ABSTRACT
This article analyzes how extensively European Union local governments are using their websites to disseminate financial information in order to evaluate whether electronic-government (e-government) is promoting convergence towards more accountable local governments. We also posit and test various hypotheses about the influence of internal and contextual factors on Internet financial reporting (IFR) practices. Results show that the public administration style, the size of the city, and the audit of financial information by private firms are significant explanatory factors of IFR practices. Our findings also suggest that multilateral organizations are overly optimistic about the possible convergence in transparency and financial accountability through the use of common modes of IFR. That is, the introduction of information and communication technologies without the corresponding institutional reform is leading to limited success of IFR.
ACKNOWLEDGEMENTS
This study has been carried out with the financial support of the Spanish National R&D Plan through research projects SEJ2007-62215-ECON/FEDER and ECO2010-17463 (ECON-FEDER).
Notes
Note: Hypothesis: There is no difference between the four public administration styles.
Hypothesis refused at 5% significance level.
*p < 0.05; **p < 0.01.
Note: VIF = Variance inflation factor.
*p < 0.05; **p < 0.01.
In this article we adopt the typical definition of transparency based on the principal–agent model. Heald (Citation2006, 27) refers to it as “transparency downwards”: the “ruled” can observe the conduct, behavior and/or “results” of their “rulers.” Similarly, the Oxford Dictionary of Economics defines transparent policy measures as “[p]olicy measures whose operation is open to public scrutiny. Transparency includes making it clear who is taking the decisions, what the measures are, who is gaining from them, and who is paying for them. This is contrasted with opaque policy measures, where it is hard to discover who takes the decisions, what they are, and who gains and who loses. Economists believe that policies are more likely to be rational if they are transparent than if they are opaque.”
For a thorough discussion of the terms openness, transparency, and accountability, see Demchak, Friis, and La Porte (Citation2000), La Porte, Demchak, and De Jong (Citation2002), Wong and Welch (Citation2004), and Hood and Heald (Citation2006).
These countries represent more than 85% of the EU population.
Guthrie, Olson, and Humphrey (1999) use this term to refer to the accounting and financial reforms introduced in public sector entities during the 1980s and 1990s (e.g., cost accounting, accruals, GPFS, consolidation, performance indicators, value-for-money audits, and so on).
For 41 cities we found the city budget data on their websites. We sent an e-mail to the remaining 34 cities (where an e-mail address or comment box was available) and, in this way, we obtained budget data for 16 additional cities. As a result, we have 18 missing values for this variable.
This index measures the degree of corruption as seen by business people and country analysts, and ranges between 10 (highly clean) and 0 (highly corrupt). As a result of the definition of this variable, we expect a positive relationship between this variable and the IFR scores; that is to say, high levels of corruption (low scores in the corruption perception index variable) would imply low levels of IFR scores.
All websites were analyzed by one of the authors with previous experience in analyzing local government websites.
See the IASB Conceptual Framework and the FASB Statement of Financial Accounting Concepts No. 2.
According to Hung (Citation2001), an equal weighting method can be used when there is no well-defined theory for other weighting methods. He notes that the importance of an accounting standard varies across countries but sees no reason why the equal weighting will bias the results.
For further details about these techniques see Serrano, Mar Molinero, and Bossi (Citation2003).
MDS maps reduce an ‘n’-dimensional space—86 dimensions in our case, because we have used the 86 items analyzed—to just two dimensions. The objective of MDS is to describe, geometrically, the relationships existing among different objects (local governments) by providing a map that depicts the position of subjects/objects according to the distances or proximities between them. Therefore, local governments are organized in a map in such a way that the distance between them is an indicator of the degree of relationship.
Pro-Fit analysis is a technique closely related to multivariate regression analysis since it attempts to relate the position of an object (local government) in the configuration to the values of the variables (financial reporting items) for this object. If a variable is related to the position of the object in the MDS configuration, there is a function that relates the variable value to its position in space (Serrano, Mar Molinero, and Bossi Citation2003). Following this reasoning, 84 ordinary least squares regressions were performed. (The two excluded variables are variable 69 Provides link to or text of public information law or regulation, since Luxembourg does not have a law of this kind and, therefore, we have 5 missing data for this variable, and variable 77 Database of financial reports, since no local government presents this feature on its website).
See Pina, Torres, and Yetano (2009).
In the MDS analysis, the value of Kruskal stress in six dimensions was 0.121, which could be described as “fair” (Kruskal Citation1964). To visualize the map, it is necessary to work with projections of the map onto pairs of dimensions. Figure 1 shows the projection of the map onto dimension 1 and dimension 2.
To determine whether multicollinearity was a problem, we checked variance inflation factors (VIF). Greene (Citation2003, 58) suggests that an individual VIF value higher than 20 is cause for concern. The VIF coefficients are reported in Table ; the average VIF for all independent variables was 1.95, and none individually exceeded 3.0. Therefore, it is unlikely the results are distorted by multicollinearity.
As can be seen in Table , only 57 cities have been included in the regression analysis. The reason is that the variable “budget per inhabitant” has missing data for 18 cities. We have replicated these regressions excluding the variable “budget per inhabitant” in order to include the 75 cities analyzed and similar results have been obtained.