public auditing in his country as a farce. Although the insti-tution is fully independent from the executive by law, in fact it is not and every-body knows that it is only an empty-shell institution. He expressed this situationwith the following proverb, A camel hides his head behind a tree and thinks nobodysees it, but in fact everybody does.

Many of the SAIs that lack independence and perform poorly can be found in suchsemi-authoritarian regimes. Thus it is my main interest to particularly explore po-

tentials for reform in (semi-) authoritarian regimes. Under what conditions do(semi-) authoritarian regimes successfully implement accountability reforms?What strategies do (semi-) authoritarian regimes employ to successfully resist re-forms?

46 For a critical discussion of this book see f.i. Hartzell; Elshtain; or Diamond, all: (2010) ReviewSymposium: The State and Violence, in Perspectives on Politics, Vol. 8, issue 1 (2010).

47 Abdukadirov (2010) explains that these regimes were originally conceptualized in the literatureas transitioning towards full democratization. Today they have come to be recognized as a sepa-

rate category, distinct from both full democracies and full autocracies (Carothers, 2002). Vari-ous authors have categorized them as electoral authoritarian (Schedler, 2006), semi-authoritarian (Ottaway, 2003), competitive authoritarian (Levitsky and Way, 2002) or hybridregimes (Diamond, 2002).


20/82

Katharina Noussi 19

Haber, Razo, and Maurer (2003) explore the question of who can threaten a dicta-tor with removal from power, and the mechanism that they use to do so. A majorthreat to dictators constitutes political entrepreneurs who lead organized groups.This presents us according to Haber (2006) with a paradox: if dictators fear politi-cal entrepreneurs and the groups they lead, why dont they simply stamp them

out? Haber (2006) suggests that there is an inevitable set of strategic interactionsamong dictators and the organized groups that launch them to power. Haber(2006) as well as Magaloni (2008) claims that there are three distinct politicalequilibriums, depending on the power struggles between the dictator and whatHaber calls the launching organization. In the first scenario, the dictator elimi-nates the power of the launching organization through terror, torture, and purges.The second strategy is one of collusion, where the dictator ensures a stream ofrents to leaders of the launching organization. A third political equilibrium is char-acterized by organizational proliferation, whereby the dictator encourages the cre-ation of competing organizations to increase the costs of collective action by the

launching organization.Each of the three strategies applied by autocrats generates quite different propertyrights systems, and each of those property rights systems has consequences foreconomic growth, distribution and accountability reforms. Dictatorships in whichthe power of the launching organization was curbed by terrorizing or co-opting itsleadership tend to allocate property rights to a restricted set of individuals. As aresult, the economies of these countries tend to grow very slowly, and public ac-countability is impossible. Dictatorships in which the power to the launching or-ganization was curbed by proliferating yet more organizations, however, tend toallocate property rights to a much larger set of individuals. As a result, the econo-

mies of these countries tend to grow more rapidly. Accountability reforms aremore likely, therefore, to be successful in those countries in which a dictatorshipwas characterized by organizational proliferation, rather than terror or co-optation. These accountability institutions can gain strength in situation of politicalconflict.

7. Organizational theories of leadership-led change48H7: Leadership commitment by heads of SAIs49 as well as long periods of

broad reform commitment foster progress.

The second theory that my argument on the development of political will builds onis organizational theory and more specifically, leadership as an important driver ofchange. I adopt the approach by Andrews et al. (2010b) for understanding the roleof leadership in the process of change. Their definition of leadership is based on anextensive literature review and a qualitative study comprising 140 in-depth inter-views covering 14 development projects in fragile countries. They define leader-

48 This title is an adaptation of Andrews et al. (2010) who entitle their report Development as

leadership-led change49 This hypothesis was not part of my original hypotheses as developed on the basis of the literature

review. Instead it was one result of the expert interviews, conducted at a later stage of the re-search project. The SAI leaders themselves pointed out that leadership abilities and commit-ment by SAI leaders is one of the main factors that explain success or failure of SAI reforms.


21/82

Katharina Noussi 20

ship as involving the setof actions that intentionally creates change space and mo-bilizes people, ideas, meaning and resources to achieve a change purpose50

This definition speaks about leadership and not leaders because the authors con-sider leader plurality as important, thus they talk more about groups than individ-

uals. This definition is driven by the functional approach to leadership, thus lead-ers are identified more because of their functional contribution to change thantheir personal traits or authority. In answering the question what leadership doesin the change process, they mainly refer to transformational, transactional and re-lational leadership models. According to their proposition, change space developswhen leadership

builds acceptance (by managing attention and meaning, framing facts,coaching, etc.),

ensures that appropriate authority and accountability structures are estab-lished (by empowering followers, delegating responsibilities, buffering the

organization from external influences, etc.) and when leadership enhances ability (by fostering new productive rela-

tionships, accessing new finances, etc.).

Change space can then emerge at the intersection of these three conditions.

Figure 1: The AAA model ofchange space (Andrews et al., 2010b:6)

Space does not have a static importance in their model, but emerges as importantin a process where different stages have different space requirements. In earlystages of reforms the major challenge may be to facilitate contextual acceptancewhile in later stages it might be to introduce new abilities or formal or informalauthorization mechanism to institutionalize the change.51 I understand this con-cept ofspace change as resembling the idea of a momentum of change as definedby the sociological term oftipping points.52

50 Andrews, et al. 2010b:1351 Andrews et al., 2010b:9, Andrews, 2008b52 Research on tipping points: Grodzins, 1958 ; Schelling, 1972; Granovetter, 1978; Gladwell, 2000


22/82

Katharina Noussi 21

According to Andrews et al. (2010b), in the development context, it is vital to askwhether the psychological issues and learning dimensions of change receiveenough attention and particularly also whether the change process addresses aproblem which has been identified by the persons who should lead the changeprocess, and thus whether there is local ownership of the process. Referring

amongst others to North (199553) this study also considers aspects of the rationalchoice institutionalism theory in that contextual factors affect the costs of exercis-ing leadership that leads change:

contexts are not always ready for the type of leadership we believe leads change....some contexts were more flexible than others and accommodated the move from

change initiation to transition more readily Where contextual factors make it toocostly to either initiate change or to ensure it is sustained, leadership will be lacking.

Where contextual factors make it less costly or allow cost sharing across a team, coa-

lition or network, such leadership will be more apparent54

Despite these context specificities, successful leadership-led change events shared

a number of key characteristics in their study. These were plurality, functionality,problem orientation and change space creation:

1. Plurality indicates that leadership is more about groups than individuals

2. Functionality is important because leaders have been identified more be-cause of their functional contribution to change than their personal traits orauthority

3. Problem orientation and change space creation defines leadership whichcontributes to the acceptance for change, which grants authority to change,and introduces or frees the abilities necessary to achieve change.55

To summarize, leadership actions are required at all stages of the reform processto expand change space and by doing so they create the political will necessary fora de facto institutionalization of SAI reforms. And this leadership most importantlyhas to come from the heads of SAIs themselves. But there can also be leadership byparliamentarians, political parties, the media, public pressure groups and of courseexternal actors such as INTOSAI bodies, donors and regional organizations. Afterhaving discussed a number of potential hypotheses which propose determinantsfor the variation in the de facto power of SAIs, I conclude with a main hypothesis:

While the institutional arrangements for government auditing, the na-ture of the source of national income, the external influence on reform,

the local demand for public accountability and the technical and organi-zational capacities of SAIs matter, the political economy of elite rela-tions and the existence of plural, functional and problem-oriented lead-ership creating reform space ultimately conditions the developmentand consolidation of the de facto power of SAIs.

In a next step, I will test this hypothesis through the exploration of statistical data.This part of the research process consists of the objective construction of the ob-

53 qtd. explicitely in Andrews et al. 2010b: endnote xxiv54 Andrews et al. (2010b:49-50)55 Andrews et al. (2010b:49-50)


23/82

Katharina Noussi 22

ject, that is, the construction and testing of an explanation from an observers pe r-spective by the use of statistical techniques. After that, chapter six will look at thepractice of reform as experienced by heads of SAIs, thus the subjects of the re-search project themselves. This type of mixed methods strategy was first proposedby Bourdieu56 and is commonly known under the name praxeological knowledge.

The objective of this mixed methods strategy is to mix quantitative and qualitativemethods in order to incorporate reflexivity into objective knowledge.57

5. Exploring statistical data

The dependent variables have been presented and discussed above (see chapter)and table D.1 (see Appendix D) summarizes the expected effect of my hypothesesas suggested in the literature and diagnosed in preliminary bivariate regressionsand graphical examinations, which I conducted on a large set of independent vari-

ables.58

Basically, the bivariate regressions largely confirmed my hypotheses. Buthow do the variables perform when put together, i.e. when testing each variablewhile holding the others constant? As I am using observational data I must beaware that what I am actually finding are only correlations between two variables.This does therefore not mean that the relationship which I am finding betweenvariables is necessarily causal. Particularly when comparing countries, the demon-stration of causality becomes very problematic. The variance between countriescould be caused by many other factors than the ones I am testing. And these factorsmight be at the root of both my dependent and my independent variables. This iswhy I consider this chapter to be an exploration of relationships more thana testof causality.

DATA ANALYSIS

The main test of the hypotheses is presented in table 3 below. Due to space limita-tions, additional regressions on the OBS2010 dependent variable and regressionson alternate dependent variables are presented in appendix D and are referred toonly for comparison purposes.

(1)Testing the institutional theories:I have created a categorical variable SAImodel_09 (please see table C.1. in Appen-dix C for a detailed description of all independent variables used) with the follow-ing categories: Board/Collegiate model, Court/Napoleonic model, Ministry/part ofgovernment and Monocratic/Westminster model. In table 3 we clearly see thatSAIs which belong to the category of the monocratic institutional model have ahigher statistical probability of ranking high on the OBS2010 SAI index than SAIswith a Board model or court model.

56 Bourdieu, 1973; Fries, 2009; qtd. in Harrits, 2011:15157 This approach is in contrast to the nested analysis as proposed by Lieberman (2005), whereby

qualitative research tests and elaborates on the quantitative findings. See Harrits (2011) for a

thorough comparison of the two approaches.58 Due to the limited space available in this paper, I am not able to comment on the individual biva-

riate regressions and on all the variables tested here. Interested authors are encouraged to con-tact the author personally for more information concerning individual hypotheses.


24/82

Katharina Noussi 23

Table 3: OLS regression of all hypotheses on OBS2010 SAI power

Regressions: (1) All (2nonOECD) (1b) (2b)SAImodel_09Board 47.79

(0.015)* 54.42

(0.000)*** 53.34

(3.457)** 56.15

(5.737)***

SAImodel_09court 42.51

(0.028)

* 51.82

(0.000)

*** 42.72

(2.819)

** 47.92

(4.926)

***

SAImodel_09monocratic 56.25(0.004)

** 64.61(0.000)

*** 60.00(3.971)

*** 64.53(6.760)

***

dpi_checks 3.14(0.006)

** 2.76(0.027)

* 4.23(4.566)

*** 4.07(4.4349

***

Oil_02.07 -0.42(0.000)

*** -0.47(0.000)

*** -0.48(-5.299)

*** -0.55(-6.045)

***

IMF_1 -13.14(0.004)

** -13.69(0.004)

** -11.37(-3.087)

** -11.10(-3.1729

**

INTOSAIregionAFROSAI-E -12.14(0.015)

* -14.51(0.009)

** -13.08(-3.356)

** -15.18(-3.8829

***

INTOSAIregionARABOSAI -22.51(0.001)

*** -26.31(0.000)

*** -22.67(-4.526)

*** -26.27(-5.222)

***

INTOSAIregionASOSAI -8.07(0.086)

. -13.19(0.019)

* -9.54(-2.5479

* -14.21(-3.5349

***

INTOSAIregionCAROSAI -25.29(0.047)

* -25.32(0.065)

. -29.78(-2.979)

** -29.98(-3.037)

**

INTOSAIregionCREFIAF -12.29(0.060)

. -16.82(0.022)

* -14.21(-2.7369

** -19.25(-3.612)

***

INTOSAIregionNRM -6.32(0.643)

NA -3.91(-0.3569

NA

INTOSAIregionOLACEFS -10.86(0.011)

* -11.61(0.013)

* -13.74(-3.953)

*** -14.01(-3.997)

***

INTOSAIregionPASAI -46.05(0.000)

*** -46.20(0.000)

*** -47.91(-6.7939

*** -47.44(-6.851)

***

battledeaths_02.08 0.003(0.001)

*** 0.003(0.001)

*** 0.003(1.118)

0.002(1.003)

log(fh_press) -5.29(0.25)

0.94(0.891)

2.37(0.5519

12.97(2.123)

*

ciri_injud1. Partially inde-pendent

9.07(0.007)

** 9.99(0.009)

** 11.53(4.329)

*** 13.04(4.692)

***

ciri_injud2. Generally inde-pendent

11.44(0.032)

* 15.08(0.016)

* 21.36(4.2679

*** 29.58(5.233)

***

SAIleader_total1 1.19(0.001)

** 0.99(0.014)

* 0.92(2.943)

** 0.67(2.1139

*

Constant (Reference categorymeasuring SAImodel_09.ministry;INTOSAIregionEUROSAI;IMF_0; ciri_injud0.Not indep.)

10.20(0.69)

-18.92(0.519)

-24.69(-1.090)

-68.37(-2.681)

*

N 85 73 79 67Multiple RAdj. R

0.82120.769

0.78060.7074

0.89290.8585

0.89020.8491

F-statistic p.value < 2.2e-16 5.777e-12 < 2.2e-16 < 2.2e-16Significance levels reported as: *** p< 0.001; ** p< 0.01; * p< 0.05 ; . p < 0.1; t-statistics in brackets. The regression models were tested asfollows: Multi-collinearity tests GVIF^2(1/(2*Df))


25/82

Katharina Noussi 24

The negative effect of the court system becomes particularly clear, once I excludeBrazil. The Brazilian SAI has a judicial function and a Board decision making struc-ture, but receives high scores in the OBS and other rankings. The tendency of apositive effect of the monocratic SAI institutional design is confirmed in all alter-nate tables, although at times without statistical significance (see Appendix D, ta-

bles D2-D6). The literature and my interviews suggested that the advantage of themonocratic system is the assignment of clear personal responsibility and thusleadership to one person. On the other hand, the problem of court/judicial systemsis that they the legislative follow-up is weaker than in the other SAI models, whichrisks causing weak follow-up to their reports. Furthermore, many courts of auditsin developing countries primarily engage in pure legality audits rather than in fi-nancial and performance audits.

I also tested, whether a plurality electoral system has a positive effect on the pow-er of a SAI, which could not be confirmed. The effect was at times even contrary towhat I expected and was in all regressions minimal.59 When testing the type of re-

gime60, parliamentary democracies fared best, as expected, but there was no clearresult concerning the difference and the extent of the effect of mixed democraciesor presidential democracies as well as between various types of authoritarian re-gimes. Thus, I suggest that the existing regimes cannot easily be classified into thevarious categories but instead there are many mixed types. Instead I opted for avariable (dpi_checks) which counts the number of veto players in both the execu-tive and the legislature, and thus the overall political competitiveness, in any re-gime. This variable showed a stable positive and significant effect. It thus suggeststhat the institutional hypothesis from the public finance literature might be true,i.e. a larger number of veto players and thus stronger political competitiveness

reduces corruption or otherwise put, increases the probability that a powerful SAIdevelops.

(2)Testing the theories on the source of national income:Table 3 shows that the effect of a large share of Oil rents (as % of GDP) is indeedassociated with weaker SAIs. This effect is highly statistically significant and verystable across all types of regressions (on the OBS2010 but also the other depend-ent variables, see Appendix D, tables D.2 and D.3.) The effect of a variable combin-ing oil rents with other rents from natural resources (particularly minerals, butalso gas) was also significant but not as strong. Interesting was also that a large

share of carbon-rents had a positive effect on the probability of a powerful SAI. Iassume that this effect is caused by the high probability that it is particularlyemerging economies and industrialized countries which extract carbon. Thus, car-bon might have the effect of being a proxy for industrialization.

59 Tested through the dpi_pluarlity and dpi_pr variables for the year 2007, as in Teorell, Jan, MarcusSamanni, Nicholas Charron, Sren Holmberg and Bo Rothstein. 2010. The Quality of GovernmentDataset, version 27May10. University of Gothenburg: The Quality of Government Institute,http://www.qog.pol.gu.se (accessed September 2011)

60 Tested through the chga_hinst variable for the year 2007, as in in Teorell, Jan, Marcus Samanni,Nicholas Charron, Sren Holmberg and Bo Rothstein. 2010. The Quality of Government Dataset,version 27May10. University of Gothenburg: The Quality of Government Institute,http://www.qog.pol.gu.se (accessed September 2011)


26/82

Katharina Noussi 25

The effect of effective taxation on the power of a SAI was not clear. I tested thishypothesis on a number of variables61, which all showed positive effects, althoughwith at times little or even no statistical significance. Tax compliance (EFW5Cvii)as measured by Economic Freedom of the World showed a positive and significanteffect on OBS SAI power. However this variable might also be a proxy for rule of la

as it actually measures the cost of tax compliance (see Appendix C and Appendix D,table D2). It was also interesting to observe a strong effect of direct taxation (as a% of GDP) in politically free Subsaharan African countries (compare with Broms,2011; see Appendix D, table D.11, graph D.3 ). They main reason why I did not in-clude any of the taxation variables in the final regression is the limited number ofobservations which were available, thus reducing the overall dataset while notadding any significant explanatory power.

Finally, I also tested the effect of various aid variables62 on the power of SAIs (seeAppendix D, tables D7 and D8, graph D.2). They all had negative effects on thepower of SAIs. However, the caveat here is that I am only measuring the effect at

one point in time and comparing the effect with countries which do not receivedevelopment aid. Probably the fact that a country is aid recipient is thus a proxyfor poverty or economic crisis. I am thus not measuring whether a particular aidmechanism has had a positive or negative effect. In order to do that, I would needto conduct time series analysis. However, the effect of PFM support under the con-dition of an IMF led Poverty Reduction and Growth Framework (IMF PRGF) or aWorld Bank Poverty Reduction Strategy Paper (WB PRSP) was positive and statis-tically significant (in the case of the IMF PRGF only). This suggests that PFM sup-port benefits SAI reforms particularly when they are part of comprehensive pov-erty reduction reforms.

(3)Testing the theories on institutional diffusion:All three parts of the hypothesis on institutional diffusion could be confirmed bymy statistical tests. First of all, the region a country belongs to does have a strongeffect on the power of a SAI. In table 3 we can see that there is a very high statisti-cal probability that holding all other variables constant, the SAIs in countries be-longing to the Pacific region have the weakest SAI. In concrete terms, the regres-sion results point out that these countries on average have a SAI score in theOBS2010 which is 48 points below the average for countries belonging to theEUROSAI region. The Pacific Region is characterized by small, economically weak

Island states, which also lack strong international influence by good performers, allof which explains their rather weak SAIs as measured in 2009. However, there arenew efforts by PASAI, the regional INTOSAI body, to strengthen the capacities of itsmember SAIs and there is thus hope that the picture might change in the nextyears.

The fact that an INTOSAI regional body can have a strong effect on its member SAIsbecomes particularly clear when looking at Sub-Saharan Africa (SSA). South Africa

61 Such as the CPIArevenues, World Bank taxation data, Tax compliance as measured by the Eco-

nomic Freedom of the World (EFW5Cvii) and data on direct taxation for Subsaharan Africancountries (see Broms, 2011).

62 Total aid (as % of GDP), total PFM aid, general budget support, IMF support and World BankPRSP support.


27/82

Katharina Noussi 26

has taken on a strong leadership role and together with INTOSAI partners has puta lot of effort in building the capacities of AFROSAI-E. And really the results sug-gest a better performance of the SAIs in AFROSAI-E than for other regions whichmight face similar challenges concerning poverty and authoritarianism, namelyARABOSAI and CAROSAI, but also ASOSAI and OLACEFS (see particularly regres-

sion 2b in table 3, graph D.4. and table D.11). Graph D.4 shows very well how acountries SAI power differs according to its membership in a specific INTOSAI re-gional group. SAIs in countries with a similarly high child mortality rate 63 performdifferently depending on which INTOSAI region they belong to.64 Members ofAFROSAI-E perform much better than SAIs from the other regions or without re-gional membership, with nearly all regions showing statistical significance.AFROSAI-E performs best according to this regression specification. This findingeven holds without South Africa, which has a very strong SAI (OBS2010=73.3).This positive effect of belonging to the AFROSAI-E region holds very robustly in allkinds of regression specifications; even after excluding South Africa itself, which

has a very powerful SAI.65

Only the tests on the IDI dataset (Annex D, tables D4-D6)suggest that ARABOSAI fares better than AFROSAI-E. However, the other regres-sions (D2-D3) show that ARABOSAI is highly statistically significant and robustand has particularly bad results.66

The second part of the diffusion theories pointed out that a post-conflict situationmight be an opportunity for institutional development as elite relationships areweakened and the influence of the international community can be particularlystrong. This positive effect is depicted in table 3 through a measure of the numberof battle related deaths during the period 2002 to 200867. Although this variable ishighly statistically significant in table 3 (and in table D.2., while not in table D.3.) its

coefficient is very low. For an increase by one battle related death, the SAI powerincreases on average by only 0.003 points. I suggested, that this vacuum of stab

institutionalizing public accountability - white paper

Documents