Similarities in high and low satisfaction LGs group

Could the satisfaction of the Estonians with their home municipality be determined by the quality of services that they receive from a local government? Local authorities manage multiple aspects of people’s daily lives such as education, public transportation, waste management, recreation and cultural facilities. It is therefore possible that variation in their performance in those multiple areas would have a noticeable impact on people’s satisfaction with living in a particular area.
ROBERT LIPINSKI, World Bank Consultant

Correlation between satisfaction and service levels

Thanks to a comprehensive dashboard of the Estonian Ministry of Finance containing statistics pertaining to the quality of work undertaken by local governments across 16 key public service areas, it is possible to investigate this hypothesized relationship between satisfaction and public services quality. The dashboard annually evaluates the service quality of each local government on a 0-9 scale, with 1-point increments, for 16 different public service areas. Therefore, the more points on average a local government scores, the higher its overall quality of public services.

In the figures below, one can observe that this overall service quality is indeed positively correlated with citizens’ satisfaction, i.e. municipalities scoring highly service quality, are more likely to have a high share of satisfied citizens. In fact, out of the 76 municipalities covered, 27 score above the national-level average on both service quality and satisfaction levels (upper-right quadrant of the figure). For example, residents of Tartu linn enjoy second highest service level quality (5.69) and 4th highest satisfaction level (81.1%). Conversely, a total of 24 municipalities score below the national average on both service quality and satisfaction level (lower-left quadrant).

There are considerably fewer municipalities scoring above the national average on one measure, but below on the other one. In 15 municipalities citizens are satisfied with their local government above the national average, despite below-the-average service quality (upper-left quadrant), suggesting that their satisfaction is boosted by other local-level advantages. Finally, only in 10 municipalities, the above-the-average service quality does not seem to translate into above-the-average satisfaction levels (lower-right quadrant).

Still, there is a considerable amount of variation in those numbers. For example, across all 76 municipalities, the residents of Muhu vald are the ones most satisfied with their local government as a place to live (90.4%), but the service quality they enjoy is evaluated as below-the-average (3.69). In contrast, Harku vald enjoys highest level of service quality (5.73), but less than 6 out of 10 of its residents are satisfied with their local government (59%) – below the national average of 61%.

This variation is present among municipalities, regardless of their geographical location – all 5 major regions of Estonia display high variation of service quality and satisfaction scores. This is indicated by a considerable dispersion of the dots representing different regions in the figure below. The figure also shows that the population size of a local government similarly determines neither the service quality nor the satisfaction level, as municipalities of all sizes are dispersed across the whole range of values in the figures below.

However, the economic conditions prevalent in a municipality appear to matter more than either the geographical location or the population size. In the figure below, we can observe that when grouped by the average wage level, municipalities tend to cluster in more discernible manner. In particular, the municipalities in the lowest quartile of the wage distribution (i.e. 25% of all municipalities where the wage level is the lowest) are located almost exclusively in the lower-left quadrant in the figure below (14 out of 19). Thus, in addition to their low wage level, they also face below-national average quality of local public services and satisfaction levels. In contrast, municipalities in the 3rd and 4th quartile tend to be located in the upper-right quadrant (9 and 10 out of 19 respectively), or, if not, not much below the national-level averages on either service quality or satisfaction.

A similar, if a bit less consistent, pattern emerges, when municipalities are grouped by their unemployment rate. Here also the municipalities scoring the best on this measure, that is with unemployment below 1.9%, tend to occur in the upper-right quadrant (11 out of 19), indicating that they also enjoy above-national average quality of local public services and satisfaction levels. Their mirror image is formed by the municipalities with the highest unemployment rate (above 2.9%). Out of 19 of such municipalities, 11 fall in the lower-left quadrant, indicating below-national average local public services quality and satisfaction levels.

The above analysis, in particular the positive relationship between service quality and satisfaction level and the clustering of observations in the lower-left and upper-right quadrants of the figure, suggests that the quality of services provided by local authorities is an important predictor of residents’ satisfaction level. However, the considerable variation exhibited across municipalities also indicates that the services quality cannot be the sole influential factor shaping people’s satisfaction with their local government. In particular, the relative influence of the municipality’s public services quality, its wage level and unemployment rate, merit further investigation.

Individual- and municipality-level drivers of citizen satisfaction

The list of factors that can potentially influence one’s satisfaction with their home municipality is incredibly large. Though the municipalities that provide on average higher service quality were found to have a larger share of satisfied residents, it was clear that they are not the only factors driving it. This is to be expected. For one, public services are only part of everyday life in a given locality. People live in municipalities of different types and sizes – from half-a-million people capital city of Tallinn, to the serene island of Muhu, with only few thousand people. Moreover, the economic conditions faced by people, like the wage and unemployment levels, also vary by locality. On a more personal level, one’s gender, age, education level, work status, or income, all have a strong potential to affect one’s satisfaction.

How to understand the impact of all those factors and weigh them against the quality of public services as drivers of people’s satisfaction? Using a statistical modelling tool, called multilevel regression analysis (for details see Box 2), we found that there are 4 key factors that play a key role in determining one’s satisfaction with their local government as a place to live – two relate to the features of municipalities in which one lives and further two concern one’s personal situation[1]. Those are:

  • Average quality of local public services (municipality-level): The trend broadly evident in the previous post’s analyses is confirmed using the more detailed statistical model. Municipalities with higher average quality of local public services tend to have residents that are more satisfied with their local government, even when accounting for 14 other municipality- and personal factors.
  • Unemployment (municipality-level): Municipalities with higher level of unemployment exhibit lower level of satisfaction. It might be both because unemployed people face multiple hardships, as do employed people who, seeing high unemployment levels in their area, might fear about the security of their job positions. This result is highly consistent with studies conducted in other countries, including Germany, Canada, United Kingdom, and United States (Clark, Knabe and Rätzel 2010; Chadi 2014; Chen and House 2016)
  • Employment status (personal-level): Self-employed individuals tend to express lower levels of satisfaction with their local government as a place to live. Across Estonia 55.8% of self-employed individuals say they are ‘satisfied’ or ‘very satisfied’ with it, in contrast to 60.7% of retirees and 64.1% of salaried employees.
  • Household income (personal-level): People in households with monthly income below 500€ per person are the least satisfied income group (55.9%). However, satisfaction with one’s local government doesn’t increase linearly with one’s household’s income per person. Controlling for all the other factors, people from households that earned above 1,000€ per person last month were not more likely to be satisfied. In fact, only the group earning between 500€ and 1,000€ was more likely to be satisfied.

Survey details
The survey was conducted online, between June and October of 2021. In total, a sample of 7,219 people from across 76 municipalities provided their responses to a two-part survey. In the first part, the respondents were able to express the feelings about their local governments, including aspects such as efficiency and responsiveness of the policy-makers. In the second part, they could use a novel survey tool, called quadratic voting for survey research (QVSR), and assume the role of local policymakers to decide how to distribute a ‘budget’ of votes for future spending on 16 different types of local public services - from pre-school education, through waste management, to sports. As a decision-making tool, QVSR is increasingly attracting the attention of both academics and policy-makers (see here). The survey was concluded shortly before recent local government elections, held in October 2021. As such, its results should not be viewed not as an evaluation of the performance of the current local authorities, whose mandate started only after they survey had finished collecting citizens’ feedback. Rather, it should be seen as a potential input in deciding on reforms and new policy directions by the incoming local decision-makers.

Predicting citizens’ satisfaction – multi-level regression model
A person’s satisfaction can be determined by his or her individual circumstances, as well as the characteristics of the environment in which he or she functions. When it comes to satisfaction with local government as a place to live, the individual-level characteristics, or variables, of relevance might include one’s gender, age or employment status. The area-level variables might in turn compromise the local wage level, unemployment rate, or the geographical location. Ascertaining the relative importance of those variables is typically done using regression models. Those are statistical tools that allow one to determine the strength of association between the variable of interest (dependent variable, in this case the satisfaction level) and the variables expected to influence it (independent variables). In addition, they allow one to determine the probability that the observed association is random or not. In a context where the predictor variables are measured at both individual and area-level, as is the case in the current context, the appropriate type of regression is multi-level modelling. The ultimate outcome of interest, i.e. the satisfaction level, is measured at the individual-level. However, the satisfaction of individuals living in the same municipality is influenced by the same area-level variables (all individuals in municipality X face the same wage level, unemployment rate and a host of other circumstances). In the technical terms, the individual-level data are therefore not independent of each other. Multi-level modelling accounts for this fact and allows one to calculate the impact of all predictor variables without introducing a bias that would arise if the area-level predictors were ignored. For the interested readers, here is a more extensive introduction to it: https://rstats-tartu.github.io/bayesiraamat/mitmetasemelised-mudelid.html

Data is available here

*Satisfaction survey was conducted by The World Bank. Relationships were found using the Data from www.minuomavalitsuse.ee website.

Last updated: 17. July 2020