You are on page 1of 74

FROM PLANNING TO EVALUATION: A systems approach to agricultural development projects by Rolando Garcia Document of The International Fund for

Agricultural Development Monitoring and Evaluation Division Economic and Planning Department REPORT No. 0431 May 1993 The views and interpretation herein are those of the author and should not be attributed to IFAD. FOREWORD Professor Rolando Garcia has an unusual curriculum. Argentine mathematician and physicist, he was from 1947 to 1966 Dean of the Faculty of Exact and Natural Sciences at the University of Buenos Aires. Meteorologist, he designed and directed the Global Atmospheric Research Programme, a joint programme of the World Meteorological Organization (WMO) and of the International Council of Scientific Unions. Epistemologist, he worked closely with Jean Piaget and wrote with him several books.' He is since 1984 Professor of Methodology and Theory of Science at the Centre for Research and Advanced Study, National Polytechnic Institute, Mexico. Professor Garcia brought his powerful analytical skills to bear on a system approach to development problems in the context of the "Drought and Man" project of the International Federation of Institutes for Advanced Study, project taking the 1972 drought in different parts of the world as a central theme of interdisciplinary investigation. He directed this project out of which three books were published by Pergamon Press, Oxford.2 As he was developing the analytical framework of the Drought and Man Project, I was, at the same period, that is from 1976 onward, developing a research project for the United Nations Research Institute for Social Development (Geneva) taking also a systemic approach to development issues, the Food Systems and Society programme. He, then, gave his own methodological contribution to this endeavour and applied it, in particular, to the Bajio region of Mexico.' After a few years at IFAD, I was increasingly being struck by the general linear thinking implicit in project documents. For example, an improved technology will improve yields, which in turn will improve incomes, which... I then thought that Professor Garcia could help us to develop a more systemic approach to development projects, the objective of such an approach being not to make projects more complex but, on the contrary, to identify the few critical linkages on which to act efficiently. Hence this contribution presented in the EM Special Studies series, an outcome of his interactions with IFAD's Monitoring and Evaluation staff. It comes at an opportune time as discussions are presently taking place within IFAD on ways and means to reduce projects' and programmes' complexity and to improve the definition of objectives and sub-objectives as well as on the potential offered by a dynamic logical framework approach. Pierre Spitz Director Monitoring and Evaluation Division 1 "Les relations causales", P.U.F., Paris 1970, "Psychogen6se et histoire des sciences", Flammarion, Paris, also available in Spanish, Siglo XXI, Mexico, 1984. Z Volume 1 "Nature Pleads not Guilty" (1981) ; volume 2 (with Jose C. Escudero) "The Constant Catastrophe: Malnutrition, Famines and Drought" (1987); volume 3 (with Pierre Spitz) "The Roots of Catastrophe" (1986). 3 "Food Systems and Society: A Methodological and Conceptual Challenge". UNRISD, Geneva, 1984. Also available in French.

Contents Page I. Introduction II. Agrarian Systems as complex systems 11 III. The design of agricultural development projects from a systemic perspective 20 IV. Diagnosis 24 Project design 35 VI. Implementation, monitoring and evaluation 41 VII. Integrated programs and sets of projects 53 Annex 1: Some remarks on data bases and statistical sampling in the context of complex systems analysis 71

I. Introduction This document arises from the need to respond to the 'concrete problems encountered in the day-to-day practice of that design, execution and evaluation of development programs for agricultural regions. The proposal made here is strongly influenced by the experience gained through case studies carried out in very different countries over many years and where the obstacles we encountered, our repeated mistakes and demonstrated failures count for as much as the instances of success. The preceding comments are intended as a precautionary measure aimed at countering possible reactions to the initial part of the proposal which may appear the fruit of an "academic environment", far from the real problems that technical assistance experts face. The preliminary remarks which follow are intended as a rationale for the particular approach adopted to the theme concerning us here. excessively theoretical The concept of an agrarian complex Development projects in agricultural regions generally refer to situations generated by a multiplicity of problem involving the physical and biological environment, production,

technology, society, the economy. Such situations are characterized by the confluence of a variety of processes whose interrelations constitute a complex which functions as an organized totality. "Complexity" here is determined,- not only by the heterogeneity of the constituent parts, whose nature usually places them within the domain of several quite different fields of science and technology, but rather by the interdefinability and mutual dependence of the functions they fulfill within the totality. An agrarian complex seldom has well established geographical limits, nor does it have a well defined number of components or elements with observableR characteristics that can b6 recorded and classified in a unequivocal, unambiguous form. What characterizes such a complex is a particular behavior i.e. a 'certain number of activities that together determine what we shall call the functioning of the complex. The activities of the complex cover a wide range and include the production of particular crops, the importation of inputs for production, the consumption of water and nutrients from the soil, the work of the peasants, migrations, economic inputs (credits), trade, and so on. All these activities are the result of interrelated processes. The fact that there is an ensemble of interrelations among the processes that take place in a complex allows us to refer to it as an organized totality. : When we study one of these complex situations, we cannot analyze "all" its elements, not just because of a material impossibility (what does all mean?), but also for practical reasons. We do not examine every square meter of soil, nor the total depth of the subsoil, nor each individual member of the population, nor every single activity. It is therefore inevitable that each study consists of establishing relationships between a limited number of elements abstracted from reality, and all abstraction implies taking into account some aspects of experience and not others. But when an element is abstracted and others put" interpretation of the element has been made. An example- may clarify this point: When we go to an agrarian community, we don't "see" peasants. The "data" we have in front of us are individuals with certain personal characteristics. The fact that they are "peasants" is the result of a conceptual elaboration concerning the productive activities of these individuals and of their relationship with production within the society to which they belong. Of course, in practice, `Acquired experience such a place, we immediately identify the individuals we see peasants. The individual is a datum of experience. The peasant is an observable in that he is an interpretation

(conceptualization) of the datum. In the same way, a sociologist can enter a town and say that he "sees" different "social classes". In actual fact what he sees (the empirical datum) - to give a crude example - are barefoot people dressed in rags, others driving small modest vehicles, and yet others in luxurious chauffeur- driven cars, etc., etc. On the basis of these data and by means of a lengthy conceptualization, the sociologist constructs the social classes as observables. _ The problem becomes more complicated when we pass from the identification of the elements ("observables", the product of conceptualization) abstracted from the agrarian complex, to the attempt to comprehend the processes that take place there. A process is a change or a series of changes that constitute the course of action of causal relationships among events (natural or man-made). They are not empirically given "data", nor are they "observables" constructed as the interpretation of data. They are relationships established on the basis of inferences. The role played by such inferences in the study of a complex makes it a central theme for the conceptual and methodological framework we are proposing in this paper. In this connection, it is important to establish the following distinction: - Some of the inferences may consist of an application of inductive generalization, from previous experience in similar

situations encountered in earlier projects or, from professional know-how. However, the inclusion of a specific situation as a particular case within a general kind of situation is a step to be taken with care. Seemingly "similar processes" that take place in different contexts may not perform similar functions. What the similarity consists of is in dealing with complex situations. We shall return to this subject in Chapter VII. - More often than not, processes taking place in complex are identified by establishing links between separate events. Such links are not "observed". They are inferred through logical deduction taking as a starting point certain premises which - as indicated above are provided by the conceptual framework of the researcher. In other terms, the set of causal relations among events in a complex, established by neans of this type of inference, are constructions where the conceptualizations of the researcher intervene as much as the "objective facts". The construction of a system The distinctions we have described between data, observables and processes lead us to pose the more general problem which is at the root of all methodologies for empirical work: the relationship between experience data and the conceptualizations made by the researcher. With the expression "agrarian complex" we have designated

a set of elements interrelated through a series of activities centered on agricultural production. With this formulation, the agrarian complex is referred to a concrete empirical reality. However, as we have pointed out above, the elements" that are "taken out" of this empirical complex and are used by the researcher are abstractions, that is, conceptualizations of the empirical data. In the same way, the causal relationshiops and the processes that intervene in the analyses are inferences made on the basis of events that are found to occur empirically. The researcher constructs a system with these conceptualized elements and these inferred relationships and processes . Strictly speaking, 'a system is thus a construction produced by the researcher in order to represent the most significant activities and actors in a given empirical complex. Nevertheless, in practice, there is a tendency to talk in an ambiguous fashion, using the tern "system" to refer to both the empirical complex and the constructed system. In the same way, we will use the expression "complex system" to refer to one or other case. When it is necessary to establish the distinction, we will use the terns E-System and C-System (E and C standing for empirical and constructed respectively). To construct" system (C-system) thus means to select the elements that will be abstracted from the empirical complex (E-System) and to identify a hertain number of relationships between this set of elements. The set of

relationships constitutes the structure of the system. It should be noted that the same elements can be used to define different systems, that is, C-systems whose structures differ in that different sets of relationships between the elements have been taken into account. The selection will depend on the objectives of the research and will be specific questions that are formulated regarding the type of situations that are to be studied. This exceptionally important point and we must look more closely at its implications. . All scientific theories, whatever their degree of formalization or explanatory level have developed historically as an attempt to explain a certain domain of phenomena and as a response to specific questions regarding these phenomena.* Scientific revolutions and the new theories which emerge from not pllc)duced so much by those who found new answers to old questions but by those who could formulate new questions about old problems (and obviously also, for the new ones). We gave the name epistemic framework" to the set of questions (or queries not always formulated as precise questions) that the researcher posed when faced with a domain of reality which he proposed studying. The epistenic framework represents a certain conception of the world and on many occasions, it expresses - albeit in vague and implicit fashion -- the `"table of values" of the researcher.

A concrete example may clarify one of the meanings of this conception. A research project designed to reply to the question "How can we increase the productivity of basic food crops in region X?" will be very different to one designed to respond to the question Why does malnutrition among the poor increase in that country?". Problems concerned with productivity and food self-sufficiency are posed in both cases, but from different perspectives. In the second case, differences in research design are also possible since they depend on the conceptions of the researcher. If he is convinced that malnutrition is a consequence of disequilibrium between production and demographic growth, he will look - and will no doubt find - information which, when taken on its own, leads him to confirm his "theory" . The research design will be very different for a researcher who conceives of malnutrition as the result of structural problems that comprehend both production and a complex set of social, economic,and political relationships. The "observables" registered as well as the chain of inferences linking them will be very different in both cases. The working hypotheses with which a systemic approach to a project begins - from the perspective offered by the proposal contained in this document - may thus be summarized in the following proposition: given a specific agrarian complex and a set of questions referred to it, a system can be defined in terms of a set of elements and the interaction, among them, and such that its structure be that which corresponds to the functioning of the complex, thus providing the basis for an explanation that answers the initial questions. II - AGRARIAN SYSTEMS AS COMPLEX SYSTEMS The objectives of the proposal contained in this document- with the background sketched out in the Introduction can be formulated as consisting of a working methodology based on a systemic analysis. Despite the apparent clarity of such a synthetic statement, confusion will inevitably ensue since it contains two highly equivocal expressions: "methodology" and "systemic analysis". If by methodology we understand a series of procedures which allow a given task to be carried out successfully, our proposal goes far beyond this definition. While we do expect that a working methodology in this restricted sense will emerge, it will be built upon a theoretical basis constituted by a theory of complex systems, together with a conceptual framework which orientates methodological issues. On the other hand, the expressions "complex system" and "complex system analysis", that we are going to use, differ conceptually and methodologically from close and distant relatives with similar names which are in frequent use in engineering and economics. There is a variety of versions of system" and quite different ways of approaching "system analysis". They go from the input-output models of engineers and econometrists, to the theory of dissipative systems and the dynamic equations of its model builders. But for all that, there is no consensus on how to tackle problems of the interactions between environment and society from a systemic standpoint. This may account for the fragmentary character of the results that have been obtained and justifies further efforts in this kind of studies. Our present proposal is intended as a contribution in this direction. However, the experience already accumulated in a number of "case studies"' is sufficiently indicative of the potential of this approach. The necessity of employing a theory of complex systems to deal with the problem area this document is concerned with is based on the following considerations. Agricultural production is the nexus of the articulation between the physical environment, which provides the nutrients for the development of biomass, and society, which provides the means of production and depends directly or indirectly on what is produced. In this articulation - in which production technology (in the wider sense of the term) plays a predominant role - physical environment/society interactions take place, generating processes of deterioration upon which technical assistance projects seek to act.

The accumulated experience shows that the growing degradation of the ecosystems, as well as the deterioration of living conditions and the malnutrition which affects hundreds of millions of human beings are closely related to each other, and that they cannot be explained with reference to simple linear causal chains of events. They are the result of the structural problems of complex systems made up, in broad terms, of a physical subsystem (soil, water, climate, natural ecology), a productive subsystem (crops, methods of production, technological inputs) and a socio-economic subsystem (social groups, economic conditioning factors, political factors). The heterogeneity of the phenomena and the components of such systems are not enough - as we have said in the Introduction - to define it as "complex" (although a very heterogeneous system may be thought of as very "complicated"). Two conditions give it the character of complex: - the functions of the elements (subsystems) of the system are not independent; this determines the interdefinability of these components; - the system, as a totality, is open, that is, it lacks rigid boundaries; on the contrary, it is immersed in a wider reality with which it interacts by means of flows of matter, energy, economic resources, regional or national policies, etc. These two conditions are an important part of the definition of an agrarian system and have important implications for the study of such a system. Interdefinability and interdependence of the subsystems. The organs of a living being have a certain structure that allows them to fulfill a specific function within the total system" that is the organism. However, the same function can be fulfilled by organs of different structures. A fish and a monkey both have respiratory apparatus. Their functions are the same: to extract oxygen from the surrounding medium (water or air respectively). But the structure of the respiratory apparatus of each one is different: the monkey has nothing like the gills of the fish. The function is always linked to a particular activity that, together with other functions of other organs (subsystems), takes part in the general activity of the total system. The latter has been referred to in the text as "the functioning" of the system. In a human being, functioning is multiple and very complex, covering a very varied range. It is clear that the functioning of an organism is not a simple sum of the functions of its organs taken in an isolated fashion, nor is there action in one direction that gives linear relations of cause to effect between the functions of organs and the functioning of the organism. On the contrary, not only are the organs interdependent in their functions, but also the total functioning can include, stimulate or modify any of the particular functions (and through it, the others). When applied to our present concerns, the above concepts allow us to visualize-each "case study" as an analysis of the functioning of a complex system whose parts (subsystems) carry out certain functions thanks to the fact that they possess a certain type of organization (that is, the structure of the subsystem which is a substructure of the total structure). This organization is not unique as we said previously, the same function can be fulfilled by different structures. Neither is that function independent of the functions carried out by other parts of the system. Our frequent reference to structures and to structural properties may require some clarification (in particular, to divorce ourselves from the still persistent controversy on structuralism among social scientists). Structural analysis is a methodological tool based on certain conceptualizations of the system being studied. It should be taken as a guideline in the search for significant relationships among the elements of the system as well as an aid in the way data is processed. The structure is simply what characterizes the system as an organized whole. The objective of structural analysis is therefore to account for the links and interactions among the set of elements that make up the system.

The are many kinds of links depending on the nature of the system and the "moment" of its evolution. Some may constitute strong ties between elements, others only weak relationships; some may relate elements that are mutually dependent and that jointly perform a given function; others may only be indicative of the coexistence of functions that are independent of each other, but which contribute to the functioning of the system as an interconnected whole. It is clear that the set of relations of each element (or subsystem) with the others depends on its own internal organization which, in turn, is the result of the evolution of internal processes over time. Thus, a careful distinction must be made here between two levels of description: a) the analysis of the processes that go on within each element (subsystem) and determine the kind of relations that it establishes with the rest of the system, and b) the analysis of the processes that go on in the system as a whole which are determined by the interrelations among the subsystems. The fact that the processes at levels a) and b) are interdependent does not mean that everything should be put together in the same bag and be studied simultaneously. Quite the contrary, in so far as we can identify a subsystem within a system means that its constituent parts have sufficient coherence, at least during a given period of time. This means that we have considered the ties within the subsystem as being of different nature than that of the relations it has with the other subsystems. In this connection, the terms "interdependency", "interrelations" and "interactions" may be very misleading. For instance, in the interactions between subsystems, action and reaction may be of quite different kinds and may be acting on quite different time-scales. The evolution of agrarian systems as open systems The evolution of an open system does not constitute a continuous or linear process. It proceeds by a succession of reorganizations, each one of which leads to period of relative dynamic equilibrium during which the system maintains a structure which fluctuates within certain limits. The fact that this type of evolution has been proven without doubt in laboratory experiences with physical-chemical and biological systems and has been demonstrated mathematically in numerous cases, has led many social scientists to reject the idea that the theory is applicable in their domain. The terminology in which it is expressed does tend to have something of a physicalist "flavor" which suggests reductionism. However, it tends to be forgotten (or not known in the first place) that Marx , in political economy (the evolution of modes of production), and Piaget, in psychology (the stages" in the psychogenetic development of the concepts employed by the child and the adolescent), discovered the discontinuous and non-linear evolution of the respective systems they studied long before the current theories which originate about 40 years ago. This notwithstanding, the degree of generality of these theories has only been recognized very recently. Their relevance and importance for the study of the systems we are considering may be briefly summarized by the following remarks. The variables on the basis of which we define the system with its characteristic structure during a given period of time, do not have static values - they permanently fluctuate in the same way as the interactions of the system fluctuate with what we have left "outside its boundaries". These fluctuations can be of two types: (a) small scale fluctuations that induce small changes which do not alter the fundamental relationships that define the structure of the system; (b) larger fluctuations which, when they exceed a certain threshold, produce a disruption in the structure. In the second case, the disruption of the structure is the consequence of its instability when faced with that particular type of fluctuation. Stability and instability are thus properties of the structure of the system, but are relative to the type of fluctuation or perturbation that it may suffer. Other concepts such as "vulnerability", "resilience" and

"elasticity" can be defined in terms of stabilities and are thus structural properties of the system. A clear and important application of these concepts is found, for example, in studies carried out in Africa'. Here, the study of prolonged droughts showed that climatic anomalies of SC similar- magnitude produced different "impacts" in 1 See the series on the IFIAS Program "Drought and Man: The 1972 case history", Vols. I and II.

19 different periods. The catastrophic consequences of some of these were not due to the greater intensity of the natural phenomenon causing the disequilibrium, but to the greater vulnerability of the system. In very fragile productive and socioeconomic systems, even a weak "perturbation'" can have disastrous consequences.

20 III - THE DESIGN OF AGRICULTURAL DEVELOPMENT PROJECTS FROM A SYSTEMIC PERSPECTIVE Governments, international organizations agencies of all sorts, that are called upon to provide solutions for critical conditions anywhere in the world, are generally urged to "act" as soon as the means are available. A medical doctor very often finds himself in similar situations. There are cases where the symptoms are so clear that the kind of medication needed may not be in much doubt. In other cases, even if the situation appears prima-facie too complex to come up with an immediate diagnosis, some emergency measures have to be taken (pain relief, for instance). However, more often than not, a good doctor will refrain from prescribing any treatment before having enough elements in hand for a sound diagnosis. This does not exclude acting on the basis of some preliminary findings that require urgent attention, before the complete etiology of the situation is known. Once a satisfactory diagnosis is obtained, the treatment may proceed on safer grounds. "Satisfactory" is not synonymous with free of uncertainties", nor does it mean that further analysis may not become necessary in the course of the treatment. The way of acting in medicine is applicable mutatis-mutandis to what seems to be

a sound approach to the problems of development in the agrarian domain. In both cases, diagnosis, action (treatment) and evaluation need to be in frequent interaction to guarantee success. It is clear that the mode of operation in international organizations precludes the direct application of ideal schemes. The task of the planner in this connection is to find the ways and means to get as close as possible to them. There are, however, some obstacles that need to be avoided or overcome. - First, as a general rule, the three phases referred to above - diagnosis, action and evaluation - are carried out by different teams that seldom have the possibility (let alone the desire!) to interact with each other. - Secondly, the project designers usually confine their attention to the actions they are going to propose, taking for granted the reliability and the sufficiency of the information they have gathered, which means that there is no real diagnostic phase. - Thirdly, both the plan of action and the evaluation(s) are often focussed on the attainment of specific objectives that are analyzed in isolation (i.e. without paying attention to the implications for other sectors of the system). In this context, diagnostic analysis of the problems involved in the agrarian system of a given region or community, the planning and implementation of necessary

22 reforms to modify existing conditions and the forecasting of the evolution of the system appear to have quite different objectives and require different methodologies. This is in fact so in most of current practice in connection with the design and execution of development projects. From the perspective offered by the complex system analysis of projects on agricultural development, such a separation does not hold. Let us summarize the arguments. - The diagnostic studies are centered on the identification of processes and mechanisms that are integrated in a chain of events that occur over a period of time. The diagnosis requires a reconstruction of the history of the system because what happens today may be best (and sometimes only) interpreted in terms of this history. - The study of a new proposal for the region, on the other hand, will have a prospective character. It is centered on the predictability of the evolution of a new system - which is a modification of the current one - that would result from the implementation of the proposed measures. This requires the identification (foreseeing) of processes that will be set in potion when the changes are introduced. And this can only be done on the basis of the kind of diagnosis indicated above. - The passage from a diagnostic study to a study of alternative policy for a region is not linear. New productive strategies, for example, can have an incidence on parts of the system that were not (or not sufficiently) analyzed in the

23 diagnosis, as they did not have an important role in the processes in action in the contemporary productive system. This leads us to the need to repeatedly return to the diagnosis in the analysis of each proposal for change, in order to study aspects not previously included. In brief, diagnosis, planning for change, forecasting possible implications for the system, checking the results of the implementation of plans, are all parts of a dialectical interplay or activities. This notwithstanding, each one of the phases involve a number of specific problems, and we need now to turn our attention to them.

24 IV. DIAGNOSIS The decision to undertake the study of a given agrarian complex usually arises from the recognition of situations or phenomena that take place in that particular geographical location and have generated (or are generating) processes of deterioration in the physical environment that is the support of production and/or the living conditions of the population involved in production. No research project begins from zero. The task begins with the identification of the geographical area where such processes are developing and which present the specific problems that, in a first approximation, define the "problematique to be studied. In general, there is sufficient knowledge of these phenomena or situations so as to formulate general questions as a starting point for the research. The long road that will lead to the definition of the "system to be studied will commence on this basis. When a researcher or a team of researchers have need to carry out such a study, the "system" never presents itself as given, awaiting observation and analysis. What is given" to the researchers is an agrarian complex. A fundamental part of the research effort is the "construction (conceptualization) of the system to be studied (C-system) which, as pointed out

25 in the Introduction, constitutes a more or less arbitrarily cut out piece of reality and does not appear with precise limits or definitions. The "construction" of the system is no more than the construction of successive models that represent the reality to be studied. It is a laborious process of successive approximations. The test as to whether a satisfactory step in the definition of the system (as a model of the reality - agrarian complex- being studied) has been taken can only be based on its capability to explain a functioning that fits the observed facts. To this end, it is not sufficient to have a model representing a clear description of the system at the moment or the period in which the study takes place. Historical studies are indispensable tools in systemic analysis. It is not a question of reconstructing the whole history of the region under study, but of reconstructing the evolution of the principal processes that determine the functioning of the system. The relation between function and structure (which is parallel to the relation between processes and states) is the key to the understanding of the phenomena. We have said that a system is not just a set of elements but is characterized by its structure. Thus, a system will be defined only when a sufficient number of relations between a certain set of elements, which can be linked together with reference to the functioning of the set as a totality, has

26 been identified. In this connection it should be remembered, as pointed out in the Introduction, that with the same elements we can define different systems, and that the selection depends on the objectives of the research and will be determined by the specific questions formulated about the type of situations that are to be studied (in our case, these are situations characteristic of the physical environment, the productive system and the socio-economic system, considered as "elements" or subsystems of a single system). Having said that, it is clear that there is a great distance separating the questions, and an understanding of the links and causal relations defining the structure of the system, and that the path between the two is far from obvious. However, the definition of a system is not yet complete and some further elements we must now consider will help to find the way forward. Levels of processes and levels of analysis The functioning of what we have called an "agrarian complex" is determined not only by the very varied activities carried out within the territory of the complex, but also by external processes which generate, condition and modify, partially or totally, these activities. What is cultivated, how it is cultivated, who cultivates it and when crop changes regulated by decisions adopted in different spheres and governed by a great diversity of mechanisms.

27 Processes intervene which originate in government policies, national and international markets, technological revolutions,` etc. All these processes differ in their genesis, their dynamic and their mode of operation, but they can be grouped in a limited number of levels, each one of which requires a special study which we will call "level of analysis. The level of analysis is determined by the level of processes. In former studies on rural systems we found it useful to distinguish three different levels that may briefly be described as follows: - First level processes are changes affecting the physical environment, the socioeconomic relationships of the population in the area and their living conditions. In general, they are associated with modifications introduced into the productive system of the region. - Second level processes or metaprocesses are the changes introduced into the productive system (such as the introduction of cash crops, industrialization, or the emergence of certain "poles" of development) which result in significant first level processes. - Third level processes are changes in national development policies, international trade, capital flows, etc., that determine second level changes. To take a simple example: a first level process such as soil salinization may be produced by a metaprocess (second level)

28 consisting of the introduction of a cash crop with a technological package that led to an overexploitation of the water resources. In turn, the introduction of the cash crop might be induced by third level processes, such as changes in the foreign trade policies of the country, as a consequence of changes in the international market. The three levels of processes require different kinds of analyses. Each level has its own dynamics and requires the consideration of data belonging to quite different scales of phenomena. Moreover, processes pertaining to a higher level act, in general, as "boundary conditions" for the systems of processes at a lower level. The implications of these considerations will be seen in the next sections. Here, we shall only point out an important consequence of these distinctions for the analysis of the functioning of a system: Two first level systems having different structures may set up different mechanisms leading to a similar functioning. For instance: there are a variety of mechanisms leading to soil salinization; there are different forms of social organization to cope with the same kind of production problems. On the other hand, similar mechanisms acting at the second or third level may induce different functioning at the first level: For instance: the same technological package may induce quite different processes in two different ecosystems;

29 the same policies of expansion of the agricultural frontier may produce quite different -effects in two different socio-ecosystems. Boundary conditions and evolution of complex systems Socio-ecosystems do not exist in a vacuum. The functioning of any one of such complex systems consists of activities interacting with activities of other systems. The interactions determine flows of very different types, in and out of the system: energy, matter, credits, people, policies. They are part of what are called "the system's boundary conditions". Neither is a complex system invariable across time. It has a history. It is constantly evolving. The critical points in the life-history of a system occur when there is a functional disarrangement generating a disruption of its structure. Whether or not a system reaches a critical point depends to a great extent on its structural properties. Vulnerability is the key property here and it is mainly determined by a combination of two factors: the kind of relations among the elements of the system (i.e. the kind of relations that define the structure of the system), and the homeostatic mechanisms that prevent disruptions in the set of relations under the influence of perturbations. . The recovery of a system after the crisis is over

30 depends on the way the crisis has been overcome. The system may find itself under the influence of new boundary conditions or might have suffered some drastic internal modifications. Once such changes reach some sort of steadiness, processes will be set up to stabilize a new set of interrelations, i.e. the system will acquire a new structure. When reorganization takes place under the action of new boundary conditions, the processes that are responsible for the re-structuring of the system go in the opposite direction to the one generally assumed. In such instances it is not the case that changes in the elements (subsystems) produce changes in the functioning of the system in a linear cause-effect relationship. It is rather that changes in the activity of the total system determine rearrangements in the elements (subsystems) that finally lead to changes in the formulative structure. These mechanisms of structural modifications, based on the functioning of the total system, in essence translate into the old dictum of biologists "the function creates the organ", taking away, however, its teleological overtones. The function does not "create" the organ ex-nihilo, nor is the "totality" an entelechy that acts magically on the elements. We shall not go into details here as to the manner in which a complex system reorganizes when its boundary conditions modify or when the internal fluctuations and changes exceed a certain threshold. We will only insist on some points:

31 - The study of the state of a system at a given moment does not explain the processes that take place in its evolution. However, the analyses of the structural properties of the system, such as vulnerability or resilience, fragility or resistance do permit an explanation of the unleashing of certain processes and of the kind of reaction to be expected when the system is subject to the action of external factors or internal modifications. - The knowledge of the interactions between the whole and the "parts" (i.e. between the functioning of the system and the functions of its subsystems) is fundamental not only for the understanding of the behavior of any complex system. In addition to being a diagnostic tool, it is an essential element to be considered in the analyses of possible actions to maintain, restore or modify the system (see Chapter V). In this respect, it is necessary to distinguish a whole spectrum of possible cases. At one extreme we find "natural systems" like geological processes or biological evolution which tend to operate in the very long term; at the other, there are systems subject to economic and political decisions which can be modified in the short term. Similar considerations are valid for the spatial scales of processes. In a complex system one may have subsystems of quite different natures and the analysis of their Interactions, when the scales are quite different, may become very difficult. However, in some systems the problem can be

32 considerably simplified. For instance, in the atmosphere we find all scales of motion, but energy is concentrated in only a few of them, and this fact has made the separation of scales with their own dynamics possible. This led to a much clearer understanding of the general circulation of the atmosphere on the basis of the interactions between those few scales of motion. The boundary conditions of a system do impose severe restrictions on the possible structures that the system may have. This means that the possibility of implementing certain solutions to the problems of the system is strongly conditioned by the possibility of altering the boundary conditions. The problems may therefore be found not in the system, but outside the system. The importance of focussing on these problems in making a diagnosis cannot be overemphasized because of their incidence on planning for actions on the system. Two extreme situations may present themselves, with a large number of intermediate cases: a) A situation where the boundary conditions are very rigid, in general a result of processes taking place at higher levels. In this case, alternative policies that are technically, socially and economically feasible may not be adopted because the introduction of the needed changes in the boundary conditions is beyond the reach (or the will) of those with the power of decision.

33 b) A situation where the boundary conditions are rather loose. Here it may be possible to explore solutions that could lead to different reorganizations of the system. An evaluation of the social, economic and political implications, as well as the implications for the environment, may show that the best solution is not the most efficient one from a purely technical point of view. It would be extremely important to have case studies of types a) and b) as well as other intermediate cases. In all cases, the definition of the boundaries of the system and the interpretation of the interplay between the boundary conditions and the "internal" elements (subsystems) require serious considerations. The interactions may be very indirect. Changes in the structure of the system may either be generated by external factors (processes of higher levels) or by internal first level processes. However, even in the latter situation, significant structural changes take place, as a rule through changes in the boundary flows. An example taken from a case study carried out in the State of Tabasco in Mexico illustrates this point. In the earlier stages of the research, petroleum based activities were conceived of as an external source of perturbation with respect to the local system. As the research progressed, it became clear that the functioning of the system after the oil boom began could not be understood unless the

34 oil industry was seen as an integrated element, interacting with the agroproductive, socioeconomic and biospheric elements in multiple ways. The system, which had become fragile and vulnerable as a result of the progressive deterioration of its environmental components, and of the processes that increasingly impinged upon the peasants' living conditions, was unable to absorb the sudden expansion of the oil-related activities, without structural changes. This destabilization could not be thought of as the result of the action of an external factor affecting the system from without, de-structuring it and then retiring. Rather, specific changes in the boundary conditions gave rise to the uncontrolled growth of some of the system's internal elements which began to interact with the rest of the system in a different way, until a structural change in the entire system came about. The most obvious impact of the expansion of the oil industry - pollution and other localized alterations of the environment and direct job generation - cannot explain the general changes suffered by the system. The latter can only be dealt with by structural analysis, taking into account the indirect effects resulting from the mediation of economic elements.

35 V. PROJECT DESIGN From the perspective of the conceptual and methodological framework outlined in this document, the projects that are included in programs of agricultural development should be expressed finally in terms of an integrated system in which a physical environment, that has reached a high degree of fragility, interacts with an unstable socio-economic organization. The formulation of a proposal for an alternative way to regional development should be orientated towards introducing 'the changes needed to overcome the current situation of vulnerability in a relatively short period of time. The proposal should satisfy two criteria that are indispensable for it to be acceptable: in the first place it should be aimed at detaining and, where possible, reversing the trends that have meant the progressive deterioration of the socio-environmental system; it should be capable of re-orientating the productive system, leading to sustainable development. The concept of "sustainable development" should not only be understood in terms of productivity and biospheric

36 environmental conservation. It should also require - as a sine qua non condition - that economic activity in the region permit raising the standard of living of the people involved in the productive processes of the region. A proposal for regional development that satisfies these conditions should necessarily be based on the diagnosis reached by the previous studies mentioned above. We cannot overemphasize the fact that a diagnosis should not be a mere description of what occurred in the region, but rather it should constitute an interpretation of the processes that have led to the current state of the system. Following this idea, the formulation of a proposal cannot be restricted to the visualization of a new state of the region (obviously better than the current one). The possible transformations of the system have to be taken into account. This requires analyzing how the processes that govern the functioning of the system at present might be modified in order to reach the new state visualized for the region. Stated in this fashion, the objective of the project is that of solving what in physics is called "a problem with initial conditions" which impose severe restrictions on the feasibility of the solutions that can be visualized. To give an example, in one of our case studies, the hydrological system of the region reached a degree of deterioration that could be detained but which would be difficult to reverse and where restoration in the short or medium terms is probably not possible.

37 On the other hand, the socio-economic subsystem appeared susceptible to change, having less restrictive conditions. However, the deterioration of the peasant sector has led to the development of family based survival strategies which may not be reversible (in order to raise living and working standards) without profound changes in the boundary conditions imposed by the regional economy. It is in this context that alternative policies should be conceived. No proposal can be acceptable unless it contemplates the effective possibility of setting in motion transformations that lead the system from its initial conditions (current state of the system) to the state of sustained development that has been formulated as the hypothetical goal. These considerations serve as guidelines for the methodology that is suggested here. The specific study of each proposal for agricultural development should be carried out in two stages. The first of these would aim towards: -understanding the nature and scope of the declared objectives, that is, the modifications that are explicitly proposed; - underscoring the implicit objectives, that is, the regional or national policies that will be required for the proposal's implementation;

38 -evaluating the resources required for implementation. The second stage, the most difficult and longest - consists of a systemic analysis of each proposal which will comprehend: a) the way the proposed changes in a sector or subsystem will affect other sectors or subsystems; b) the new interactions between the subsystems as a consequence of the modifications introduced; c) the characteristics of the new structure that the system will adopt (structural properties); d) the required modifications in the boundary conditions so that the new system can function. Following this analysis, the initial system (arising from the diagnosis) should be reformulated. As a result there will be a "reconstructed system" that represents the state of the system after the implementation of the proposal. After this, the problem consists of identifying the evolution of the reconstructed system. The projection of the future of a bio-social-environmental system is not easy. The difficulty resides in the interplay of interactions between processes with different dynamics and different development time scales. In each case this requires some estimate of the speed of development of the processes generated in the subsystems, the period of time required for a given process to reach critical values that could introduce potential instability into the system and finally, the possibility of

39 the occurrence of instability fluctuations arising from changes in the boundary conditions. The study of projections in time of the significant processes of each "reconstructed system" and its interrelations -obviously a diachronic study- should be complemented by a synchronic analysis of the behavior of the global system at various moments in the future, if the projections are correct. This means making time cross sections with intervals that are suggested by the dynamic of the processes. Each "cross section" at a given moment in time leads to the recomposition of the system, then to showing how it would function at that moment before proceeding to the type of systemic analysis mentioned above. The recomposed system, for each time cross section, will be referred to as a scenario, adopting terminology found frequently in the literature with variations according to the specific context. No integral solution to the problems of a complex system can be found in the simple addition of partial solutions formulated in isolation. The interdependence of sectors requires firstly the reorganization of the partial solutions in order that they can complement one another, and then their integration with supplementary measures that permit the articulation of a coherent system. The systemic analysis of the proposal contained in the project and the study of the possible scenarios" that it might lead to in the future, must be complemented with the

40 analysis of its practicality. This requires: - the evaluation of the technological feasibility of execution given the technology already available in the region and the possibility of technological changes, adaptations or innovations that are adequate to the specific problems that have to be solved; the evaluation of the resources available in the physical, social, economic and financial environments; - the consideration of local and national interests that maybe favored or affected, as well as the possibilities offered by the international context. Boundary conditions are of special relevance and require particular attention in this analysis, with their characteristics of flexibility or rigidity and of stability, as well as the range and periodicity of their fluctuations.

41 VI. IMPLEMENTATION, MONITORING AND EVALUATION The two phases of project planning we have described above - diagnosis of the situation and project design - have been formulated in such a way that they contain the principles that should govern the subsequent phases of implementation,, monitoring and evaluation. Within the framework of our proposal, the order established in the succession of phases is only valid in that it refers to an order of precedence at the beginning of each phase: a project, which has been previously designed, is implemented; the project's progress is monitored; etc. This does not mean that each phase has an identity of its own which develops independently of the others. On the contrary, the content -and even the method- of each is, integrated with all the others. As we have indicated, the design of the project, in which corrective actions for the situations to be modified are established, is baled on a previously elaborated diagnosis of the situation. But the systemic reaction to each action

42 proposed involves interactions with sectors different from those the actions were conceived for. Thus, there is the need to return to the diagnosis in order to go into some aspects in greater depth, aspects which were not considered to be particularly important during the initial analysis, or which were perhaps set aside as not pertinent to the declared objectives of the project. These considerations should continue into the implementation phase, with the additional requirement that at this stage, adjustments should be introduced into the design when difficulties in application are encountered, or when unforeseen possibilities arise as the project progresses. This, in turn, implies monitoring with a certain degree of continuity in order to detect both the difficulties and the new possibilities that open up. Finally, it is unnecessary to emphasize the fact that monitoring can only be based on a succession of evaluations, even when these are not expressed in a formal report with the term "evaluation" on the front cover. From this point of view the (final) evaluation constitutes the synthesis and the ultimate balance of the partial evaluations involved in the process of monitoring. The interrelations between phases may seen rather obvious, but they are rarely thought of as an inherent part of the planning of projects. The explanation for this should be sought in profound reasons linked with the conceptualization .

43 of each phase. The latter is, in turn, linked with the form in which the objectives of the project are conceived. If the goal pursued is the modification of an indicator , the different phases are defined with very precise and limited objectives and the interrelationships between them become trivial. For example, if the idea is to increase the productivity of a crop which predominates in a given region, the evaluation is reduced to the truism of verifying whether the production per hectare increased or not (and by how much). Here we must return to two important points we made earlier on. In the first place, as we have indicated before, it is the questions posed by the researchers (or the experts responsible for the project) that determine the content of the monitoring and the evaluations. When the questions are aimed at determining whether the modifications introduced into the productive system are "sustainable" (in one of the acceptable meanings of this term) for the physical environment, or if they contribute - and to what extent - to the improvement of the quality of life of the peasantry, productivity is no longer the significant observable parameter. If we accept the insufficiency of a methodology based exclusively on the verification of the values of certain specific parameters at given moments, the challenge for a systemic conception is that of offering an alternative form of evaluation. In order to give a foundation to our response, we will once again look at some of the basic elements of our

44 proposal which we will summarize in the following six points. i) Agricultural development projects have' (or should have) the overall goal of achieving sustainable development in rural areas. Arriving at this goal assumes that three objectives have been achieved: a) to control, detain and, as far as possible, reverse the deterioration in the physical environment affecting production; b) to improve the quality of life of the peasant population involved in production: c) to establish conditions of social, economic and environmental stability. ii) The three objectives indicated do not refer to states of the system but to processes, that is, to temporal successions of interrelated events, in causal sequences, that - as we have said repeatedly - are not observables . iii) The set of relationships that determine the processes Constitute the systemic structure. Thus; the situations of social, economic or environmental deterioration correspond to problems linked with the structural properties of the system (fragility, vulnerability, resilience). iv) The critical periods in the evolution of a system present themselves when the processes of deterioration exceed the resilience of the structure, or when the system is destabilized by strong fluctuations. v) The vulnerability of a system is not distributed uniformly throughout its structure nor is instability simultaneously triggered in all the subsystems. The disruption

45 of a system usually begins in certain specific areas where processes occur which in other publications we have called - in accordance with a conceptualization proposed by Pierre Spitz - "structural disjunctions". vi) A systemic evaluation of a project for sustainable development should give priority to the analysis of the evolution of the processes linked with the structural properties we have indicated. On the basis of these six points it is now possible to formulate the problem posed by a systemic evaluation with greater precision. It is a question of establishing which types of observation, measurement, analysis, information from different sources, etc. are necessary to be able to infer the processes linked with physical, social and economic deterioration, and those tending to stabilize or destabilize the- system. : With this formulation we can redefine the evaluations as diagnoses focussing on certain critical problems. The evaluator can, as a consequence, find himself in two different situations. If the planning of the project has been carried out with s systemic approach, the diagnosis done initially will have already included the identification of the structural problems of the system. In this case, monitoring will consist of following the processes involved and will provide the necessary bases for an adequate evaluation of the effects of the actions of the project. In the opposite case,

46 the evaluation will have to replace the absent diagnosis and retrospectively identify the changes that have occurred in the critical processes from the project's inception onwards. The latter presents additional difficulties. Observables and Processes We now take up again two points that are at the foundation of our conceptual framework and should be kept in mind if the following is to be clear: - there are no pure observables; every observable" contains an interpretation of empirical "raw" data. - processes are not observables, they are inferred on the basis of observables. Some examples taken from case studies carried out in Mexico may serve as illustrations of how these two principles are to be taken as guidelines for the analysis of a system, whether at the stage of diagnosis or of evaluation. Example I At the beginning of the research on the introduction of cash crops in the region known as E1 Bajio , interviews were carried out among peasant communities. There was unanimous agreement that one of the most serious problems in the region was a climatic change reflected in progressively prolonged droughts. This was a fact accepted on the basis of the personal experience of the farmers, that is, supposed to be

"the product of observation". The climatological analyses we carried out did not, however, show any change in the frequency or the duration of droughts in recent decades. There were variations but they wee due to normal climatic fluctuations, without a particular trend. The reasons for this discrepancy became clear as the research progressed. The irrigation district had been expanded when the new crops were introduced by drilling wells on a /large scale. The overexploitation of the aquifers led to a marked descent of the water table. This, in turn, led to the rapid absorption of rainwater by the subsoil. What the peasant farmers "observed" was not more prolonged drought but longer periods when the soil was dry. The process which occurred was inferred on the basis of four types of observables: - the peasants' "experience". - pluviometric data. - soil analyses. - data on the descent of the water table. Example II In the study carried out in the Etla Valley in the state of Oaxaca, we found that the river that provided water for irrigation purposes had undergone a change in regime in recent decades. The peasant experience (even more than the official data) was conclusive: "when I was young I could bathe in the river all the year round; now there are long periods when the

48 river is dry". Once again drought was to blame. Further research showed that the rainfall regime had not changed. The lack of water in the river was due to deforestation on the upper areas of the hillsides surrounding the valley. The origin end nature of the process is described in the project report. It should be emphasized that in this process events corresponding to very diverse 'spatial and temporal scales combine, which, in turn, are very different to the spatial and temporal scales of the project. Example III In the next chapter we shall briefly describe a project in a region of Mexico called "La Laguna , showing how the "real processes that were responsible for the deterioration of the physical environment and the living conditions of the peasants were "hidden" behind the "empirical data provided by the official statistics. Example IV In the same region of La Laguna, we undertook more recent studies outside the irrigation district. There we also carried out retrospective research on the quality of life, using the variation of malnutrition rates in communities with different productive histories as our indicator (employing Duval's method): Two of the results obtained are particularly pertinent to our theme: On the one hand, the groups whose production was linked with the regional market had a less stable nutritional

49 state, they were more affected by changes in general economic conditions and showed higher levels of malnutrition. Meanwhile, the groups with least variability, greater ability to resist economic fluctuations and with the lowest rates of malnutrition were found in the ejidos orientated towards subsistence farming and with diversified production. 0n the other hand, when we analyzed the impact of the general economic crisis of the 1980's, we found that the different grciups had reacted differently and the evaluation showed again more favorable conditions in the case of the ejido peasants with more diversified production. These evaluations can only be made with a systemic approach such as the one we are proposing here. The process of evaluation and the evaluation of processes In the preceding pages be have tried to demonstrate that the evaluation of a project from a systemic perspective is not an "act" which is accomplished by analyzing statistics, but a process accompanying the evolution of the project. This process is none other than a succession of diagnoses or, in other words, the monitoring of the most significant processes in the agrarian complex which is the object of the project -in particular, those processes responsible for socioeconomic and environmental deterioration. Here, as we have said, it is clear that the diagnosis is not "something which is done and got over with", but is reformulated as the project progresses.

50 However, more often than not, monitoring and evaluation are thought of as independent activities, and in such cases there is a single evaluation at the end of the project. Furthermore, these cases are usually of projects that were not conceived or developed with a systemic approach. This being the case, is there any sense in trying to carry out a systemic evaluation? What would such an evaluation consist of? The examples we have given above, as well as those we will analyze in the following section, in our view give sufficient reason to insist on the application of a conceptual and methodological approach of a systemic nature in all cases where projects of sustained agrarian development are to be evaluated. The difference should not be in the approach. The difference will be found in the difficulty obtaining adequate information ex-post-facto and the relatively short period of time that tends to be set aside for evaluation. We will look at three reasons (which are not independent): - It is not infrequent that a new medication detains the specific illness it is aimed at treating, but the so- called "secondary effects" generate problems that are as bad or worse that the illness itself. The sick person dies "as the result of some other cause". For this reason there is a long list of prohibited "cures". . An agrarian development project can also successfully solve problems; but it can also create more serious ones. The history of the La Laguna region in Mexico mentioned above

51 provides a paradigmatic example in this sense. - The sustainability of the solutions achieved by a project of agrarian development are not reflected in the momentary trends of a few indicators of production or peasant income. Once again, what is involved is the problem of the predictability of the processes generated or modified by the project and their interactions with the other processes that determine the evolution of the system. The projects of agrarian development do not take place under conditions of controlled experimentation where certain variables are modified and the rest are kept constant or stationary. The changes in the boundary conditions during the period of duration of the project can be considerable. In these cases, the distinction between the effects of the actions of the project and those generated by the boundary conditions is not immediate, nor is it always easy to evaluate by means of conventional indicators. Given these considerations, no evaluation acquires reliability if it is limited to the verification of the changes that have occurred in the variables that constitute the focus of the project's action: A-priori definition of the elements that should constitute a systemic evaluation is not possible since projects are implemented in agrarian complexes, each with their own particular characteristics. Generally speaking, it could be said that the evaluation should include an analysis of the structural properties of the system when

52 the project reaches its conclusion (vulnerability, resilience, fragility, etc.); with particular reference to the possible existence of "structural disjunctions". On this basis, a retrospective reconstruction of these properties should be undertaken, backwards to the moment the project commenced. If the diagnosis and monitoring of the project is carried out in the conditions we pointed out at the beginning of this section, the evaluation is reduced to a synthesis and balance of this process.' In other cases, the problem centers on how the reconstruction should be carried out. In some cases and for some variables, statistics can be used. But this is an exception. It is well known that, for example, "where there is malnutrition, there are no good statistics and where there are good statistics, there is no malnutrition". Even when statistics are available, in a systemic analysis they should be used with care as we will see in Annex l. In the absence of reliable statistical data, the retrospective state can t be constructed from indirect indicators. In the "Interdisciplinary Research Program on Complex Systems" we have successfully used a method for the ` reconstruction of nutritional levels in peasant populations, developed by Dr. Guy Duval, J based on anthropometric measurements of social groups selected with certain sociological and anthropological criteria. A review of this and other indirect indicators is to be

53 found in the Annex 2 when we shall present tentative bases for establishing frames of reference for systemic evaluations. VII. INTEGRATED PROGRAMS AND SETS OF PROJECTS Projects of agricultural development in deteriorated regions of the world are carried out, as a general rule, in isolation. The experience gathered from the different projects is piecemeal. The lessons learned, summarized in each "final report" in the shape of "conclusions", are seldom accumulated. They may increase the personal experience of the researchers and field operators involved, but without leaving significant traces at the institutional level. The sets of projects do not constitute integrated programs. Research that we have carried out in various countries within a program called "Interdisciplinary Research on Complex Systems" included projects in which direct application was made of a complex systems approach, with the intention of obtaining - as a by-product - some insight into common conclusions that might be drawn from them. It was obvious from the beginning that this could not be obtained by means of any sort of inductive generalization" with the exception of some rather superficial analogies. The preliminary question to be asked was therefore: what are the criteria of comparability that might be used in order to look for common features between projects? The accumulated experience of the case studies referred

54 to above showed that, while each project had its particular characteristics differentiating it from preceding studies, there were profound similarities in the behavior of the various systems. They cannot be identified by just comparing data that describe the state of the systems at given moments. Here again a complex systems approach based on the analysis of processes became an indispensible tool: The lessons obtained from the comparison of case studies carried out with the approach mentioned above ate more fruitful than those that arise in evaluations made in isolation in each case. Moreover, the accumulated experience of the program mentioned above shows that new case studies can be undertaken with tools that surpass the possibilities of classic sectorial studies. Once more, this supports the assertion that both planning and evaluation reinforce each other by a judicious application of a complex systems approach when the projects are established within the perspective of an integrated program. An integrated program would consist of a series of projects referred to concrete situations and that were established in accordance with agreed criteria. The regions where the projects would be located may have quite different environmental characteristics and may differ from each other in their social, economic and cultural history. This notwithstanding, such a"'' diversity of projects nay- be put together in a single integrated program only when they have

55 been conceived as having very specific shared objectives. The fulfillment of such common objectives requires, in turn, the adoption of a common basic approach. Thus, while each project is meant to study and to solve specific problems in specific locations, the program as a whole would try to find connecting links among the variety of phenomena and situations peculiar to each project, which would permit a certain degree of generalization of their findings. This points to the need for a conceptual and methodological framework to be agreed upon by the participants in the program to allow for comparability of the conclusions that may emerge from the local studies. The possibility of attaining this goal may seen to face insurmountable difficulties. However, it can be done without ignoring the wide diversity of situations or the specificity of local conditions. Neither does it require that the researchers abandon their own `way of carrying out their work in order to follow strict rules of procedure. It does call, however, for `a serious effort of coordination and a firm commitment to include some specific features in the design of all the projects on the basis of which an integration of the program becomes possible. We do not expect the above statements to be readily accepted by all those involved in development planning. Objections will come from those who believe that some sort of integration of projects may be possible only at the level of results. Experience gathered from various international

55 efforts has provided enough evidence to show that this is not the case. Programs that have been established by putting together a certain number of loosely connected projects have remained as such in all cases we know. Attempts to carry out an ex-post-facto integration have hardly gone beyond writing some "general conclusions" that include the analysis of each project's conclusions plus some general comments pointing out similarities and differences between them. A deeper objection will come from researchers who believe that the search for similarities in widely different sets of case studies is a futile task. At the level of generalizations they will say that "conclusions" lose all practical value. There is a certain amount of truth in these claims. More often than not generalizations of results obtained in specific cases studies do end up in the same bag with philosophical speculation. In extreme cases they are merely tautologies or disguised definitions. That "the natural resources are exhausted by overexploitation" is just a trivial implication of the definition of "overexploitation". . On the other hand, although not tautological, the existence of poor peasants or of the overexploitation of resources in different places is very obvious and its verification could not be the objective of the program. At a more sophisticated level, the verification of the existence of certain kinds of 'soils in various parts of the world that could be made more productive (or just productive, if now they

57 are not) may undoubtedly be a valuable result if it could lead to an evaluation of the possible increase of agricultural production in the world that might be obtained with the present technological means, but it could hardly be considered to be the objective of the program either. Our first task is, therefore, to make clear what is meant general results" (and thereby what might be expected from the "generalizations"). The search for generalizations Generalizations that are merely a sunning up of conclusions shared by various projects or else consist of assertions of such wide scope that they lack any explanatory power cannot be the aim of an "integrated" program. More sophisticated versions usually refer to a standard or "dictionary definition" according to which generalizations are "the forming of general propositions obtained by induction". The Baconian idea of science is based on this definition and goes on to assert that it is through this kind of generalization that we get scientific laws, and that by some sort of association of laws we arrive at scientific theories. This inductivist conception of science is, unfortunately, still very persistent, sometimes implicitly, in the mind of many researchers. Curiously enough, the scientific revolution already

58 underway during Bacon's lifetime (although he was not aware of it) put scientific research on an entirely different road. Scientific theories have not been obtained by the mere association of inductive laws (which belong to the most elementary level of scientific statements); nor is a proposition considered as having the status of a scientific law in the strict sense, unless it is part of a theoretical corpus (although it might have been suggested on the basis of some kind of induction). There are therefore sufficient theoretical reasons for excluding inductive generalizations as the way to arrive at "general conclusions". We may add, however, purely pragmatic objections that might be more important for those who insist on applying Baconian methods: a ) In the kind of program we are concerned with, there is no hope `of having enough cases "of the same kind" (whatever that means) to arrive at statistically significant results. b) Even assuming that one can actually carry out enough case studies, or that some hasty generalizations could be made on the basis of a few cases, general propositions obtained from the analysis of "isolated" sets of variables, in a complex system, may be, and usually are, highly misleading. If the processing of the values of independent variables (empirical data) does not provide sufficient elements by itself to obtain the kind of generalizations we are looking for, what else can then be done?

59 Our contention is that an answer to that question can be obtained by a reformulation of the problems 'from the perspective of the systemic approach propounded in this paper. Comparative studies The lessons to be learned from a comparison of case studies` each case taken in isolation. It is the way to get meaningful general results" and "generalization's*. However, the notion of "comparing" is related to the concept of "similarity", the meaning of which, within the context of the complex systems that we are dealing with in this paper, is not obvious. From our point of view, it is a key question to be clarified before an agreement can be reached on the formulation - and even the meaning of an integrated program. ''We may be forgiven if we start with some elementary remarks. When we refer to a transformation of anything (a system, an object, a mathematical function, a theory) it is clear that we are considering different states of a certain "totality", some components of which remain constant whereas others have changed. If everything changes, we don't talk about the transformation of a system: it is the substitution of one system for another. Transformations are modifications that leave some "invariants". Mutatis-mutandis, when we refer to "comparative cases", we mean cases that have something in common, but that differ

60 in some specific features. Meaningful comparative studies concerning complex systems do require careful identification of similarities and differences. With these considerations, let us now return to the subject of "meaningful" comparative studies. What is in fact at stake is the way of making comparisons between systems. However, direct comparisons that look for similarities in the structures seldom provide solid grounds for useful generalizations. The reason is that comparative case studies tend to focus on one or another of the subsystems. Similarities found in their structure may hide profound differences in the role they play within their own systems. This may require some clarification. Let us now assume a comparison to be made between a system A, with subsystems X1, Y1, Z1, and a system B with corresponding subsystems X2, Y2, Z2. If we focus the comparison on the subsystems X and X= we may find that they are quite similar in their structural characteristics (same structure of the productive subsystems, if X refers to them: or same characteristics of soil and climate, and so on). This notwithstanding,- when there are great differences in the subsystems Y and Z interacting with the corresponding X in the functions performed by X1 and X2 may contribute in a quite different form to the functioning of A and B respectively. Likewise, as it is pointed out in Chapter II, quite

61 different structures of corresponding subsystems may perform similar functions in each one of the systems they belong to. What then may we infer from such "structural similarities" that might lead to statements having some degree of generality? It seems that this road will not take us very far and that the sources of generalizations are to be found elsewhere. -Summing up: meaningful comparisons of complex systems as well as the very possibility of non-trivial and practical generalizations will require shifting the comparative analyses from structures to functions, processes and mechanisms. Examples taken from former studies We shall summarize very briefly conducted in Mexico, having particular characteristics which make then suitable for the purpose ` of illustrating the concepts introduced in the preceding sections. CASE I The region called E1 Bajio" was known as "Mexico's Granary . Since the colonial period it was an important producer of wheat and maize for towns in the mining areas in the center of the country, and for Mexico City. This situation continued until the 1950's when the cultivation of sorghum suddenly appeared on the scene, and within two decades displaced maize and leans from the best agricultural land: Sorghum was presented as the ideal type of crop for the region

62 given both the soil characteristics and the climatic conditions. Moreover, the growing development of livestock farming in the country assured the crop a good price on the market. At a time when the world economy was governed by a distribution of production based on so-called comparative advantages, sorghum appeared to be the best choice for El Bajio. Sorghum cultivation had the support of a system of credit which included a technology package with improved seeds, new types of chemical fertilizers, pest control and weed killers. In the region of E1 Bajio, the introduction of sorghum as a cash crop began in the irrigation district and in a few years the crop had extended to the lowlands where rainfed crops (maize and beans) had been grown. These areas were provided with water as a result of the unrestricted drilling of wells: It is ironic to find that sorghum was introduced because of its "comparative advantages", one of which was its greater resistance to drought than maize. Notwithstanding this argument, the first effect was that of the considerable expansion of the irrigation system. The net result was the substitution of maize by sorghum on the lowlands. On the highlands, new rainfed areas were opened up for the cultivation of maize and beans. These new lands were fallow lands which could be more intensively farmed by using fertilizers. But the process did not stop there. The growing demand for sorghum and the economic success of its

63 producers, stimulated a new advance by the crop during the 1970's, this time towards the plateaus. The mechanism was the same: the monopolization of land and the drilling of wells. Social effects: Displacement of effective ejido cultivation of maize and beans towards marginal areas and reduction of the size of area. The production of large `' groups of peasants was restricted to subsistence levels. A sharp decline in employment possibilities since sorghum cultivation is very mechanized. The demand for labor diminished by more than 50%. A process of proletarization and semi-proletarization of ejido members. Many of them emigrated. - Deterioration of the living conditions of the displaced groups. The impoverishment of these groups was clearly reflected in the measurements taken by the project team in order to determine the nutritional status of different social groups. For instance, the peasants who lost the effective use of their land or who did not work their farms, have a rate of malnutrition of 45% as against only 27% in the case of the peasant groups continuing to cultivate their land. Effects on the physical environment The introduction of sorghum -find the extension of the cultivated area generated a greater demand for irrigation water. The surface water provided by the irrigation district was insufficient. Thus, the uncontrolled exploitation of the

64 underground water began by means of the usually clandestine drilling of wells without any regard for the effects on the environment. The water table went down from 8 m in 1955 to more than 200 m in 1982. The soil of El Bajio is of volcanic origin. At such a depth, the water extracted is hot and contains large quantities of dissolved salts which precipitate on reaching the surface and contribute to the increase of the process of salinization which is appearing in various places in the area. The recovery of salinated soil requires an increasing quantity of water in order to wash out the soil, leading to deeper and deeper wells. Thus, salinization and the descent of the water table are two processes which reinforce and speed each other up. In parallel with this management of the water, which also produces the drying out of the soil, the substantial increase in mechanization (large tractors and harvesters) which accompanied the development of sorghum, led to soil compactation and hardening that demands the use of still heavier machinery, creating a vicious circle which is difficult to break. CASE II The agricultural history of the region called "La Laguna" is quire different. It began in 1830 with the cultivation of cotton a and wheat. From 1865 onwards the area ' dedicated to cotton increased as a consequence of international and

65 domestic demand. The cotton exports from this region became one of the most important sources of foreign currency for the country. Thus, what has been called the "cotton economy was created. Urban centers grew as a result of the services required by the expansion process. In the middle of the 1950's, the international price of cotton fell drastically. Prices did not cover production costs. The private sector began to abandon cotton cultivation. The opening up of dairy farming in the region offered a highly promising opportunity for capital and the private sector turned to dairy production. However, it was necessary to exhaust the means, firstly to maintain and then to increase the volume of cotton produced since it continued to be the country's principal source of foreign currency and, moreover, sustained a textile industry that could not be left abandoned. The only solution was to compensate for the decline in the area growing cotton with increased physical yields. Only the State could provide this solution and it did. The ejido sector was largely tied to the cultivation of cotton through credit and inputs. By 1975 the participation of the ejido sector in the cotton growing area had gone up to 89%. The intervention of the State saved the production crisis. Social effects - The ejido sector paid the cost of modernization with its own

66 impoverishment, albeit in an irregular manner and despite having been left with the principal area of production traditionally generating wealth in the region, having achieved the highest historical production levels and having used the most advanced technology. The social deterioration has two aspects. On the one hand the system expels peasants. Such emigration, forced on the peasantry by the need to rent or sell land illegally, is a structural problem, not one generated by the vegetative growth of the population although it was aggravated by the latter. On the other hand, those that remain cultivating their land do not receive an income allowing capitalization - and thus independence. Neither can they sustain a standard of living in accordance with the wealth they generate. The latter leads to periods with high malnutrition levels. Effects on the physical environment The hydrological system of the region has suffered the effects of two deteriorating factors. - The "optimization" of the use of surface water, obtained mainly by lining the channels and redistributing the irrigated area, meant a decrease in the recharge of the water-bearing layers (greater efficiency in transportation implied less water filtered through to the subsoil). - Increased productivity (meaning greater biomass and, thus, greater evapotranspiration) added to the increase of the total

67 cultivated area by the introduction of the forage crops, implied an increased demand for water, which was satisfied by greater extraction from underground sources. Due to the combined effects of both factors, after 1972, the water table fell on average 2 n per year. At present, the wells reach depths which vary between 200 and 400 m. When a well stops producing because the water table is beyond its reach, drilling deeper and the new installations require investments that are no longer income-producing: When the number of shut down wells increases, agriculture will tend to be limited to those areas that can be irrigated with surface water, producing the collapse of the system. Experts in the region consider that such a moment is very close. Comparative analysis Both case studies were centered on the changes which have taken place in the agroproductive subsystem associated with processes of agricultural "modernization" (metaprocesses or second level processes). The two cases are very dissimilar in their structural characteristics: they refer to regions whose physical features, history and type of production differ considerably. Despite such differences, both studies reveal profound analogies that illustrate and support our emphasis on the functional aspects (processes and mechanisms) in comparative studies. A very general conclusion can be established, that the

68 first level processes resulted in deterioration of two kinds: on the one hand, a degradation of the physical environment is produced, due to the abusive application of a technology (in the wider sense) unadapted to regional characteristics; on the other hand, a concomitant process of marginalization of large and the degradation of their living population groups standard occurs. -A The most significant conclusions for the purpose indicated above are the following: a) Deterioration of the physical environment - particularly the hydrological system - took place in both regions. This was not due to "peasant ignorance", or to a careless administration of the resources, or to "corruption" (the three factors most frequently used to "explain" these processes). It was due to a depredatory use of resources in maximize production. In both cases, the deterioration will put an end to the present productive system. This will be the result of the same process in both cases: abusive utilization of water resources. But the mechanisms of action will be different: salinization in one case; economic inaccessibility of water in the other. b) Deterioration of the living conditions took place in : the peasants were victims of similar processes of modernization. Both regions appear to be examples of great prosperity achieved thanks to the adoption of the most advanced agricultural technology and of a type of production

69 selected with the criterion of "comparative advantages . There was a constant increase in production, yield and high economic profits for each one of the regions as a whole. There was, however, a process of impoverishment of large groups of peasants. But the mechanisms were different. - In La Laguna, the ejidatarios were tied to cotton production when it was no longer profitable for the private sector. The peasants were provided with credits and inputs and did join the "Green Revolution", obtaining large increases in yields. But,, in fact, the peasant sector indirectly subsidized the productive system. Their income (i.e. what was left after they had paid for the increasing cost of the major technological package) was extremely low and the family helped to support itself through the proletarization of some of its members. The nutritional studies corroborated the deterioration of 'their living' conditions. In El Bajio, the production of sorghum was highly profitable but most peasants did not get the means to join the "Green Revolution". Thus, they were displaced to marginal areas with the results mentioned above. In both regions the mechanisms of social marginalization were different but led to the same processes of emigration and/or proletarization of members of the peasant families. And here again, the expulsion mechanism is the same. The comparison of the case studies presented above provides room for some general remarks:

70 1.- The above case studies are examples of economic growth leading to a deterioration of social conditions. In those cases, while regional wealth increased, according to conventional indicators, a substantial part of the population got poorer. It is hot "poverty" in the abstract sense that is responsible for the environmental and social deterioration. 2.- Agricultural "modernization", imposed without regard for the cost in terms of the physical environment and the : society, functions in that way: with impoverishment of the peasant; with migrations; with proletarization; with the overexploitation of natural resources. 3:- The viability of those systems cannot be maintained for a long time. Part of the damage is barely reversible. Alternative 'paths that could be attempted would require a complete reorganization of the system (i.e. its subsystems and the boundary conditions). The socioeconomic implications are such that the problems would require a reformulation of the productive role of each region within the national context with due regard to the social implications for the local population.

71 ANNEX 1 SOME REMARKS ON DATA BASES AND STATISTICAL SAMPLING IN CONTEXT OF COMPLEX SYSTEMS ANALYSIS1 At the beginning of the 1950's, the United Nations proposed a set of indicators for measuring the standard of living of a population. This initiative with the discussion, current at -the time, as to the conceptual differences between growth and development and can be interpreted as an attempt to add an additional module" to the system of national accounting. During the 1970's, the U.N. Office of Statistics organized a series of lectures which discussed the possibility of constructing a system of social and demographic statistics that could be incorporated into the system of national accounting. In September of 1988, the United Nations University hosted in Tokyo "The International Symposium on Hunan Response to Global Change", jointly organized by the International Federation of Institutes for Advanced Study (IFIAS), the International Social Science Council (ISSC), and the United 1 This Annex is based on a synthesis of a joint paper with Fernando Cortes that arose from the weekly discussions held by the authors during 1989, and has not yet been published. 2 Report on International Definition and Measurement of Standards and Levels of Living, United Nations, New York, 1954.

72 Nations University (UNU). Among other activities, the meeting recommended: - "the extension of the existing inventory of global physical data bases to include those available in the socioeconomic area and -preparation of a report examining current levels of demand and supply for socioeconomic data". The report of the Symposium also pointed out that "it is important to recognize the need for new data collection systems and data bases in the social science area". Several meetings held after the Tokyo Symposium still left the problem unsolved as to what to collect and how to do it. In this section we shall examine the way data bases should be conceptualized and the role as well as the limitations of statistical sampling in the kind of studies we have been dealing with from the perspective of complex systems. 1.- Data bases The term data base is used to designate a set of indicators together with an integrated statistical system. The idea of constructing data bases is an almost natural derivation of an empiricist point of view. If data constitute the starting point of all knowledge, they can be directly

73 perceived and are neutral3. Therefore scientific progress in any field would be guaranteed when a large enough quantity of data can be gathered. The recurrent failures of initiatives of the sort indicated above, during the last 20 years, should have led to the suspicion that there was something amiss with this formulation. Every research project contributes new observables that are constructed by the researcher on the basis of initial observables and theory. In turn, each observable is a composite made up of inseparable elements arising from the new material provided by empirical data and the interpretations of the researcher. The very idea of constructing a data base could be doubted if the epistemic frameworks are made relative, since the edges" of each would then require its own set of observables. Obviously this in no way denies the fact that the development of science has generated large sets of observables shared by the scientific community (added to those of common knowledge). What is meant by the above comments is that it is impossible to construct a data base which can be of general use by researchers in all projects. In the hope of finding the interrelations between the Sts of complex systems analysis AM the, structure of the
3

Rolando Garcia, Conceptos basicos para el estudio de sistemas complejos", in Enrique Leff, (coordinator), Los Problemas del Conocimiento y la Perspectiva Ambiental del Desarrollo, Mexico City, Siglo XXI, 1986.

74 data base, let us recall some of the characteristics of the former. Systems are complex because they are composed of subsystems which operate in an articulated fashion (linked by mechanisms) at different levels. The metaprocesses are second level processes that influence the first level processes lows that define the boundary conditions. The evolution of the system depends both on its own dynamic and on the fluctuations of the boundary conditions. They both act on structural change. From the perspective of complex systems, a data base should not be thought of as the systematic collection of variables on the basis of which one will proceed to fully describe the state of the system at a given moment. The idea is rather to attempt to organize a set of data (observables) that: (i) allows hypotheses to be formulated as to the interplay between elements in the system (ii) provides the basis for identifying critical points in the evolution and, on the basis of these, (iii) orientates the analyses of the dynamic of the systems. In other words, the data base should contain the variables that, in addition to the objective of describing its functioning, permit visualization of the periods in which structural chanqes occurred in the system. The non-empiricist use of a data base, but which is consistent with the analysis of complex systems, relocates the role of the datum (observable) to the extent that it uses it

75 as an indicator of the stages in the process of analyzing the system's evolution. Moreover, the data base includes the set of initial observables on the basis of which the new observables will be identified. Any data matrix can be thought of as a data base if it is (potentially) useful to a fair number of studies. However, this does not mean that the initial observables required by any complex systems analysis could be fully covered by the data base. Depending on the characteristics of the specific project, it will be necessary to obtain additional information, different in each situation. The analyses of the initial variables normally allows the identification of some regularities and also anomalies in the series which cannot be explained simply with statistical reasoning. We thus have the first clues for a preliminary exploration of the field guided by some hypotheses with respect to the processes that might have produced them. This exploration does not only require a physical reconnaissance of the site but also a search for existing information in archives and specific publications, as well as carrying out interviews with key informants. All the material that has been gathered together should be seen from the outset from a systemic perspective. This means that special emphasis should be placed on the information that may help explaining the relationships between the subsystems, both in the synchronic and the diachronic

76 dimensions. Interest centers on the study of the relationships and their evolution especially around the critical points which will have been proposed hypothetically on the basis of the first analysis of the data. Based on the hypothetical critical moments in the evolution of the system, one proceeds to explore possible indicators that 'point td the boundary-' conditions and to collect additional information in order to find more evidence supporting the initial assumptions concerning the way in which the boundary conditions are related to the productive system.. Once (i) the relationships between the subsystems have been identified, (ii) the hypotheses regarding the critical points have received some confirmation, then a systemic explanation may be put forward. The validity of this explanation consists in endowing observables from different theoretical fields with coherence, constructed on the basis of very different sources and instruments of observation: census data, sample data, information generated by earlier studies, interviews, archives, newspaper libraries, etc. The process we have described, which has the objective of constructing a complex system, would seem to be very distant from the prescribed rules for statistical inference whose base is sampling., Before we discuss this issue, we must first describe the way we conceptualize qualitative research.

77 2.- Some remarks on qualitative (soft) analysis and quantitative (hard) analysis ... The difference between quantitative < research and qualitative research does not arise from the nature of the observables. Does it originate with measurement then? To measure is most usually taken to mean expressing the observable attributes of objects' in a standardized language whose syntactic rules allow operations to be carried out with them. The numerical system is most frequently used as the standard language, but logic is used on occasions. To the extent that the numerical system can be translated to that of logic but not vice versa, the first can be reduced to the second. If we call "quantitative studies" those that use the numerical system for measurement, and "qualitative studies" those that use logic, then the former can be expressed in the language of the latter but not vice versa. Thus it could be concluded that there are differences of degree between one or other type of study, but not essential differences. This argument assumes that the earlier development achieved in the field of study has reached a level in which the concepts used and the object of the study are very clear for the researcher. This is not usually the situation in pioneer research which, precisely because it is pioneering, seeks not only to define new` concepts but also constitute new objects of study and research problems. In the early stages of

78 the constitution of new fields of study, some of the problems that have to be dealt with are the construction of new observables and the elaboration of a standard language (measurement) adequate to the conceptual development and the type of questions that arise from the research problem. Thus, trivial notions of qualitative research should be excluded: those which reduce it to metric measurements, but used at the ordinal or nominal levels; and those which establish the differentiation according to the qualitative or quantitative character of the reality. There are two assumptions underlying this position: (i) that quantitative and qualitative processes "really exist" and (ii) that the researcher can directly and "objectively" observe them. We reject both of them. 3.- Statistical sampling and complex systems The distinction between processes, metaprocesses and boundary conditions pose clearly differentiated challenges as regards information gathering. Statistical sampling, whatever the design used, assumes the independent selection of the sampling units and takes as its problem the estimation of the basic parameters (means and variance) of unidimensional frequency distributions. The very definition of what a process is, excludes the possibility of applying random sampling the state of the process at t depends on its state at one or more previous

79 units of time. That is, the search for information to reconstitute the different moments of a social process cannot be obtained on the basis of the independent selection of the sampling units. On the other hand, the application of the complex systems approach is also concerned with reconstructing the metaprocesses, and the mechanisms that link do with the social and the physical processes at the first level. We have shown in Chapters IV and VII how the same process can be set in motion by different mechanisms. It is not clear how, on the basis of sampling theory, one can sample processes and mechanisms, fundamental elements for "explaining" physical and social processes. It is even difficult to imagine the constitution of a sampling frame (that is, a list of all the processes and mechanisms that make up the population) and how selection could be carried out. The above does not exclude the possibility of using random sampling for the units that interact at the first level. Here, two types of unit should be distinguished, on the one hand the physical-spatial areas and, on the other, the social units (that could be of different levels, for example, towns, communities, socioeconomic systems, domestic groups, individuals, etc.). The accumulated research experience shows that the diversity of physical-spatial and social units actually interact in ways that correspond to a rather reduced number of

80 metaprocesses, processes and mechanisms. This feature, which is peculiar but not exclusive to the studies carried out from the perspective of complex systems, restricts the possibility of taking random samples subject only to the rules of sampling theory. For this reason, the selection of the units should be made on the basis of theoretical, methodological and empirical considerations which will provide - the possibility of identifying the relationships between the observables, processes and mechanisms. This strategy for selecting the cases is quite often used in the social sciences, especially in the comparative studies of nations. The theoretical and methodological, situation of these .-studies is totally equivalent to that which characterizes the complex systems perspective: few socio- spatial units, affected 4y an infinite and small set of processes (think of the case of the stages of economic growth, demographic transition and epidemiological transition, etc.). The pertinence of sampling in the selection of the "social units" will depend, among other things, on the units of analysis that the processes being studied remit the researcher to. Probably the mechanisms provoke much more differentiated impacts if the units of analysis are domestic groups or individuals than if one is dealing with communities or socioeconomic systems with wide spatial coverage.

You might also like