LEARNING AND POLICY CHANGE: THE CASE OF SWEDISH INTERNATIONAL DEVELOPMENT COOPERATION AGENCY (Sida) 1980-1995
LAUREAN JOSEPHAT PHILOMENA NDUMBARO
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF'THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1998
Laurean Josephat Philomena Ndumbaro
To my parents--Josephat and Philomena Ndumbaro--for
their dedication and support to my education, and to my late brother, Deonatus Ndumbaro, for his encouragement and prayers.
I would like to thank God for enabling me complete this program successfully. My sincere regards go to my supervisory chairman, Prof. Goran Hyden, whose guidance, encouragement and patience inspired me to explore my fullest intellectual potential. I also thank him for being a teacher, mentor and, above all, a friend.
I would like to extent my specialithanks to my
committee--Dr. L. Brown, D. Hedge, C. Andrew and W. Francis for their generosity in sharing their knowledge and time. In addition I would like to thank my professors at the University of Florida for helping me to develop my intellectual abilities.
Special thanks go to the Swedish Agency for Research and Cooperation with Developing Countries (SAREC) for financial support, Prof. 0. Elgstrom for his insightful comments, Prof. R. Mukandala for selecting me to join this program, Dr. K. Mease, Dr. S. Snook and Menghua Zeng for their comments in the initial preparation of this study.
I would also like to thank Sida for all the support
they gave to me during my work in Sweden and all the people iv
whose responses make up this dissertation. Special thanks go to Lars Johansson for making my work in Sweden smooth, to Mama Melania for her motherly support, and to my colleagues in the department of political science and R. Uttaro and S. Franco, for being such good friends.
I would like to express my sincere gratitude to my parents--Josephat and Philomena--for their support, encouragement and prayers throughout my entire school life. To my wife, Dolorosa, my daughter, Bridget, and my son, Ivan, I owe them my most profound thanks for their love, support and patience.
TABLE OF CONTENTS
ACKNOWLEDGMENTS . . . . . iv
LIST OF TABLES . . . . . x
LIST OF FIGURES . . . . . xi
KEY TO ABBREVIATIONS . . . . xii
ABSTRACT . . . . . xiii
CHAPTER 1 INTRODUCTION .. . . . .
Background . . . . . 1
Significance of the Study . . . 3
Development Assistance: An Overview . . 5
Multilateralization . . . . 9
The Swedish-Tanzania Aid Relation . . 10
Hypotheses . . . . . 17
Dissertation's Organization . . . 18
CHAPTER 2 THEORETICAL FRAMEWORK . . . 21
Why Learning Theory? . . . . 21
Alternative Models . . . . 23
The Class Theory . . . 24
Pluralist Theory . . . 26
Contextual Model . . . 29
Rational Choice Theory . . . 31
Bureaucratic Politics . . . 32
Learning: Conceptualization . . . 37
Learning: Defined . . . 38
Organizational Learning . . . 48.
Learning vis-a-vis Criteria for Information
Processing for Decision Making . . 54
Learning and Policy Change . . . 60
Learning Model . . . . 66
Feedback Knowledge . . . 66
Consensual Knowledge .......... 68
Institutional Processes .. .... ........71
CHAPTER 3 METHODOLOGY AND DATA COLLECTION .........78
Case Studies .......................78
Justification for the Case Study Method .. .82 Limitation of the Case study Method .. .....84
Reliability versus Validity .............86
Sampling and Data Collection. ............89
Data Collection. ............... 91
CHAPTER 4 THE DOMESTIC CONTEXT OF THE 1980s SIDA POLICY DECISIONS.........................105
Swedish Aid Administration ..............105
The Cabinet ................... 106
The Ministry for Foreign Affa.irs ........108 Sida .......... ...............109
The Budget Process...............110
Sida's Structural Change ............112
Swedish Aid Politics .......................114
Swedish Aid: The Guiding Principles. ......116
Swedish Aid Versus Changing Government
Swedish Aid: What the Public Says,.........124 Swedish Economic Situation ..............129
CHAPTER 5 LEARNING: FEEDBACK, CONSENSUAL KNOWLEDGE AND INSTITUTIONAL PROCESSES .................136
Sida and Learning..................136
Sida: Is it a Learning Organization?. ........ 137
Knowledge Seeking: From Within .........139 Knowledge Seeking: From Outside. ........ 142
Sida: How do they Consider their Work and What
are the Criteria for Decision Making? .143
Why Policy Shifts in the 1980s .......... ...146
Feedback From Tanzania ............. 154
The Rise of Cons'ensual Knowledge in the
Development Field and its Influence on
Sida Understanding of Reality .. 165
The IMF and The World Bank In Closer
Association: Elevating the New Thinking
in Third World Development . 168 Sida, Tanzania and the New Consensual Knowledge 172
Learning And Sida's Institutional Processes 180
Is Sida a Slow L.earner? . 180 Sida's Bureaucratic Politics and Leade;ship 183 Rules and Procedure for Decision Making 186 Program and Strategies . . 187
Budgetary Routines . . . 189
Ideology . . . . 191
CHAPTER 6 LEARNING: ITS IMPACT OF SIDA 193
Shifts in Norms and Priority in Aid Objectives 193
Norms . . . . . 193
Goals . . . . . 197
Mainstreaming its Approach . . . 200
Changes in Nature and Volume of Aid . 203 Shifts in Strategies . . . 208
Rural Development . . 208 Non-Governmental Organizations (NGOs) . 209 Institutional Capacity Building . 210 Import Support . . . 212
Shifts in Organizational Routines . . 213
Structural Change . . . 213
Evaluation ... . . . 214
Reporting . . . . 215
Consultants . . . . 216
Budgetary Routines . . . 217
Shift in Technical Competence . . 218
CHAPTER 7 CONCLUSION . . . . 221
Sida and Policy Change . I . . 221
The Learning Model . . . 231
Swedish Aid: The Future, and Its Implication to
Policy Makers . . . 232
APPENDICES . . . . . 237
Appendix: 1, Value of Tanzania's International
Trade (Millions Tshs) . . . 237
Appendix: 2, Official Government RevenueExpenditure 1975-85/86 (Millions Tshs) 238
Appendix: 3, Tanzania: Exports of Major Crops
1970-84 by Volume (tons) . . 239
Appendix: 4, Focus Group Scores . . 240
Appendix: 5, Tanzania's Share of Swedish Aid to
Program Countries Of Africa . . 241
Appendix: 6, Questionnaire . . . 242
REFERENCES . . . . . 245
BIOGRAPHICAL SKETCH . . . . 256
LIST OF TABLES
1.1 Sweden's Share of Bilateral ODA to Tanzania on
Percent of Total DAC Aid............14
1.2 Aid Disbursed to Tanzania by Majon Donors (in
4.1 The Allocation and Disbursement Aid as
Percentage of GNP Versus the Type of the
4.3 Swedish Economic Situation 1960-1984. ........130
5.1 Factors in the 1980s Sida Policy Shifts .......148
5.2 The Weighing of Variables in Order of
Importance ..... ........ ..... 151
6.1 Average Sectoral Allocation to Tanzania in
6.2 Tanzania's Share of Swedish Aid to Program
Countries of Africa 1975-1995. .........207
LIST OF FIGURES
Figure 2.1 Sida's Learning Model . . . 76
Figure 4.1 Swedish Public Opinion Toward Foreign Aid
1981-1995 . . . . . 126
Figure 7.1 State Commitment to Macro-Economic Reform
and its Implication to Policy Makers . 233
KEY TO ABBREVIATIONS
BITS- The Swedish Commission for Technical Cooperation.. CK -Central Committee for Swedish Technical
Assistance to Less Developing Areas. DAC Development Assistance Committee.
DANIDA Danish International Development Agency. DCO Development Cooperation Office.
FINIDA Finish International Development Agency. GNP Gross National Products.
HESAWA Health through Sanitation and Water. IMF International Monetary Fund.
IMPOD The Import Promotion office for Products from
Developing Countries. LP The Liberal Party.
MP The Moderate Party.
NGOs Non-Governmental Organization.
NIB Board for International Assistance.
NU The Swedish Board for Education in International
ODA Official Development Aid.
PP- Policy Framework Paper.
SAL Structural Adjustment, Lending.
SAREC. -. The Swedish Agency -for Research and Cooperation
with Developing Countries.
SASDA Secretariat for Analysis of Swedish Development
SDLP Social Democratic Labor Party.
Sida Swedish International Development Cooperation
SIDA Swedish International Devel9pment.
SIDO The Small Industries Development Organization.
SIP Sister Industries Program.
SWEDFUND -The Swedish Fund for Industrial Cooperation with
Developing Countries. UK United Kingdom.
us United States.
Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy
LEARNING AND POLICY CHANGE: THE CASE OF SWEDISH
INTERNATIONAL DEVELOPMENT COOPERATION AGENCY (Sida) 1980-1995
Laurean Josephat Philomena Ndumbaro August 1998
Chairperson: Prof. Goran Hyden Major Department: Political Science
This dissertation examines how learning relates to
policy change, using the Swedish International Development Cooperation Agency (Sida) as a case study. It argues that faced with dramatically different conditions for foreign aid in Africa, feedback and consensual knowledge were the most important variables for a major policy shifts in Sida in the 1980s. Whereas feedback from Tanzania signaled the necessity for change in Swedish aid policies, the neoliberal policy outlook of the IMF and World Bank provided the most viable alternative. These policy shifts led to multilateralization as manifested in changes in Sida's operational norms, aid objectives, aid approach, nature and volume of aid, organizational routines and technical competence.
Background to the Study
One of the most notable changes in international development assistance since the 1980s has been the introduction of economic and, more recently, political condit ionalities for providing foreign aid. This has significantly changed the way both bilateral and multilateral donors operate. This study focuses on the Swedish International Development Cooperation Agency (Sida) in an attempt to explain those changes'. Sida is particularly relevant because the changes it has had to undergo have been both more far reaching and painful than those experienced by other bilateral agencies. This is particularly true with regard to Sida's relationship with Tanzania, a favorite recipient of Swedish aid since the 1960s.. In fact, Sida was among the last donors to change its aid policy toward Tanzania.
'The Abbreviation for Swedish International Development Cooperation is in small letters "Sida."
T he importance of Sida to this study is further
heightened by the fact that the bulk of Swedish aid to Tanzania has been allocated by Sida, and because Sida has traditionally been the biggest donor in Tanzania. Between 1965 and 1984/85 most (if not all) Swedish aid to Tanzania was allocated by Sida; in the period between 1985/86 and 1994/95 more than 90 percent of Swedish aid to Tanzania was
disbursed through Sida
Employing a learning perspective, this study attempts to explain why Sida changed its aid provision behavior to Tanzania in the 1980s. It'also tries to show the implication of those changes within Sida with regard to the agency's objectives and ideology as well as its technical competence and organizational routines. The nature and volume of aid to Tanzania is also examined. Although this study is based on learning perspective, alternative explanations are discussed with the view of assessing their influence on the 1980s Sida policy shifts and the extent to which the policy change is a consequence of organizational learning.
2 Computed from Sweden's Development Assistance in Figures and Graphs 1994195.
Significance of the Study
This study attempts to accomplish two objectives. Most studies which deal with the changes in the 1980s have paid particular attention to issues of policy and institutional changes in the Third World countries. Little has been done to account for changes which have taken place within the international development agencies themselves since the 1980s. This study intends to not only fill this gap, but also contribute toward an understanding of the contemporary aid giving process.
At the theoretical level, this study attempts to show how international development agencies fit into organizational learning theories. Determining why international development aid agencies changed their policies in the 1980s is very important for the understanding of the operation of both multilateral and bilateral agencies. Was it simply a result of power politics among concerned interest groups, or classes? Was it caused by the need to maximize self interests? Was it the product of domestic and/or international processes that happened in the 1980s? Or, was it due to politics internal to these agencies?
One important point worth noting here is that
international development agencies are neither prisoners of
the structures surrounding them, nor are they always acting on their own will. While they are constrained by other institutions, international development agencies have a number of choices available to them. In cases where international development agencies employ knowledge from experiences and/or epistemic communities in making their choices, the learning perspective is likely to provide a more convincing explanation than others'.
Most studies in international relations that have been grounded in learning theories have focused on general changes in foreign policy (Levy 1994). Those that have dealt with learning theory in international organizations have mainly dealt with multilateral organizations such as the World Bank (Haas 1990 and Le Prestre 1995) and the European Union (Brown 1995). International development agencies, such as Sida, whose policy shifts have considerable impact on the development processes of a number of Third World countries have not been given sufficient attention in the context of learning theories. By examining Sida (a bilateral agency), this study hopes to throw new light on the usefulness of learning theories.
'For the definition of. epistemic community see Chapter Two page 70.
Develo=ent Assistance: An Overview
Development assistance as a conscious and purposeful instrument of intervention in the socio-economic and political processes of developing countries of Africa, Asia,. and Latin America is a post World War II phenomenon. The Marshall Plan is its logical precursor (Arnold, 1979). Although development aid has declined in recent years, throughout the 1980s it grew in volumeland importance. In some developing countries, for example, development aid accounts for more than 50 percent of the national development budget. Under such circumstances, it is difficult to envisage a situation wherein these countries could undertake development without foreign assistance.
In respect to the countries of Sub-Saharan Africa, development assistance has become their life and blood to the extent that it is sometimes assumed that most, if not all, cannot operate without it. Indeed, most sectors of African economies are in one way or another financed by foreign aid. Unfavorable world trade, particularly for primary commodities, and the unavailability of commercial loans explain in part African governments' dependence on foreign aid (Sandbrook 1993). In fact, Africa today not only is more dependent on foreign aid than it was at the time of independence but also is relatively as poor as it
was thirty years ago (World Bank 1989 and Morton 1993). The latter may be attributed to the fact that much of the aid disbursed in the Cold War period was not used for development, but rather for ensuring the economic and political survival of governments in power (Arnold 1979, Commons 1988, and Bayart, 1993).
In most cases, that portion of foreign aid used for development was not very effective largely because development assistance during the post World War II era operated on the assumption that states in Third World countries would have the capacity to continue to efficiently run development projects once aid was terminated. In fact, the capacity was not there. As a result, most of these projects were either under utilized or simply died following donor termination. An interesting point here is that the ever growing aid dependence makes what donors do of great signi fi'cance to the socio-economic and political processes of Africa.
Donor agencies have shifted their attention from one
sector of the economy to another since the 1950s in response to changes in the social, economic, and political processes of Africa as well as changes in global power relationships (Sandbrook 1993). Consequently, donor perception of
Africa's problems has changed over time as well (Radetzki 1992, Hyden 1994, and Morton 1993).
In the 1950s and 1960s, for example, more emphasis was put on industrialization (the import substitution approach) and "modernization." In the early 1970s, attention shifted from industrialization and modernization to agriculture under an "integrated rural development approach" (Hyden, 1994),. This approach inco.aitporated a more explicit "poverty. focus" (basic needs strategies) than the previous one.
The world-wide economic recession of the 1980s was felt more severely in Sub-Saharan Africa than in any other part of the world (World Bank 1989) The 1680s economic crisis in Africa was partly seen as the failure of the previous approaches to live up to expectations (World Bank 1989 and Hyden 1994) This failure was attributed mainly to the problems of macro-economic policy in these countries (World Bank 1989, 1994) As a result, in the 1980s policy reform came to head the agenda for Third World development.
The policy reform agenda was not introduced easily.
There were considerable economic and political constraints to the introduction of such reforms. In fact, economic and political conditionalities for providing foreign aid had to be introduced to facilitate the acceptability of macroeconomic and sectoral policy reforms such as trade
liberalization and privatization in most of the Third World countries. In the case of Sida, for example, Hyden at al, (forthcoming, p.453) write, "... Sida lias fallen in line with the 'Washington consensus'. In its relation with Tanzania in recent years, Sida has followed it quite strictly so as to add pressure on the Tanzanians to take the necessary reform measure." The introduction of macroeconomic and sectoral reforms significantly altered the thinking and the operation of both bilateral and multilateral donors.
These changes were more important than the previous
ones as they paved the way for a fundamental change in the donors' perceptions of their aid goals, Ipolicies, and strategies. It was not just a shift of attention from one sector to another as with previous changes. Changes in the 1980s called for the redefinition of the entire aid program including the reasoning therein. The underlying assumptions of the role of the state in the development process and the ideological logic of development assistance were brought to the test of history in the'1980s (Jackson and Rosberg, 1986,' Migdal, 1988, Evans, et al 1985, and Evans, 1989) The results contradicted existing theories of development and practices.
One of the most significant manifestations of the
changes of the 1980s was the "multilateralization" of what had been bilateral relations in the aid-giving process. The move to conditionally involving a third party--the international financial institution--and embracing its policy outlook in the previous bilateral aid provision process is referred to as "multilateralization" in this study.4
The Swedish-Tanzania aid relationship was
multilateralized beginning in the mid 1980s. Swedish aid to Tanzania began to depend upon not only the agreement between Tanzania and Sida as before, but also the signing of an agreement between Tanzania and the International Monetary Fund (IMF). In other words, Tanzania was required to show a certificate of approval from the IMF before Sida would commit most of its funds (Campbell and Stein 1991).
4 The provision of bilateral aid to Third World
countries, which prior to the eighties was "purely" a matter of agreement between a donor and a recipient (bilateral), experienced the inclusion of important third parties (the IMF and the World Bank) in the matters of policy and conditionality. Aid could not be given to most of the Third World countries without a certificate of approval from that third party (Campbell and Stein, 1991, Lele, 1992, and Radetzki 1992).
Other donor agencies, such as the Danish International Development Agency (DANIDA), Finnish International Development Agency (FINIDA), and the United States Agency for International Development (USAID), also introduced similar changes in strategy. It is interesting to note, however, that bilateral donors responded differently. USAID, for example, was quick to embrace the IMF's conditionalities while Sida was among the last to do so. Why this happened the way it did will be discussed later.
To Sida, multilateralization manifested itself in a number of ways, the most significant ones being shifts in Sida's ideology, priorities among aid objectives, and the volume and nature of bilateral aid.5 In this respect, multilateralization also connotes a new vision in Sida's thinking about the aid provision process.
The Swedish-Tanzania Aid Relation
Tanzania won its independence in 1961. Ever since, Tanzania has been one of the major beneficiaries of development aid. In fact, its share of total global development assistance has been around 2.5 percent (SASDA
'Multilateralization discussed here is different from
the process of giving aid through multilateral organizations such as the Bretton Wood institutions and United Nations' agencies. Sweden has always been giving aid through such multilateral agencies, but such support did not require a certificate of approval from either of these agencies.
Report, 1994). According to SASDA (1994), Tanzania received
8 percent of all Development Assistance Committee (DAC) assistance that went to Sub Saharan Africa in the period between 1965 and 1991. In monetary terms, Tanzania received about US $9 billion (current) during that period. Among the major bilateral contributors of development assistance to Tanzania is Sweden.
Swedish aid to developing countries officially started in 19"52 with the establishment of the Central Committee for Swedish Technical Assistance to Less Developed Areas (CK). This Committee was formed by non-governmental organizations (NGOs) Forty-five NGOs of different orientations became members of the CK. They ranged from religious, industrial, and labor to cooperative unions and athletic organizations. CK had two major responsibilities. The first was to establish a Swedish aid program to developing areas. The second was to run programs to help raise public awareness and support for increasing Swedish participation in Third World-development. These goals were implemented through public education efforts and fund raising campaigns. Two national fund raising campaigns and education were organized by CK during a decade of its existence.
Thus, Swedish development aid was initially an
activity of non-governmental organizations. Missionary
groups were the Swedish pioneers of the aid endeavor. In this respect, moral considerations and solidarity with the poor have been at the heart of Swedish aid to developing countries. It is not surprising then that foreign aid in Sweden is seen as a moral obligation (Karre and Svensson, 1989) This perception of foreign aid could partly explain why the first decade of Swedish aid was both less systematic and more guided by moral and idealistic principles (Elgstrom, 1992) Moreover, it could also explain why Swedish commercial interests in most cases have taken a back seat in Swedish aid to developing areas, particularly before the late 1970s.
By 1961, the Swedish government saw the need to play a dominant role in the aid process. The'Board for International Assistance (NIB) was created to replace CK. Whereas CK was a Committee for NGOs, the NIB became a separate administrative unit within the Ministry of Foreign Affairs dealing with all bilateral aid. Within four years of its existence, the NIB found itself immersed in managerial and organizational problems which led to its dissolution in 1965 (Jellinek, et al, 1984) The Swedish International Development Authority (SIDA) was formed instead. Thirty years later SIDA changed its name to the Swedish International Development Cooperation Agency (Sida)
following its merger with five other Swedish aid agencies6 Unlike the NIB, Sida was created as anlautonomous government agency under the auspices of the Ministry of Foreign Affairs. Sida is responsible for planning, implementing, and monitoring all bilateral cooperation.
While Sida is responsible for administering all
bilateral aid, multilateral aid is handled by the Ministry for Foreign Affairs which has overall responsibility for all government aid. Since its establishment, Sida remains the central agency for Swedish bilateral aid. The relationship between Sida and the Ministry for Foreign Affairs was not particularly harmonious at the beginning partly because the boundary of authority and responsibility between the two was fluid. Over the years these boundaries have been redefined and the relationship between the two has become amicable (Elgstror, 1992).
With respect to Tanzania, Swedish aid officially began in 1964 under NIB. Since then, Tanzania has been one of the main beneficiaries of Swedish aid. For example, in the
'Before 1995 Sweden had five small agencies dealing with specific issues--The Swedish Agency for Research Cooperation with Developing Countries (SAREC), The Swedish Fund for Industrial Cooperation With Developing Countries (SWEDFUND), The Swedish Commission for Technical Cooperation (BITS), The Import Promotion Office for Products from Developing Countries (IMPOD), and the Swedish Board for Education in International Development (NU).
period between 1965 and 1995 Tanzania received a total of 20,258,843,000 Swedish Kronor (in 1995 prices). Of this amount, about 19,407,865,000 was disbua sed through Sida. Compared with other countries receiving Swedish aid in this period, Tanzania was given more from the Swedish aid budget than any other country, Tanzania received about 16.6 percent of all Swedish aid disbursed to program countries of Africa between 1965 and 1995.7
Table 1.1 Sweden's Share of bilateral ODA to Tanzania in percent of Total DAC aid
Year Share Year Share Year Share Year Share 1962 0.05 1971 21 1978 19 1985 13
1963 10.80 1972 31 1979 20 1986 21
1964 4.09 1973 36 1980 14 1987 11
1966 6.90 1974 25 1981 16 1988 13
1967 9.90 1975 23 1982 15 1989 13 1
1969 30 1976 24 1983 16 1990 18
1970 118 11977 122 11984 13 11991 19
Source: SASDA Report No. 5, 1994, p.5.
Table 1.1 above further demonstrates the contribution of Sweden's bilateral Offi'cial'Development Aid (ODA) to Tanzania as a percent of total Development Assistance Committee's (DAC) aid. The table shows that Sweden has made
7 Computed from Sweden's Development Assistance in Figures and Graphs 1994/95.
a significant contribution to Tanzania's development process. As can be seen in Table 1.1, Sweden's share to Tanzania increased from an average of 8 percent in the 1960s to over 20 percent in the 1970s. It then declined to about 15 percent in the 1980s and rose again to about 20 percent in the early 1990s.
The rapid increase of Swedish aid to Tanzania in the
1970s could be explained by developments which took place in both Sweden and Tanzania. The Ujamaa and self-reliance policy adopted by Tanzania in the late 1960s emphasized development, equality, "we-lfarism," support for liberation movements, and non-alignment. These policies were not only perceived as progressive by Sweden but were in line with Swedish aid objectives. The fact that Swedish aid objectives coincidentally agreed with Tanzania's development aspirations fostered a spirit of friendship and mutual understanding. This spirit explains the generous Swedish contribution to Tanzania's development efforts. Tanzania, as a result became a favorite recipient of Swedish aid in the 1970s and early 1980s. Table 1.2 below shows Swedish aid to Tanzania in comparison with various donors.
Compared with other donors, Sweden has been Tanzania's largest bilateral supporter(Table 1.2). Between 1965 and 1992 Swedish aid averaged about 12.14 percent of all
bilateral aid to Tanzania, as shown in table 1.2 above. Sweden exceeded the second placed Netherlands by 4.07 percentage points of the total bilateral aid to Tanzania, and the United States by more than 10 percentage points.
Table 1.2 Aid Disbursed to Tanzania by Major Donors (in US Dollar).
Country 1970s 1980s 1990-92 Total Percent
Sweden 422 780 386 1588 12.4
Nether 246 607 202 1055 8.07
Germany 252 533 205 990 7.57
Norway 150 566 271 987 7.55
Denmark 194 475 263 932 7.13
UK 102 441 197 940 5.66
Italy 4 486 210 700 5.35
Japan 51 453 171 675 5.16
Canada 177 300 96 573 4.38
Finland 70 291 127 488 3.73
IUSA 1 167 1 174 1 101 1 442 1 3.38
Source: SASDA Report, no. 5 1994, p.20. Key: *= Percent of the total aid received since the 1970s
The two tables above show the importance of Sweden to the Tanzanian development process. They also show that the Nordic countries, together with the Netherlands, contributed about 38.53 percent while the contribution of the major western powers--the United States, UK, and Germany--is 16.61
percent. It is worth noting that Sweden, both because of the size of its aid to Tanzania and its influence among other donors, particularly the Nordic countries, is a case of particular interest to the student of foreign aid.
The following propositions regarding the 1980s Sida policy shift are based on the discussion provided above'. The propositions that guide this study are as follows.
1) Sida initiated policy changes in the mid-1980s
a) its own projects and program portfolio had
failed to achieve acceptable results (the
b) the IMF's and World Bank's policy outlook was
increasingly more persuasive for Sida (the
Consensual knowledge criterion);
c) Sida's institutional processes slowed the
learning process (the institutional criterion).
'Feedback knowledge, consensual knowledge and
institutional processes are theoretical basis from which these propositions are derived. For detailed account see Chapter Two, page 66-78.
2) Sida's response to feedback and consensual knowledge
a) a marked shift in priorities among aid
b) growing acceptance of "mainstreaming" its own
3) The impact on Sida's organizational processes has
a) a relative shift in organizational routines;
b) a shift in technical competence.
Dissertation's OrganizationChapter 2 presents a literature review and discusses the theory to guide this study. Learning theory, which gained momentum in explaining policy change and choice in the 19803, provides the lens through which this study is analyzed. Chapter 2 sets the parameters for discussion by presenting the theoretical and empirical justification for the applicability of learning theory in this study, reviewing alternative theories, defining key learning concepts, levels of analysis, types of learning, and describing some of the debates surrounding learning. The learning model that guides this study is also presented.
Chapter 3, discusses methodology and data collection. It shows the strength and weakness of case study, and the justification for using this method in social science research. The whole process of data collection and problems encountered from sampling, interviews, document analysis to focus groups are discussed. This chapter also explains why this study used interviews, focus groups, and document analysis as techniques for data collection. The strengths and weaknesses of each technique and how the combination of these three data collection methods helped to reduce problems associated with each one also are explained.
Chapter 4, examines the Swedish economic and political environment in order to explain the domestic context of Sida's 1980s policy shifts. In particular, this chapter describes how Swedish economic performance, changes in Swedish public opinion about foreign aid, and changes in political leadership have affected Sida aid policy in the 1980s. The Position of Sida within the Swedish government's hierarchical structure is also presented.
Chapter 5 portrays Sida as a learning organization. It explains how feedback knowledge from Sida's programs and projects in Tanzania affected its understanding about development endeavor. In addition, it focuses on why and how the new consensual knowledge in the development field
came about in the l980s and the role it played in Sida's understanding of reality. This chapter also discusses the role played by Sida's institutional processes in the 1980s policy shifts.
Chapter 6 presents the impact of learning on Sida as an organization. It argues that lessons drawn by Sida from both feedback and consensual knowledge impacted not only Sida's norms, institutions, and principles but also the nature and volume of aid. Chapter 7 provides conclusions, prospects for Swedish aid, and future research questions.
Why Learnina Theory?
This chapter presents*learning theory as the lens through which this study is analyzed. The stage for discussion is set in this chapter by presenting a theoretical and empirical justification for the applicability of learning theory to this case study, reviewing alternative explanations, defining key learning concepts and levels of analysis. The chapter ends by providing a learning model which guides this study.
The incentive for the selection of learning theory is partly provided by the nature and intensity of the debate about the Swedish aid process and in-depth aid performance analysis which was carried out by Sida in the period between 1979-84 (Elgstrom 1992). Prior to 1980, there was a political consensus on aid among different actors in Swedish politics. The missionary spirit and solidarity with the poor were the cornerstones of this consensus which ruled out both the possibility of any political challenges from within Swedish politics as well as "interference" in the political 21
processes of developing countries. Under such circumstances, policy shifts had to come through learning.
Moreover, the 1980s saw multilateral institutions such as the IMF and World Bank, and bilateral institutions working closer together in development policy than at any other time in history. This closeness created an optimal environment for both the development of consensual knowledge about development aid, and the globalization of that knowledge. This manifested itself in the way development language has been increasingly standardized and key prescriptions for development problems have been homogenized across different aid agencies in the last ten years. These developments, as will be discussed later in this study, provide important variables for organizational learning.
A number of research studies-- Claesson (1982), and
Ekengren (1984), Forss (1985) Lele (1992), Elgstrom (1992)-were initiated in the 1980s and 1990s to evaluate aid performance and to suggest lessons which could be drawn from thirty years of experience in development aid. The underlying logic for initiating these studies was not only that organizations are capable of learning but also that they are devoted to improving their understanding.
The utility of learning theory has been demonstrated by scholars from different fields and methodological
backgrounds. Breslauer (1987), Mendelson (1993), Moltz (1993) and Stein (1994), for example, successfully employed learning theories in an-attempt to explain shifts in Soviet foreign policy. Others, such as Leng (1983) and Reiter (1994), explained changes in the Super Powers' perception of international security and cooperation through learning theory. Haas (1990), Brown (1995), and Le Prestre (1995) have also successfully used learning theory in explaining policy changes in international organizations. Argyris and Schon (1978) attempt to provide a theory of action in organizational learning. Thus, the above mentioned issues provide not only the stimulus but also an empirical and theoretical justification for learning theory.
Learning is not the only theory which is relevant to explain policy change and choice. Scholars from a variety of disciplinary and methodological ori nations have shown increasing interest in explaining policy change and choice in different perspectives. Two major groups can be isolated: those who provide society-centered and those who provide state-centered explanations. Society centric theories include class theory and pluralism and the contextual model, while in state centric theories,
bureaucratic and rational choice have been particularly prominent.
The Class Theory
Class theory asserts that policy and policy change is rooted in the relationship of power and domination among
classes'. According to this view, the dominant relations of production in a society sets parameters for both classes and power relationships. Societal conflicts, whether social, political, or economical, are all explained in terms of class, based on economic relationships among groups of people. Policy is seen as instrumental in reconciling class antagonism. An overall objective for any policy, in this view, is to safeguard the interest of the dominant class and/or ensure the persistence of the dominant system of production.
Within this paradigm, policy change is explained in
terms of shift in the dominant class coalition or as a means for ensuring the survival of the capitalist system (Poulantzas 1973). The relative autonomy of the state (visa-vis other classes) grants it the flexibility to deal with any threat arising from subordinate classes (Poulantzas,
9Although class explanations have seemingly been
relegated to the museum of antiquity, some people in both Sweden and Tanzania showed interest on it. For them we include this explanation in this study.
1973). In situations where the survival of the capitalist system is threatened, state agencies can initiate a policy or policy change that seemingly contradicts the interest of the ruling class. According to Poulantzas (1973), this is only possible because the relative autonomy of the state allows it to nurse the myth that policy change serves the interest of the society as a whole while in reality the survival of the system is for the interests of the dominant coalitions. As Lenin (1976:45) also, writes:
People always have been the foolish victims of
deception and self deception in politics and they will
always be until they have learnt to seek out the
interests of some class or other behind all the moral, religious, political and social phrases, declarations
In the case of Sida's policy shifts during the 1980s, Class explanations seems inadequate. The social democratic government under which Sweden's development aid policies were established in the 1960s and early 1970s was replaced by a conservative government in 1975-82. No significant policy shifts in development aid were witnessed with this change of power and concurrently there were no discernible shifts in patterns of class stratification, i.e. class relationships remained constant during this period. In fact, the 1980s Sida policy shifts came when the government which established it was in power. In other words, the
class-which was in power when policy shifts occurred in the' 1980s was the same class which established it. Class explanations would be more convincing if those policy shifts had occurred with a change of government.
Moreover, scholars who have dealt with Swedish aid agree to the fact that until the 1980s Swedish aid was mainly driven by moral and solidarity convictions (Elgstrom 1992). These moral and solidarity motives crisscross Swedish class boundaries and formed the basis for Swedish aid. Class explanations cannot adequately address these issues. Moreover, the deterministic character of Class explanations obscures not only learning but also the role of choice in policy change.
Pluralist theory, on the other hand, situates policy
and policy change within the context of conflict, bargaining and coalition formations among societal interest groups. According to this view, society is made up of a variety of groups which are organized to safeguard the special interests of their members. These groups compete and unite behind common policy issues.
To pluralists, policy change and choice depends on how different interests from the society are mobilized,
aggregated, and articulated. Baumgartner and Jones (1993.:1.90) write:
The mobilization of interests changes over time, and
with those changes come difference in the likelihood of
certain issues to hit the public agenda. Interest
groups play an important role in formulating questions, affecting public opinion and defining the terms of the
Thus, shifts in interest group coalitions and changes in their bargaining competence are instrumental for policy change. In short, the source and content of policy and policy change in this perspective are societal groups. The major function of state institutions, according to the pluralist view, therefore, is to guide and direct interest group competition into a particular policy resolution.
Contrary to this view, scholars including Stepan, (1978), Evans, Rueschemeyer, Skocpol, (1985), and Nordlinger, (1987) argue that state agencies are not arenas lacking interest or power of their own. Skocpol (1982:4), for example, argues that states do 11 ... formulate and pursue goals that are not simply reflective of demands or interests of social groups, classes or society." By denying governmental bodies' capacity to formulate independent policies, pluralists implicitly obfuscate the capacity of the state agency not only 'to make independent policy but also to learn.
Like Class explanations, pluralists tend to obscure the role of learning in the policy process in favor of power politics. Our findings show that learning more than anything played an important role in the 1980 Sida policy shifts.1 Findings show that consensus existed among societal interest groups in support of existing aid levels and policies particularly, the non-interference policy. In such circumstances it is unlikely that pluralism can offer adequate explanation. Instead, the relative autonomy which Sida and other Swedish agencies enjoyed was sufficient enough to change policy based on learning without having to seek the approval of domestic interest groups. Moreover, the knowledge base which Sida has in international development, and the multiple, heterogeneous contexts in which it operates, limits the explanatory power of pluralism in regard to the 1980s policy changes". This conclusion is supported by Jellinek, et al (1984) and Elgstrom (1992) who argued that interest groups played a minimum role, if any, in the 1980s Sida policy shifts.
1For more detail see Chapter 5.
"For detailed discussion see Chapter 4.
A general consensus exists among organization theorists that a sound explanation of an organization's actions and decisions cannot-be drawn from.internal factors alone (Moe,. 1985, Tipple and Wellman, 1988, and Henry, 1995). The consensus is more apparent when it comes to public organizations than private organizations.
Henry (1995:67), for example, identifies two factors which make the context of public organization an important variable in understanding organizational decision making.
The high importance of the demands being made on the
public organization by its task environment relative to
internal management needs; the unique impact of the
task environment on the public organization's structure and bureaucracy; its decision making; the motivation of
public managers, their satisfaction with their work;
the public organization's internal management; and even
the performance of it-s administrators.
This is particularly true because public organizations are the product of the political process. As such, they are largely shaped and controlled by the political and economic environments in which they operate. As Gabris (1983:143) emphasizes "Public agencies are subject to and cannot escape their political environments." The impact of environment on the inner workings of public agencies, therefore, is considerably more significant than in private organizations (Chubb and Moe, 1988 and Henry, 1995). This study would
fall short of a convincing explanation without paying particular attention to Swedish politics. As a result, an examination of Swedish politics is presented in Chapter 4.
Actions and decisions made by public agencies attain their legitimacy not only from legislation and internal expertise, but also from their domestic constituents (the public). In fact, as public agencies grow, the need for public political support becomes an important issue. Public opinion concerning the actions and decisions of public agencies is very important for their survival. This study, therefore, examines Swedish public opinion on development aid in an attempt to assess its influence in the 1980s Sida's policy shifts.
As most public agencies draw their funds from
government budgets, the state of the national economy becomes an important factor in their actions and decisions. Reduction of government expenditure inIcertain budget areas can affect agencies concerned. An examination of the Swedish economic situation in the 1980s is important in assessing its influence in the 1980s Sida's policy shifts.
The structure of the government and the position of an agency in the governmental hierarchy are also important in agency's decisions., The examination of Sida's position within the government structure and Sida's decision process
enables this study to establish the influence of the government and the parliament in the 1980s Sida's policy shifts.Thus, this study examines four variables--aid
administrative structure, Swedish aid politics, Swedish public opinion about development aid, and the Swedish economy--to establish the extent to which contextual factors influenced the 1980s Sida's policy shifts. Rational Choice Theory
Rational choice theory focuses on the individual or
organization acting as rational beings in explaining policy formulation and policy change (Frohock, 1979). Rational choice tends to confine the rationality of decision makers to maximization of their self-interests. While, rational choice can mainly explain policy change in a situation of poor performance in terms of inefficiency of the means, it cannot sufficiently deal with the learning conveyed by feedback and the role played by epistemic actions in policy change. In the 1980s Sida policy shifts where lessons from experience and the consensual knowledge in international development field played an important role rational choice cannot adequately address these issues.
Moreover, rational choice provides limited insight into how factors such as contextual factors, historical
experience, ideologies, and alliances, enter decision making circles and shape or even determine decisional outcomes (Grindle and Thomas, 1991). These issues are important in explaining the 1980s Sida policy shifts as chapter 4 and 5 show.
Rational choice theory also has a tendency to disregard the dimension of power. As a result it tends to trivialize institutional processes which not only define power relationship within the organization but also constrain choices. As chapter 5 reveals, Sida's institutional processes, particularly the leadership played an important role in setting and opening parameters for debate which led to the 1980s policy shifts. By belittling or ignoring the dimension of power, rational choice theory cannot adequately account for the role played by institutional processes in the 1980s Sida policy shifts.
Moreover, in the context of Swedish aid, where
altruistic and moral factors played an important role in aid decisions, rational choice theory falls short of a convincing explanation.
Bureaucratic politics explain policy change in terms of competition among bureaucratic entities and actors who are constrained by their organizational roles and capacities
(Allison, 1971, and Grindle and Thomas, 1991). The resource base for their participation in policy making is the position they hold in the organizational hierarchy, their access to and control of information, and their bargaining competence12
Strategies and tactics employed by proponents and
opponents of the policy under discussion are instrumental for policy decision. Allison (1971:145), for example, in support of this view argues that a policy decision is the result of "... not simply the reasons that support a course of action, or routines of organizations that enact the alternative, but the power and skills of proponents and opponents of the action in question."
All these factors are important for explaining policy change, however, the importance of each one depends upon the issue in question. Three scenarios could be drawn in which one of the factors plays a.more important role than others. The first scenario.is the situation where evidence does not overwhelmingly support or preclude a certain course of action. In such a situation, proponents' and opponents' skills and power are likely to play aniimportant role. The
12 Bureaucratic politics in the context of this study is limited to power relationship and competition of ideas among bureaucratic entities and actors, and the way different bureaucratic entities and actors influence policy decisions.
second scenario is a non-controversial policy situation. Here, organizational routines are likely to play an important role. The third scenario is a situation where feedback knowledge or new knowledge regarding a particular issue or both 'are overwhelmingly in favor of a new course of action.' Under such circumstances knowledge that supports a' new course of action is likely to play a more important role.
The last scenario is more evident in the 1980s Sida's policy shift. As this study will show both feedback knowledge from the field and new knowledge in the development field in the 1980s fostered a redefinition of goals, strategies, and policy to address issues raised by that knowledge. In Sida, for example, feedback knowledge from Tanzania, coupled with the new dominant consensual knowledge in the development field in the 1980s, made even those who would have otherwise been opponents of the 1980s Sida's policy shift to Tanzania (for political reasons or emotional attachment to a project in Tanzania or otherwise) accept policy change (Elgstrom, 1992, dnd personal interviews)."
While bureaucratic politics is important for
understanding what is happening within an organization, it
13Questionnaire number 20.
is inadequate for this study because it gives only a superficial explanation about ways knowledge is obtained from experience, and how historical contexts shape and influence policy change. Moreover, bureaucratic politics provide "... little analytic capacity to deal with issues such as professional biases and alliances or consensus on technical analyses that cut cross bureaucratic boundaries" (Grindle and Thomas 1991:30), issues which played an important role in the 1980s Sida policy shifts. Learning theory addresses some of these deficiencies. Findings presented in Chapter 5 reveal that in the 1980s Sida policy shifts bureaucratic politics was more of a catalyst than a cause for change. In this respect, bureaucratic politics alone cannot successfully explain the situation.
Further, the study argues that all bureaucratic policy decisions are in one way or another grounded in bureaucratic politics. In this respect, bureaucratic politics is seen as an institutional process which is not only found in all bureaucratic organizations but whose influence on a particular policy also depends on a policy situation. This study, therefore, conceptualizes bureaucratic politics as an institutional process rather than a model for policy making. In this way, learning as a source of policy change is not obfuscated.
Generally speaking, the above explanations can be
grouped into two major camps. State-centric models base their explanations on people working within government institutions. These perspectives view contextual factors as well as historical experiences as playing a limited, if any, role in the initiation of policy or policy change. Society based explanations, on the other hand, mainly focus on societal groups or classes in explaining policy change. In fact, "... most or nearly all public policies are understood as responses to the politically weightiest societal expectations, demands and pressures; the state is almost invariably unwilling or unable to act upon its preferences when these diverge from society's" (Nordlinger 1981:43). The roles and interests of the people working within the government, therefore, are belittled by these explanations. The lens used in this study takes the middle road between the two major camps.
Learning theory asserts that members of an organization do initiate policies and policy change. In the process, they are influenced by both internal and external factors. Internal processes, including organizational routines, leadership, historical experience and ideology, and external factors, inter-alia, feedback and consensual knowledge from
the organizational environment, all have bearing on an organization's policy process. The importance of each factor, however, is issue and context specific. Learning theory, as discussed in the context of this study, attempts to address these issues.
Learning as a concept has been a subject of controversy among scholars. Scholars from different disciplines and methodological orientations have defined learning in different ways. Learning is defined as: a change of behavior in a manner that enhances effectiveness (Porter, 1980, Argyris and Schon, 1978), a change of behavior as a result of acquisition of information or new knowledge (Jarosz and Nye 1993), the understanding of a particular policy question (Reiter, 1994), a change in belief (Levy, 1994), a change in cognitive structure (Bennett, 1990), increase in cognitive complexity (Etheredge, 1985), a process of specification of causal relationship in new ways (Haas, 1990).
Tetlock (1991), draws.'five definitions from the learning literature with each definition rooted in a specific approach. He identifies the definition of the neorealists, the psychologists and cognitive theorists, the cognitive structuralists, the organizational and political
cultural theorists, and the efficiency definition of learning.
As a theoretical framework, learning is also a subject of contention. The controversy is complicated by the lack of a unified theoretical trend, a variety of types of learning (e.g. individual, organizational, governmental, political, social)", different levels of analysis used by different scholars, different levels of learning (single loop vs double loop, simple vs complex, peripheral vs fundamental). Other contentious issues include learning vs unlearning, intentional/conscious vs unintentional/unconscious learning, causal linkage between learning and policy or behavioral change and the distinction between learning and adaptation. Given these contending views, this section discusses each aspect of the controversy in an attempt to locate the current study in a proper perspective.
Porter's (1980) conceptualization of learning as an intentional or conscious process is criticized by Huber (1991) who argues that learning need neither be intentional
"For individual learning see Moltz (1993) and Bennett (1990). For political learning see Heclo (1974) and Hall (1987) for social learning. For governmental learning see Etheredge (1985) and Bennett (1990).
or conscious nor does it necessarily increase the learner's effectiveness. Argyris and Schon (1978) link learning with increasing effectiveness, however, in their 1996 publication, Argyris and Schon agree that learning need not always increase effectiveness. They recognize the distinction between learning that increases effectiveness and learning that does not. They call the former "'productive learning" and it is the center of discussion in their work.
Despite recognizing the distinction between coming to see things in new ways and behavior change, Jarosz and Nye, (1991), and Argyris and Schon (1996) argue that for learning to be of any theoretical and practical utility it need be conceived in relation to behavior change. To Friedlander (1983), learning can result in change of one's understanding or cognitive structure without requiring a change of behavior. In fact, in certain circumstances learning results in reinforcing an existing behavior.
Levy (1994) sees the conceptualization of learning in a generic sense as having.'more theoretical and practical utility than one that is limited to behavioral or policy change. According to him, this allows students of learning to explain "... when individual learning gets translated into policy and when learning gets blocked by institutional
or*po-1-itical constraints" -(Levy 1994:290). 1 agree with Friedlander (1993) and Levy (1994) that there is no need for linking 1 earning with behavior or policy change a priori. The linkage need remain as an empirical question. In other words, the conditions under which learning leads to behavior or policy change needs to be a subject of empirical investigation.
Reiter (1994:492) defines learning as "... the
application of information derived from past experiences to acquire understanding of a particular policy question." This definition is too narrow for this study. As argued above, learning is more than acquisition of an understanding of specific policy questions. Learning can involve coming to see things in new ways without focusing on a particular policy issue. Besides, learning need riot always require the use of information obtained from past experiences. It may also include the application of information derived from the experiences of others (Hedberg, 1981 and Rose, 1991), new knowledge (Haas, 1990 and Jarosz and Nye, 1991) and study (Bennett, 1990).
Levy (1994:283) defines learning 11 ... as a change in belief (or degree of confidence or the development of new skills or procedures) as a result of observation of and
interpretation of experience." Not all change in beliefs as
a consequence of observation or experience involve learning. Individuals or organizations can be forced to change their beliefs for fear or many other reasons besides learning.
While this definition may explain changes in foreign policy in general, it falls short of explaining changes in international aid organizations. International aid agencies have specific tasks to fulfil. The overall goal of Sida, for example, is helping to improve the living standard of poor people. The way these agencies carry out their programs depends very much on their understanding of the recipients' socio-economic and political processes. In fact, unfamiliarity with and uncertainty of the recipients' environment make development aid agencies operate their programs mainly on an "experimental" basis (Hedberg, 1981).15 International development agencies repeatedly engage in testing, construction, and reconstruction of their
knowledge. In this respect, the definition of learning that focuses on change in beliefs does not capture the reality underlying international development aid agencies.
"This study uses uncertainty in the same meaning that is used by George (1980). The conditions of uncertainty which he identifies include inadequate information surrounding the issue in question or limited available knowledge for assessing the expected outcomes of'different course of action. It seems to this writer the situation under which international development agencies make policy decisions fits George's definition of uncertainty.
One important aspect which this study borrows from Levy's definition is "change in a degree of confidence." The uncertainty and unfamiliarity with Third World environments are adequate justifications for this study to include "change in a degree of confidence" in a definition of learning involving international development aid agencies or individuals working in them.
To Bennett (1990:100), learning consists of 11 ... change in cognitive structure ... as a result of experience or study...". When it-tomes international development aid agencies or individuals working in them (as discussed above), change in cognitive structure is just one side of the learning coin. The other side is consolidation of those structures. For example, when a program run by an aid agency is successful, the existing cognitive structures are reinforced; when the program fails, cognitive structures change.
Furthermore, Bennett's (1990) emphasis on study as an important source of information for learning is very relevant to international aid organizations and individuals working in them. Most of these agencies (individually, or in cooperation with others) sponsor a number of studies with the view toward gathering information or knowledge on a particular issue. The use of consultants by these agencies
to gather this knowledge falls under information derived from studies.
'Etheredge (1981) sees learning as a process by which
learners increase their intelligence arld sophistication in a way that enhances their effectiveness. To him, learning involves an increase in sophistication in cognitive complexity, evaluative complexity, or cognitive integration. By cognitive complexity, he means a variety of different arguments underlying a set of beliefs. Valuative complexity consists of both a level of inconsistency among them and the manner-into which the inconsistencies and value trade-offs of different arguments are dealt with in a larger framework. In this view, simple policy changes are not a part of learning.
According to Levy (1994), the problem with this
conceptualization is the implicit assumption that increased sophistication in cognitive structures leads to an improved understanding of reality or better skills and effectiveness. In reality, increased cognitive complexity does not necessarily result in learning or in increasing effectiveness. This conceptualization produces also some methodological problems--simple mind versus complex mind. How do we identify and measure simple cognitive structures or complex cognitive structures? Is the complexity or the
simplistic cognitive structures across-context or transissue-s or is it constraine-d by-the'context or issue under question? In other words, is the complexity or simplicity of cognitive structures universal or issue and context specific?
Another concept of learning is provided by Haas (1990). In his view, learning is a "... process by which consensual knowledge is used to specify causal relationship in new ways so that the results affect the content of public policy" (p.23). Specification of causal relationship is an important ingredient in linking learning to policy change. For the purpose of this study, however, this definition is inadequate. The major limitation of Haas' conceptualization is that it sees epistemic communities as the sole producer of causal explanations. In this view, members of organizations are passive recipients of new knowledge from epistemic communities.
While this may be true in some contexts, in others such as the development context, organizations work hard to produce new knowledge and establish causal relationships among the variables with which they work. More often than not, an international aid development agency, (or its department), individually, or in collaboration with other agencies (or departments) initiates studies with the view of
establishing causal relationships among different issues. Thus it is not necessarily the epistemic community which can provide a causal-linkage.
Drawing from the above discussion, learning for the purpose of this study is defined as a -process whereby information or knowledge derived from experiences, epistemic community or study significantly affec-H the understanding of realiLy. By "significantly" I mean at least part of understanding of reality is reevaluated or confirmed. By experiences I mean own experience and the experience of others. By study I mean own study and studies done by others. This conceptualization recognizes the distinction between acquiring new understanding and translating these new insights into actions. In doing so, it addresses one of the major controversies in the learning literature concerning the relationship between learning and adaptation.
Anderson (1991:101) defines adaptation as "... a switch from one behavioral routine in the repertoire of an individual or organization to another routine." Haas (1990:33) goes further to qualify this change of behavior; he views adaptation as "... the ability to change one's behavior so as to meet challenges in the form of new demands without having to evaluate one's entire program and the reasoning on which that program depends." According to
this view, then, if one evaluates just part of the program and the reasoning in which that part is based, it does constitute learning. 16
While scholars such as Hedberg (1981), Haas (1990) and Anderson (1991) maintain the distinction between learning and adaptation, Tetlock (1991) and Levine (1991) contend that the distinction is unimportant. Levine (1991: ), for example, defines learning as 11 ... adap ation to changing circumstances." More cautiously, Tetlock (1991) argues that adaptation can be considered as an outcome of learning. While this study does not equate learning with adaptation as Levine does, it views adaptation as one of the methods organizations use to put their lessons into actions.
In fact, adaptation involves the acquisition of
competency in certain organizational activities. Changes in the environment can force an organization to make simple adjustments to increase competency in its strategies or programs without questioning the underlying assumptions of
"For the purpose of this study, whether the
acquisition of an understanding is a product of the evaluations of ones' whole program and the underlying assumptions, as Haas (1990) maintains, or part of it and that of the reasoning involved, all involve learning. However, as we show in our discussion between learning and policy change (later in this chapter), the first level of learning, or simple learning, in most cases leads to adaptation while the second level of learning, fundamental learning, often leads to basic policy change.
those strategies. Sometimes, when an organization faces a unique situation it adjusts to this specific situation to create a fit between itself and its environment without necessarily evaluating part or the entire program; this is an adaptation.
1t. is also important -at this juncture to explain about unlearning. According to Hedberg (1981:18), unlearning is IV ... a process through which learners discard knowledge. Unlearning makes way for new responses and mental process." Unlearning is a big problem with most international development aid agencies. This is partly because it takes years before these agencies accumulate experiences. Likewise, it requires a highly cumulative convincing counter-evidence to make an organization discard its learned knowledge.
It seems to me, when an organizational world view, or myth (Hedberg, 1981), or theory of action (Argyris and Schon, 1978, 1996) reaches a "dead end", an unlearning process is spontaneous as the organizational efforts are directed toward creating new vision, nqw policies, and new organizational routines. Otherwise the process is slow. I agree with Hedberg (1981:19) that "... fund shortages, falling revenues, actual losses, diminishing popular support, or public criticism" and the failure of an
organization to achieve acceptable results can trigger organizational unlearning. This study, however, goes beyond H-edberg's (1981) contention. It argues that it is the emergence of a problematic situation and/or a new consensual knowledge in an organization that can switch on the unlearning process.
Organizational learning is more controversial than the concept of learning itself. Argyris and Schon (1996) identify three challenges to organizational learning. The first challenge is presented by scholars who maintain that
H..the very idea of organizational learning is
contradictory, paradoxical, or devoid of meaning" (p.xx). The second challenge comes from scholars who accept organizational learning as a meaningful term but who doubt that real world organizations actually engage in it or can be capable of doing so" (p.xx) The third challenge is presented by researchers who "..accept organizational learning as a meaningful notion and agree that organizations do some times learn; what it denies is that organizational learning is always or ever beneficent" l(p.xxi). This study relates primarily to the third challenge. It works under the assumption that organizations do learn but whether what
is learned is positive or not is left as an empirical question.
Another controversy in organizational learning is the unit of analysis. Organizational learning theorists are divided on the unit of analysis for organizational learning. Scholars such as Levy (1994) maintain that organizational learning takes place at the individual level. In other words, organizations learn primarily by institutionalizing individual learning into organizational learning practice. As Argyris and Schon (1978) argue, individuals
... reflect on and inquire into previous context for
learning ... They discover what they did that
facilitated or inhibited learning, they invent new
strategies for learning, they produce these strategies,
and they evaluate and generalize what they have
produced. The results become encoded in the individual
images and maps and are reflected in organizational
practice (as quoted by Levy 1994:286).
However, individual learning by itself cannot cause change in organizational behavior or policy, unless such learning is institutionalized in a way that not only reaches policy makers but also is accepted among the dominant coalition of policy makers. This process requires collaborative inquiry and effort rather than individual searches. It is difficult to envisage a situation where an individual in an organization single-handedly can make an organization learn. This is basically because the logic of an organization's operations works against it. Thus this
study designate the organization as its unit of analysis. This is particularly important in Sida's case because not only do organizations learn as collective entities but like Sida, they work in a disaggregate fashion through committees.
Organizational learning differs from individual learning. They learn by encoding collectively learned inferences from experience and new knowledge into organizational routines. As Argyris and Schon (1996:16) contend:
In order to become organizational, the learning that
results from organizational inquiry must become
embedded in the images of the organization held in its
members' minds and/or in epistemological artifacts (the
maps, memories, and programs) embedded in the
To Ravel (1978:28) an organization learns by ".
imposing upon the structure and process of policy choice a set of decision rules ... that will dispose the system to respond in certain ways--presumably better than before--to future contingencies." According to Hedberg (1981:3) organizational learning takes place when "... organizations interact with their envir-on Iments: organizations increase their understanding of reality by observing the results of their acts. ... by imitating other organizations' behavior, or by accepting others' experiences and maps of the environment." Lovell (1984:135) maintains that
organizational learning takes place when 11 ... policy experiences become assimilated into organizational doctrine, structure, decision-making procedures, personnel systems, and organizational commitments." To us, unless individual inquiry and effort are translated into collaborative inquiry and efforts no organizational learning will take place.
Thus, the work of learning agents (whether an
individual or group) is unfinished business until the result of the inquiry, that is, the discoveries, inventions and evaluations, become encoded in the organizational processes. If they are not encoded, individual members will have learned, but not their organization. In other words, what members have learned remains an unrealized potential for organizational learning (Argyris, 1978).
While Haas (1990) sees learning as a result of
failures, Argyris and Scho.n (1996:16) treat learning as a result of a problematic situation. In their words:
They experience a surprisingly mismatch between
expected and actual results of action and respond to
that mismatch through a process of thought and further
action that leads them to modify their images of
organization or their understandings of organizational phenomena and to restructure their activities so as to
bring the outcomes and expectation into line ...
In this study, the failure of an organization to
achieve acceptable results, the emergence of a problematic
situation, or achievement of success could all result in organizational learning.
Bennett (1990:100) defines organizational learning as "... changes on organizational equivalents of beliefs." To Haas (1990:36), organizational learning is a process whereby IV... an organization is induced to question the basic beliefs underlying the selections of ends.." For the purpose of this study organizational learning is a process whereby collaborative inQuiry by members of an organization (as a result of feedback knowledge or consensual knowledge) generates information or knowledge that significantly affects the organization's understanding of reality.
The common denominator for these definitions is the fact that learning involves information processing and change. Here change is defined in a more generic sense to include change in a degree of confidence in an organization's understanding of reality. In this sense, even learning that may reinforce or consolidate an existing organization's understanding of reality is addressed. This is because even under such circumstances, changes, at least in the confidence level of organizational understanding, will occur. This is more so for development agencies which operate on the condition of uncertainty than for those organizations which operate in a more stable environment.
At this juncture, it is appropriate to note that
neither does all information processing involve learning nor is all change is a product of learning. I agree with Bennett (1990) who maintains that a particular way of information processing that assumes decision makers as naivee scientists" who use their "knowledge structures" and "judgmental heuristics" to infer about the socio-economic and political environment fall under the learning paradigm. According to Bannett (1990), the word "scientist" is used to express the fact that when confronted with a problem, policy makers in most cases will employ the same methods used by trained scientists. Policy makers, like trained scientists, sample data, make generalizations, observe covariation, formulate causal linkages, and use them to predict and test strategies against goals. The word "naive" connotes the fact that judgmental heuristics used by policy makers carry with them certain biases which sometimes work against statistical assumptions (Bennett, 1990).
In processing information-for decision making,
organizations employ different criteria which are couched in different assumptions. An understanding of these criteria or procedures and their underlying assumptions in organization's decision making processlis important because it enables this study not only to explain how the process of
data collection of these models fit in the learning paradigm but also establish which criteria for information processing fit the 1980s Sida's policy decisions as well as the learning model. Scholars of decision making (e.g. Simon, 1957, Lindblom, 1968, Etzioni, 1964, March and Olsen, 1979, among others), have developed a variety of decision making methods over the years. These methods include the comprehensive rational model, incrementalism, bounded rationality, mixed scanning, garbage can and groupthink. Models developed by these decision making scholars focus on two issues--information processing for decision making and the decisional outcome, i.e. incremental or fundamental change. The Former, in addition to the underlying assumption of decision making model is of interest to this study and will be a subject of discussion in the following section.
Learning vis-a-vis Criteria for Information Processing for Decision Making.
The classical model of rationality assumes decision
making occurs under stable conditions where decision makers have perfect information, goals, means are hierarchically arranged and the cost of decisions are under estimated (Frohock, 1979). The criteria for a rational comprehensive model are considered to be ideal, however, and its
unrealistic idealized assumptions, inter alia, unlimited human or organizational capacity, and costs involved in terms of time and attention in gathering all information, have made it less practical (Simon, 1957 and Lindblom 1968). The uncertainty, risks, and information problems associated with the development environment make its utility in the development field highly questionable. In fact, the way Sida policy decisions are made (as shown in Chapter 5) casts doubt on the applicability of this model in policy decisions for Sida.
The incremental model advocated by Lindblom (1968) assumes decision making as a bargaining process among different interested parties. Each interested participant aims at improving his/her resources. Under this model, changes in programs or policies are done incrementally--step by step. Fundamental changes that depart from the status quo are difficult to envisage under incrementalism. This model does not suit this study because the 1980s Sida policy shift was a fundamental departure from its previous aid policies. Besides, the 19'80s policy shifts included not only tangible programs and projects but also abstract goals and policy statements with which incremental solutions have limited capacity to deal (Gortner, et al. 1995).
Groupthink ...refers to a mode of thinking that
people engage in when they are deeply involved in a cohesive in-group, when the members' strive for unanimity override their motivation to realistically appraise alternative courses of action" (Janis, 1972:9). Tn this view, high group cohesiveness, structural faults of the organization, provocative situational context and pressure toward concurrence seeking are the basis for the emergence of group think. This leads to the deterioration of a group's mental efficiency, reality testing, and moral judgement. The consequences are-a disasteYt:
... a distorted view of reality, excessive optimism
producing hasty and reckless policies, and a neglect of
ethical issues. The combination of these deficiencies makes these groups particularly vulnerable to initiate
and sustain projects that turn out to be fiascoes ('T
Janis (1972 and 1982) delineates eight symptoms that manifest the occurrence of groupthink in an organization. They include the illusion of both unanimity and invulnerability, collective rationalizations, direct pressure on dissenters, self appointed mindguards, self censorship, belief of inherent morality of the group, and stereotypes of out groups.According to Elgstrom (1992), the debate which took
place within Sida before the policy shifts in the 1980s was characterized by openness. Different viewpoints were
allowed to compete in that debate. There was no cens worship of deviant arguments and the leadership was more or less impartial. These, together with the mode of decision making employed by Sida, warrant us to conclude that the criteria for a groupthink model was not likely present in the 1980s Sida policy shifts.
The garbage can model contends that the decision making process is ambiguous, unreliable, and anarchic. According to this view, decision making does not fit the assumptions of the more classical decision making models because decisions take place in situation where ... preferences are problematic, technology unclear or participation is fluid (Cohen, March and Olsen, 1988:321). To these scholars, the policy making process serves more as an arena for individual and group expression of conflict, values, myths, friendships, and power, rather than problem solving. Knowledge has little, if any, role in the decision making process. As Cohen, March and Olsen, (1988:296) write:
The mix of garbage in a single can depends on the mix
of cans available, on the labels attached to the
alternative cans, on what garbage is currently being
produced, and on the speed with which garbage is
collected and removed from the scene.
The underlying assumption of this model is that
decision making is a matter of accident or coincidence. As Cohen, et al (1988:297) go on to say, a decision is an
outcome or interpretation of several relatively independent streams within an organization." As will be revealed in Chapter 5, the 1980s Sida policy shift involved the interpretation of not only relatively independent streams within the organization but also other independent streams outside the organization.
The mode of policy decision used by Sida, together with the intensity of the debate about Swedish aid within and outside Sida, and critical aid performance analysis which was conducted by Sida in the early 1980s, make it unlikely that this model could explain Sida's aid policy shift in that period. Moreover, the major role played by knowledge in the 1980s Sida policy change make the garbage can model less useful. Another limitfation of the garbage can model in the context of this study is that it does not provide a useful explanation about the way organizational structures define inter- alia who should participate in the decision making process and what issues should come to the agenda (Masuch and LaPotin, 1989). In the 1980Os policy shifts the Sida leadership played a dominant role in setting parameters for debate and by sending a number of missions to Third World countries to get first hand information and evidence for their policy decisions."7
'7For more detail see Chapter 5.
The satisficing model, like incrementalism, was
developed as a reaction to the limitations of the classical rational model. The model assumes that goals are not hierarchically arranged, information a-Vailable to decision makers is incomplete, and collection of information is costly. In this view, decision makers always seek a-,course of action that is minimally satisfactory. To reach this end, decision makers do not search every possible choice. Bounded by their own intellectual capacity, time and the resources available to them, decision makers select only a sample 'of possible choices--bounded rationality. An alternative that meets minimum criteria or standards of acceptability is chosen (Simon 1959). In other words, the alternative selected is based on "satisficing."
According to this view, organizations develop routines which simplify information search and issue assessments. Decisions are made after inquiring into issues that seem to be the most relevant for a particular choice. For that matter, decision makers do not seek to exhaustively analyze every choice.. March and Simon (1959) argue that, the level of knowledge of policy makers and information available to them are among the key factors that bound a decision maker. In this respect, this model is likely to fit into learning theory.
The satisficing model underlies the way most research is conducted and fits the learning paradigm (Bennett, 1990). When faced with a problem, organizations (particularly development agencies) would in most cases employ a trained scientist or use the same criteria as used by trained scientists. Organizations will define !the problem, collect and analyze data that are relevant to the issue under question, formulate causal linkages and use them to choose an appropriate solution. Organizations, like individual researchers, are bounded by time, resources, and intellectual capacity which make the rationality of their outcome, if at all, bounded. As will be indicated in Chapter 5, the criteria for the satisficing model fits the empirical reality underlying Sida's 1980's policy decisions.
Learning and Policy Change
Learning is by no means the only possible explanation for policy change. Changes in the external environment or political leadership, for example, can lead to policy shifts. More often, organizations change their course of action in an adaptive fashion (Haas, 1990) or through coalition shifts (Anderson, 1991) rather than learning. In this regard, the question of what type of change can be attributed to learning becomes problematic, because a loose conceptualization of learning can encompass any new policy
initiative in an organization while a more restricted conceptualization can exclude policy initiatives which in reality involve learning.
Within this problematique, three-major theoretical
trends that try to explain the relationship between learning and policy change in the learning paradigm can be identified. The first is based on analogical reasoning. It combines the ideas of knowledge structure and judgmental heuristic (Reiter, 1994).*18 The second trend, which is couched in consensual knowledge, emphasizes both the conceptual framework through which the interpretation of experience is done and the political process through which the epistemic communities try to shape the actions and decisions of dominant actors (Levy, 1994). 19 The third trend puts more emphasis on the relationship between means and goals. Feedback knowledge and organizational routines are central variables."0
It is a common phenomenon in the learning literature on organizational theories and international organizations to
"8For the discussion and application of this theoretical trend see Bennett (1990).
'9For discussion and applicability of this theoretical trend see Haas (1990).
20Fo discussion and application of this theoretical trend- see Argyris and'Schon (1978).
assume that "... organizations have a target level of performance or aspiration level to which they compare their actual performance; in each period, they determine whether they have performed above or below this aspiration level" (Lant and Mezias,,1992:48).. Organizational performance is measured by the difference between actual performance and the accepted goals or aspiration level. International development aid organizations, like any other organizations, have their own goals or aspiration levels and means or strategies to realize those goals. These postulations are very important in the organizational learning literature because it provides the cornerstone for the organizational learning theoretical framework. In fact, most literature on organizational learning and international relations distinguishes between two levels of learning. On one hand is learning associated wit:h means; on the other is learning associated with goals. In most cases, learning atzeach level is translated into different policy actions
The first level of learning focuses on the means rather than ends. At this level, for example, members of an organization individually or collaboratively inquire into the underlying assumptions of their tactics and strategies after repeated failures (Argyris and Schon, 1978) or an emergence of a problematic situation. In other words,
participants in an organization come to understand that the existing organizational processes (in terms of organizational framework of action and routines)* do not make it possible for the expected goal(s) to be achieved.Learning which occurs at the level of means is
sometimes called simple learning (Nye, 1987), peripheral learning (Breslauer, 1987), or single-loop learning (Argyris and Schon, 1978 and 1996) At this level, members of the organization collaboratively become the principal agents of learning as they invent new techniques and strategies based on feedback knowledge. The latter is necessary for this level of learning to take place. This level of learning is particularly important in organizations like Sida whose major goal is to support development in an unfamiliar and uncertain environment. Learning at this level of means is likely to lead to simple incremental change or adaptation.
Learning which occur at the second level--goals or
ends--is likely to lead to fundamental policy change. This is particularly true because organizational inquiry is often geared not only toward resolving incompatible organizational norms, goals, and policies by developing new priorities, as well as establishing new weights but also by rethinking the whole process and the reasons therein (Nye, 1987, Breslauer, 1987, and Argyris and Schon, 1978 and 1996). Learning at
for organizational learning is the subject of empirical test.
While Argyris and Schon (1978 and 1996) give examples where organizations learn and change their policies as a result of feedback knowledge, Haas (1990) presents cases where organizations learn and change their policies as a result Pf new consensual knowledge. A more interesting issue is that in all these cases the organization's institutional processes have either facilitated or hampered organizational learning.
This study provides a case where gn organization learns and changes its aid policy as a result of a combination of feedback knowledge and new consensual knowledge in the development field. The above discussion reveals that a successful organizational learning model needs to include organizational institutional process. Thus, this study takes organizational institutional processes as an intervening variable.
Unlike Schechter (1990), who views learning as a "topdown" process, the conception employed in this study views learning as both a "bottom-up" and a op-down" process. Learning which occurs at the first level is essentially a "bottom-up" process while that associated with the second level is primarily a "top-down" process.
This study draws "feedback knowledge" from
organizational theorists, "consensual knowledge" from students of epistemic community, and "institutional processes" from students of institutions to construct a model of learning for Sida in particular and international development agencies in general. These three variables are at the center of-the organizational learning process because they enable us to underscore not only how learning occurs, but also the source of those lessons, and their outcomes. In order to locate the learning model used in this study in its proper perspective each of the three variables is examined in detail below.
In the context of this dissertation, feedback knowledge is viewed as a relatively formal and established fact, referring to insights or conclusions about effects of past decisions which an organization can use to inform its subsequent decisions.' Organizational learning could simply be viewed as a cyclical process whereby feedback knowledge and/or consensual knowledge leads to collaborative inquiry which in turn leads to collective efforts to change the organization's policies, procedures and strategies leading to further feedback.
Although feedback knowledge is a necessary factor for organizational learning, it is not sufficient by itself. For feedback knowledge to play an effective role in the organizational learning process, it requires an open window in the organization's institutional processes, otherwise its impact will be felt slowly. This is so because institutional processes in an organization can either block or facilitate the flow of feedback knowledge.
In addition, institutional processes structure and
institutionalize mechanisms for obtaining and interpreting feedback knowledge (Argyris 1993) In fact, "... Good feedback is unlikely when critical members of the system are motivated to distort or downplay bad news which might be taken as indicative of poor judgement in prior policy selection" (Bobrow and Dryzek, '1987:2O4).
Feedback knowledge is also important because in most cases organizations insulate themselves from the new consensual knowledge unless feedback knowledge has begun to undermine the existing organization's understanding of reality. In other words, consensual knowledge will be more convincing and easily accepted by an organization if feedback knowledge has begun undermining the existing organizational understanding of reality. In this respect,
organizational institutional processes intervene with both feedback and consensual knowledge.
The importance of feedback in the organizational
learning process is also recognized by Hedberg (1981) who argues that "Feedback can also challenge ruling myths in cases. where organizations -expect outcomes which do not occur and in cases where organizations misinterpret outcomes which actually should have confirmed the strategies and myths" (Hedberg, 1981:12).
This dissertation studies the way Sida received feedback information from Tanzania and the way this knowledge was used in the 1980's Sida policy shift. The main concern will be the evaluation process employed by Sida to assess, the performance of its own specific projects in Tanzania and the Tanzanian economy in general, and how the knowledge obtained from this process helped Sida to multilateralize its bilateral relations with Tanzania. Consensual Knowledgfe.
H-aas (1991:65) defines consensual knowledge as:
Generally accepted understandings about cause-effect
linkages about any set of phenomena considered
important by society, provided only that the finality
of the accepted chain of causation is subject to
continuous testing and examination through adversary
procedures. Cause-effect chains are derived from
information, scientific or nonscientific, available
about a given subject and considered authoritative by
interested parties-although the authoritativeness is
According to Haas (1991:66), organizations are exposed to consensual knowledge through the medium of epistemic communities. An epistemic community is ~* network(s) of professionals with recognized expertise and competence in a particular domain and an authoritative claim to policyrelevant knowledge within that domain or issue-area" (Haas 1992:23),. It is P... composed of professionals (usually recruited from several disciplines who share a commitment to a common causal model and a common set of political values" (Haas 1990:41). It can be a group of scientists or academicians or analysts from varieties of disciplines or organizations. The common denominator for members of an epistemic community is
shared causal beliefs, which ... serve as the basis for
elucidating the multiple linkages between possible
policy actions and desired outcomes; ... shared notions
of validity--that is, inter subjective, internationally
defined criteria of weighing and validating knowledge
in the domain of their expertise; and ... a common
policy enterprise--that is, a set of common practices
associated with a set.'of problems (Haas 1992, as quoted
by Brown 1995:54)
The success of an epistemic community in shaping
organizational learning depends on two characteristics .
(1) the claim to truth being advanced must be more persuasive to the dominant political decision makers than other claims, and (2) a successful alliance must be made
with the dominant political coalition" (Haas 1990:42). According to this view, when new consensual knowledge commands an upper hand in the bureaucratic politics in particular and the political process at large it leads to the institutionalization of changes in both goals and causeeffects relationships (H-aas, 1990 and 1991) This institutionalization of change necessarily leads to policy shifts. Therefore, an epistemic community plays-an important role in the occuftence of the second level of learning.
In respect to this study, the 1980s witnessed an increasing acceptance of the IMF and World Bank policy knowledge in the development process. This was in part due to the failure of Keynesian economics and state socialism to stand the test of time. These paradigms claimed that extensive planning under active state management of the economy could lead to rapid economic development in Third World countries. The 1980s worldwide economic crisis, which was more pronounced in countries under state-centric economies, proved otherwise. As a result of this economic crisis, "State worship in its various forms--national integration theory, nationalization, development economics, central planning--suffered a series of hammer blows ... "(Young, 1994:42). The World Bank and the IMF came
up with the Berg Report and structural adjustment programs respectively, as a response to this crisis. Reaganism, Thatcherism, the 1982 socialist retreat of Mitterand, economic reforms in both the Soviet Union and People's Republic of China worked for, and in favor, of the IMF' and World Bank new development policy outlook (Young, 1994).
In fact, these organizations represent a dominant consensual knowledge in the development field since the 1980s. That is why in the model below the consensual knowledge in foreign aid is represented by the IMF' and the World Bank's aid policy outlook. Thus, the IMF' and World Bank policy community represent the dominant international epistemic community in Sida's learning process. This study identifies the lessons Sida drew from the international consensual knowledge and the impact of those lessons on Sida's aid policy shifts in the 1980s. Institutional Processes
Students of institutions have placed emphasis on
different issues in defining institutions. North (1990), for example, puts more emphasis on constraints. He defines
institutions as ". .. humanly devised constraints that shape human interactions" (North, 1990:6). To Haas (1990), institutions can be arenas for innovation as well as constraints. I agree with Commons (1934) who sees
institutions not only as a constraining force but also as a liberating and expansive force (McDermott and Adrew, forthcoming). In fact institutions can "... forbid, permit or require some actions or outcomes" (Ostrom, 1990:5). Ostrom's (1990) definition of institutions captures the three dimensions of institutions--forbid, permit or require-which are important to this study. Thus, Ostrom's
definition of institutions is employed in this study.
Institutional processes, for the purpose of this study, includeorganizational routines (rules, procedures, and programs), leadership, budgetary routines, bureaucratic politics, and organizational doctrine through which conceptions, interpretations, and questioning take place. In terms of structure, for example, decentralized authority, broad discretion, and open interface with other parts of the organization tends to stimulate the learning process. According to Sabatier (1988), a decentralized organization facilitates learning by acting as a catalyst for policy experimentation and realistic points of comparison for evaluating different policy instruments. In the case of
22 She defines institutions as "...a set of working rules that are used to determine who is eligible to make decisions in some arena, what actions are allowed or constrained, what aggregation rules will be used, what procedures must be followed, what information must or must not be provided and what pay-off will e assigned to individuals depends on their actions" (Ostrom, 1990:51).
Sida,.El1gstrom (forthcoming) argues that "..Sida's own structure contributed initially to an acceleration of the growth of Swedish aid to Tanzania and subsequently as a break on policy change."
Doctrine, which according to McDe~nott and Andrew
(forthcoming, p. IV-12) "... consists of a philosophy; what personnel hold to be true; what personnel value; their work norms; and attitudes toward their work, the clients, the program, and institution itself", could either hamper or catalyze an organization's learning process. Organizational doctrine which provides the incentive for risk taking, innovation and discourse, enough latitude or space to minimize fear of failure or error, for example, is conducive for organizational learning because it creates an organizational environment that is inf~rmation-rich (Hitt, 1975, Alexander, 1982).
Sida's development doctrine prior to the 1980s was very much influenced by the ideology of the left (Elgstrom, 1992). Ideology is crucial for this study. It is important because ideology guides actions and defines policies, programs and strategies to pursue. As it embodies values and myths, ideology not only sets the basic principles but also deals with the question of means and ends.
program and budgetary routines are also important
variables of institutional processes. Program involves what is done,where it is done, how it is done, and how successful it is (McDermott and Chris, forthcoming) The last two program attributes should enable us to understand to what extent the program's mode--project or country programming--as well as intervention level--macro or micro or both--facilitated or hindered Sida's learning process.
In terms of budgetary routines as an institutional
process, this study mainly focuses on the criteria used to approve Sida's budget and how the availability of funds impinged on its learning process. Institutional processes are somewhat related to what Goldmann (1988) calls "policy stabilizers.". They have often a tendency of reinforcing previous policy. This may help explain why Sida made its decision at the time it did.
Institutional processes are also important because they define responses open to an organization in respect to its performance. An organization can deny error, externalize it, or embrace it (Korten, 1980). According to Korten, individuals have inclinations toward one or the other. However, an organization has a tendency to institutionalize one among the three. In those organizations which treat errors as indications of personal incompetence, the
personnel will develop excellent skills for hiding errors. This will be the dominant norm in the "self-deceiving" type of organization.
."Self-defeating"-organizations are those that
externalize their errors. They tend to speak openly and in detail about their organization's errors, showing the impossibilities or difficulties involved, in accomplishing their goals. Learning organizations, on the other hand, embrace error. Members of this type of organization take error as a source of data for learning and making the necessary adjustments. They discuss their errors with the view of improving their performance by seeking solutions to those errors. Thus institutional processes will enable this study to situate Sida in terms of its capacity to learn.
Institutional processes are important variables for
successful learning because "The kind of knowledge, skills, and learning that members of an organization will acquire reflect the payoff--the incentive--embqdded in the institutional constraints" (North, 1990:74). Therefore, understanding Sida's institutional processes will reveal the role these processes played in Sida's learning process. The foregoing discussion leads us to Sida's learning model, as illustrated in the chart below.
Figure 2.1 Sida's Learning Model
SWEDISH ECONOMIC & POLITICAL SCENES
FEEDBACK KNOWLEDGE (Performance of aid in Tanzania)
CHANGE OF NORMS, PRIORITIES AMONG OBJECTIVES AND IDEAOLOGY
SIDA'S CHANGES IN THE
INSTITUTIONAL NATURE AND
PROCESSES VOLUME OF
SHIFTS IN SIDA'S TECHNICAL COMPETENCE AND ORGANIZATIONAL ROUTINES
CONSENSUAL KNOWLEDGE (IMF/World Bank policy outlook)
Independent variables = Feedback knowledge & consensual knowledge Intervening variable = Sida's institutional processes Dependent Variable Multilateralization
The model above represents a learning model for Sida.
In the context of this study feedback knowledge is seen as a relatively formal and established fact, referring to insights or conclusions about effects of past programs and projects performance in Tanzania which Sida used to inform its 1980s policy decisions. Consensual knowledge is represented by the IMF and .World Bank macro-economic reforms which became the dominant consensual knowledge for Third World development in the mid 1980s. Sida's institutional processes include organizational routines (rules, procedures, and programs), leadership, budgetary routines, bureaucratic politics, and ideology through which conceptions, interpretations, and questioning take place. We assume that the combination of the above factors and pees led to multilateralization which manifested itself in change of priorities among Sida's aid objectives and ideology, changes in the nature and volume of its aid, and shifts in Sida's technical competence and organizational routines.
METHODOLOGY AND DATA COLLECTION Case Studies
This chapter deals with the process and techniques used to collect and analyze data for this study. It shows the strengths and weaknesses of the case study design, and justifies the use of this-design. It also explains why interviews, focus groups, and document analysis were chosen as techniques for data collection. This chapter explains not only the usefulness and shortcomings of each of the above techniques for data collection, but also how combining these three techniques helped reduce the problems associated with each one. The process of data collection, including problems encountered in the field, are also explored.
According to Lijphart (1977) and Yin (1994), it is important for researchers to understand types of case, studies and their potential contribution to theory building before they choose a case study design. While Lijphart (1977) identifies six types of case studies,--atheoretical, interpretive, hypothesis-generating, theory confirming, theory infirming, and deviant-- Yin (1994) identifies three78
-exploratory, descriptive, and explanatory. It is difficult to fit a research effort into a single type of case study because most research efforts tend to overlap. For example, an atheoretical case study and a hypothesis-generating case study also can fall within either Yin's exploratory or descriptive case studies. Similarly, a case study can be both exploratory and descriptive. For this reason, I agree with Lijphart's (1977) contention that these categories represent "... ideal types, and any particular study of a single case can fit more than one ..." (Lijphart, 1977:63).
An atheoretical case study, according to Lijphart (1977), is a single country or case analysis. It is descriptive and not guided by a formal theory. It is useful for basic data-gathering operations which can-indirectly contribute to generalization. In conducting this type of case study, a researcher is mainly motivated by an interest in that case situation rather than by theory construction.
Interpretive case study, like atheoretical case study, is also selected due to the researcher's interest in a particular case rather than theory-building. However, it differs from the former in the sense that the latter explicitly uses established theoretical propositions. Generalizations, if at all, are applied only to that specific case. In fact, the major objective of this kind of
study is to throw light on the specific case rather than to make generalizations. Interpretive caqe studies are studies in "...'applied science'. Since they do not aim to contribute empirical generalizations ... "(Lijphart, 1977:64).
The hypothesis-generating case study, theory
confirming\infirming case studies, and deviant case analysis, are all selected for the purpose of theory building. Hypothesis-generating case studies, a kind of inductive theory generation, are useful for building theories in a virgin area. In applying hypothesisgenerating case study a researcher is likely to start with a more or less vague hypothesis and work on developing hypotheses to be tested in a large number of cases. Hypothesis-generating case studies have significant theoretical value.
Theory confirming and theory infirming case studies
involve the analysis of a single case within the parameters of established theories. Theory confirming and theory infirming fall under the dqductive method. Hypotheses are drawn from existing theories and tested on that particular case. It can be limited to a single variable or a number of variables that relate to existing generalizations. If it confirms, it strengthens the proposition and if it infirms,
it weakens the generalization. The impact, however, is marginal.
Deviant case analyses "..are studies of single cases that are known to deviate from the established generalizations" (Lijphart 1977:64). These types of cases are chosen to reveal the reasons behind that divergence. These studies can have great theoretical value, particularly when used to reveal important variables overlooked in past studies, as well as reoperationalizing some or all the variables used in previous studies. They .can weaken the original proposition or suggest a modified proposition that may be stronger than the original one. However, for the new proposition to be valid, it must stand a test of more cases. A deviant, case study, according to Lijphart, may be likened to an experimental group with the more or less similar cases constituting the control group. This can only be successful if there is a clear definition of the deviant case in terms of its position with regard to both the variables under consideration and the more or less similar cases.
Thus, hypothesis-generating and deviant case studies have the greatest value irn theory building. While the former serves to generate new hypotheses, the latter refines and sharpens existing hypotheses (Lijphart, 1977). This study falls within the theory confirming/infirming category
as it builds its hypotheses from existing theories of learning.
Justification for the Case Study Method
This study uses a case study design because of its
utility in explaining a complex phenomenon. This is in line with Lijphart's (1971), Feagin, et al's (1991), and Yin's (1994) assertion that a case study design is more appropriate when a researcher seeks to explain a complex course of events as it avails the researcher with a detailed knowledge of the process. According to Orum, et al (1991), one of the major advantages of case study design is that it gives-a holistic analysis of complexes of action and meaning. They argue that investigators who use the case study approach have the advantage of a close view of their subjects. In such situations it is easier to get a sense of what drives their actions and decisions. They write: "A good case study can provide a full sense of actor's motives that eventuate in specific decisions and events" (p.10).
The complexity of Sida's policy shifts in the 1980s has been appreciated by most scholars who have dealt with that process (see Fruhling, 1986, Karre and Svensson, 1989, Elgstrom, 1992, Larsson, 1994, and Hyden and Mukandala, forthcoming). To be able to understand and explain this complexity, an in-depth analysis of processes which led to
these policy shifts is necessary. In this respect, case
study analysis is an appropriate design for this. study.
Moreover, case study is an important design,, not only for the development and evaluation of public policies, but also for testing theories (Lijphart, 1971 and Yin, 1994). According to Johnson and Joslyn (1991:121), the ". strongest case studies start out with clearly identified theories that are expected to explain events. Case studies are particularly useful for testing hypotheses deduced from existing theories ....." As the previous chapter attests, hypotheses which guide this study are based on consensual
knowledge of the epistemic community theorists, feedback knowledge of the organizational theorists and institutional processes of the students of institutions. In this respect, this study fits Johnson and Joslyn's (1991) contention as it attempts to explain Sida's 1980s policy shifts by testing hypotheses extrapolated from existing ]earning theories.
The importance of case study design for this
dissertation is further increased by the qualitative and historical approach utilized in this design. Its sensitivity to time and history makes case study a valuable design. This is partly because case studies allow the
researcher to 'examine patterns of continuity and' change and uncover. the historical dimension of the phenomenon or,
setting under study. By tracing the development of Sida and its policy process over time, this study has been able to generate insights which otherwise would have been impossible. These insights, as will be revealed in later chapters, have enriched our understanding of why and how Sida, having drawn lessons from its own experience and the experience of others, changed its aid provision policy in the 1980s. Conclusions drawn in this 9tudy have also gained more weight from those insights. Limitations of the Case Study Method
Despite its merits, case study design has been
conceived as inferior in the social sciences (Lijphart, 1977 and Yin, 1994). In fact, researchers who have used this method have been accused of conducting investigations with
in-sufficient precision (that is quantification), objectivity, and rigor" (Yin, 1994:xiii).
Case study design always comes face to face with the issue.of generalization. This problem,, however, can be reduced if we take case study to be a process of adding knowledge to existing knowledge in the field under study. As Yin (1994:10) writes:
Case studies, like experiments, are generalizable to
theoretical propositions and not to populations or
universes. In this sense, the case study like
experiment, does not represent a "sample" and the
investigators goal is to expand and generalize theories
(analytical generalization) and not numerate
frequencies (statistical generalization)."
On the issue of objectivity, more often than not, case study researchers have been sloppy and their findings and conclusions have been based on biased equivocal evidence (Yin, 1994). It is not the case design which inherently
allows biased- views to influence researchers I findings and conclusions but rather the way in which investigators execute the method. Biased views are not limited to case study alone. They can enter any research design. What is important, however, is the need for researchers who use case studies as well as other designs to report all evidence fairly and objectively (Yin 1994).
A third problem associated with case studies, according
to Yin (1994: 10) is that they take too long, and they result in massive, unreadable documents. I agree with Yin (1994) that people who accuse case study research of this
flaw, confuse specific data collection techniques with case study design. Data collection techniques such as ethnography or participant observation naturally take
23 According to Yin (1994:30), in statistical
generalization, an inference is made about a population (or
universe) on the basis of empirical data collected about a sample. While analytical generalization is a method in IV ... which a previously developed theory is used as a template with which to compare the empirical results of the case study. If two or more cases are shown to support the same theory, replication may be reclaimed."
considerably longer and call for more Aetailed accounts of observations than other techniques. Case study design does not necessarily depend upon ethnography or participant observation data collection techniques. This study, for example, took only four months and it employs interviews, focus groups and document analysis.
Reliability Versus Validity
Two questions which are pivotal in all social science
research (case study inclusive) are questions of reliability and validity. Reliability, according to Carmines and Zeller (1979), deals with:
the extent to which an experiment, test, or any
measuring procedure yields the same results on repeated
trials .... The most consistent the results given
repeated measurements, the higher the reliability of
the measuring procedure; conversely, the less
consistent the results, the lower the reliability" (as
quoted by Johnson and Joslyn, 1991:68).
According to Johnson and Joslyn (1991:70 & 76) the
reliability of the measurements used by political scientists is frequently suspect and seldom demonstrated. Most measures of political phenomena... are partially accurate." Despite this acknowledgment, the reliability of case study design is more questionable than other designs. This is partly because case study depends on a small sample. The likelihood of getting the same results by replicating the
study in a different setting is very small. Recognition of this fact by case study researchers is very important for avoiding becoming a victim of "ecological fallacy" (Bernard, 1994).2
Validity, according to Bernard (1994:38), is the accuracy and trustworthiness of instruments, data and findings in research." In terms of validity, case study design is superior to other methods of investigations such as experimental method and-surveys (Orun, et al, 1991). This is so particularly because this approach uses a number of data collection techniques to prove the same case. This study, for example, uses interviews, focus groups, and document analysis as its data gathering methods. It is this "triangulation of sources", if I may use Orum, et al's words, which improves the validity in the case study design.
Having chosen a case study method, a crucial issue for the success of this study remains the elimination of competing explanations. With regard to the study of Sida, two approaches--the, contextual approach and bureaucratic
"4Ecological fallacy is the result of making conclusion about the wrong unit of analysis. In most cases, it occurs by drawing generalizations about people from data about groups (Bernard, 1994). It is important for researchers to ensure that the unit of analysis of the measures and the hypothesis are matched (Johnson and Joslyn, 1991).