Review of Sharpe, L.J.. & Newton, K.: Does Politics Matter? The Determinants of Public Policy. Oxford: Clarendon Press, 1984.

Much if not most of what is denoted by 'public policy analysis' is qualitative reasoning. If at all based on empirical argument, the evidence employed is typically derived from case studies. Too little effort has been devoted to quantitative reasoning aimed at testing explicit models by means of econometric techniques. Modelling the policy process in terms of clear cut equations based on hard data would no doubt bring more rigor and clarity to this branch of political science, where it is often far from clear what problem is being discussed, which solutions are being suggested and how far the evidence supports the proposition. The main strength of the Sharpe-Newton public policy book is to be found in its skillful employment of quantitative reasoning. Various models are derived from theoretical arguments that are tested against a precisely defined set of data. And great effort has been made to display to the reader how the findings are arrived at and interpreted. The Sharpe book resembles the excellent Danziger public policy study from 1978; both books employ quantitative reasoning to solve problems about the nature of public policy and they refer to the same empirical reality. However, their findings are as different as their public policy analysis is similar.

The output approach deals with whether politics matter in determining public policy. In their study of service expenditures for all top tier local authorities in England and Wales over a 16 year period (1956/7 to 1972/3), Sharpe and Newton set themselves firmly in the 'politics are important' camp. Rejecting the transmission model and the 'black box' view of the policy-making process and exposing the largely futile and meaningless nature of the politics versus the socio-economic environment battle, they develop one approach based on the primacy of politics.

Drawing on Tore Hansen's conception, they point out that socio-economic variables are simply inert descriptive statistics until they are transformed into inputs by the political actors who are, in turn, influenced in their perception and selection by their values, ideologies and assumptions. Arguing from this perspective, Sharpe and Newton maintain that policy outputs should be viewed as a product of the political system. They modify the traditional output method by the incorporation of more refined political variables, particularly in relation to party control and party systems, into their regression analysis. Their findings suggest that political parties are important in determining local service outputs; that there is a relationship between Left party majority and higher expenditure, particularly with regard to redistributive and ameliorative services; and that,,

contrary to previous findings, party competition, at least in the British situation, does not lead to higher expenditures than does party monopoly. Their study highlights the need for rigor in the definition and selection of political variables and reveals how, in past studies, the use of cross-sectional and aggregate data has masked the importance of politics.

The authors undertake a more rigorous and detailed examination of the political variables for inclusion in their regression analysis than is commonly the case in output studies. Emphasizing the need to include relevant and meaningful political variables, they suggest a number of refinements to improve the effectiveness of the political variables. These include the use of time series data which minimize the possibility of aberrations, offers the possibility of constructing a more realistic measure of political control, length of control, and permits the recognition of the fact that policy change takes time. Although there are problems involved in using longitudinal data such as ensuring that the unit of analysis remains the same, the advantages gained by using such an approach seem to outweigh the disadvantages. They graphically illustrate the dangers of only looking at total expenditures as a measure of party effect and introduce the use of disaggregated data in order to explore the extent to which parties discriminate between different services. They introduce a different perspective on party competition, arguing, at least in the British context, that parties will try to implement party ideology when in control, only modifying this when forced to do so by close competition from a rival party. Considerable attention is paid to the party effect on policy outputs and both urban (county boroughs) and rural (county councils) are included in the analysis in order to broaden the range of party systems considered.

Two analyses are carried out for both types of authority. The first, consisting of simple correlations between the political variables and service expenditures, shows that generally there is a relatively strong association between party colour and expenditure variation (although the results are somewhat patchy). A regression analysis using the most powerful social, economic and political variables, produces a general, if weak, confirmation that there is an association between political variables and expenditures as compared with socioeconomic variables.

The left party effect and the party system effect are then explored in more depth. Local authorities are divided into five categories according to strength on the council. Their results produce an almost perfect pattern of Labour high spending and Conservative low spending. From detailed analysis of a selected group of services they find a direct positive relationship between the degree of left control and per capita expenditure for redistributive services and the reverse for the two non-redistributive services examined (highways and police). A closer examination of the political control of the county borough over the length of the study period yields the following results: authorities which stayed

continuously under Labour control increased their spending more than authorities under continuous Conservative control; authorities which switched from Labour to Conservative cut their spending in relative terms, but more slowly; Conservative authorities which changed to Labour increased their spending more rapidly. The conclusion drawn from this is that there is a clear party effect on spending and that the Labour high spending effect is stronger after a change of control than the Conservative low spending effect.

This is the first theme of the book; the second theme is quite different — perhaps the lack of similarity between these two basic arguments is larger than the authors are aware of. The second theme follows on the introduction of another explanatory dimension, the 'unitary model' which relates to a city's holistic characteristics stemming from its role in the wider economic system. The authors then apply a number of conventional approaches to their data. These include the Tiebout hypothesis, the demographic model, factor and cluster analysis and the incrementalist approach. However, none of these methods achieve more than, at best, a very modest success.

Following the failure of these methods to explain expenditure variations, Sharpe and Newton elucidate their own research strategy. They begin by ranking authorities from the highest to lowest spenders on each service and then consider what these 'leagure tables' show about variations in local spending. This simple device produces a number of interesting observations relating service expenditures to a city's position in the urban hierarchy, its main economic function and its geographical location, as well as to cultural and political factors. The next stage involves the development, in relation to the urban local authorities, of two aspects of the 'unitary model' based on a city's function as a service centre and as a city type. The former draws on central place theory as a means of assertaining a city's position in the urban service hierarchy. A city's role as a central place is seen as impacting on local service expenditures: the higher a local authority's position as a central place, the more its per capita expenditure is likely to be on specialised services such as art galleries, museums and special education facilities. Demands are also placed on other local services such as police, highways and public transport as a result of people visiting the city to avail themselves of its central place facilities (both public and private).

The second explanatory strand, city type, relates to a city's main economic role. Existing city type classifications were found to be unsatisfactory and, using stringent criteria, Sharpe and Newton construct their own typology of nine city types. This is notable in that, despite being comprehensive, it contains only one 'mixed bag¹ or residual category. A third factor, tied in with the previous, is the nature of a city's location. Whether a city lies within an urban or rural setting has implications for service expenditures, reducing or raising costs depending on the type of service.

Although the combination of a city's function as a service centre and a city type performs well in explaining expenditure variations on indivisible services, it is essentially an 'urban' explanation. Attempts to develop a similar model for the county councils meet with little success largely due, it would appear, to their more heterogeneous nature. This raises the issue of how well this approach would function in relation to the new local authorities which, in many cases, are of a more mixed urban/rural character than the old small authorities. In 1974 the number of authorities in England and Wales was reduced from 1392 to 422. A further question is raised by this approach, namely, what happened to the importance of politics? The unitary model appears to ignore the question of politics altogether, basing its explanation of variance on a city's role and importance as a central place and its main economic function. There would seem to be a danger that this model reiterates the same kind of determinism (albeit of a more geographic/economic nature) for which the authors have so rightly criticized the traditional output approach.

It would seem possible to treat a city's holistic functions in the same way as its socio-economic structure, i.e. as contextual variables imposing limitations on the political decision-makers' scope for choice but which, nevertheless, have to be perceived and translated into inputs by the political actors. Thus, within the parameters set by these constraints, possibilities for choice will exist and as decision-makers are not value-free, objective mechanisms simply translating contextual factors into policy inputs, their choices will be affected by the values, beliefs and ideologies they hold.