1.1 What is the PAF Frame of Reference?
This chapter analyses in depth the concept, role and the general use of a plan frame of reference (as basis of alternative scenarios) as an inherent tool for the programming approach itself. Frischâs âplan frameâ, which we have encountered in the previous parts of the Trilogy, represents the principal tool of the programming approach. Frischâs plan frame will be referred to as the planning accounting frame (PAF), in order to be more precise, emphasising its quantitative aspect and nature.
Following on from what has been said in Vols. I and II on the programming approach of Frisch and that of other planning economists and decision theorists, the programming approach is not only a process that allows us to analyse phenomena and behaviour in a decision-centred manner,1 but one that requires a holistic, systemic and simultaneous vision of all phenomena of interest and concern to public management.
To bring this vision to fruition we must build a frame, a scenario2 through which to enlarge the scope of the decisions to configure a system of decisions that are consistent and compatible. This frame also consists of the articulation of several alternative frames or scenarios, which will allow us to make final decisions (as when we want to provide more than one complete scenario that is either based solely on the variations of alternative variables and or on a revision of the entire model).
In such a way, the building of planning or programmatic frames or frames of reference becomes the preliminary component of the programming approach (and therefore, of a good form of planning that has been described in this book).3 We shall call the result of this construction, the essential tool of the programming process, âthe planning (or programmatic) accounting frameworkâ (PAF) for public policy.
The PAF is a framework composed of meaningful numerical values, such as measures, indicators and variables, of facts and phenomena (expressed in the relevant technical unit as long as it is quantifiable) that are significant to the welfare and development choices and programmes of the communities concerned.4
The first step of the PAF is therefore to include the phenomena (and related quantified values) that conform to the choice of those building the PAF and whose data should be in principleâwhether such data is already available or whether it must still be researched or createdâa function of their role as representatives of the objectives of the programmes.
The second step is to include the temporal evolution of the programmes. Therefore, the data included in the PAF is (virtually or factually) dynamic, concerning the future rather than the past.5 In this way they express programmatic objectives for changing a situation from a given starting point (the present state) towards an arrival point (the future state).
The third requirement of the PAF to express the programmes for which it is the instrument, in addition connecting the phenomena that belong to the system to which the programme will apply. The PAF, therefore, must express a capacity for ultimately adapting and fitting in the phenomena included in the larger framework.
Concerning policies, programmes and plans at a national scale, the PAF must be understood (without losing its programming character) in accordance with the articulation of the significant phenomena upon which the justifications for policies and programming decisions on an international and supranational (or global) scale depend.
The PAF, therefore, is an articulated system of predetermined accounting number and values of significant phenomena, in terms of:
- the intra-relations (or internal relations) upon which the policies and programming decisions at the institutional level are made concerning the phenomena and data that the programme controls and monitors;
- the external relations among the phenomena and data which the system must (since it is an open system) inevitably maintain with co-determinant relations.
Without information (or assuming the basis of knowledge of such information) about such systemic relations, any kind of planning or programming becomes precarious and loses intelligibility and utility, becoming candidate for failure.
This would lead to planning (or programming) without systemic bonds; a kind of pseudo-planning.
1.2 The PAF and its Models
In its configuration, or constellation, the PAF implies a model. Any PAF is the computed solution of a decisional model, provided that its variables and parameters (outlined in Vol. I, Chap. 7) are based on choices and decisions computed on forecasts (and not simply results, naturally conferring an important role to the past).
The PAF is then a decisional model of phenomenological relations based on intentions and preferences; and then on programmatic choices and decisions; and, at the same time, a decisional model with which to monitor and test the compatibility and feasibility of such intentions and preferences.
However, insofar as it models an ex ante programmatic framework, the PAF expresses feasible relations in a time perspective, from which we can ascertain the technical and political feasibility of a plan: or rather, its construction and its use constitute an instrument for making political decisions, since it allows users to control the feasibility of the different decisional hypotheses over time ex ante.
The PAF, therefore, is the instrument that allows politics to be efficient in its purposes, and not merely untested declarations of goals and purposes.
The PAF is the instrument for the construction of programmatic scenarios or reference frameworks needed for making conscious political decisions in light of the complex effects of each partial choice; decision-making that is far more necessary than the programme needs for democratic consensus (in the given institutional forms); and even all the more necessary as implementing the programmesâ needs for a negotiated consensus, both of the operators and the beneficiaries of whatever type.
1.2.1 Temporal Flexibility of the PAF
In Leontiefâs work on national economic planning quoted above, he makes some very simple and useful basic observations that are suitable for tracing how the reference PAF must be considered in general terms:6
In its published form, a national economic Plan, or rather the statistical appendix to its text, can be visualized as a detailed, systematic annual survey of manufacture and agriculture, of transportation, and of trade and the federal and local budgets. However, it describes the state of the economy not for a given past year [âŚ] but rather for five years in advance and, in a more summary form, for a much longer interval of time stretching into the future. This does not mean that a plan must be rigidly adhered to over the entire period of, say, four or five years. On the contrary, the plan should be revised each year in the light of past experience and newly acquired information and pushed out as a moving average one year ahead. (Leontief 1976a, pp. 150â151)
Frisch often emphasises the concept of the moving average:
This simply means that each year we work out a new dynamic optimum decision analysis for the planning horizon (say five or seven years) which is adopted, taking account, of whatever fresh information has become available. This means inter alia that in the plan which is worked out in any given year, we have to include in the set of non decisional elements (i.e. in the set of already-committed-to elements) those things that were decided upon in the analysis of the preceding year. (Frisch, Preface to the Oslo Channel Model (1961e), Memo of DE-UO, p. 44)
Frisch often uses the terms moving or sequential planning. And in the following passage, he notes:
It is customary to speak of, say, a one year plan, a five or seven year plan, a twenty year plan, and so on. This kind of division is a practical necessity because many concrete aspects of the work must be different according to the length of the time horizon.But this practical necessity must not lead to the idea that a plan of a given time lengthâsay a five year planâis something that is to be worked out at a given date and then to be petrified and stuck to for the coming five years, regardless of what is going to happen in the course of these five years. This would be a dangerously naive procedure.In a dynamic and living world the planning work must be flexible enough to absorb and utilize all the new information that is constantly pouring in. And it must also be flexible enough to take account of changes that might occur in the policy makersâ desiderata. For many years I have advocated the view that the only rational way to introduce this sorely needed flexibility, is to put the whole planning work on a moving basis.Consider a plan of a given time length, say a five year plan. Such a plan should be elaborated every year and each such plan should cover the next five yearsâthis is the principle of moving planning.Each year there will be certain commitments that have been made and certain actions that have been taken on the basis of a previous five year plan. Some of these commitments and actions may perhaps later have turned out to be ill-considered, others perhaps very wise. But in any case these commitments made and actions taken will have to be considered as data when elaborating next year the new five year plan in which all new information and any possible change in the policy makersâ desiderata are to be incorporated.This way of proceeding is far from being haphazard and undecided. It means pursuing the decisional analysis and the action program in the most rational and persistent way.This viewpoint leads to the important distinction between two categories of magnitudes: The already-committed-to magnitudes and the decisional ones. The latter are the magnitudes which have to be decided upon at the time when the new plan is to be elaborated. In previous and more detailed papers I have used a special notation for distinguishing between these two types of magnitudes. (Frisch, An Implementation System for Optimal National Economic Planning [1963a, b Memo of DE-UO, p. 40], rep).
The planâs flexibility is therefore implicit within the same logic of planning. It is only from total ignorance of the literature that vulgar objections that plans are rigid can arise.
1.2.2 Flexibility of the PAF (according to Johansen)
Leif Johansen, for instance, says that the basic decision-theoretic scheme of planning (referenced in Vol. I, Chap. 7, Sect. 8) is too simple unless it is interpreted in relation to its time dimension.7 The components that represent the various time periods provided by the application of the scheme can be obtained by calculating:Johansen explains the reasoning in this way:
- 1.the set of policies;
- 2.the set of exogenous factors of decision schemes;
- 3.the set of the results or states of the economy resulting from it (which make up the basic scheme itself).
We now consider a series of time periods designated by t = 1, 2, âŚ, T, where T is the âhorizonâ â the last time period under consideration. In most of what follows we shall not specify this explicitly.[Previously] a policy was represented by the symbol a, and the set of all possible policies by A. An element a could be thought of as a vector indicating the values assigned to all available instruments of policy, perhaps supplemented by qualitative descriptions if not only quantitatively different policies were under consideration. We shall now interpret an element a to be a description not only of a policy for one specific p...