R.J. Johnston and C.J. Pattie, "The Impact of Spending on Party Constituency Campaigns at Recent British General Elections," Party Politics , 1 (April, 1995), 261-273.

First Paragraph:
The funding of political parties is once again on the political agenda in the UK and has been the subject of an investigation by the House of Commons Home Affairs Committee (1993, 1994). As with most academic work on this topic (for example, Pinto-Duschinsky, 1981; Bogdanor, 1982), virtually all of the Committee's attention focused on the funding of parties nationally and on the national general election campaigns. Although very substantial sums are spent on the separate constituency campaigns( 10,168,462 in 1992 in Great Britain), commentators almost entirely ignore them, believing that they are largely irrelevant to the election outcome. Butler and Kavanagh (1992: 244-5), for example, argue that data in the returns of candidates' election expenditure are of dubious value ('creative accounting...is universally acknowledged to occur in expense returns') and that the consequences of spending variations are minimal - 'it is hard to locate evidence of great benefits being reaped by the increasingly sophisticated and computerized local campaigning'.

Figures and Tables:
Table 1: Constituency campaign spending as a percentage of the maximum allowed: summary statistics for all constituencies in Great Britain.
Table 2: Constituency campaign spending as a percentage of the maximum allowed, according to constituency marginality.
Table 3: Parameters of the regressions of the inter-party vote percentage ratios at successive elections.
Table 4: Values of R2 after each stage of the three-stage regression modelling.
Table 5: Regression coefficients for spending variables at stage III in the models.
Table 6: The simulated impact of parties spending to the maximum allowed in constituency campaigns.

Last Paragraph:
The conventional wisdom is that such campaigning is largely ritualistic only, having no substantial impact on the election result. Statistical analyses reported here counter that view. First, it is clear that parties are rational in their campaigning activity: they raise and spend more money for the campaign in the seats that they are defending and in the more marginal constituencies. Second, the pattern of activity indicated by spending levels is clearly related to the distribution of votes: in general terms, the more that a party spends (relative to the maximum allowed) the better its relative performance. Finally, because of variations in how much parties spend, increasing expenditure to the maximum could influence the outcome in a not insignificant number of seats. The ritual, it seems, has both purpose and consequence.