4.1. Conceptual Framework and Definition
In order to determine the significant socio-economic diversity of rural development in Poland, we needed to start by defining the concept that was fundamental for this study. First, we assumed that “rural development” is an analytical category of “socio-economic development”; secondly, it can progress in different directions and depends on local conditions (including previously shaped social and economic structures). It is understood as a process of rural structure evolution towards a resident-friendly social environment enabling residents to obtain socially accepted income from work, the best possible access to public services, and labour markets, giving them a sense of agency and enabling them to fulfil their aspirations. Its measurement should reflect the extent to which residents’ needs, thus, defined, are met, or, to put it differently, the degree of advancement of processes that contribute to increased well-being. Of course, the “extent to which needs are met” can be the same or similar in very different conditions, in the presence of very different combinations of economic and social structures. There is no single path along which all rural areas travel in such processes.
This understanding of development was used as the basis for a definition that is not universal but corresponds to the current situation in rural Poland, i.e., it is related to the current stage of changes in rural areas and reflects the rural reality under analysis. Relating this definition to the unique reality of present-day Poland, involved specifying the main areas requiring action and enabling rural residents’ well-being to improve, especially measures overcoming the major barriers to development [
8,
66]. It was concluded that these are the barriers related to the diminishing role of agriculture as a source of income, which also implies the development of non-agricultural economic functions and the related labour markets as well as changes in farming aimed at an increased work output [
67,
68]. The shortage of jobs in rural areas causes migration from the countryside, which is highly selective and has a negative impact on rural demographic structures. Furthermore, it is taking place parallel to a decreasing demand for labour in farming. In the social sphere, barriers to development emerge in connection with rural residents’ relatively low qualifications, a deficit in their sense of agency related to the changes taking place, and a lack of faith in the effectiveness of self-organisation in local communities for solving their problems. Almost half a century of a “real socialism” economy in Poland discouraged rural local communities from showing initiative and taking local affairs into their own hands. The predominant view was that the government would solve all their problems. The collectivist ethic promoted under communism disapproved of individual initiatives and successes, and quite effectively suppressed the emergence of local leaders. Moreover, the official slogan of “socialist rural transformation” and the ideological treatment of peasants as a social group that did not fit in with the Marxist class structure model and was, thus, condemned to disappear, did not contribute to a sense of stability either, and did not encourage social (community) activity [
30,
69].
At present, agriculture is still a major, and in some regions the predominant, economic function in rural Poland; it employs about 10% of national labour resources (22% of rural resources) and generates 2.6% of the GDP (Gross Domestic Product) [
70]. Of all the post-communist countries, Poland was the only one that had not undergone collectivisation. However, family farms have their specificity; they are prone to be guided by the evaluation not of their market results, but their benefit to the family [
71,
72]. In the period of transition from a centrally planned economy to a market economy, after 1989, these farms absorbed some of the people who had lost their jobs as a result of the restructuring of non-agricultural workplaces, pursuing not the maximisation of economic indicators such as the production value as their main goal, but social objectives (chiefly the economic security of family members). This is why the proposed definition of rural development as applied to Polish reality underlines the problem of the deagrarianisation of the economy (the development of non-agricultural jobs, limiting the role of agriculture in rural residents’ sources of income) and problems of societal activity, including the self-organisation of local communities.
The application of the concept of “rural development” proceeded in two stages. In stage one, 11 components covering its scope were specified. However, these components were not empirical. In stage two, several empirical indicators were selected for the individual components (
Appendix A presents the set of indicators for each component). This was justified by the fact that the components usually corresponded to complex issues impossible to characterise with the help of a simple empirical indicator. In practice, this meant establishing 11 measurement scales to determine the intensity of a given issue in individual spatial units. Next, this was used as the basis for developing the typology of rural areas. The study procedure is shown schematically in
Figure 1.
The first of the 11 components was called “spatial accessibility”, because it is related to a fundamental quality of rural areas, namely, their spatial extensiveness or dispersion. Development requires the negative aspects of this rural specificity to be overcome; such aspects include problems with the accessibility of the units under consideration (gminas/communes) from other areas as well as the accessibility of a given gmina’s centre from the other villages within it. We measured this by the distance from major labour markets but also from regional cities, by the accessibility of public transport, etc.
Components 2–6 are related to economic development. The “local economy deagrarianisation” component was especially important here. By deagrarianisation we mean the progressively growing role of non-agricultural functions in providing residents with sources of income, i.e., shifting away from the domination of agriculture in the rural economy. The need for the rural economy’s deagrarianisation lies at the foundation of the concepts of multifunctional rural development, lasting development, and sustainable development. This makes it an extremely important component, which is also connected with social development. These connections are generated by the development of public services and increased employment in them, the development of the social and technological infrastructure, and growing opportunities for employment compatible with the aspirations of rural residents. The component creates conditions for structural changes within the agricultural function, especially if we consider agricultural over-employment in the family-farm sector (hidden unemployment). Components assessing the situation in the “agricultural sector” have been applied separately from those for the “non-agricultural sector”.
The next component, “local public finance”, does not just characterise the gmina budget situation, but is interpreted much more broadly. On the one hand, it shows what budget income the local economy can generate, and, on the other, what the actual capacity for improvement within the gmina’s own tasks is, i.e., infrastructure development, improving the availability of public services, etc. In addition, the gmina budget’s income provides indirect information on the earlier level of development, especially the value of assets accumulated from earlier periods, with which the gmina is equipped (e.g., infrastructure).
“Labour market balance” is a component covering issues related to the employment of the farming and non-farming population as well as those linked to shaping the structure of labour resources in the gmina. This means that the study considered not only the volume of registered unemployment, but also indicators related to hidden unemployment in agriculture, the ageing of labour resources, and the attractiveness of the local labour market for migrants.
The next group of components, from 7 to 10, describes elements of social development. Among these, “demographic issues” deserve special attention, including how the effects of internal (domestic) migrations influence the make-up of the local community. It is worth noting that mainly young, well-educated, resourceful, and enterprising people migrate. As a result, distinctive changes in the demographic structure take place in the migrants’ areas of origin: an ageing of the population, a shortage of well-educated people, especially young, educated women. The low level of human and social capital in areas of emigration causes difficulties with the development of non-agricultural functions and a diminished interest in technological and social innovations.
The final, 11th component is “living conditions”. This complements the infrastructure elements that were partially covered by “spatial accessibility”, some related indicators also being covered by “local public finance”. The intention was to interpret living conditions not in terms of quality of life, but in terms of infrastructure for the population (water supply, sewerage, etc.).
4.2. Data Sources and Indicators
The study of the level and structure of socio-economic development was carried out at the level of Local Administrative Units (LAU); thus, covering all 2173 rural gminas and rural areas of urban–rural gminas in Poland. On average, a rural gmina comprises several to a dozen or so villages (depending on their size, which can be very different in different regions) and is inhabited by approx. 7000 people.
The study is distinguished by a unique set of diagnostic features that were the basis for constructing 47 empirical indicators using the 11 components to explain rural development (
Appendix A lists the empirical indicators, the data sources and the median). The components were selected with the following criteria in mind: significance for the phenomenon being analysed, accuracy and unambiguity, exhausting the scope of the defined phenomenon, the logical nature of connections, maintaining proportionality, and the availability of information for all the
gminas in the study. The data were obtained from the Local Data Bank (BDL) of Statistics Poland (GUS), but also from the unpublished databases of government and communal institutions. These databases include the Agricultural Social Insurance Fund (KRUS), the Central Examination Commission (CKE), the Office of Electronic Communications (UKE), the National Electoral Commission (PKW), the Agency for Restructuring and Modernisation of Agriculture (ARiMR), the Central Register of Vehicles and Drivers (CEPIK) affiliated with the Ministry of the Interior and Administration (MSWiA), the Ministry of Finance (MF); the Ministry of Family, Labour and Social Affairs (MRPiPS); the Ministry of Investment and Development (MIiR), and 16 Marshal Offices (LEADER+ Programme), as well as original primary data collected by
gmina polling offices. For instance, the Gmina Survey was conducted using a software tool, and then computer-assisted telephone interviews (CATI) were conducted in every
gmina and were the only comparable source on the accessibility of public transport in the country. The data collection was in accordance with Polish law, through access to public infor-mation. All the data were obtained under the Rural Development Monitoring (MROW) pro-ject. The typology results were based on indicators from the last complete MROW 2018 database, while
Appendix A lists the latest data already collected for 2019. This study has been conducted cyclically by the European Fund for the Develop-ment of Polish Villages and the Institute of Rural and Agricultural Development of the Polish Academy of Sciences (IRWiR PAN) since 2010. The database was used to pro-duce reports more: [
73,
74], which discuss the methodology in detail.
4.3. Method
All of the empirical indicators underwent a statistical normalisation procedure, which was conducted according to the unitarisation method. They were made comparable, i.e., normalised in the range (0, 1). The higher-the-better normalisation was affected by subtracting the lowest number in a given set from its basic value (before normalisation) and dividing the result by the range (the difference between the lowest and highest value in a given set). The procedure for the lower-the-better normalisation was similar, only the number in the numerator was the highest instead of the lowest.
The next stage of the study involved the hierarchical classification of the
gminas according to a synthetic measure calculated for each component. Of the many existing methods of linear ordering [
75,
76], in this study we chose the above-zero unitarisation sums procedure [
77], one of the taxonomic patternless methods. In this study, the evaluation of a variable that characterises the
i-th object is called a “synthetic variable”. The synthetic variable obtained from the following Equation (1) assumed values within the range (0, 1):
where
a’ij is the normalised value of the
j-th feature in the
i-th object (after the lower-the-better characteristic was changed to higher-the-better),
n is number of objects, and
mi is the weight factor of the
i feature.
At the stage of calculating the synthetic value of a given component, the empirical indicators were weighted, their weights being based on an analysis of the importance of a given indicator (subjective criterion), and the existence of formal statistical relationships was treated as an extra criterion [
45,
78].
Next, the synthetic indicators of the 11 components served as the basis for non-hierarchical grouping of the
gminas with the aim of forming homogeneous types by similarity of features. The iteration method was chosen for this stage of the study, based on the gravity model proposed and described by Edwin Diday in 1973, called dynamic clouds clustering [
79]. In the most general terms, the purpose of this method is to group units in order to form their homogeneous classes (clouds) by internal combinations of the criteria being studied, where, at the same time, the classes are maximally different from one another.
Diday’s algorithm is deterministic, it operates in a multidimensional space. In the first step, the units (
gminas) form as many classes as they number. The mass of a unit in our study was its population. This distribution of classes defines the inertia to which all the units contribute. The initial cloud (2173 classes =
gminas) is then reduced to a new cloud of
k class centres, spread around the cloud’s global centre. Let us add that this potentially optimal number of subsets (
k classes) was determined using Ward’s agglomerative method [
48]. Each cloud has its centre of gravity
S, and the distances between individual points (
gminas) are counted as Euclidean distances [
80]. The initial (temporary) partition with a specified number of
k classes (clouds) has randomly assigned centres of gravity (seeds). Its quality is then improved iteratively, by moving some elements from one class to another if this increases the value of the objective function. Next, the centres of these classes, i.e., their centres of gravity, are calculated and then replace the initial centres (seeds). The assignment procedure is then repeated; the centres are recomputed, and so on. At each iteration, some units change class. The process continues until a final configuration, which cannot be further improved through a local reassignment of units, is reached. The partition obtained is a local optimum: this means that small changes in the allocation of the units to the groups are unable to improve it. Of course, we cannot be sure that the optimal partition (
optimum optimorum) has been found, as the method is heuristic. It yields a good-quality partition, but not the absolutely best one. However, for combinatorial reasons, we had to accept this [
79].
Once a satisfactory configuration (optimisation phase) was achieved, a profile control procedure began for the individual classes = types. The profile of each class had to be compared with the overall profile in order to assess which variables characterise it more, because of their significantly higher or lower than the overall average values. This analysis was performed in order to explain the structure (complexity) of each type. Hence, the location of individual centres of gravity was determined for clouds xm
(j,i) (types of similar
gminas) in relation to the centre of gravity of the whole system xg
(j). Variable (j) was relative to cloud (i) when the difference xm
(j,i) − xg
(j) was far from zero (in+ or in−). To estimate the significance, this difference (characterisation of a given type) was compared to the standard deviation S(
j). The indicator’s value was calculated using Equation (2).
xm(j,i)—average value of variable j in cloud i (type i);
xg(j)—overall average value of the same variable j;
S(j)—standard deviation of variable j.
The strength of the individual variables’ influence on development processes was also studied with the help of correlation (mutual dependence of features). It was assumed that the coefficient exploring the relationship between empirical variables would be based on Pearson’s linear correlation (r). All of the empirical variables were described on quotient scales, i.e., scales that are logically stronger compared to those describing synthetic measures. These, in turn (i.e., the measures of the components), were described on ordinal scales. This was why nonparametric statistics were used: Kendall’s
tau (T), which should be interpreted in terms of probability [
81,
82].