Development of mathematical models for the Finnish state administration

Development of mathematical models for the Finnish state administration

Development of mathematical models for the Finnish state administration Jorma PIHLATIE Planning Secretariat, Ministry of Finance, SF.00121 Helsinki 12...

814KB Sizes 0 Downloads 43 Views

Development of mathematical models for the Finnish state administration Jorma PIHLATIE Planning Secretariat, Ministry of Finance, SF.00121 Helsinki 12, Finland Received May 1979 Revised January 1980

The planning and budgeting system and the use of mathematical models in the planning of the Finllish State Administration are briefly surveyed. It is noted that one area where mathematical modelling is relatively well established is medium-term econometric modelling. Despite a few early successes in the traditional OR application areas, there has been. until recently, a conspicuous l:~ck of successful models at the ministry or agency level. Against this background, the paper describes a training and consultation program that has been conducted by the Ministry of Finance since 1974. The pro:~ram has aimed at increasing the use of mathematical models at the ministry avd agency levels with the ultimate aim of thus improving Central Government efficiency. Each year some 6 project groups have been selected in the program. Training has been given to the groups in 2 to 4 sessions of one week each to support critical phases of their modelling work. The paper surveys the results of the program and concludes by summarising the experiences obtained. This paper does not seek to be an objective and scholarly description of the history or present state of the art cf modelling within the Finnish Central Government. Instead, it is a personal evaluation of the subject matter.

1. G o v e r n m e n t

planning and budgeting system

Planning at all government levels is conducted with three different time spans (a comprehensive account is presented in [4]):

The author is indebted to his colleagues for many helpful comments on the text. Special thanks are due to Mrs. Ralli Knuuttila, Head of the Bureau of Planning Methods, and Mr. Reino Niinivaara, both from the Planning Secretariat, as initiators of the Government activity reported in this paper.

North-Holland Publishing Company European Journal of Operational Research 6 (1981) 333-339

- medium-term planning (usually 5 years), - annual budgeting, - short-term planning (e.g. operational planning and cash flows). Naturally, in some sectors plans are also drawn up for longer time spans. These plans and planning processes do not have fixed institutional forms and are not discussed here. The pervading structure in all of the above plar~s is the line-item structure of the annual budget. It m~,.y be said, however, that medium-term plans are program-structured to a considerable extent, while shortterm plan structure largely depends on the govern° ment sector and level. Relationships between the different plans are roughly as follows: Medium-term financial plans are used to derive expenditure ceilings for the next year's budget. The fiscal year for which the annual budget is applied coincides with a calender year. The budget appropriations as approved by Parliament, on the other hand, form a fiscal framework within which spending in short-term plans must be confined. Comprehensive medium-term ~lanning was initiated in 1967. The planning process starts with general instructions issued by the Ministry of Finance to all other ministries. These again issue the agencies and other admhdstrative bodies in their s, tor more specific guidelines and schedules for their plan preparation. The sectoral plans are consolidated at the respective ministeries before being forwarded to the Ministry of Finance for analysis. The Ministry of Finance prepares feedback reports for each ministry and also compiles a summary of the plans to be included in the next year's budget proposal to Parliament. •Development work in planning and budgeting is continuously carried out mainly in the form of separate projects in ministries or agencies (see [3]). Some of the means used in this work are: - introduction of new management techniques (MBO) mainly at agency level, - improvement of the planning process, e.g. by increasing the dialogue on setting and attaining targets between different levels of administration, - improvement of the data base used in planning

0377-2217/81/0000-0000/$02.50 © North-Holland Publishing Company

334

J. Pihlatie / Mathematical models for the Finnish state administration

e.g. by developing standards and methods in cost accounting and output measurement, - development of methods to evaluate plan alternatives and to allocate and manage resources more efficiently. It is naturally the last item in the above list which nearest approaches modelling. Although development projects may arise from widely varying points of departure, much of the work is associated with medium-term planning. Efforts have been made at all levels of administration - from regional authorities to Parliament _ to make medium-term planning a major instrument of policy formulation (strategic planning). Another line of development is to avoid a tendency inherent in *.he planning process of preparing growth-oriented plans which cannot be realized given the growth rate of the economy as a whole. Hence, efforts are being made to increase realism in medium-term planning. The plan structure has also been subject to a continuous though moderate development. As an example, some influence from zero-based budgeting is reflected in the instructions issued by the Ministry of Finance in recent years. Much of this development work has been carried out by the Planning Secretariat at the Ministry of Finance. This is a small staff unit of some 20 analysts organized directly under the Permanent Secretary of the Ministry. The bulk of the work of the Secretariat is carried out in the form of expert advice and consultant services to other ministeries and agencies. It is believed that this approach can lead to better results than by resorting to the traditional administrative practice, i.e. authority in the form of instructions and regulatioas. Modelling expertise is a relative newcomer to the 'product line' of the Secretariat which includes manegement development, planning and accounting systems, cost accounting and cost-benefit analysis. 2. U ~ o f models within the Government

It must be initially noted that there is no tradition of public sector model-building in Finland to speak of. A significant exception~ however, is econometric modeUing which has been practised at universities and research institutions as well as within the state administration since the beginning of the sixties. The most solidly established single model application is the forecasting and planning system KESSU

for tile Finnish economy [ 1]. This system has been developed from a macro-economic medium-term model constructed at the Finnish Economic Planning Centre during the 60's. From 1973 onwards the model development has been entrusted to the Economics Department of the Ministry of Finance. The main use of the system is to support the Ministry in coordinating the medium-term plans drawn up by other ministeries and adjusting them to the resources of the central government sector and the whole economy. This medium-term assessment also forms a basis for work on the following year's budget. The system is also increasingly used in short term economic policy preparation. As for micro-economic models and operations research, things are not so well. There are, however, a few cases since the beginning of the sixties where successful models have been constructed within the administration. Naturally, a number of OR studies pertaining to Government have also been made at various universities. It is only to be regretted that these have not generally been utilized in decisionmaking. The earliest OR application areas were the operational problems of transportation. For example, some successful simulation studies were made at the State Railways during the sixties. Some modelling has continuously been going on there, even though no OR group has been establighed. The military also developed a few interesting though rather insignificant models. During the seventies meaningful modelling has been carried out for example in energy administration and research institutions. A major modelling exercise in the beginning of the seventies was a transporation study aimed at determining 'optimal' policies in allocating the commodity flows of Finnish imports and exports to alternative routes and modes of transportation. Special emphasis was placed on traffic to and from northern Finland and the fact that shipping by sea is hindered by ice, particularly in the northern ports. A port simulator was constructed to determine theoretical cargo handling capacities and unit costs at different ports. These data ~,ere used in an elaborate transportation model which was used to evaluate the consequences of alternative policies in traffic allocation. It is perhaps interesting to note that the most concrete result of the study was that a seriou~ work for developing common cost accounting standards for different modes of transportation was initiated by the Ministry of Communications at the respective agencies.

J. Pihlatie ~Mathematical models for the Finnish state admhzistration

In general, very little in the way of successful aid to aecision-making had been achieved by modelbuilders by the middle of the seventies. To some extent, this can be attributed to the fact that recruitment policies of government bodies do not favor people with science or engineering degrees. As appropriate forms of co-operation between Government and universities have been lacking, it has been difficult even to recognize potential OR application areas. Because of this situation, most of the models have been developed in one of the following ways: - by engineers or economists, who are newcomers i~. the administra,ion, on their own initiative, or by external consultants, with the blessing of an agency's top management, on a contract basis. Probably the most important single reason for limited success of modelling has been due to these wrong processes of model-building. The first approach fails because of insufficient participation of management in model-building. In the second case, the model is hard to sell to its actual users. The probability of incorrect problem identification is also very high in both cases.

3. A training and consultation program The transporation study mentioned in the previous section became, in a way, a turning point in Government model-building. It still is, and probably will remain, the largest single exercise in applying traditional OR to Government problems in Finland. Even though the use of the study results in decisionmaking were limited, they showed the potentiality of OR in policy formulation to a wide and rather highlevel audience. It also was a decisive lesson to those involved in the actual model construction at the State Computer Centre. The large application potentiality as well as the necessity of active participation of both management and line personnel in the model construction were revealed. Against this background, a training and consultation program was started in 1973 to promote the use of quantitative planning methods in Central Government. Initially, the program was run by the State Computer Centre, but since 1974 it has been directed by the Planning Secretariat of the Ministry of Finance. Some of the basic principles from the very beginning of this program have been: - The best way of learning about models is by construtting and using one.

335

- Model-building is group work. The group leader must not be an OR expert but an administrator or planning officer in the problem area in question. - Appropriate levels of management must be represented in the group or, if this is not feasible, a steering group must be appointed. - The modelling work must be so planned and the problem so constrained that pilot tests will have been carried out by the end of the training program time span (less than a year). - The work of each group in the program is stimulated by requesting that progress reports and a final report be presented to other groups. Some 5 or 6 projects have normally been selected in tile program each year. Each project group of 3 to 5 persons is assigned a consultant from the Planning Secretariat or State Computer Centre. Steering groups are usually also appointed with one member from the Planning Secretariat. The training within the program has been conducted by the State Training Centre as part of its general training program. The contents and timing of the seminars are prepared at the Planning Secretariat to support critical phases of the modelling work as well as possible. in the first years of the program, training was provided in four sessions of one week each. Since then, the training component has been reduced first to three weeks (which perhaps is an ideal amount) and thea to two weeks. This has resulted partly from the monotony of repeating essentially the same courses from year to year. Additionally, as more and more important problems have been attacked, more senior staff with fewer opportunities for extended absence from daily duties have participated in the training. An essential part of the program has been a oneweek intensive workshop during which the groups work out the model structure. An indicative schedule of the 'institutional' part of the program in the form it was run in 1977 is the following: 1st training session (February). Model-building: process and methods. Teaching is based on a carefully prepared simplified version of an existing application which is worked out stage by stage in teams. - Workshop (April). Construction of the mathematic,,! model in each group. 2nd training session (May). Project administration. Information systems. Statistical estimation. 3rd training session (November). Each project group presents its final report. Examples of appli-

-

-

336

J. Pihlatie / Mathematical models.for the Finnish state administration

CONTENT OF WORK (MINISTRY OR AGENCY)

TRAINING (STATE TRAINING CENTRE)

METHODOLOGICAL SUPPORT (MINISTRY OF FINANCE) ii

TRAINING

]

SEMINAR FOR MANAGEMENT (3 days)

l

STEERING GROUP FOR THE PROJECT

GROUP OF CONSULTANTS

Chairman [

From

Members I

the agency

Leader: Chief of the Planning Methods Bureau

Project consultant

I

Secretary: project leader

) i

PROJECT GROUP SEMINAR FOR PROJECT GROUPS (3 x 1 weeks)

project leader (project secretary) project members consultant

/

- project consultants

Project consultant from Finance Ministry or the State Computer Centre I I

I

1

I,,i

Other resources from the ministry or agency

ADP personnel

ADP personnel

Fig. 1. Organization of the training and consultation program.

cations. OR methods. In recent years, members of the steering groups, most of whom are senior policy makers, have been invited to a three-day seminar for maganement. The overall organization of the program is illustrated in Fig. 1.

4. Program results During its existence, some 250 civil servants about half of whom are from senior or middle management level have been trained in the program. About 40 models of diverse structure have been constructed with varying success (see e.g. [2,5 ]). The following is a sample of the more successful applications: - A production planning model of the Basic Atlas of Finland for the National Board of Survey.

- A computer r ame for training rescue organizations in the even~ of a catastrophic accident. - Teacb.:i demand models for upper primary and secondary schools. - A freight train scheduling model for the State Railways. - A model to estimate shares of Local Governments in monthy total tax receipts. - A medium-term financial planning model for the Ministry of Social Affairs and Health. As can be seen from this sample, not all of the models can be directly associated within the general framework and goals of planning development pre. sented in Section 1. This is largely due to the fact that the projects almost always arise from the initiative of the ministry or agency. Hence, no overall programming of this development work is possible. On the other hand, it is strongly felt that no modelling exercise is successful without the users being in

J. Pihlatie / Mathematical models for the Finnish state administration

charge of the effort all the time. When projects are selected in the program, considerable attention is naturally given to their relevance from the point of view of planning and economy in general. Many of the models do not reach their final form during the time span of the training program, and technical support is often required in their further development. It is felt essential, however, that at least some concrete results are obtained as soon as possible. These will provide in most cases much better hints for further development of the model than detailed construction of a large model without a possibility to experiment with its early versions. A clear tendency since the inception of the program has been a shift from advanced techniques to simple, nonoptimizing models. In the first years of the program, problems often were forced to lend themselves to a linear programming formulation. The philosophy was that with intensive training and a concrete application example of a problem which has at least some real-life appearance, one can reveal the application potential of OR methods. After this it would simply be a matter of time to let important and extensive modelling applications emerge. Without reflecting here why, it suffices to say that this was not the case. In the first attempts, the best models - which is to say those for which the problem the model was built to solve had a great deal in common with a real-life problem - had some impact on decision-making provided that the interest of appropriate policy-makers was sufficiently drawn to the exercise. More often than not, however, models where optimization - even in goal-programming fashion - was used, were never understood well enough by administrators. There was another problem which turned out to be a much more serious obstacle to successful model implementation than one would have expected. After the dissolution of the group which constructed the model, responsibility for its further development and maintenance was usually transferred to a certain person or group in the line organization. Computer runs were usually made at the State Computer Centre. In surprisingly many cases, it turned out to be far too demanding a task for the line organization to cope with estimation of new parameter values for the model, determine necessary changes in its structure and arrange for the technical execution of these tasks to be performed at the Computer Centre. In this way many 'good' models were soon forgotten in the pressure of daily duties of the persons involved.

337

It was realized that problems in the technical use and maintenance of models - as insignificant as they may appear compared with problems in systems analysis and model construction - s h o u l d receive serious attention. Naturally, this is predominantly a problem of models being used in organizations without OR groups of their own. It is a fact, however, that even though OR expertise exists here and there, there are no groups specializing solely in OR within the Finnish civil service. As one would expect that the environment in this respect is very much the same in many organizations in Finland and elsewhere, it is surprising that the problem has received so little attention in the literature on the OR process and methodology. There are methodologies and administrative procedures - developed largely by ADP specialists - which make it possible to construct data processing applications costing incredible sums of money but contributing little or nothing to the efficiency or effectiveness of the organization as a whc.le. The usual recipe is to create a separate organization to use, maintain and develop the system. Once such an organization has been set up, it is extremely difficult to question its existence or performance. This is not so with models. Most useful models are small and simple; they do not bring in much money to a Computer Centre. In addition, they require more from their end users than a conventional ADP application. A standard solution would be to establish an OR group wherever models are used. It is doubtful, however, whether this would always lead to the most efficient overall use of resources which indeed, is what OR itself claims to be advancing. Hence the approach in Finland has been to promote model-building in existing organizations, planning units or line organization. An effort towards a solution to the problems discussed above was acquiring a planning software system which would be simple enough to allow nonexperts to learn to build simple models largely on their own within a reasonable period of time. For this purpose, a survey of existing financial planning systems was made in 1976 and detailed information was obtained on all major program products in this field. After testing the best alternatives, a Swedish product called Mercur was eventually selected and installed at the State Computer Centre. The main criteria were simplicity, range of application and programming power. Since 1977 most of the models constructed within the training and consultation program have been coded using Mercur. In addition, other applica-

338

I. Pihlatie/ Mathematical models .for the Finnish state administration

tions of highly varying type have been set up, including some models which initially were tailor-made in Fortran.

5. On the role of modelling in Government As has been said earlier, there are no groups within the Finnish Central Government which could be called OR groups. Perhaps the Bureau of Planning Methods with a staff of four at the Planning Secretariat comes closest to being one. Clearly, establishing such groups would bring about many obvious advantages. On the other hand, there are many arguments against forming such groups. The first years of the seventies witnessed a proliferation of various planning units at different levels of the Finnish State Administration. These units have frequently been blamed for having devised planning systems and procedures which require far too much effort compared to the benefits gained from employing them. Automatic data processing has also brought about many applications which are not economically justifiable. H-',, can one ensure that OR groups would avoid such criticism, justified or unjustified? Is there not a danger that OR groups will themselves assume the responsibility of constructing and using models to become an isolated island in the organization? Is it not the very nature of OR that an interdisciplinary approach be applied in problem solving? Last but not least, there is the fact that with a sharp decline in the growth of government revenue during the last few years, the Government has adopted a very stringent policy in establishing new vacancies within the civil service. Consequently, the opportunities for establishing new organizational units have been very limited. The philosophy of promoting modelling has thus been to support such activities within existing organizations. This means that, as a rule, the model has no organization to support it or keep it alive if ~t isnot accepted by its end user. More often than not, this approach leads to simple models that do not require extensive data collection. It has also been a part of the philosophy to rely as much as possible on existing data. This is based on the assumption that with the same effort, much more can be achieved by using existing data and extracting relevant information in a condensed form from these as compared to using the same effort to collect new data. It is suprising to see how limited is the use of many large government data bases in decision-making, even

in cases where they have been created explicitly for that purpose. Perhaps models can contribute to bridging the gap between such data bases and policy making at higher government levels. At least the program described in this paper has consciously avoided creating new systems for collection of basic data. As regards the role of modelling in the development of planning in general, it is perhaps illustrative to divide the applications into two categories. This division is based on a rather clear distinction of the organizational counterparts in the two categories. (1) Budget-structured planning. The users of this type of models are general administrative organizations or budget divisions. The models are used primarily as calculation automata to facilitate comparions of different plan alternatives. Typically, the application could cover consolidation of. medium-term financial, personnel and output plans based on bids from several organizational units. Models for quarterly or monthly allocation of resources within the frames of annual budget appropriations also are typical in this category. The models usually have to adapt to tight schedules of preparing official plan documents. Therefore, the techniques should provide for user-oriented ADP, hopefully with a possibility of interaction of the user with the model. It is not believed that models would be very useful in annual budgeting except in the case of a few large expenditure items. (2) Analysis o f individual programs~problemoriented planning. Here the model users would typically be planning officers or policy makers responsible for the content of a program. The models tend to be more complicated in structure, depending on the complexity of the real system. If the model is constructed for medium-term planning, it typically is built to evaluate consequences of alternative policy options. Shortterm models in this category tend to deal with allocation and management of resources and may be oriented towards optimization. Naturally, there are many models which cannot easily be assigned to either of these categories. The main purpose of presenting the above division is to emphasize the importance of pursuing model building in both of the two areas. Even though models of the first category may not appear intellectually very challenging, they often contribute quite as much to decision making and improvement of government effectiveness as models in the second category. Efforts will continue in the Finnish State Administration to promote the use of models in both of the above areas. It perhaps goes without saying that most of the work in this field still remains to be done.

J. Pihlatie /Mathematical modelz for the Finnish state administration

References [ 11 E.-L. Kaski, K. Mannermaa and R. Myhrman, Tile forecasting and planning system "Kessu" for the Finnish Economy, Economic Commission for Europe, Document No. EC.AD]SEM.2/R. 104 (1974). [2] I. Nousiainen, PERKA - a model describing the use of resources in the second cycle of the comprehensive school, Paper presented at the Sixth Nordic Congress on Operations Research, Turku (1977). [31 Planning Secretariat, Ministry of Finance, Development

339

of planning and budgeting systems and methods within the Finnish State Administration, Report presented at the conference of the European Group of Public Administration in Tampere (1976). [4] A. Saviaho, R. Sailas and T. Ylinen, Julkisen talouden suunnittelu ja budjetointi (Planning and Budgeting of Public Finances (in Finnish)) (Gaudeamus, ltelsinki, 1976). 151 K. Turtiainen, Planning model for the welfare of mi~lors, Paper presented at tile Sixth Nordic Congress on Operations Research, Turku (1977).