The empirical analysis was concerned to examine three propositions about the impact of participation. The first concerned the impact on the service delivered to the consumer, the second concerned the impact an attitudinal outcomes relevant to the bank and the third concerned the relationship with the performance of the customers business. In general, these propositions are evaluated using simple multiple linear regression with the dependent variable being the outcome which is of interest and levels of participation entering as part of the set of explanatory variables. With proposition 1 in mind , the first set of findings concern the impact of participation on the terms and conditions on which finance is provided.
Table 3 summarises the results of the regression analyses for rate of interest paid (points over base) and collateral ratio on overdraft limit. These two variables were selected as being key aspects of the lending decision. In addition to the measures of participation a number of firm specific variables were also incorporated where these might be expected to be relevant (eg size, age, profitability, self perceived risk etc). The model for interest rates (points above base) is significant with the variables measuring the absence of fear and the degree of monitoring both tending to have a negative impact (ie reducing interest rates payable). Given the role of interest rate in accommodating risk, these results suggest that firms with better personal interaction and better monitoring may be able to increase both the volume and quality of information available to bankers and thus benefit from lower prices as a result of more accurate assessments by the bank of the risk potential of the business. In this context it may seem surprising that information sharing does not have an impact. However, this may reflect the fact that those firms in the model are all firms with overdrafts (lines of credit) and the nature of the overdraft is such that active sharing of information is expected.
Perceptions of positive behaviour by the bank also tend to be associated with lower rates of interest. Other variables which are significant include size of overdraft, size of business (as measured by sales revenue), age of business (which proxies the length of the banking relationship and the businesses track record) and profit. The model for collateral ratio is significant overall but with very poor explanatory power and none of the measures of participation appear to have any significant impact on this particular dimension of the core product.
With respect to both propositions 1 and 2, it is desirable to examine certain other outcome variables For customers, these outcomes are service quality while for banks, these outcomes are satisfaction, perceived value and loyalty. Regression models for the first three of these are reported in table 4. In the case of quality, the measure used is an aggregate3 over all the dimensions described earlier although the results for the five individual dimensions provide similar results
Once again, we find that the participation effect is significant with positive behaviour being the most important element of bank behaviour and the absence of fear/personal interaction being the most relevant aspect of firm behaviour.
Finally, in relation to proposition 2 the relationship between behavioural loyalty and participation is examined. As the level of bank switching in the UK is low, a proxy variable is used as a measure of behavioural loyalty and this proxy is whether or not the respondent had considered changing bank in the previous year. Mean scores on the participation variables were compared for potential switchers and non-switchers. the results of this comparison are shown in table 5.
Once again, it is apparent that participative behaviour produces more positive outcomes and that from the perspective of the banks, firms involved in participative relationships are less likely to consider switching bank. Indeed the only aspect of participative behaviour which does not appear to differ between the two groups is the monitoring activity of the firm.
Thus far, there is evidence to suggest that banks may benefit from participative relationships in the form of greater customer satisfaction and retention. Equally, there is evidence to suggest that customers benefit through an enhanced quality of service and through more favourable financing terms and conditions. If, as argued earlier, finance is a potential constraint on business performance then there might be grounds for suggesting that firms in more participative relationships may be less constrained and thus business performance may be better when compared with firms in less participative relationships. Formally testing such a problem is difficult, particularly in a cross sectional data set, because of the time lags involved in the relationship and the problems of determining the direction of causality.
These problems notwithstanding, a preliminary attempt to identify associations between participation and performance was undertaken. The two performance measures were profit as a percentage of sales revenue for the last accounting year and rate of growth measured on a six category ordinal scale4. First, Table 6 reports partial correlations of participation measures with profit (controlling for age) and age (controlling for profit). Subsequently Table 7 reports differences in the distribution of firms by relationship type across a compressed 3 category growth scale. In this case the relationship categories were identified using a simple k-means clustering procedure, with four clusters emerging as the most logical solution.
It is apparent that the relationship between participation and profitability is virtually non-existent with the exception of a weak correlation between the absence of fear in the relationship and the level of profit. A similar pattern emerges with respect to growth rate. Consideration of differences across growth categories by relationship type suggests a slightly clearer pattern in that relationships which are characterised by high levels of participation from both parties are associated with higher proportions of firms in fast growth categories while the reverse holds for relationships characterised by low levels of participation by both parties.
3The first two measures were based on a 5 point scale and although this raises questions about the applicability of OLS, inspection of the residuals did not suggest significant problems. The aggregate service quality measure is simply an importance weighted average across the 18 dimensions of service quality.
4Strictly speaking a pearson correlation is not appropriate for data of an ordinal nature but is used in this instance to provide a simple and easily available indicator.
© 1997 Babson College All Rights Reserved
Last Updated 06/06/98