Joumnl ofManagement Studies 32:3 May 1995 0022-2380

BUDGET ALLOCATION IN A UK UNIVERSITY: CONTRASTING PERIODS OF RESOURCE AVAILABILITY WITH RESOURCE SCARCITY* MAHMOUD EZZAMEL Manchester School of Management, UMZST

MICHAEL BOURN Universip of Southampton

ABSTRACT

This paper seeks to identie variables which explain variations in budget allocations in a UK univirsity using a pooled database relating to two periods: one of resource availability, the other of resource scarcity. The results over both periods indicate, consistent with previous research, that a combination of workload and power proxies explains a very high proportion of budget allocations. However, analysis of the two periods revealed that: (i) incremental budgeting, which was observed to be in operation during the period of resource availability, was not supplanted by comprehensive budgeting during the period of resource scarcity, (ii) the impact of power proxies on budget allocation did not materially change in response to changes in levels of resource availability, and (iii) compared with internally based power, externally based power (as proxied) had neither a greater impact on resource allocation during the period of resource scarcity nor a lower impact during the period of resource availability as hypothesized. INTRODUCTION

The pervasiveness of large organizations in modern society means that significant resource allocation decisions occur within organizations rather than being restricted to trading between organizations (Boulding, 1963; Pondy, 1970). Our understanding of how and why resources are distributed to various functions and sectors within society would be better informed by an appreciation of the mechanisms and processes of intra-organization resource allocations. The means of resource allocation range from incremental budgeting, in which allocation proceeds from an existing base and attention is focused on the increment/decrement of growth, to comprehensive budgeting systems, such as zero base budgeting (ZBB) and planning, programming and budgeting systems (PPBS) which fundamentally challenge previous allocations. Under ZBB each manager has to justie the entire budget request as a complete programme in competition with Address f o i reprink Mahmoud Enamel, Manchester Schnol of Management, UMIST, PO Box 88, Manchester M60 lQD, UK.

0 Basil Blackwell Ltd 1995. Published by Blackwell Publishers, 108 Cowley Road, Oxford OX4 lJF, UK and 238 Main Street, Cambridge, MA 02142, USA.

314

MAHMOUD EZZAMEL AND MICHAEL BOURN

other programmes as users of scarce resources. For each activity a ‘decision package’ is carefully prepared in which the manager identifies a minimum level of spending and the expected cost-benefits associated with additional expenditure levels on that activity (Phyrr, 1973). ZBB therefore ‘compares alternative funding’ and might be thought of as ‘manifesting vertical comprehensiveness’ (Wildavsky, 1986, p. 320). PPBS emphasize objectives and output and seek to attain rationality in strategic resource allocation. They are constructed by reference to functional groupings of activities with a common objective (i.e. programme) rather than the conventional organizational sub-units (e.g. divisions). Under PPBS it is argued that budgets are more directly related to plans and total system objectives as the programme structure is dictated by the objectives of the entity (Ezzamel and Hart, 1987). PPBS compare ‘programs’ and could be viewed as ‘embodying horizontal comprehensiveness’ (Wildavsky, 1986, p. 320). Several researchers have sought to identify resource allocation models used in universities as a particular type of organization. Most of the previous studies have been conducted in situations where the resource base is expanding or stable. T o date, there have been few studies on decision making, particularly those involving allocation of resources, in situations of resource scarcity (Levine, 1980; Rubin, 1980). This is a particularly pertinent issue since for about the last fifteen years the university sector in the UK has been subjected to severe financial strictures imposed by the state. The main purpose of this paper is to examine the impact of the extent of resource availability on the parameters driving budget allocations. This is performed here at two levels. First, the paper explores whether or not incremental budgeting systems, which are assumed to be prevalent in situations of resource availability, are supplanted by different, and in particular by comprehensive budgeting systems, in situations of resource scarcity. Secondly, the paper investigates the extent to which power proxies in general compared with workload proxies assume different levels of importance depending upon the extent of resource availability. An even finer question which the paper seeks to address is the extent to which there is evidence of a differential impact on resource allocation for externally based and internally based proxies of power in situations of differing resource availability. Hence, at a general level this paper seeks to contribute to the study of universities as organizations in which important resource allocation decisions are made. The results reported here are based on analyses of a pooled database (1975-84) extracted from the files of an established, above average-sized UK university. The paper has five further sections. The next section provides a brief overview of previous work and develops the research hypotheses. The second section describes the research method used. This is followed by a presentation of the results and then a discussion of these results. The final section draws together the main conclusions.

BUDGET ALLOCATION UNDER VARYING LEVELS OF RESOURCE AVAILABILITY AND DEVE1,OPMENT OF HYPOTHESES

Much of the literature on budget allocation in organizations has focused on situations of growth in which resources are in abundance. Such emphasis may be

8 Basil Blackwell

Ltd 1995

BUDGET ALLOCATION IN A UK UNIVERSITY

315

a reflection of what Scott (1974) has called ‘the pervasiveness of a growth ideology’ which has its roots in the Keynsian economics notion of continuous and unlimited growth that was popular in Western economies until relatively recently. These growth-based assumptions, however, have been undermined, particularly during the last two decades, by a severe and persistent economic recession resulting in a pressing fiscal crisis in Western economies. In contrast to the preoccupation of scholars with budget allocations in growth settings, precious little attention has been devoted to this topic in settings of resource cutback and organizational decline (see Boulding, 1975; Levine, 1980). Below, we examine the literature dealing with resource allocation in both growth and decline situations. There is some convergence in the literature with respect to the prevailing pattern of budget allocation in growth situations. As Scott (1974, p. 245) has pointed out, ‘organization growth creates organizational abundance, or surplus, which is used by management to buy off internal consensus from the potentially conflicting interest group segments that compete for resources in organizations’. Similarly, Cyert and March (1963) have suggested that in periods of resource availability allocation decisions tend to be based on the self-interest of organizational sub-units. Incremental budgeting is assumed to persist in such settings of resource availability (David and Kantor, 1980; Greenwood, 1983; Schick, 1980; Wildavsky, 1975, 1986). Several arguments can be deployed in support of the presumed prevalence of incremental budgeting when resources are not in short supply. First, labourintensive organizations, such as universities, are characterized by: (a) long-term labour contracts, and @) highly organization-specific labour assets (Williamson, 1985). The combination of these two characteristics renders incremental budgeting less costly than comprehensive budgeting as the latter could result in expensive redundancy schemes. Second, incremental budgeting is more politically acceptable because it preserves the previously negotiated status quo; it allows for the construction of (possible arbitrary) rules for resource allocation to maintain the relative positions of power of coalition members (Cyert and March, 1963; Meyer and Zucker, 1989). Hence, incremental budgeting reduces conflict by restricting bargaining to incremental resources rather than extending it over the whole budget. Third, in organizations with a large professional staff component, such as a university, financial considerations have tended until recently to have a less pronounced impact on decisions compared with professional autonomy (Becker and Gordon, 1966; Ezzamel and Bourn, 1990) and hence incremental budgeting may be more suitable compared to comprehensive budgeting. The above arguments are underpinned by a particular decision-making mode. Wealth, in the form of resource levels, provides a ‘cushion, which takes time to dissipate’ (Wildavsky, 1986, p. 15). In these cases, decisions tend to be made on the basis of ‘habit, intuition, snap judgements, and other forms of informal analysis will suffice for most decisions because the costs of making mistakes can be easily absorbed without threatening the organization’s survival’ (Levine, 1980, p. 15). But what makes such an informal decision-making mode possible is the realization that the bulk of resources is allocated through incrementalism in the form of ‘a base’, so that intuition and snap judgements are applied only to the margin. The above discussion leads to the following hypothesis:

0 Basil Blackwell Ltd

1995

316

MAHMOUD EZZAMEL AND MICHAEL BOURN

Hypothesis I: Incremental budgeting systems are likely to be prevalent in periods of resource availability. In situations of resource scarcity, the ‘wealth cushion’ is no longer there and organizations become appreciably more uncertain with respect to available resources compared with likely demands for spending (Wildavsky, 1986). Hence, decision makers become preoccupied with establishing ‘inclusionary mechanisms to prevent organizational participants from avoiding the sharing of “public bads” (increased burdens) that result from the depletion of the common pool of resources. . . . These mechanisms need to be carefully designed to penalize and constrain “free exiters” and cheep exits at the convenience of the employees while still allowing managers to cut and induce into retirement marginally performing and un-needed employees’ (Levine, 1980, p. 16). Similarly, Cyert and March (1963) have argued that in periods of resource scarcity allocation criteria become more explicit and resources tend to be directed more towards solving organizational problems and less towards providing prestige for the sub-units. Loose resources are gathered up to help cope with contingency, and resources may then be re-allocated from shrinking to growing sub-units (Rubin, 1980). One of the administrative mechanisms that could be introduced is comprehensive budgeting, whether in the form of PPBS or ZBB. For example, following severe and continuing fiscal problems in the municipal government of Wilmington, Delaware in the USA, ZBB was introduced in 1976 to cope with the situation (Singleton et al., 1980). This and other examples were part of a wider concern with the fiscal crisis in the USA a decade or so earlier which led to calls for the use of comprehensive budgeting systems in order to ameliorate the crisis by comparing the costs and effectiveness of alternative means of attaining public policy objectives (O’Conner, 1973). Furthermore, the sparse literature on the management of organizations in decline (e.g. Levine, 1980) suggests that such organizations should follow a number of scenarios to manage their decline, including cutting or abandoning low prestige programmes, and installing rational choice techniques such as comprehensive budgeting. As Levine (1980, p. 24) has suggested, comprehensive budgeting such as ZBB ‘gets such high marks as a method for making decisions about resource allocations under conditions of decline’, because it is future-directed, allowing for analysis of existing and proposed new activities rather than being premised on an ‘inviolate-base-plusincrement’ method, and it permits trade-offs between activities below their present funding levels. Wildavsky (1986, p. 323) has also noted that only poor countries (which may resemble organizations facing fiscal crisis) ‘come close to zero-base budgeting, not because they wish to do so, but because their uncertain financial position continually forces them to go back on old commitments’. The above discussion leads to the following hypothesis:

Hypothesis 2: Comprehensive budgeting systems are likely to be more prevalent in periods of resource scarcity. Inehtably, the budgeting process is political as many commentators have come to recognise (e.g. Cyert and March, 1963; Pfeffer, 1978; Wildavsky, 1975, 1986).”] Budget allocations involve aggregating incompatible needs and preferences and

0 Basil Blackwell Ltd

1995

BUDGET ALLOCATION IN A UK UNIVERSITY

317

this renders the budgeting exercise a political problem. Even comprehensive budgeting systems, which are advocated in the name of rationality and economic efficiency are inherently political, for whenever decisions are made to allocate cuts in resources a trade-off has to be made between equity (ie. equal sacrifice for all sub-units) and efficiency (i.e. assigning the greatest cuts to those sub-units deemed less valuable to the organization). As Levine has pointed out: ZBB allows a ranking of decision packages by political bargaining and negotiation so that attention is concentrated on those packages or activities most likely to be affected by cuts. As a result, ZBB allows both analysis and politics to enter into cutback decision making and therefore can incorporate an expression of the intensig of need for resources by participating managers and clients while also accommodating estimates of how cuts will affect the activig levels of their units. (Levine, 1980, p. 24; emphasis in original) Writers prescribing ‘effective’ modes of managing financial crises have suggested that one of the tactics to smooth out decline is to cut programmes run by weak sub-units (e.g. Levine, 1980). Furthermore, empirical evidence (see below) suggests that budget allocation in organizations is driven not only by sub-unit workload but also by its power. Under incremental budgeting, much of the power effects once negotiated become embedded into the base budget and hence in subsequent periods power dynamics only affect the distribution of periodic increments. In contrast, it could be that when resources become more scarce there is a greater scope for sub-unit power to influence the whole budget when comprehensive budgeting is used. Salancik and Pfeffer (1974) reported that power was more closely associated with allocation decisions involving scarce resources compared to decisions dealing with plentiful resources. In a later study, Pfeffer and Salancik (1977) reported that, when resources were allocated through committees of departmental representatives, political advocacy for using criteria favourable to the sub-unit was positively associated with allocations, particularly for more powerful sub-units. However, for general fund allocations made by unrepresentative groups, including the administrative component, political advocacy paradoxically tended to be associated with decreased resource allocations for more powerful departments. They suggested that allocations may be used by administrators ‘to reward and punish sub-units according to their unwillingness not to challenge the structure of control and norms in existence’ @. 654). Pfeffer and Salancik (1974) and Pfeffer and Moore (1980) reported that representatives on committees, and grant contract funds were significant explanatory variables of allocations, but Hills and Mahoney (1978) found no support for the influence of committee representations. Further, Hackman (1985) reported that environmental power (ability of a subunit to acquire external resources needed by the organization) and institutional power (relative influence of a sub-unit within the organization) together explained half of the variance in internal resource allocations. Ashar and Shapiro (1988) reported that the degree of sub-unit centrality had a significant effect on budget allocation particularly in periods of financial stress. There is also limited evidence to suggest that different forms of power can have different impacts on allocation depending upon the extent of resource avail@ Basil Blackwell Ltd 1995

318

MAHMOUD EZZAMEL AND MICHAEL BOURN

ability. Thus, Hills and Mahoney (1978) have suggested that the predominant influence on budget allocations when resources were scarce was externally based power whereas under conditions of resource abundance precedent (as reflected in incremental budgeting) was the predominant factor. When resources are scarce the organization could become more prone to environmental pressure. Sub-units which are proactive externally can shelter the organization from the effects of environmental turbulence and they could also enhance the external legitimacy of the organization. Moreover, as internal slack diminishes, such sub-units could exploit opportunities in the external environment to secure additional resources for the orcganization. The organization may in turn reward sub-units with a high external profile with an increased share of scarce resources. The preceding literature seeking to document the effect of sub-unit power on budget allocation is rooted in the conventional view that power can accrue to, and be possessed by, sub-units because they: (i) have crucial knowledge (e.g. Crozier, 1964); (ii) contribute significantly to organizational mission (e.g. Thompson, 1967); (iii) absorb uncertainty facing the organization (e.g. Hickson et al., 1971; Perrow, 1970); and (iv) command scarce and critical resources (e.g. Pfeffer and L o n g , 1977).r21These arguments relate both to scarcity and centrality; the more scarce are the resources (be they in the form of knowledge, money, etc.) the more able the sub-unit in possession of these resources to demonstrate its centrality to the organization, and the more the organization is likely to concede more of its internal resources to the sub-unit. The above discussion leads to the following two sub-hypotheses: Sub-hyiothesti 3a: Sub-unit power has a greater influence on budget allocation in periods of resource scarcity compared with periods of resource availability. Sub-kypothesis 36: Externally based power has a greater influence on budget allocations compared to internally based power in periods of resource scarcity.

'I'HE RESEARCH METHOD

This section describes the models used, the data and its organizational context, the choice of the dependent and independent variables, and the method of analysis. ?he Models The first model employed is:

where:

Bi is the budget allocated to sub-unitj in period t, whether as an absolute sum or as a percentage, 0 Basil Blackwell Ltd

1995

BUDGET ALLOCATION IN A UK UNIVERSITY

319

U{ is the vector of workload criteria relevant to sub-unitj in period t, pi‘ is the vector of power variables relevant to sub-unitj in period t, and u is the error term. In (l), workload criteria and measures of sub-unit power have a direct influence on budget allocation. It is possible that workload can influence expenditure patterns indirectly through their influence on sub-unit power; for example, subunits with greater size may command greater power (Pfeffer and Moore, 1980). This possibility is investigated in the results section. An additional model applied in this paper is:

where:

Bj-1 is last year’s budget for sub-unitj. n2e Unwersig Context The UK university used as a research site (henceforth ‘the University’) comprised academic faculties (n = 8) with 40 departments, academic supports groups (n = 4), and service groups (n = 5). From 1974/75, it operated a scheme of budgetary devolution. Each faculty, support group, and service group was a separate budgetary group. Block sums (i.e. one-line budgets) were allocated by the University Council annually to each budgetary group, on recommendation of the policy and finance committee of Senate and Council (advised by its budgets and development sub-committee), after deducting a given amount (< 10 per cent of the total available) ‘off the top’ for general commitments. The system was introduced in order to devolve financial responsibility and to provide a framework for planning, which would permit faster and more flexible responses to changing conditions than the previous method of centrally planned staffing establishments and recurrent expenditure. The University enjoyed strong student demand and had a good research record. In common with many UK universities, it had experienced strong growth from the early 1950s up to the mid 1970s when the real rate of growth in the UK university sector slowed over a period of five years (Scott, 1984). The University had deliberately engaged in a policy of deficit budgeting by over 2 per cent per annum over the period 1978-81. In July 1981 the University suffered a reduction of 11 per cent in real terms in its state-based funding spread over a period of three years while retaining the same funded student targets; it also had to save its own prior overspending. By 1984/85 the immediate financial crisis was securely negotiated. For the purposes of this paper, the overall period of study is divided into two sub-periods: the period of resource availability (1975-80) and the period of resource scarcity (1981-84). The first sub-period was characterized by both over-spending through deliberate deficit budgeting and expansion in the resource base of the University. During that period not only were posts that became vacant filled, but many new posts were created. During the second sub-period the University had to manage a reduction in its resource base amounting to 11 per cent in real terms in addition to the 0 Basil Blackwell Ltd

1995

320

MAHMOUD EZZAMEL AND MICHAEL BOURN

accumulated deficit in its budget over the previous four years, which was well in excess of 10 per cent. The Data Data collected related to budget allocations, workloads and power for each of the 40 academic departments. Although the study covers a period of ten years, data was collected for eleven years (1974-84) in order to allow for lagging the explanatory variables by one year. Of these academic departments, 34 had data covering the eleven-year period. Of the remaining six, one had data for ten years, one for seven years, two for six years, one for four years, and one for three years. The missing years resulted from either the addition of new departments or the decomposition of large departments into smaller departments. Data sources included University Revenue Account papers, University calendars, minutes of Senate and other committees, and internal papers. No data were collected for academic support or service groups. The share of total expenditure for these groups declined slightly during the latter part of the period of study. The dependent variables. The dependent variables used in this paper relate only to expenditure by academic departments from budget income allocated by the Centre of the University each year. This choice was deliberate; in order to uncover the role of power in budgetary allocations the dependent variable(s) were restricted to only those common funds for which different budget units competed internally. It should be noted that the data are for actual expenditures made, and not for the budget allocations. Over the full ten-year period there was very little difference in total for any budget unit between actual expenditure and budgets (however determined), although there may have been over or under-expenditure relative to budget from year to year. Actual expenditure therefore can be deemed a satisfactory proxy for budget allocations. Three variants of the dependent variable have been used in this paper: (1) total expenditures by each department; (2) the sums expended by each department as a proportion of total expenditure of all departments; and (3) annual expenditure increment (decrement) for each department. The first variant has been used by most previous studies. Pfeffer and Moore (1980) used the second variant in order to remove the effects of inflation, budget growth, and systemwide constants. With this measure, only changes in the share of expenditures of a given department relative to all academic departments would be observed. The third variant was used to focus more fully on the impact of incremental changes in expenditure caused by incremental shifts in the explanatory variables.

Ihe explanatoy variubtes. These include three sets of variables: workload, sub-unit power and previous year’s budget. Workload. Budget allocations may depend, wholly or partly, upon the department’s contribution to the organization’s total workload. In this paper, student load proxies workload. It is the number of full-time equivalent (fte) undergraduate and postgraduate enrolled students taught in a department. Workload thus reflects teaching patterns rather than recruitment patterns. The figures used here are those calculated by the University Centre. This measure has been used 0 Basil Blackwell Ltd

1995

BUDGET ALLOCATION IN A UK UNIVERSITY

32 1

by many previous researchers as a proxy for either workload (e.g. Hills and Mahoney, 1978; Pfeffer and Salancik, 1974) or teaching output (e.g. de Groot et al., 1991). Sub-unit power. For the purposes of operationalizing our parameters, power may be classified in terms of source, conceptualization, and measurement. Power may emanate from internal or external sources; conceptually, it may be potential or enacted (Wrong, 1968; Provan et al., 1980), and it may be measured using subjective or ‘objective’ measures. The proxies for power used in this paper are: membership of University committees; grants and contracts, in terms of both amount of money and number of grants; number of academic publications; and representation on outside bodies. According to the above classification, membership of committees is an internally derived, potential, objective proxy for power. Numbers and amounts of grants and contracts are externally derived, enacted, objective proxies for power. Number of publications and representation on outside bodies are externally derived, potential, objective proxies for power. This classification can be problematic. For example, it could be argued that membership of committees reflects enacted power in the sense of being an outcome of power relations. There may well be an element of positive feedback. Committee memberships may reflect past power relations, whilst influencing potential future power. It was not possible to derive subjective measures of power (e.g. peer group reviews, and internal surveys of views about prestige and power) for the period examined in this paper, although more recent developments in internal and external audit and assessment make this a much greater possibility in future studies. Membership of University committees has been used in previous studies as a measure of sub-unit power (e.g. Hills and Mahoney, 1978; Pfeffer and Salancik, 1974). The measure used in this study was the number of memberships of important University committees held by members of a department. The committees selected were: Joint Policy and Finance, Budgets and Development, Advanced Studies, Staff Policy, Library, and Computer Services. Representation on outside bodies was used as a proxy for power derived from advisory relationships of a sub-unit with the external environment (see also Hills and Mahoney, 1978). Fifteen external bodies (e.g. the Boards of Research Councils) were used in the study. It is arguable that these are a result of past power, not a ‘cause’ of anything. A measured association doesn’t determine cause/effect, of course, but we do here hypothesize one. It has been assumed that the greater are a department’s resources from research grants and contracts, the more internal prestige it commands, and the more favourable are the Centre’s decisions towards the department. Furthermore, the more non-UGC (University Grants Committee) related funds, particularly Research Council funds, which a department can raise, the more important it may become to the University, and the greater is its relative power (e.g. Cohen and March, 1976). Grants and contracts include funds from Research Councils, government agencies, and industry, as well as gifts or donations. Amounts of grants and contracts are likely to be strongly related to, but not identical to, research expenditure in any given year. This latter measure defines away variations in productivity (Brinkman and Leslie, 1986). Therefore, number 0 Basil Blackwell Ltd

1995

322

MAHMOUD EZZAMEL PlND MICHAEL BOURN

of publications was preferred to research expenditure as a proxy for research output, and this proxy has been used by previous researchers (e.g. de Groot et al., 1991). Clearly, comparisons of publications produced in different fields are problematic since there are variations in research traditions and publication outlets between different disciplines. However, conducting the analysis over a tenyear period within the same university should permit valid comparisons of relative patterns of activity through time. It is worth noting that the demarcation line between workload and power proxies is not clear-cut. For example, in so far as publishing academic work is an integral part of an academic’s job, number of publications could be viewed as another proxy for workload. It is classified here as a power proxy because of its presumed significant impact on reputation and hence on internal (and possibly external) power. However, a significant tranche of funding (approximately 35 per cent) has been allocated to, and within, the University since 1992 specifically by reference to research achievement. This is based heavily on the quantity and quality of publications, which are presumably being treated as quality-adjusted ‘workload’ in such a system, but this lies outside the period examined here. Previous year’s budget. Budget allocations may be based on a uniform proportional change in the previous year’s budget. This may be closely reflected in departmental expenditures. Davis et al. (1966) reported on the practice of uniform proportional budgeting, and Pondy and Birnberg (1969) referred to ‘fair share’ budgeting usually associated with incrementalism. To capture such proportional allocations the previous year’s expenditure by each department has been used here in separate estimation equations (see (2) above) as the onb independent variable. This is similar to a measure developed by Hills and Mahoney (1978). The results of the estimated equations will be used to reveal the underlying pattern of budget allocation, be it incremental or comprehensive. Allocations will be deemed incremental if the results show that previous year’s budget explains an extremely high proportion of the budget for the following year. In contrast, a comprehensive allocation pattern will be assumed if previous year’s budget explains a relatively modest part of the budget for the following year.

Method of Anabszi The data were analysed by estimating regression equations for each of the ten years 1975-84, where variants of departmental expenditures were treated as the dependent variables and regressed on the independent variables. Regressions were performed using both non-lagged and lagged (by one period) independent variables in order to test whether the explanatory power of the model is significantly improved through lagging variables. The lagged variables were generally those which would have been available to those responsible for budget decisions. Further, pooled regression analysis was performed on the whole data set, as were tests of differential intercepts and slopes. To test for differential intercepts, the following two-way fixed effect model was used:

r,i=a;+ p x ; +

E;

(3)

where, i indexes departments, i = 1,2,. . . . ,40, and t indexes time, t = 1,2,. . . , 10. 0 Basil Blackwell Ltd

1995

323

BUDGET ALLOCATION IN A UK UNIVERSITY

The analysis assumes that there is a vector of Kindependent variables, p is the vector of constant coefficients, a2 is the constant term for the ith department, and & is the constant term for the tth year. Hence, in this regression, there is no overall intercept in the model, rather the intercept is allowed to vary across departments and time periods. Standard econometric F tests were employed to examine the stability of differential slopes Uohnston, 1972).

RESULTS AND INTERPRETATION

Initial Results Table I contains some of the descriptive statistics and table I1 shows the matrix of correlation coefficients between the variables used for the pooled data. It is clear from table I that there are considerable variations in the levels of the variables studied. Table I1 shows that: (i) each independent variable is highly correlated with the dependent variable, and (ii) there is a high degree of multicollinearity between several independent variables; this is expected since they are all size-related. In the subsequent analysis the effect of such multi-collinearity on the robustness of the results is explored.

Table I. Descriptive statistics (n = 376)

VaTiabk

std dev.

Mean

Budget K) Budget '10 Std load No. of grants No. of publications External committees University committees

330,790 2.52 162.02 8.20 38.29 0.56 5.73

M&imU?SV

38 1,930 2.69 127.33 14.71 58.38 1.27 5.63

Marimum

8,711 0.04 2.00 0.00 0.00 0.00 0.00

2,937,000 17 650 117 442 8 31

Table 11. Pearson correlations matrix of variables (n = 376)

Y1

x1 x2 x3 x4 x5

YI

XI

x2

x3

x4

x5

1.oo

0.7 1 1 .oo

0.78 0.38

0.90 0.59 0.84

0.69 0.43 0.56 0.64 1 .oo

0.83 0.74 0.65 0.77 0.63 1.oo

1.oo

1 .oo

Y 1 is budget as a percentage of allocations to all academic departments XI is departmental student load X2 is number of grants obtained by department X3 is number of publications per department X4 is number of members of each department on outside bodies X5 is number of members of each department on University committees

0 Basil Blackwell Ltd

1995

324

MAHMOUD EZZAMEL AND MICHAEL BOURN

Main Resultr Hypotheses 1 and 2 state respectively that the use of incremental budgeting is associated with resource availability, while the use of comprehensive budgeting is associated with resource scarcity. These two hypotheses were tested by estimating annual regression equations for nine years (data on amounts of expenditure were available only from 1975 onwards) by regressing each annual expenditure level on the previous year’s expenditure level. The results indicated that the previous year’s expenditures alone explained more than 99 per cent of annual expenditures throughout the whole nine-year period. While this may offer strong support for hypothesis 1 during the period in which resources were available (1975-80), hypothesis 2 is not supported for the period of resource scarcity (1981-84). The results suggest that no significant changes took place in the budget allocation system during that period. In order to explore this issue further, additional analysis of the data was undertaken by performing ten annual regressions (1975-84), and also by pooling the data over the whole period. Initial annual regressions were performed using as dependent variables amounts of expenditure and the percentages of departmental expenditure to total expenditure by all academic departments. The independent variables were student load, number of grants (which out-performed amount of grants as an explanatory variable), number of publications, representations on University committees, and representations on outside bodies. After correcting for heteroscedasticity, the results of using expenditure percentages and amounts of expenditure as dependent variables were virtually identical; therefore only the former are reported here (see table 111). The results of pooled regressions are also reported in table 111. Student load (the proxy for work load or size) was significant in each of the ten years studied, and in the pooled regression analysis. In contrast, evidence on the significance of power proxies was mixed. For example, number of publications was highly significant in every year and in the pooled regression analysis. However, representation on University committees was significant in only two of the ten years (1982; 1984) but was significant in the pooled regression (notice that although the regression coefficients were negative the correlation coefficients between this variable and the dependent variable were positive: +0.79 for 1982, +0.73 for 1984, and +0.83 for the pooled data). Representation on outside bodies and number of research grants were significant in three and five of the ten years studied respectively, and both were significant in the pooled regression analysis. This may have signalled a decline in the impact of some power proxies vis-u-vis others. Further, workload may have gained greater relative importance in determining expenditure patterns as resources became scarce. But this still offers no support for comprehensive budgeting, since all these proxies may themselves be used to foster incremental budgeting procedures in order to preserve the status quo of the political coalition in the University. Hence, even though statistically significant shifts in the intercept and slope parameters occurred, they did not seem to be sufficiently strong to result in a major change in the resource allocation formula used in the University. The potential impact of interactions between the explanatory variables on the results was examined by focusing on student load and number of publications. To control for the interactions with size, four interaction terms were created by 0 Basil Blackwell Ltd

1995

35

35

35

36

38

38

39

40

40

40

1975

1976

1977

1978

1979

1980

1981

1982

1983

1984

0.1731 (2.901)-

0.2790 (2.202)** 0.2752 (1.784)* 0.3218 (2.366)0.1499 ( I .076) -0.1 713 (-0.098) 0.2 130 (1.299) 0.3449 (2.733)*** 0.3608 (2.634)*# 0.2542 (2.353)0.1230 (1.188)

Constant

0.0052 (7.164)****

0.005 (5.397)*** 0.0053 (3.235)*** 0.0069 (3.444)**** 0.0057 (4.988)**** 0.0045 (1.985)** 0.0063 (3.151)*0.0069 (6.892)**** 0.0077 (6.425)*** 0.0064 (7.649)**0.0082 (7.323)****

Student had

* p < 0.10; “ p < 0.05; ***p < 0.01; ****p < 0.001.

Pooled regression 1975376 1984

n

Year

~

Table 111. Results of annual regressions

0.0311 (3.404)****

0.0504 (2.637)*** 0.0492 (2.019)** 0.025 1 (0.942) 0.0484 (2.233)** 0.0499 (2.894)**+ 0.0355 (1.512) 0.0256 (2.015)** 0.01 39 (0.907) 0.0093 (0.625) 0.0151 (0.754)

No.of grants

variables

0.020 (7.252)****

0.0372 (4.395)**** 0.0391 (3.650)**** 0.0423 (7.145)**** 0.0285 (5.814)*#* 0.0237 (2.394)** 0.0278 (8.139)*0.0267 (6.747)*** 0.0282 (5.225)*#* 0.027 1 (7.1 34)**** 0.0285 (7.746)****

Publuahonr

I*endent

0.2952 (3.166)-

0.3604 (2.558)** 0.6817 (3.232)*** 0.7604 (2.465)** 0.1840 (1.343) 0.3941 ( I .597) 0.21 14 (1.209) -0.0363 (-0.341) 0.0441 (0.389) -0.0058 (-0.079) 0.0296 (0.269)

Outride bodm

0.942 -0.0050 (-0.107) -0.0893 (- 1.585) -0.1434 (- 1.834)* -0.0042 (-0.16 1) 0.0514 (1.112) -0.0419 (-0.963) -0.0472 (- 1.815)* -0.0946 (-2.124)** -0.0452 (- 1.172) -0.0838 (-2.843)*** 0.05 17 (2.036)**

he1

0.895

0.958

0.957

0.939

0.951

0.9 17

0.892

0.936

0.908

0.907

AdJlcsted

Untverng commttkm

326

MAHMOUD EZZAMEL AND MICHAEL BOURN

multiplying student load by each of the remaining explanatory variables, but none was found to be significant. However, the introduction of the interaction terms affected two of the original explanatory variables; both number of research grants and membership of University committees became insignificant. Similarly, to examine the effect of research reputation on the results, three interaction terms were created by multiplying number of publications by each of number of research grants, memberships of University committees and representations on outside bodies. A pooled regression model was estimated (n = 376; adjusted R2 = 0.920; F, 8, 367, significant at 0.00000) and the results showed that: (i) of the original five explanatory variables, membership of University committees became insignificant, and (ii) two interactive terms turned out to be significant (number of research grants X publications; p = 0.00000) and (memberships of University committees X publications; p = 0.005).c31Another pooled regression model was estimated with all seven interactive terms included (n = 376; adjusted R2 = 0.94; F, 12, 363, significant at 0.00000) and the results were identical to those of the previous two steps except that the interactive term number of grants X student load became highly ~ignificant.'~' The remaining two sub-hypotheses, 3a and 3b state that power proxies, and in particular externally based power proxies, are more important determinants of budget allocations in periods of resource scarcity compared with periods of resource availability. Examination of the detailed regressions of individual years (table 111) indicates that hypothesis 3a is not strictly supported since: (a) number of publications (as a proxy for power) was significant in all years irrespective of the extent of resource availability, (b) number of grants and representation on outside bodies were significant only in the period of resource availability, and (c) representation on University committees was significant only during the period of resource scarcity. Hypothesis 3b is not supported because the impact of some externally based power variables (representation on outside bodies and number of research grants) and of internally based power (representation on University committees) appears to have operated in the reverse direction to that hypothesized. Number of research grants and number of publications contribute to departmental research reputation. Their strong impact indicates that these dimensions of power are critical to the study of resource allocation in universities. Funds generated by departments from external sources, such as research grants, have several interesting implications. These funds are normally channelled direct to the department earning them, and hence they are outside the control of the University Centre (although an overhead charge may be levied centrally). Furthermore, they may be used by the department head as a lever with which to extract disproportionately favourable allocations from centrally controlled funds. Cohen and March observe that: Outstanding faculty at major institutions control substantial resources directly through research grants; they can and do use that control and the threat of departure (with consequent loss of funds and reputation) as a device for influencing the allocation of other resources. In recent years at major universities the teaching load in physics has rarely been as great as the teaching load in French. (Cohen and March, 1976, p. 182) 0 Basil Blackwell Ltd

1995

BUDGE?’ ALLOCATION IN A UK UNIVERSITY

327

Although both representation on outside bodies and representation on University committees were significant in the pooled regressions, their influence in the annual analysis was not persistent. ‘This finding offers weak support to the work of Hills and Mahoney (1978), Provan et al. (1980), and Hackman (1985) in relation to the impact of externally based power, and is also partly consistent with the results of Salancik and Pfeffer (1974) and Pfeffer and Moore (1980) in relation to the influence of internally based power. It does appear that power emanating from both internal and external sources has some influence on budget expenditure patterns. However, as indicated earlier, the results of the annual regressions indicate that the influence of representation on outside bodies was significant only during the period of resource availability, whereas representation on University committees was significant only during the period of resource scarcity. To examine this further, pooled regressions were performed for the two sub-periods: 1975-80 and 198184. For the first period (n = 217; adjusted @ = 0.887; F, 5, 21 1, significant at O.OOOOO), all variables except representation on University committees were significant at p < 0.0021 or less). In contrast, for the second period (n = 159; = 0.952; F, 5, 153, significant at 0.00000) all variables except adjusted representation on external bodies were significant at the 5 per cent level or below. These results do not support hypothesis 3b, and are counter to those of Hills and Mahoney (1978) as they suggest that externally based power has a predominant influence in periods of resource scarcity. It may be that during the period of resource scarcity, departmental internal power was used in the university studied here to shelter the allocation base of the departments, possibly in the widespread expectation that the sudden onset of relative scarcity of funds would be a one-off phenomenon, albeit with a long shadow. Departments with prestige reflected in representations on external bodies but with no representations on internal committees have less scope to influence the amounts allocated directly. In order to gain additional insights into the relative impact of workload and power criteria regression equations were estimated for each year by regressing the first difference in expenditure on the first differences in the explanatory variables, i.e.:

d

The results reported in table IV reveal some interesting patterns. During the years before the financial crisis, the first differences of workload (student load) and of two power proxies (number of grants and, to a lesser extent, number of publications) were significant; the first differences in representation on committees, whether internal or external to the University, were not found to be significant. In stark contrast, in the aftermath of the financial crisis (1981-82 onwards) the adjusted ps were much lower and only one variable in only one year (representation on outside bodies, 1983-84) was found to be significant. It seems then that ‘negotiations’ over budget increments (or decrements) in periods of resource availability were driven by both workload and some power criteria, but that in periods of resource scarcity most of these variables had little impact on incremental allocations. This is inconsistent with hypotheses 3a and 3b. 0 Basil Blackwell Ltd 1995

n

35

35

35

36

37

37

38

39

39

35

34

37

Years

1976 75

1977-76

1978-77

1979-78

1980-79

1981 80

1982-81

1983-82

1984-83

1984- 75

1980- 75

1984 80

Fu blicatwns -244.364 (-0.780) 1,120.159 (2.579)** 295.014 (1.149) -372.120 (- 0.506) 3,150.976 (4.286)**** -482.6 16 (- 1.596) 436.462 (0.874) I , 161.301 (1.472) 933.060 (1.490) 6,911.746 (5.229)**** 3,554.854 (4.02I)**** 770.566 (1.058)

890.68 1 (3.842)*#* 42 1.895 ( I . 109) 689.220 (4.019)**** 2,366.362 (2.240)** 1,325.009 (1.521) 978.516 (3.107)474.493 (1.168) - 296.239 (-0.408) --885.560 (- 1.027) 2,396.173 (2.855)*" 3,645.163 (4.739)*** 999.359 (1.452)

16,247.420 (3.026)*** 18,351.236 (4.027)**** 22,362.267 (5.624)**** 41,399.726 (3.942)**** 63,960.056 (6.754)**** 8,58 1.948 (2.088)** 6,043.065 ( I . 196) 1 1,466.973 (1.307) 3 1,355.337 (4.099)*#*

125,422.374 (3.808)**** 90,028.374 (3.493)*** 38,069.442 (2.709)**

Constant

student load

'I'able IV. Analysis of differences

~

-

844.133 (0.091) -70.241 (-0.012) 12,255.440 (-3.091)*** -

10,059.685 (-0.230) -60,267.044 (- 1.573) -48,809.838 (- 1.720)*

-

16,61 1.843 (-2.333)** -8,640.625 (-2.065)** 4,749.91 1 (1.255)

-

-

370.430 (0.250) 1,610.7 10 (0.432) - 1,171.014 (-0.778) - 2,12 3.427 (-0.329) 876.793 (0.141) --2,181.516 (--0.693) - 2,183.320 (- 1.169) -2,501.774 (-0.358) -2,442.655 (-0.420) ~

UniL'ersiQ committees

5,772.353 (0.342) 1,302,097 (0.110) -26,825.954 (- 1.738)* -14,337.471 (-0.320) 1 1,908.058 (0.623) --2,603.836 (-0.225) 14,393.980 (- 1.400) 8,526.979 (0.317) 312,501.978 (3.168)*%

Outside bodies

3,011.012 (2.108)** 3,999.603 (- 2.539)** 5,531.267 (6.484)*** -9,714.376 (-5.324)#** 1,914.596 (-0.877) 1,900.740 (- 2.845)*** 1,527.879 (1.283) -2,752.429 (- 1.229) 1,939.382 (0.797)

.f grants

.Yo.

Independent variables

0.463

0.750

0.780

0.344

-0.036

0.189

0.4 14

0.412

0.509

0.705

0.448

0.364

21

Ad'usted R he1

BUDGET ALLOCATION IN A UK UNIVERSITY

329

DISCUSSION

This paper has sought to examine the determinants of budget allocations as reflected in relative expenditure patterns, in a UK university in two differing periods, one characterized by relative resource availability and the other by relative resource scarcity. The results reveal that: 1. As hypothesized, an incremental budgeting system was found to be in operation during the period characterized by resource availability. 2. In some degree of departure from hypothesis 2 the incremental budgeting system continued to be used during the period characterized by resource scarcity, although it may have been elaborated by a weak form of non-incremental budgeting. 3. There was no evidence to suggest that power proxies were more important determinants of relative expenditure patterns in the period of resource scarcity compared with the period of resource availability. 4. There was no evidence to suggest that during the period of resource scarcity (availability), externally based power had a more (less) pronounced impact on budget allocations compared with internally based power. These results attest to two important phenomena relating to budget allocations in public sector, not-for-profit organizations; the first is the prevalence of incremental budgeting practices and the second is the robustness of incrementalism in the face of change situations. Evidence on the use of incremental budgeting in the University is supportive of the voluminous literature testifying to the prevalence of incrementalism in public sector organizations (e.g. Bailey and O’Connor, 1975; Davis et al., 1966; Greenwood, 1984; Hills and Mahoney, 1978; Wildavsky, 1964; Wildavsky and Hammond, 1965). But why is incrementalism so prevalent in the face of change situations? As Wildavsky has pointed out: From the time the caterpillar of budgetary evolution became the butterfly of budgetary reform, the line-item budget had been condemned as a reactionary throwback to its primitive larva. Budgeting, its critics claim, has been metamorphosed in reverse, an example of retrogression instead of progress. . . . Yet, despite these faults, real and alleged, the traditional budget reigns supreme virtually everywhere, in practice if not in theory. (?Vildavsky, 1986, p. 313) The robustness of incrementalism in the face of change situations may in part have to do with the reality of managing organizations in decline. Organizations are organic social systems that function as a totality, and hence they cannot be ‘cut back by merely reversing the sequence of activity and resource allocation by which their parts were originally assembled’ (Levine, 1980, p. 15). Many organizational activities therefore may not be sufficiently divisible to implement comprehensive budgeting. To an extent, recognition of the indivisibility of organizational activities is reflected in the range of strategies which are recommended in the literature on managing organizational decline. Such strategies are ones which aim to smooth out, rather than engage in whole mass, change such Basil Blackwell T.td 1995

330

MAHMOUI> I‘ZZAMEI, AND MICHAEI, BOURN

as limiting lower managers’ discretion, imposing a hiring freeze across the whole organization, encouraging voluntary early retirement, selling surplus assets and leasing back when needed, and so on (Levine, 1980). Managers of sub-units seemingly realize that in the face of financial distress the best they could hope for is to ‘protect the level of the budget, but not to increase it’ (Rubin, 1980, p. 177) because the likelihood of their increasing their budgets is very small. Greenwood (1984) has also suggested that the pervasiveness of incrementalism can be explained by two main arguments: its functional effectiveness and its ideological underpinnings. The functional advantages of incrementalism may be summed up by what Wildavsky (1986, p. 327) calls ‘the virtues of its own defects’: ease of calculation which makes possible the meeting of timetables despite the cognitive limitations of personnel; its capacity to accommodate a variety of policies given its neutrality to policy issues; its contribution to the engineering of conzensus given its preservation of past allocation patterns as ‘history provides a strong base on which to rest a case’ (Wildavsky, 1986, p. 327); its adaptability enshrined in its ability to perform reasonably well (though not brilliantly) on many criteria; its conformity with clear lines of authority and responsibility; and its minimization of serious dislocation of service provision (see Cyert and March, 1963; Greenwood, 1984; Miller and Freisen, 1980; Normann, 1971; Starbuck, 1983; Wildavsky, 1975, 1986). Moreover, incrementalism facilitates operation in the face of multiple, conflicting goals and ambiguous means- ends relationships by decoupling individual interests, which are already reflected in incremental systems (Beyer, 198 I), and by focusing attcntion on existing programmes rather than on strategic reorientations (Normann, 1971). Ideological underpinning is represented by the organization’s underlying beliefs, values, and assumptions. Thus, in their study of British government, Helco and Wildavsky (1974) noted that there was a close correspondence between the values of thc ‘village community’ and the decision rules of incrementalism; the budget system focused on handling potentially disruptive disagreements, and consulting and reaching consensus on budget allocations to avoid damaging conflict. Incrementalism fosters consensus by engendering flexibility through norms such as ‘fair shares’, promise of future concessions, and ability to allocate increments in favour of disadvantaged Lgroups (Starbuck, 1983; Wildavsky, 1964, 1978; Wildavsky and Hammond, 196.5). Hedberg (1981) and Starbuck (1 976, 1983) have argued that organizations interpret their environments in terms of organizational ‘myths’ or ideologies which incorporate assumptions concerning the organization’s appropriate domain and the proper way of organizing. Prevailing myths and ideologies are robust, and as a consequence they hinder organizational change. Miller and Friesen (1980) have noted that ideologies develop ‘momentums’ which are difficult to arrest even if a significant gap develops between real world events and the organization’s mapping of those events. Because environmental conditions are interpreted through closely held ideology/myth, the ‘enacted’ environment may fail to capture changing conditions affecting organizational pcrformance (Mitroff and Kilmann, 1976). There is also the crucial issue of power relations and the extent to which different budgeting forms impact on these relations. Incremental budgeting systems in the main aim to take budgetary decisions out of the realm of politics (and power) and convert them into bureaucratized procedures. In contrast, compre(0Basil Blackwell

I.ld 1995

BUDGE?' AL,I,OCA'I'ION IN A UK UNIVERSITY

33 1

hensive budgeting systems seek to challenge and fundamentally change existing power relations. Yet such systems by themselves cannot do so because of the dynamics of power relations between budget constituencies which institutionalize previous allocation patterns. Thus, Wiedenbaum (1970, p. 233) has suggested that the process of resoiirce allocation is: 'hardly deliberate and systematic choice among alternatives that economists try to envision. Rather, it is a fragmented and cornpartmentalized affair. Many of the key decisions are not made during the budget process or within the budgetary framework at all.' Similarly, plausible arguments drawn from the case study used in this paper may throw some light on the phenomenon of the persistence of incrementalism. Such explanations may, of course, be both partial and complementary. Thus, our personal observations at the University studied suggest that: (i) financial devolution (as distinct from incremental budgeting) was felt to fit the University's culture by allowing a greater degree of localized, academic control which observed unavoidable financial constraints; (ii) the method for implementing devolution was incremental and this was seen as an 'eflective response to complex problems', particularly because it allowcd for serious reconsideration of baselines at the level of cut (over 6 per cent p.a. in real terms for three years); (iii) in doing this it achieved continuity by sustaining power relations in several dimensions between academic groups, between academic and other budgetary groups, between Senate (ie. academe) and Council (i.e. external lay governance), and between the University as an employer and its staff of employees (who were the budgetary groups); (iv) the balancing act between employer and employees became enshrined in a policy statement that compulsory redundancy would be a policy only of the last resort, a position reached only after much serious aggravation, but immediate measures were taken to impose a freeze on all-posts falling vacant and to promote voluntary early retirement. This suggests that the contrast between incremental and comprehensive budgeting should not be drawn too distinctly. In the case of the, university under study, the standard incrementalist approach had always provided for an ad hoc baseline review for any budgetary group facing changed circumstances. At the time of the transition from relative plenty to relative scarcity in this study,,this procedure appears to have been extended more widely to achieve an acceptable new set of baseline relativities to be maintained by subsequent incrementalism.

This paper had two main purposes; the first was to test the widely perceived assumption of the prcvalenre of incremental budgeting as a resource allocation mechanism in periods of resource availability; and the second was to explore whether incremental budgeting may be supplanted to any significant extent by a more comprehensive form of budgeting in periods of resource scarcity. Additional hypotheses were also formulated to examine the roles of power variables as against workload variables, and externally based power as against internally based power during periods of resource scarcity compared with periods of resource availability. Data relating to a n above average sized UK university were collected and analysed to test these hypotheses. $? Basil BlackweU Ltd 1995

332

MAHMOUD EZZAMEL AND MICHAEL BOURN

‘The results are supportive of the prevalence of incremental budgeting in periods of resource growth/availability. Expenditures in the previous year succeeded in explaining over 99 per cent of expenditures in the following year. It is plausible to suggest that both workload measures and power proxies operate through last year’s budget, i.e. much of the information relating to both workload and power variables is embedded in the last year’s budget. This seems to be supported by the very high and significant correlation coefficients between these variables and last year’s budget. One possible implication of this is that once relative power positions have been bargained for they tend to be stabilized for a number of years in order to avoid the disruptive consequences of continuous bargaining. The results relating to the period of resource scarcity, however, showed that there was no significant shift from the incremental practices prevalent in the period of resource growth. Last year’s budget continued to dominate as the major explanatory variable of the following year’s budget allocations. These results are supportive of much of the literature on organizational change, which indicates that prevailing ideologies are extremely robust, and that they develop ‘momentums’ which are difficult to arrest. Moreover, what are commonly called incremental and other budgeting styles simply reflect variations in the depth and frequency of base-line review, which is the key activity involved. Finally, the results reported here do not support the hypothesis that power proxies have a greater impact on budget allocations in periods of resource scarcity compared to periods of resource availability. Budget allocations in both periods of resource availability and scarcity were explained by a vector of workload and power variables. Similarly, contrary to some previous results, the impact of internally based power on budget allocations was more pronounced during the period of resource scarcity whereas externally based power had a greater effect on allocations during the period of resource availability. The analysis presented in this paper can be extended in several ways. First, more refined proxies for size or workload and power criteria need to be developed to overcome some of the potentially problematic classifications used in this paper and in previous research. Second, proxies for outputs, as opposed to inputs or throughputs, need to be developed in order to sharpen the focus of this area of research. Third, other variables can be included in estimating the regression equations, particularly those related to subjective measures of power, and the extent of paradigm development. Fourth, the results of studies of this kind can be complemented by more qualitative and interpretative studies which are best suited to capturing not only the outcome but also the process of organizational change.

NOTES

*The authors wish to acknowledge the research assistance of Alex Bourn and Judith Brodie and the advice of David Holmes in setting up the database. Earlier drafts of this paper were presented at staff seminars at the Universities of Sheffield, Leeds and StrathClyde and at the American Accounting Association Conference, August 1992, Washington. We are grateful to the participants of these seminars and conference and to Mike Basil Blackwell Ltd 1995

BUDGET ALLOCATION IN A UK UNIVERSITY

333

Gibbins, Hilary Ingham, David Peel, Steve Thompson and Rob Watson for the numerous constructive comments they made, and to the three anonymous reviewers.

[I] There are some writers who view the budgeting process as a purely economic exercise devoid of any political character (e.g. Smithies, 1955), but this view has for some time now been strongly challenged. [2] The literature on power is fairly controversial. For more radical views on power see Foucault (1980), Lukes (1974) and Clegg (1989). [3] All the pooled regression estimates reported in this paper were re-estimated using an unbalanced panel estimator (Arellano and Bond, 1991) and the results remained robust throughout. [4] Tests were performed to assess the extent to which model parameters were stable over the period of study. These tests dealt with differential intercepts, differential slope vectors, and overall homogeneity of the model (Johnston, 1972). The F value for each of these tests was significant at below the 1 per cent level, indicating overall instability in the intercepts and slopes of the regressions as well as lack of homogeneity in the model. These results are supportive of the initial impressions gained from examination of the results of annual regressions (see table 111).

REFERENCES

AREILANO, M. and BOND,S. (1991). ‘Some tests of specification for panel data: MonteCarlo evidence and an application to employment equations’. R e v h of Economic Studies, 58, 277 97. J. Z. (1988). ‘Measuring centrality: a note on Hackman’s ASHAR,H. and SHAPIRO, resource-allocation theory’. Administrative Science Quark+, 33, 275-83. BAILEY, J. J. and O’CONNOR, R. J. (1 975). ‘Operationalizing incrementalism: measuring the muddles’. Public Administration Review, Uanuary-February), 335-56. S. W. and GORDON, G. (1 966). ‘An entrepreneurial theory of organization, part BECKER, 1 : patterns of formal organizations’. Administratiue Science Quarter&, 11, 3 15-44. BEYER,J. M. (1981). ‘Ideologies, values and decision making in’ organizations’. In Nystrom, P. C. and Starbuck, W. H. (Eds), Handbook of Organizational Design. New York: Oxford University Press. K. E. (1 963). The Organizational Revolution. New York Harper. BOULDING, ROULDING,K. E. (1975). ‘The management of decline’. Change, June, 8-9 and 64. BRINKMAN, P. T. and LESLIE,L. L. (1986). ‘Economies of scale in higher education: sixty years of research’. The Review I$ Heher Education, 10, 1, 1-28. CLECG,S. R. (1989). Frameworks of Power. London: Sage. COHEN,M. D. and MARCH,J. G. (1976). ‘Decisions, presidents, and status’. In March, J. G. and Olsen, J. P. (Eds), Arnbiguig and Choice in Organizations. Bergen, Norway: Universitetsforlaget, 1 74-205. CROZIER, M. (1964). ‘The Bureaucratic Phenomenon. Chicago: University of Chicago Press. CYERT,R. M. and MARCH, J. G. (1963). A Behavioral ;Theory of the Firm. Englewood Cliffs, NJ: Prentice-Hall. DAVID,D. M. and KANTOR, P. (1980). ‘Political theory and transformations in urban budgetary areas: the case of New York City’. In Marshal, D. R. (Ed.), Urban P o k y Making. London: Sage. M. A. H. and WILDAVSKY, A. (1966). ‘A theory of the budgetary DAVIS,0. A., DEMPSTER, process’. American Political Science Review, 60, 529-47. DE GROOT,H., MCMAHON, W. W. and VOI~KWEIN, J. F. (1991). ‘Cost structure of American research universities’. 7 3 e Review ofEconomics and Statistics, LXIII, 3, August, 424-3 1.

0 Basil Blackwell Ltd

1995

334

MAHMOUD ELZAMEI, AND MICHAEL BOURN

EZZAMEL, M. and BOURN, M. (1990). ‘The roles of accounting information systems in an organization experiencing financial crisis’. Accounting, Organizations and Socieg, 15, 5, 399-424. EZZAMEI., M. and HART, H. (1987). Advanced Management Accounting: An Organizational Emphasis. London: Cassell. FOUCAULT, M. (1 980). Power/Kinowhdge, Gordon, C. (Ed.), Brighton: Harvester Press. GREENWOOD, R. (1 983). ‘Changing patterns of budgeting in English local authorities’. Public Administration (Summer). GREENWOOD, R. (1 984). ‘Incremental budgeting: antecedents of change’. Journal of Public Policy, 4, 277 -306. HACKMAN, D. J. (1985). ‘Power and centrality in the allocation of resources in colleges and universities’.Administrative Science Quark?$, March, 30, 61 77. HEDBERG, B. (1981). ‘How organizations learn and unlearn’. In Nystrom, P. C. and Starbuck, W. H. (Eds), Handbook of Organizational Des@n, Vol 1. New York Oxford University Press. A. (1974). The Private Government $Public Money. London: HKLCO,H. and WIIDAVSKY, Macmillan. HICKSON, D. J., HININGS, C. R., LEE,C. A,, SCHNKK, R. E. and PENNINGS, J. M. (1971). ‘A “strategic contingencies” theory of intraorganizational power’. Administrative Science Quark& 16,216-29. HILIS,F. S. and MAHONEY, T. A. (1978). ‘University budgets and organisational decision making’. Administrative Science @ark+, 23, 454 -65. J. (1972). EconomehiG Methods, 2nd edn. London: McGraw-Hill. JOHNSTON, LEVINK, C. H. (1980). ‘Organizational decline and cutback management’. In Levine, C. H. (Ed.), Manuging Fiscal Stress: The CrisZr in the Public Sector. New Jersey: Chatham House, 13-30. S. (1 974). Power: A Radical Viw. London: Macmillan. LIJKES, L. G. (1989). Pennanent!y Failing Organizations. Newbury Park, MEYER,M. W. and ZUCKER, Cal.: Sage. MIILKR,D. and FRIFWN,P. H. ( I 980). ‘Momentum and revolution in organizational adaptation’. Academy of Management Journal, 23, 591 6 14. R. (1976). ‘On organizational stories’. In Kilmann, R. et al. MITROFF,1. and KIIMANN, (Eds), The Managment of Organizational Deszgrl: Vol 1. Strategies and Imphentation. New York: Elsevier, 184-208. NORMA”, R. (1 97 1). ‘Organizational innovativeness: product variation and reorientation’. Administrative Science Quarterb, 16, 203 15. J. (1973). 7he Fiscal CGiS ofthe State. New York: St Martin’s Press. O’CONNER, PERROW, C. (1970). ‘Departmental power and perspective in industrial firms’. In Zald, M. N. (Ed.), Power in Organizations. Nashville: Vanderbilt University Press, 59-89. PFEFFER, J. (1 978). ‘The micropolitics of organizations’. In Meyer, M. W. (Ed.), Environments and Organizations. San Francisco: Jossey-Bass, 29 50. PFEFFER, J. and LEONG,A. (1977). ‘Resourcc allocation in United Funds: an examination of power and dependence’. Social Forces, 25,637 53. PFEFFER, J. and MOORE,W. L. (1980). ‘Power in university budgeting: a replication and extension’. Administrative Science Quark+, 25,637 ~ ~ 5 3 . PFKFFER, J. and SALANCIK, G. R. (1974). ‘Organizational decision making as a political process: the case of a university budget’. Administrative Science Quark+, 19, 135.51. J. and SALANCIK, G. R. (1977). ‘Administrator effectiveness: the effects of advoPFEFFER, cacy and information on achieving outcomes in an organizational context’. Human Relations, 30, 2, 641 56. PHYRR, P. (1973). zero-Based Bu4eting. New York: Wiley. ~

~

c) Basil Blackwell Ltd 1995

PONUY,L. K. (1970). ‘Towards a theory of internal-resource allocation’. In Zald, M. N. (Ed.), Power in Oyunilahons. Nashville: Vanderhilt University Press, 270 - 3 I 1 . P o m v , L. K. and BIRNWKG, J. G. (196Y). ‘An expcrinicntal study in the allocation of financial resources within small hierarchical task ,groups’. Administrative Science Quart&, 14, 192~-201. PKOV~U K., G., RKYIIR,J. M. and KKUYIXOSC:H, C. (1980). ‘Environmental linkages and power in resource-dcpendencc relations between organizations’. Administrative Science QuarterJy,25,200 25. decision making under conditions of reduced RURIN,I. (1 980). ‘Universities in strc resources’. In I,c:vinc, C. H. (Ed.), Manag@ Fiscal Stress: ?he Cniis in the Public Sector. New Jcrscy: Chatharn house, 167--78. SAIAVCIK, G. K.and PFEFFER, J. (1974). ‘The bases and use of power in organizational dccision making: the case of’ a university’. Adnlznirtrative Scknce Quarte~b,19, 453 73. SCHICK, A. (1 980). ‘Budgetary adaptation to resource scarcity’. In Ixvine, C. H. and Kubin, I. (Eds), Am1 Stress and Public Policy. London: Sage. SCCYI-I., P. (1984). 7 h p Crisis ofthe U7iiuersi9. London: Croorn Helm. SCOIT,W. G. (1 974). ‘Organization theory: a reassessment’. Academy 4Munqement Journal, Junr, 242 53. SINGI.I~;I~ON, U. W., SMII’H, B. A. and C:I.F.AVELWU, .J. R. (1980). ‘Zero-based budgeting in Wilmington, Delaware’. In Levinc, C. H. (Ed.), Munaging Fiscal Stress: The Ckisis in the Public Sector. New Jersey: Chatham House, 179 -93. SMITHIES, A. (1955). 7?ie Bu&tary Process in the IJnited States. New York: McGraw-Hill. STARBI:CK, W. H. (1976). ‘Organizations and their environments’. In Dunnette, M. D. (Ed.j, Handbook vf Industnu1 and Oyanizational Pgxhnloo. Chicago: Rand McNally, 1069 123. STARBI.CK, W.H. ( I 983). ‘Organizations as action generators’. American Soczologzcal &view, February, 9 1 102. THOMPSON, J. D. ( 1 967). Chganizalions in Action. New York: hlcGraw-Hill. WIMXNBAUM, M. (1 970). ‘Institutional obstacles to rrallocating government expenditures’. In Havernan, R. and Margolis, ,J. (Eds), Public Expenditure and P o k y Anabsis. Chicago: Markham. WII.I)A\.SKY, A. (1964). cfie Politics oJ the Budgetary Process. Boston, Mass.:,Little, Brown. WII.DAVSKY, A. (1975). Budgeting: A Comparative Thee?? of the Budgetary Process. Boston, Mass.: Littlc, Brown. W I l I ) A V S K Y , A. (1978). ‘A budget for all seasons?: why the traditional budget lasts’. Public Administration RevWw, Novembcr/December, 50 1 9. WII.IIA\~SKY, A. (1 986). ‘Buiiyeting: A ComparativP ‘Theory OJ‘ the Budgetary Process, 2nd edn. New Rrunswick and Oxford: Transaction Publishers. WILIMVSKY, A. and HAM.ZIONI>, A. (1 965). ‘Comprehensive vcrsus incremental budgeting in the department of a,yiculture’. Administrative Science Quarterb, 10, 32 1--46. WIIJ.IAMSON, 0. E. (1 Y85j. ‘The Economic Institutions oj‘ Capitalism. New York: Free Press. WRONG,D. H. (1 968). ‘Some problems in defining social power’. American Journal of Sociology, 73,673 81. ~

f; Basil Blarkwdl I.td 1995