Different governments should (in theory) respond differently to the uncertainty and complexity which characterises their economic environment. In our model, government sizes adapt to an uncertain economic environment differently; depending on their risk aversion and the level of overall changes occurring in the wider micro-economy (the γ1 and β1 in the model).39 Governments also grow or shrink depending on the complexity of economic transactions – reflecting government’s isomorphism with the
macroeconomic environment and its role in facilitating that growing complexity (the γ2 and β2 in our model).
Our statistical analysis points to seven kinds of governments. Figures 29a and 29b show the results of regression analyses performed on clusters of countries -- which show the ways different groups of countries’ governments respond to changes in their
organisational environment. The statistical procedures we used chose these clusters of countries based on statistical similarities (which are described in the Figure and shown in Appendix III). Only about 20% of the governments we analysed seem to follow a
strong-form of our model – adapting to changes in the uncertainty and complexity of their organisational environment (from group 3). Of these high-adapters, all comprise high-income countries. Almost double that proportion – almost 40% of the countries analysed – have governments whose size seems to respond almost exclusively to changes in spendable resources (from groups 1,2,4 and 5) . These governments seem to fit mostly closely the resource-based view of organisational strategy. Another 20% of the governments in our sample adapted their size in response to the uncertainty of their organisational environment (as proxied by changing sectoral weights in overall GDP). For these countries, only one of the two environmental variables we analyse seemed to explain changes in government size (group 6). For another set of countries (again about 20% of the sample), these countries’ governments grew with the increasing complexity
of their economies (group 7).
39 In Appendix I, we derive the formal model and show more specifically under which circumstances governments with different preferences are likely to respond to changes in their organisational environments.
Figure 29: Statistically Suggested Groupings of Governments According to their Preferences for Responding to Changes in their Organisational Environment Group 1: Advanced Resource-Based Growers
These governments grow primarily in response to changes in the resources they have (or can borrow). However, they “match” the complexity of their organisational environment.
Members: Belgium, Cyprus, Greece, Italy and Jamaica
b-value se R2=0.75
Revenue to GDP 0.87 0.11
Debt to GDP 0.10 0.17
Geometric Effect for the Complexity Proxy **
Group 2: Tax and Grow Economies
The governments’ size seem bounded strictly by the amount of resources they are able to raise. The heterogeneity of these countries makes further generalisation about these countries impossible – and their different variance profile excludes them from Groups 4 and/or 5. .
Members: Bhutan, Brazil, Canada, El Salvador, India, Moldova, Pakistan, Papua New Guinea, Philippines, Senegal, Spain, Tunisia, Turkey, Uganda, Ukraine, Uruguay
b-value se R2=0.87
Revenue to GDP 0.81 0.04
Group 3: The Environmental Adaptors
These economies best exemplify the model presented in this paper. The levels, size and rates of change in all measurable parts of the government’s organisational environment (complexity and uncertainty) reflect in changes in the size of government.
Members: Austria, Denmark , Finland , France , Hungary, Netherlands , New Zealand, Norway , Poland , Portugal , Spain , Sweden
b-value se R2=0.62
Complexity Proxy 1.79 0.43
Uncertainty Proxy 0.09 0.04
Debt to GDP 0.23 0.02
Geometric Effect for the Complexity Proxy **
Interaction between the Uncertainty Proxy and Revenue to GDP ** Interaction between the Uncertainty Proxy and Debt to GDP **
Interaction between Revenue to GDP and Debt to GDP **
Group 4: Low-Income Tax and Grow Governments
Governments in these low-income economies -- like their counterparts in the rich countries (see Group 5) - - respond only to changes in their revenues.
Members: Cote d'Ivoire, Kyrgyz Republic, Madagascar, Mongolia, Sri Lanka
b-value se R2=0.67
Revenue to GDP 1.05 0.20
Group 5: High-Income Tax and Grow Governments
These governments’ size mainly appears tied to revenues – expanding slightly less than their colleagues in the low-income countries.
Members: Germany, Switzerland, United Kingdom, United States
b-value se R2=0.96
Revenue to GDP 0.88 0.08
Note: The groupings of countries shown in the figure reflect k-clustering – a statistically procedure which groups data according to the similarity of the data’s variance. The procedure looks at variation in the entire dataset and constructs groups which minimise the variation in the data. We use these clusters of countries in order to test our model for each group of countries in order to estimate the way that changes in government size differs between groups of countries. The b-values show the change in government size (expenditure as a percent of GDP) for changes in the variables shown in the figure. Differences in these b- values indicate differences in the way that government size responds to the variable. For example, governments in group 1 expand 6% more for a similar increase in revenue-to-GDP in Group 2. Asterisks indicate the presence of a geometric effect. R2 represents the proportion of variance in the size of government “explained” by the model. We report b-values rather than beta coefficients in order to show the exact relationship (rather than only magnitudes of importance) between variables.
Figure 29: Statistically Suggested Groupings of Governments
(continued)
Group 6: Risk-averse social insurance governments
These governments’ size correlates strongly with changes in the sectoral distribution of national output (and the amount of resources these governments have to spend). Such a correlation suggests an
organisational strategy which seeks to minimise the economically disruptive effects of asymmetric shocks and/or the shift in resources between sectors.
Members: Australia , Belarus , Czech Republic , Estonia , Ireland , Latvia , Lithuania , Russian Federation and Slovak Republic
b-value se R2=0.55
Revenue to GDP 0.99 0.15
Uncertainty Proxy **
Interaction between Uncertainty Proxy and Revenue to GDP **
Interaction between Uncertainty Proxy and Debt to GDP **
Group 7: Technology-adaptors
These economies grow bigger as their economies are becoming more complex.
Members: Bangladesh , Georgia , Guatemala , Indonesia , Kazakhstan , Mauritius , Mexico Oman , Peru , Thailand
b-value se R2=0.82
Complexity Proxy -0.80 0.21
Revenue to GDP 0.82 0.06
Note: The groupings of countries shown in the figure reflect k-clustering – a statistically procedure which groups data according to the similarity of the data’s variance. The procedure looks at variation in the entire dataset and constructs groups which minimise the variation in the data. We use these clusters of countries in order to test our model for each group of countries in order to estimate the way that changes in government size differs between groups of countries. The b-values show the change in government size (expenditure as a percent of GDP) for changes in the variables shown in the figure. Differences in these b- values indicate differences in the way that government size responds to the variable. For example, governments in group 1 expand 6% more for a similar increase in revenue-to-GDP in Group 2. Asterisks indicate the presence of a geometric effect. R2 represents the proportion of variance in the size of government “explained” by the model. We report b-values rather than beta coefficients in order to show the exact relationship (rather than only magnitudes of importance) between variables.
Hypothesis 3: Different countries will adapt reactively, strategically or