Performance Management, Management Quality, and Government Performance in OECD Countries: An Empirical Analysis

Data Analysis

Table 4.1 presents the means, standard deviations, and correlation coefficients for the research variables. We conducted a path analysis of the relevant variables to determine the sequential relationships among the sets of hypotheses presented earlier. The factor analysis was conducted through Structural Equation Modeling (SEM) using SPSS-AMOS software, which also allowed us to compare several explanatory models and test for mediation.

Main Findings

Table 4.1 shows that on average the researched countries adopted performance management practices to a medium extent (mean = 0.55 on a percentage scale; sd = 0.13), but undertook decentralization and

Table 4.1 Multiple correlation matrix and descriptive statistics for the research variables (Pearson’s r)


Mean (SD)




4 5

1. Performance management (PM)


0.55 (0.13)

2. Management quality and government effectiveness (MQGE)


0.67 (0.13)


3. Decentralization (DC)


0.64 (0.10)



4. Resource availability (RA)


0.23 (.05)




5. Coordination initiatives (COOR)


0.81 (.22)





6. Trust in government (TRUST)


0.42 (0.14)





NS, not significant; *p < 0.05; **p < 0.01

coordination initiatives more intensively (mean = 0.64, sd = 0.10 and mean = 0.81, sd = 0.22, respectively). These three variables indicate that during 2008-2011 the researched countries implemented structural changes and revised management practices to a medium-high extent. However, the inputs in terms of available resources for government services were relatively low (mean = 0.23; sd = 0.05). On average, in 2012 the researched countries achieved relatively high levels of management quality and government effectiveness (mean = 0.67; sd = 0.13). However, in 2014 the average level of citizens’ trust in national government was below 50% for the researched countries (mean = 0.42; sd = 0.14).

The correlation matrix presented in Table 4.1 shows that performance management is correlated only to management quality and government effectiveness (r = 0.41, p = 0.01), which is consistent with our hypotheses. In other words, the inputs and activities included in the performance management mechanisms are correlated to outputs but not to outcomes such as TRUST. TRUST is also correlated only to management quality and government effectiveness (r = 0.58, p = 0.000), implying that this latter variable may somehow connect performance management and TRUST. The variable, management quality and government effectiveness, is also correlated to another input variable - resource availability (r = 0.37, p = 0.03), while resource availability is correlated to decentralization processes (r = 0.45, p = 0.03).

To examine the research hypotheses further and to encapsulate the empirical relationships between the five theoretical hypotheses presented earlier within a single framework, we used an SEM model created by the AMOS software. Figure 4.2 illustrates the empirical findings and the relations between the variables. The empirical model supports only some of the hypotheses. For example, it marginalizes the role of structural changes such as coordination and decentralization in explaining management quality and government effectiveness and the trust in national government. The fit of the path model is good. The model has a x2 of 3.96 with 5 degrees of freedom (p = 0.556), CMIN/DF = 0.792, NFI is 0.893, RMSEA is 0.000 [90% confidence limits (CL) 0.000, 0.2], CFI = 1.000, and TLI = 1.142. (CMIN - Minimal Chi-Square; DF - Degrees of Freedom; NFI - Normed Fit Index; RMSEA - Root-Mean-Square Error of Estimation; TLI - Tucker-Lewis Index; CFI - Comparative Fit Index) In other words, the model is not significantly different from the data we collected regarding the researched countries and reflects the empirical answers strictly and properly.

The empirical model portrays a relatively simple picture in which PM is related to MQGE (fi = 0.39, p < 0.004) and MQGE is related to TRUST

Research findings (standardized coefficients in SEM model using SPSS- AMOS)

Fig. 4.2 Research findings (standardized coefficients in SEM model using SPSS- AMOS)

(в = 0.62, p < 0.000). These findings support H1 and H5, respectively. The model also shows that COORis not related to MQGE or to any other variable in any meaningful way, meaning that H4 is not supported. This result is interesting, because it seems that the governments in the researched countries have invested a great deal in coordination initiatives (mean value: 81%). However, that effort was not reflected in improved outputs or outcomes.

The empirical model also shows that DC and RA are related to MQGE but not to TRUST. There are direct relations between RA and MQGE (в = 0.30, p < 0.05), which support H3. However, the model shows that the direct relations between DC and MQGE are not significant, meaning that H2 is not supported. Instead, DC and RA are related to each other (в = 0.39, p < 0.012), meaning that DC and MQGE are related to each other through the mediation of RA. To verify this argument, we built a direct SEM model positing a direct effect between DC and MQGE. The findings, presented in Fig. 4.3, indicate that there are significant direct relations between DC and MQGE (в = 0.29, p < 0.05) when RA is not involved in these relations (see model fit in Table 4.2). The comparison between the models presented in Figs. 4.2 and 4.3 shows that when the relations between RA and MQGE are added, the relations between DC and MQGE become insignificant. This analysis

Direct model positing a direct effect between DC and MQGE (standardized coefficients in SEM model using SPSS-AMOS)

Fig. 4.3 Direct model positing a direct effect between DC and MQGE (standardized coefficients in SEM model using SPSS-AMOS)

Table 4.2 Comparison of the SEM models

The SEM model







The research findings (Fig. 4.2)

3.96 df= 5 (p = 0.556)



  • 0.000
  • (0.000; 0.2)



The direct model positing a direct effect between DC and MQGE (Fig. 4.3)

7.04 df= 6 (p = 0.317)



  • 0.068
  • (0.000; 0.232)



proves our argument that RA mediates the relation between DC and MQGE.

Thus, the findings support the core idea of our research model that inputs and activities influence outputs, which then influence outcomes. Performance management mechanisms and activities designed to create structural changes primarily influence managerial quality and government effectiveness, which then gradually impact TRUST. Since the level of TRUST probably also reflects the subjective evaluation of the quality of governmental services and the confidence in the good conduct and intentions of the government, we may apply our findings to a broad set of outcomes rather than solely referring to TRUST.

< Prev   CONTENTS   Source   Next >