¹ The literature on this question has become quite extensive. Seminal pieces include: Wildavsky, Aaron, The Politics of the Budgetary Process (Boston: Little, Brown, 1964)Google Scholar; Crecine, John P., Governmental Problem-Solving: A Computer Simulation Model of the Municipal Budgeting Process (Chicago: Rand McNally, 1969)Google Scholar; Davis, Otto, Dempster, M. A. H., and Wildavsky, Aaron, ‘Toward a Theory of the Budgetary Process’, American Political Science Review, LX (1966), 529–47Google Scholar; Wanat, John, ‘Bases of Budgetary Incrementalism’, American Political Science Review, LXVIII (1974), 1221–8Google Scholar; Barber, James, Power in Committees: An Experiment in the Governmental Process (Chicago: Rand McNally, 1966).Google Scholar A second common question in the research is: how is this particular level of allocations to be explained?

² These concepts are developed in Cyert, Richard and March, James, A Behavioral Theory of the Firm (Englewood Cliffs, N.J.: Prentice-Hall, 1963).Google Scholar There are exceptions to the use of only one budget-making system, particularly Crecine, , Governmental Problem-Solving.Google Scholar

³ The major works underpinning the incremental approach include Braybrooke, David and Lindblom, Charles, A Strategy of Decision (New York: Free Press, 1963)Google Scholar; and March, James and Simon, Herbert, Organizations (New York: Wiley, 1958).Google Scholar

⁴ Wildavsky, Aaron, ‘Political Implications of Budgetary Reform’, Public Administration Review, xxi (1961), 183–90.Google ScholarBraybrooke, and Lindblom, argue that ‘non-incrementalalternatives, even if desirable in some sense, are often politically irrelevant’ (p. 89).Google Scholar More dramatically, they declare, ‘Nonincremental alternatives usually do not lie within the range of choice possible in the society or body politic… Political democracy is often greatly endangered by nonincremental change, which it can accommodate only in certain limited circumstances.’ Braybrooke, and Lindblom, , A Strategy of Decision, p. 73.Google Scholar

⁵ Wildavsky, , The Politics of the Budgetary Process, p. 17.Google Scholar For an illustrative case study, see Wildavsky, Aaron and Hammond, Arthur, ‘Comprehensive versus Incremental Budgeting in the Department of Agriculture’, Administrative Science Quarterly, x (1965), 321–46.Google Scholar

⁶ Wildavsky, , The Politics of the Budgetary Process, p. 14.Google Scholar

⁷ Bailey, John and O'Connor, Robert, ‘Operationalizing Incrementalism: Measuring the Muddles’, Public Administration Review, xxxv (1975), 60–6.CrossRefGoogle Scholar

⁸ One might also examine directly whether there is stability in the process of budget creation. While one could infer process stability from the outputs, stronger confirmation would require case-study analysis. In the intensive case studies upon which this paper is based, only one of the four county boroughs maintained a highly stable budget-making process during the 1960s. The problem-solving routines for the budgetary process were altered very substantially in two of the boroughs and some restructuring occurred in the fourth.

⁹ Examples include Davis, , Dempster, and Wildavsky, , ‘Toward a Theory of the Budgetary Process’Google Scholar, Crecine, , Governmental Problem-SolvingGoogle Scholar and Cowart, Andrew, Hansen, Tore and Brofoss, Karl-Erik, ‘Budgetary Strategies and Success at Multiple Decision Levels in the Norwegian Urban Setting’, American Political Science Review, LXIX (1975), 543–58.Google Scholar

¹⁰ Crecine, , Governmental Problem-Solving, pp. 111–12Google Scholar and Wanat, , ‘Bases of Budgetary Incrementalism’, pp. 1222–3.Google Scholar

¹¹ Natchez, Peter and Bupp, Irvin, ‘Policy and Priority in the Budgetary Process’, American Political Science Review, LXVII (1973), 951–63.CrossRefGoogle Scholar

¹² Natchez, and Bupp, , ‘Policy and Priority in the Budgetary Process’, pp. 959–61.Google Scholar The score could also be calculated in terms of Base ALLO – that is, the ratio of service category to total expenditure. Here, Natchez and Bupp's focus at the program level has been retained.

¹³ The strict incremental model, the fair share model and the base budget model are explicated in Crecine, , Governmental Problem-Solving, pp. 119–20.Google Scholar

¹⁴ Wildavsky, , The Politics of the Budgetary Process, p. 18.Google Scholar

¹⁵ It should be noted that in the case where an allocation category receives a ‘parity’ fair share, the base budget model would also be confirmed. Thus the base budget model is a special case of the fair share model; but neither the fair share nor the strict incremental model necessarily presume that the level of change will preserve the same proportion of the base.

¹⁶ Potentially important exogenous variables include both local, configurative characteristics (e.g. demographic factors, the political system, the nature of the budgetary process) and also broad, system characteristics (e.g. the national political system, with its control of grants and its broad authority over service provision, and the national economic system, manifest in the rate of inflation and in changes in commodity and labor prices). A fully elaborated model of resource allocation changes would need to specify all these sets of variables. Currently, most research takes a more limited perspective, attempting to examine a certain subset of the complex of variables that might provide plausible explanations. One stream of research examines the impact of economic, social and political variables upon the levels of resource allocations. For the county boroughs, these studies include Danziger, James N., ‘Comparing Approaches to the Study of Financial Resource Allocation’ in Liske, Craig, Loehr, William and McCamant, John, eds., Comparative Public Policy: Issues, Theories and Methods (New York: Wiley-Sage, 1975)Google Scholar; Boaden, Noel, Urban Policy-Making (Cambridge: Cambridge University Press, 1971)Google Scholar; and Alt, James, ‘Some Social and Political Correlates of County Borough Expenditures’, British Journal of Political Science, 1 (1971), 49–62.Google Scholar

¹⁷ This last observation is not accepted by many scholars of British local authorities. It is beyond the purposes of this paper to deal adequately with the complex question of the level of autonomy enjoyed by county boroughs. There is, however, a growing literature which supports their autonomy, including Danziger, , ‘Comparing Approaches to the Study of Financial Resource Allocation’ and Boaden, Urban Policy-Making.Google Scholar

¹⁸ The selection of two differentiated sets of county boroughs is not particularly relevant to the research questions addressed in this paper. The two pairs were chosen because of other theoretical concerns in the larger study from which these data have been taken, Danziger, James N., ‘Expenditure Variations and Spending in English County Boroughs’ (unpublished Ph.D. dissertation, Stanford University, 1974).Google Scholar The four county boroughs are given Greek letter names to respect certain assurances of privacy made in the field research. Measures representative of the differences between the two sets of county boroughs are these:

While the four boroughs were not randomly selected, there seems no reason to suspect that the findings are unrepresentative of other county boroughs. The findings might be generalizable to other governmental budgeting contexts. But it is important to note that this research has not replicated the framework in most American-based studies because there is an important difference in the budgetary processes.

In the American-style process, budget-making is factored into a distinct set of episodes with discrete, measurable and public decisions, including the spending unit request, the pruning unit's revision, and the legislature's decision. In contrast, county borough budgeting is better characterized as a flow of transactions through time rather than a discrete set of actions. Over a four-month period, there is continual formal and informal bargaining between spenders and pruners at various levels. Negotiations with each spending unit are independent and all figures are strictly confidential. Only the final estimates presented to council, which are normally authorized without much revision, are reliable and appropriate data. Thus the multi-stage quantitative models used for American-style systems are not applicable to county boroughs. For use of the American-style system in a Norwegian municipality, see Cowart, , Hansen, and Brofoss, , ‘Budgetary Strategies and Success at Multiple Decision Levels in the Norwegian Urban Setting’.Google Scholar

¹⁹ The category ‘all rate and grant born services’ includes all those services that the county borough finances out of its general rate fund account. This includes a substantial proportion of the total spending of the county borough on governmental goods and services; but it excludes most spending on housing, on any municipal utilities, and on some other local undertakings for which a direct charge is applied. ‘Total education services’ include spending on nursery, primary, secondary, special and further education, on administration, on the school health service, on meals and milk, on teacher training and on youth centers and youth employment. The ‘local health services’ include mother and child care, midwifery, health visiting, home nursing, the ambulance service, health centers, prevention, care and after-care, etc. ‘Highways and streets’ spending includes highways, public lighting, street cleansing, road safety, etc. Representative spending figures (net) are these, for Kappa, in financial year 1969:

²⁰ For the regression analyses, the tables report the significance of the regression coefficient as measured by the F score, given the appropriate degrees of freedom. Although the primary commitment of the paper is to decide upon substantive significance, the following levels of statistical significance are appropriate:

For brevity, the tables do not report the F score, standard error of estimate or regression equation. These data are available on request from the author.

²¹ It might be that pooled data hide interesting year-to-year variations in the magnitude of allocation changes. To assess this, the data were disaggregated by borough and year for each category in Table 2. The time frame and number of services is too limited to support strong generalizations; but the figures did suggest some clusterings. For example, ‘relatively incremental’ increases and decreases seemed to cluster in particular years. Interestingly, there were few similarities between boroughs in any year. The opposite finding would have suggested the impact of macro-political or macro-economic forces.

²² Wanat, , ‘Bases of Budgetary Incrementalism’, 1221–4.Google Scholar

²³ Natchez, and Bupp, , ‘Policy and Priority in the Budgetary Process’, p. 963.Google Scholar Natchez and Bupp speculate that greater fluctuations might be expected in more specific expenditure categories. Analysis of the county borough data revealed that fluctuations were substantially greater with each further breakdown of expenditure categories. This is not a meaningless breaking-out of data in the county boroughs. In Theta, for instance, the standing orders required that each line-item in the budget be separately calculated, authorized and monitored. If any line-item (e.g. wages and salaries for home nurses within the home nursing sector of the local health services) was overspent by more than a minimal amount, an entire re-authorization procedure was required for a supplementary allocation.

²⁴ Due to amalgamations in Lambda and Iota, the financial years 1965 to 1966 values for this model are treated as missing values in the analysis.

²⁵ Crecine found some negative relationships in two of the three American municipalities but did not expand upon the finding. Crecine, , Governmental Problem-Solving, pp. 197–9.Google Scholar Applying Crecine's method (autocorrelation of ALLO,) to the county borough data, over half the correlation coefficients are statistically significant, many are quite high, and twenty-four (of twenty-eight) are negative.

²⁶ The category of net total expenditure is not included in these figures because the strict incremental model is the only applicable naive model.