This paper identifies classes using evidence from surveys taken in 2009 among nationally representative samples of approximately 2000 households from each of the three South Caucasus countries –Armenia, Azerbaijan and Georgia. Information on employment, education and income is used to identify clusters of both individuals and households, in each case a middle class, a working class and a lower class. Using individual classification, the lower classes are the largest classes in each country, which therefore appear to have pyramid-shaped class structures. However, classifying households gives the class structures a diamond shape, with larger working classes than either middle classes or lower classes. The evidence shows that the main differences between classes of households lie in their standards of living, and are driven by differences in income levels. The main differences between classes of individuals are driven primarily by education, which result in differences in attitudes and (to some extent) individual lifestyles. It is argued that in economic and socio-political terms there are as yet just two real classes among actual and potential employees in the South Caucasus –middle classes and lower classes –and that although these classes differ in their standards of living and political dispositions, these are unlikely to become bases for conflict between them.

This paper presents the results of a secondary analysis of a data set suitable for identifying emerging classes in three new market economies – Armenia, Azerbaijan and Georgia. This requires us to engage in a set of long-standing and perennial debates about whether classes are best conceptualised categorically (as ‘boxes’) or gradationally (see, for example, Prandy 1998; Rose 1998), and the unit that should be classified – the individual or the family-household. Our analysis does not prejudge these issues. Rather, it opts for a categorical rather than a gradational concept of class because the evidence indicates the formation of distinct clusters. As regards individual versus household classification, our evidence illustrates the advantages of using both simultaneously. Our conclusions cannot be extended beyond the countries from which the evidence is drawn, but a similar analysis would lead to comparable if dissimilar findings if conducted in other post-Soviet countries, and possibly beyond.

The paper proceeds by describing the data. We then situate our analysis within the long-running debates about how and who to classify, and previous research into class formation in former communist countries. We then present our findings which show that individual and household classification both indicate the formation of clusters that we label ‘middle’, ‘working’ and ‘lower’ classes, though the proportions of the populations allocated to each class vary depending on whether the unit treated is the individual or the family-household. We then show how household and individual processes result in classes developing distinctive lifestyles and other socio-cultural attributes. In our view, these are best treated as part of what classes ‘are’ rather than as effects or correlates. The paper concludes with a discussion of the likely political implications of the classes that are being formed in the South Caucasus.

2.1. Data

The Data Initiative Survey (DIS), renamed the Caucasus Barometer in 2010, is a series of annual surveys based on representative samples of around 2000 households in each of the three South Caucasus countries (Armenia, Azerbaijan and Georgia). The series began in 2004 with samples from the capital cities only (Baku, Tbilisi and Yerevan). In 2005, a specimen region in each country was added. Since 2006 the surveys have been based on nationally representative samples for which there is now a progressively extending time series which began in 2004 in the capitals.

These are household surveys, but just one respondent per household is interviewed. The interviewees are selected using a procedure which ensures that the respondents are representative of the populations aged 16 and over in each of the countries (for full details, see www.crrccenters.org). The informant supplies basic information about all household members and the household's resources, and a great deal more information about the informant's own work, lifestyle and attitudes.

These surveys are conducted and funded by the Caucasus Research Resource Centres, whose own baseline funding is from Eurasia and the Carnegie Foundation, and whose offices are in Baku, Tbilisi and Yerevan. As the title implies, these are resource centres, and their resources, including the DIS data sets, are accessible by all bona fide researchers. The authors of this paper were given access on these terms, and the views expressed here are solely the authors’.

Occupation is one item of information that the surveys have never collected. This absence may appear both surprising and a bar to class analysis, but a reason for the absence is that, due to their turbulent histories throughout the 1990s, the countries had yet to develop their own Standard Occupational Classifications via which, where they exist, survey respondents are normally placed in occupational classes. Without a SOC it is impossible to allocate occupations to classes accurately. However, the surveys do separate employees from the self-employed (who are separated into those with and without their own employees), and collect information about the business sectors where everyone works. There is information about whether each household member is self-employed, employed or not employed, his or her education, total post-tax household income per month, and total household expenditure. There is also information about the sources of the household's income, and respondents provide details of their own monthly incomes after tax.

Occupations, when this information is available and can be coded, are typically used as sole indicators of class positions for a number of excellent reasons. Individuals tend to remain in the same types of occupations throughout their employment careers, and these types (classes) of occupations are good indicators of people's long-term positions in their society's systems of economic production and distribution. The positions of most occupations have remained stable over time, and are similar in all societies with a system of market capitalism. Also, occupations are known to be significantly associated with correlates of class such as educational achievement. Similar work and market situations (and sometimes status situations, also) are the criteria normally used to group occupations into classes (see Lockwood 1958 for the original formulation). Without an occupational variable it is impossible to establish a person's ‘work situation’, except that in the DIS employees and the self-employed can be separated, and ‘workers’ can be distinguished from persons without any paid work situations. ‘Market situations’ can be identified from the information (on income) that is collected in the DIS, and we shall see that by far the major division by ‘work situation’ in the South Caucasus is whether individuals (and entire households) have any regular (see below) paid employment. In the following analysis, classes are identified using information about whether individuals were employed or not employed, whether or not they were higher-education graduates (already known to be a good predictor of individuals’ chances of entering non-manual occupations; see Roberts and Pollock 2009), and their incomes. We shall see that, using these indicators simultaneously, the samples divide into a small number of clusters which, in our view, fulfil the normal criteria to be called classes (demographic stability, similar positions in the systems of economic production and distribution, and similar standards and styles of life).

2.2. Issues

To the best of our knowledge this is the first attempt to identify socio-economic classes from evidence gathered locally in the South Caucasus, and it is one of the few attempts in any of the ex-Soviet republics. Local commentators have postulated the formation of new classes, usually factions of the new middle classes, but not on the basis of survey evidence from representative samples (see, for example, Khmelko 2002; Kutsenko 2002; Piirainen 1998). Western researchers have sometimes used Western class schemes when analysing occupational stratification in the new market economies (for example, Evans and Mills 1999). This practice is likely to become increasingly common since the development of a version of the Goldthorpe class scheme (as it is known in the UK) suitable for gathering harmonised data from all European Union member countries (see Rose and Harrison 2007). Local researchers may use this or other occupation-based measurements of class that are deemed suitable for cross-national use (see Ganzeboom et al.1992; Hoffmann 1999). Others have chosen to stratify their post-Soviet samples into income bands, or according to respondents’ own assessments of their households’ positions vis-a-vis other households in their districts or countries (see Bogomolova 1998; Caucasus Research Resource Centre 2005; Oksamytna and Khmelko 2004; Tilkidjiev 1996). Both types of evidence have been gathered in, and can be used to analyse the DIS data sets. We know that the populations in these countries are class aware. If invited to do so they will name the classes that have been formed during the post-communist transition. They invariably identify ‘the rich’ and ‘the poor’, and are also likely to distinguish a middle class, and will place themselves into one of these classes (see Roberts and Pollock 2009). The classes that are identified in the analysis that follows are quite similar to these everyday lay conceptions, but the DIS has not sought information about subjective class identities, and our procedures for identifying classes are entirely different.

Our analysis does not pre-suppose that there are categorical classes in the South Caucasus. Although sociology has many theories about class, and different class schemes and scales which purport to identify the main class divisions or differences in Western capitalist societies, there is broad agreement to restrict the term ‘class’ to the analysis of inequalities with an economic base. Thereafter, in our view, the class terminology (as opposed to socio-economic status) becomes justified only if these inequalities separate the populations into demographically stable clusters or clumps of similar positions in the systems of economic production and distribution, if individuals typically remain within the same clusters throughout their working lives, and where ideally there is substantial inter-generational continuity. In the absence of any clustering there would be a situation of classless inequality; maybe socio-economic strata but not classes. Our analysis allows either verdict, though in practice we shall show that clusters are easily identified, and we apply the term ‘classes’ to these clusters.

The question then arises as to whether the members of these classes share additional social and cultural characteristics, in which case we can speak of ‘social’ rather than purely economic classes. If classes are socio-cultural in addition to economic entities, they may develop common conceptions of their interests and political orientations, which may lead to political action. However, economic classes will not necessarily develop in these ways, and even if they do so they will not necessarily hold conflicting world views. It is possible, in principle, to have classes without class conflict, and in our analysis none of these possibilities are foreclosed a priori.

2.3. The unit to classify

During the 1980s there was a sometimes acrimonious debate in Western sociology about the unit to treat in class analysis. The alternatives, advocated by different protagonists, were the individual and the family-household, then, if the latter, whose positions to take into account – all adults or just one main earner (see Abbott 1987; Britten and Heath 1983; Charles 1990; Erikson and Goldthorpe 1988; Goldthorpe 1983; Leiulfsrud and Woodward 1987; Stanworth 1984). This debate was never resolved, but subsequently the standard practice in Western sociology has been to opt for individual classification. The case for individual classification in Western countries has become increasingly persuasive alongside rising rates of labour-force participation by women leading to the normalisation of the dual-earning couple, the spread of cohabitation as an alternative or prelude to marriage, upward movements in typical ages of first marriages and child births, and rising rates of separation and divorce. The trends over time have been similar in the South Caucasus, but typical ages of first marriages and parenthood are still much lower than in Western Europe and North America, and cohabitation, divorce and separation are far less common (see Roberts et al. 2009a). Even more significant in our view, the three-generation family-household has remained a normal residential unit. This arrangement is traditional, remains a cultural preference to some extent, and has otherwise been enforced under communism by housing shortages, and subsequently by the price of dwellings rising beyond the means of most young singles and even young dual-earning couples.

The proportions of households in the 2009 DIS where three (occasionally more) generations co-resided were 23% in Azerbaijan, 25% in Georgia and 35% in Armenia. To these we can add the single- and two-generation households where there were additional members to parents and their own children: 9% in Armenia, 11% in Georgia and 15% in Azerbaijan (see Table 1). All told, therefore, 35% of the households in Georgia were extended-family households, 38% in Azerbaijan and 44% in Armenia.

TABLE 1. 
Household types by country (in percentages)
ArmeniaAzerbaijanGeorgia
1. Singleton 12 17 
2. Couple only 11 12 
3. Single parent and child 
4. Couple and own children 25 45 26 
5. Other (than a couple) one generation 
6. Other (than couple and own children) two generation 13 
7. Three or more generations 35 23 25 
N 2078 2011 1824 
ArmeniaAzerbaijanGeorgia
1. Singleton 12 17 
2. Couple only 11 12 
3. Single parent and child 
4. Couple and own children 25 45 26 
5. Other (than a couple) one generation 
6. Other (than couple and own children) two generation 13 
7. Three or more generations 35 23 25 
N 2078 2011 1824 

The likelihood of living in a three- (or more) generation household varied during the life-cycle (see Table 2). At the time of the 2009 DIS, majorities in all three South Caucasus countries were not living in three- or more generation households. However, most were likely to have lived in such households at some stage during their lives. Fifty-six percent of all households that contained at least one child up to age 15 were such households. This dropped to 36% when at least one 16–30-year-old was present, rose again to 43% when a 31–45-year-old was a household member, declined again to 32% when someone in the 46–60 age range was present, then rose once more to 46% when at least one person aged over 60 was in the household. The dips and ascents would have been due to grandparents dying around or after the time that new generations of children were being born and reared, then the children marrying and becoming parents, leading to a repeat of the cycle. The key fact, when considering the inter-generational transmission of class positions, is that in 2009 in the South Caucasus most children were being reared in three- or more generation family-households, and were continuing to live with their own parents after they themselves became married and produced their own children (see Roberts et al. 2009a). In this context, individuals may move upwards or downwards inter-generationally, but they simultaneously assist in promoting or demoting their entire family-households.

TABLE 2. 
Household types containing members in specific age groups (in percentages)
Up to 1516–3031–4546–60Over 60
1. Singleton 12 
2. Couple only 16 
3. Single parent and child 
4. Couple and own children 36 39 36 38 
5. Other (than a couple) one generation 
6. Other (than couple and own children) two generation 13 10 12 11 
7. Three or more generations 56 36 43 32 46 
N 3687 5815 4507 4667 3673 
Up to 1516–3031–4546–60Over 60
1. Singleton 12 
2. Couple only 16 
3. Single parent and child 
4. Couple and own children 36 39 36 38 
5. Other (than a couple) one generation 
6. Other (than couple and own children) two generation 13 10 12 11 
7. Three or more generations 56 36 43 32 46 
N 3687 5815 4507 4667 3673 

Our contention is that individual and household classification need not, and in the South Caucasus should not, be treated as alternatives. We will show that in the South Caucasus, and maybe elsewhere, there are advantages in using both approaches to class analysis complementarily.

In households that contain more than one adult generation it is difficult, and in our view counter-productive, to try to identify one ‘main earner’. The more prosperous family-households are likely to owe this position to maintaining several streams of income, and accordingly we have taken account of all adults’ (operationally defined as 25–60-year-olds) positions in identifying classes of households. We shall see that the apparent shapes of the class structures in the South Caucasus countries vary depending on whether the unit that is classed is the individual or the household. We shall argue that the ‘individual in the family-household’ is in fact the unit that portrays the shapes of the class structures most accurately.

3.1. Individual classification

We use the same three indicators to classify both individuals and family-households: employed versus not employed, education and income. In Table 3 the individuals who were in employment at the time of the 2009 DIS are labelled ‘Emp’. Those educated to at least BA level are labelled ‘Ed’. The samples are also split into the relatively ‘Poor’ and ‘Affluent’ (Aff) depending on whether their personal post-tax monthly incomes were up to or over the equivalent of $250.

TABLE 3. 
Individual class
ArmeniaAzerbaijanGeorgiaCapitalsRegionsMaleFemale
1. Ed Emp Poor 
2. Ed Emp Aff 1 1 
3. Emp Poor 28 17 18 13 24 27 17 
4. Emp Aff 10 
5. Ed Poor 15 18 11 
6. Ed Aff 
7. Poor 49 53 51 38 55 40 59 
8. Aff 
N 1906 1771 1860 1486 4055 2454 3075 
ArmeniaAzerbaijanGeorgiaCapitalsRegionsMaleFemale
1. Ed Emp Poor 
2. Ed Emp Aff 1 1 
3. Emp Poor 28 17 18 13 24 27 17 
4. Emp Aff 10 
5. Ed Poor 15 18 11 
6. Ed Aff 
7. Poor 49 53 51 38 55 40 59 
8. Aff 
N 1906 1771 1860 1486 4055 2454 3075 

It is evident from Table 3 that the main division by employment and labour market situations was whether or not individuals had any regular employment. Majorities in the capitals, and in the other regions in all three countries, were not in employment at the time of the survey (classes 5–8 in Table 3). In all three countries completion of some type of full secondary education had become normal before communism ended. During the 1990s participation rates in higher education rose to around or above 50% in Yerevan and Tbilisi, and around 30% in Baku. In the regions of all three countries the higher education participation rates were substantially lower than in the capital cities (see Roberts et al. 2008a). However, since the end of communism the main educational divide everywhere has been whether or not young people complete a higher-education programme. Our income threshold of $250 a month simply divides the employed samples into higher and lower earners. The income figures are as reported by the respondents and may not always be entirely accurate, but they correspond closely with reported household expenditure in the DIS, and look realistic when compared with official and anecdotal information about earnings in 2009.

As already stated, we have no occupational variable, but the DIS gathered information about the samples’ types of employer and the economic sectors where they worked (see Table 4). So it is not difficult to deduce that the large proportions of higher-education graduates who were employed in the state sector, in health, social care and education, would have been medics and other health care professionals, social workers and teachers.

TABLE 4. 
Individual classes and types of employment
Ed, Emp, PoorEd, Emp, AffEmp, PoorEmp, Aff
Type of employment  
Self-empl, no employees 10 41 19 
Self-emp with employees 
Small firm 
Larger firm 15 21 15 26 
State 63 50 23 26 
Foreign firm 
Non-profit org 
Other 10 
N 376 280 1151 281 
Industry  
Agr, fishing 36 
Mining 
Manuf 
Energy/water 
Construction 20 
Trade 10 10 11 
Hotels/catering 
Transport 11 
Banking/finance 
Real estate, renting 
Gov and defence 18 
Education 43 18 
Health care/social work 10 
NGO – – 
Other 14 13 16 
N 377 282 1149 284 
Ed, Emp, PoorEd, Emp, AffEmp, PoorEmp, Aff
Type of employment  
Self-empl, no employees 10 41 19 
Self-emp with employees 
Small firm 
Larger firm 15 21 15 26 
State 63 50 23 26 
Foreign firm 
Non-profit org 
Other 10 
N 376 280 1151 281 
Industry  
Agr, fishing 36 
Mining 
Manuf 
Energy/water 
Construction 20 
Trade 10 10 11 
Hotels/catering 
Transport 11 
Banking/finance 
Real estate, renting 
Gov and defence 18 
Education 43 18 
Health care/social work 10 
NGO – – 
Other 14 13 16 
N 377 282 1149 284 

The DIS shows that the best-educated members of the samples were the most likely to be employed, and when in employment they tended to earn more than the lesser educated. The distributions of the respondents in Table 3 confirm the findings from previous research in the South Caucasus (and everywhere else): higher-education graduates were more likely than non-graduates to be in jobs, especially the better paid jobs in their countries. Throughout Eastern Europe market reforms appear to have strengthened typical labour-market returns to higher education (see Domanski 2000; Treiman 1995).

The relationships between education, employment chances and incomes resulted in the samples forming clusters when grouped simultaneously using these three indicators. The largest cluster, labelled simply ‘poor’ in Table 3, were not higher-education graduates, they were not in regular employment, and unsurprisingly had post-tax personal monthly incomes of less than $250. The data in Table 3, and all the evidence that follows, is from 25–60-year-olds only; that is, persons of normal working age, and above the age when higher-education graduates would normally have commenced their labour-market careers. Between 49% and 53% of the respondents in each of the countries were in the ‘poor’ group. Another 8% of respondents in Armenia, 4% in Azerbaijan and 15% in Georgia were non-employed and income poor despite being higher-education graduates. In Azerbaijan another 4% were ‘affluent’ (personal monthly incomes above $250) despite reporting that they were not in employment. We will return below to how it is possible to be ‘affluent’ but not in employment. Respondents are unlikely to have described themselves or other household members as ‘employed’ unless they had regular jobs, with a contract, or were self-employed on an ongoing basis, and thus in receipt of regular and predictable incomes. Other respondents and household members are likely to have been working and earning irregularly. In total, 57% of 25–60-year-olds in Armenia, 61% in Azerbaijan and 66% in Georgia were reported as not in employment at the time of the DIS. Forthwith we call these their countries’ lower classes. The employment rates were low in all three countries. Azerbaijan's oil and gas revenues were probably responsible for employees in Azerbaijan tending to earn more than employees in the two other countries, but the Azerbaijan employment rate was not being boosted. Across all three countries, 50% of the males and 71% of the females were lower class on our criteria. It was 50% in the capital cities and 62% in other regions of the countries. The gender difference is consistent with other evidence of women being among the socio-demographic groups who have been losers during market reforms (see Bridger et al. 1996; Predborska 2005; Trapido 2007).

The next largest classes in Table 3 are composed of employed non-graduates, mostly ‘poor’ by our criteria, meaning that they had personal post-tax monthly incomes of less than $250. We provisionally call these the working classes. They accounted for 32% of the 25–60-year-olds in Armenia, 26% in Azerbaijan and 21% in Georgia; 37% of males and 19% of females across all three countries; 20% in the capital cities and 31% in the regions.

The third cluster is composed of employed higher-education graduates, who tended to be better paid than non-graduates, but those qualifying as affluent by our criteria amounted to a bare majority of employed graduates in Azerbaijan and were minorities in Armenia and Georgia. We call these groups the middle classes. They amounted to 11% of the 25–60-year-olds in Armenia, and 13% in both Azerbaijan and Georgia. Fourteen percent of males and 10% of females were in the middle classes on our criteria, and 20% of everyone in the capital cities and 9% in other regions.

We are using three variables – employment, income and education – to identify class clusters, and we are handling this information differently that when measurements of occupation, income and education are combined by awarding points for each to construct linear, gradational scales of socio-economic status. Our clusters are formed because being in employment (and types of occupation), levels of income and education were inter-related. Thus the clusters are categorical. Employment, income and education are all treated as determinants and indicators of positions in the systems of economic production and distribution. The inter-relationships create clusters, but the connections are imperfect, and in any case, all the separable indicators of class (including income, and authority and autonomy in work situations) are distributed gradationally. Hence all classes have fuzzy borders. The categories are conceptually distinct and identify clusters, but in actual social life there are no boundaries, analogous to ditches or fences, separating one class from another.

Table 4 lists the types of employers and the economic sectors where different classes of respondents worked. The higher-education graduates, especially the lower-paid graduates, were more likely than non-graduates to be employed in public-sector occupations. Forty-three percent of the lower-paid graduates were employed in education, 10% in health and social care, and a further 9% in other types of public-sector jobs, a grand total of 62%. Fifty percent of the higher-paid graduates also worked in the public sector, but were more likely than the lower paid to be in public administration rather than health, education and social care. The higher-paid graduates were also more likely than the lower paid to work in the private sector, in banking, manufacturing, trade and transport. Even so, despite all the countries having become market economies, it is clear that their new middle classes were based mainly in public sector jobs.

Non-graduate employees, provisionally (see below) called the working classes, tended to earn less than graduates. As many as 41% of the lower paid were self-employed without employees, and 36% worked in agriculture or fishing. Higher-paid non-graduates were more likely to work for private firms, in construction and transport, and if they were self-employed, they were more likely than the lower earners to have employees, though most of the higher-income self-employed did not employ anyone else.

The classes that we are identifying do not separate the self-employed into a separate class, a petit bourgeoisie, as do the occupation-based categorical class schemes developed and used in Western countries. Our argument is that self-employment in the South Caucasus (and most likely in many other new market economies) is different, and that the imposition of Western class schemes (which is technically straightforward if occupations are coded into SOCs) misrepresents the social realities. The largest group of self-employed in this research were farmers. Under communism they had worked household plots alongside main jobs in local factories or state or collective farms. Under post-communism they just had their (possibly enlarged) household plots. They would retain independent farming as their main occupation only for as long as they were unable to obtain proper jobs. Very few of their children had any desire to become farmers (see Roberts et al. 2008b). The urban self-employed were typically petty traders and taxi drivers. These were their main occupations only because, or for as long as, they were unable to obtain what they regarded as proper jobs. The self-employed in the South Caucasus do not exhibit the basic demographic stability to be treated as a distinct class. They are workers, mostly poor, with an affluent upper fringe.

Another difference from Western class structures, as represented in their class schemes, is that in the South Caucasus it is impossible to identify a separate service class, a salariat, that is distinct from other white-collar, non-manual, office employment. Higher-education graduates are typically recruited into the basic, lowest grades, then (try to) make career progress (Roberts et al. 2010). Employed graduates in all occupations are situated at different levels, close to or more distant from the cores of their countries’ emerging middle classes.

3.2. Household classification

We identify household classes using exactly the same variables as when identifying individual classes – education, employment and income – but take account of the education and employment or non-employment of all household members aged 25–60, and use total household income, divided by the number of 25–60-year-olds, splitting the households into relatively poor and affluent groups according to whether their per (25–60-year-old) capita post-tax monthly incomes were less or greater than $150. The bottom two classes of households listed in Table 5 (classes 7 and 8) had no-one in employment and no-one with higher education. The two classes immediately above (classes 5 and 6) had no-one in employment but at least one 25–60 year old with higher education. Both are split into relatively poor and affluent groups (the latter are tiny). We call all these groups the lower-class households. We describe the top two classes listed in Table 5 as middle-class households. All had someone employed and someone with higher education, and at least a half of their 25–60-year-olds were either higher-education graduates or in employment. Again, this middle class is divided into poor and affluent strata. The households in groups 3 and 4 in Table 5 had at least one member in employment, but did not meet all the criteria necessary to qualify as middle class. We provisionally call these middling groups the working-class households.

TABLE 5. 
Household classification
ArmeniaAzerbaijanGeorgiaCapitalsRegionsMaleFemale
1. Aff MC 11 13 13 26 13 12 
2. Poor MC 10 
3. Aff WC 17 28 11 25 17 19 18 
4. Poor WC 38 26 31 20 36 33 30 
5. Aff Ed Non-emp 
6. Poor Ed Non-emp 
7. Aff Non-empl 10 
8. Poor Non-emp 21 19 25 11 25 21 22 
N 1549 1576 1354 1174 3305 2100 2372 
ArmeniaAzerbaijanGeorgiaCapitalsRegionsMaleFemale
1. Aff MC 11 13 13 26 13 12 
2. Poor MC 10 
3. Aff WC 17 28 11 25 17 19 18 
4. Poor WC 38 26 31 20 36 33 30 
5. Aff Ed Non-emp 
6. Poor Ed Non-emp 
7. Aff Non-empl 10 
8. Poor Non-emp 21 19 25 11 25 21 22 
N 1549 1576 1354 1174 3305 2100 2372 

The distribution of the samples between the middle, working and lower classes differs considerably depending on whether individuals or households are classified. Treating households as the units, the working classes become the largest classes in each of the countries: 55% of households in Armenia against 32% when individuals are classified, 54% against 26% in Azerbaijan, and 42% against 21% in Georgia. The middle classes also become larger when households are classified: 17% of all households in Armenia compared with 11% of individuals, 15% compared with 13% in Azerbaijan, and 23% compared with 13% in Georgia. The lower classes look considerably smaller when households are classified: 29% of households against 57% of individuals in Armenia, 30% against 61% in Azerbaijan, and 36% against 66% in Georgia.

We acknowledge that both the pyramid shapes of the countries’ class structures when individuals are classed and the diamond shapes when households are the units are partly products of the criteria that we have used to allocate the units (individuals and households). However, in both cases the lower classes are identified on the basis of a total absence of employment. The apparent shapes of the countries’ class structures change basically because only 41% of poor individuals (not in employment, no higher education, and personal monthly incomes beneath $250) lived in poor households where no 25–60-year-old was employed or self-employed, no-one had higher education, and where the per (25–60-year-old) capita monthly income was beneath $150. Most non-employed individuals were members of households where someone was in employment, usually a person without higher education. In contrast, 74% of affluent middle-class individuals (employed graduates earning at least $250 a month) were members of affluent middle-class households, and 62% of poor workers (without higher education, earning less than $250 a month) were living in poor working-class households.

Individuals with the same class characteristics were tending to co-reside in the same family-households, and higher education was deeply implicated in the processes that were responsible. As in all countries where the relevant information has been available (see Blossfeld and Timm 2003), like was tending to marry like in the South Caucasus: higher-education graduates tended to have married one another. The children of well-educated parents were the young people who were most likely to progress through higher education. These relationships have been observed in all the countries where the relevant evidence has been assembled, and the DIS data shows that the South Caucasus countries are not exceptions. Like-with-like clustering in family households will be reinforced by the role of informal connections in local labour markets and business networks, a role that has been noted in research in all parts of the world including ex-Soviet countries (see Clarke 2000; Yakubovich and Kozina 2000). Family members who are in employment are in better positions to help each other to obtain (better) jobs than are the members of employment-less households. However, the tendencies for like to marry like, for children to mirror their parents’ educational achievements and for employment to breed further employment are just tendencies to which there are exceptions, and the net outcomes of the exceptions in the South Caucasus were a smaller proportion of households with no-one in employment than non-employed individuals, and increases in the proportions of the populations in working-class and middle-class households. We shall see that we need to take account of both individual and household class distributions in order to explain the character of the social and cultural dimensions of class in the South Caucasus.

The DIS collected information about the levels and also the sources of each household's income. Table 6 gives the proportions of households in each class that reported receiving any income from the listed sources. Some households in all classes were receiving financial support from non-residents, usually from abroad. Non-employed households were particularly likely to be receiving such assistance. In some cases it was the households’ main source of income, and the reason why some non-employed households were affluent (on our criterion) (see Table 7). ‘Remittances’ were also a source of income in around 6% of all households where someone was in employment.

TABLE 6. 
Percentages of households receiving any income from different sources
Poor Non-empAff Non-empPoor Ed Non-empAff Ed Non-empPoor WCAff WCPoor MCAff MCTotal
Any household income from the following sources 
Family/friends in country 11 14 17 
Family/friends elsewhere 14 24 35 10 11 
Family/friends anywhere 22 33 19 51 12 14 13 15 17 
Sale of agricultural products 39 27 14 17 41 22 29 12 31 
Other earned income 10 30 13 24 52 74 70 78 49 
State benefits 58 50 63 45 39 36 35 39 44 
Other 17 18 24 29 16 18 14 20 18 
N 904 283 82 70 1377 824 256 540 4336 
Poor Non-empAff Non-empPoor Ed Non-empAff Ed Non-empPoor WCAff WCPoor MCAff MCTotal
Any household income from the following sources 
Family/friends in country 11 14 17 
Family/friends elsewhere 14 24 35 10 11 
Family/friends anywhere 22 33 19 51 12 14 13 15 17 
Sale of agricultural products 39 27 14 17 41 22 29 12 31 
Other earned income 10 30 13 24 52 74 70 78 49 
State benefits 58 50 63 45 39 36 35 39 44 
Other 17 18 24 29 16 18 14 20 18 
N 904 283 82 70 1377 824 256 540 4336 
TABLE 7. 
Main sources of income in households in different household classes
Poor Non-empAff Non-empPoor Ed Non-empAff Ed Non-empPoor WCAff WCPoor MCAff MCTotal
Main source of household income 
Family/friends in country 11 
Family/friends elsewhere 10 17 29 
Sale of agricultural products 26 14 28 16 19 
Other earned income 24 12 19 43 65 61 70 42 
State benefits 44 28 56 23 14 11 13 13 22 
Other 11 13 14 
N 904 283 82 70 1377 824 256 540 4336 
Poor Non-empAff Non-empPoor Ed Non-empAff Ed Non-empPoor WCAff WCPoor MCAff MCTotal
Main source of household income 
Family/friends in country 11 
Family/friends elsewhere 10 17 29 
Sale of agricultural products 26 14 28 16 19 
Other earned income 24 12 19 43 65 61 70 42 
State benefits 44 28 56 23 14 11 13 13 22 
Other 11 13 14 
N 904 283 82 70 1377 824 256 540 4336 

At least 35% of all households in each of the employed household classes were in receipt of state benefits. Unsurprisingly, non-employed households were the most likely to be receiving such benefits, and these were typically their main source of income. Other households could qualify for benefits in other ways – a pensioner resident or qualifying for child benefit, for example. Some ‘affluent’ households named state benefits as their main source of income. The wide distribution of state welfare payments is the background to complaints that, in so far as poverty reduction is the priority, state largesse could be much better targeted (as discussed, for example, by Falkingham 2005).

Most of the employed households reported that earned income (from employment or the sale of agricultural products) was their main source of income. The proportions of households where no-one was employed but who reported some earned income, sometimes as their main source of income, is one indication of the size of the unofficial economies in the South Caucasus. Another indication is that some households with members in employment reported that their main income was from ‘other’ unspecified sources.

Western sociologists who treat the household as the unit to be classified may persist in the view that placement should be determined by the attributes (usually the occupation) of just one household member (see Goldthorpe 1983 for the classic defence). Irrespective of whether this procedure is defensible in Western Europe and North America, it simply does not match social reality in the South Caucasus where so many households are extended families in which there may well be more than one main earner, or no-one in employment. Households position all their members in the prevailing system of economic distribution, and differences between households in the South Caucasus cannot be properly assessed in terms of the attributes of a single selected member.

3.3. Socio-cultural dimensions of class

The main differences between household classes were in their standards of living, and income levels were responsible for most of these differences. These may appear to be statements of the obvious, but they need to be highlighted. An individual's standard of living depends less on his or her own education, job (if any) and personal income than on the resources of the entire household of which he or she is a member.

With rates of employment and levels of education within and between households held constant, per-capita income differences were related to whether households were equipped with DVDs, auto-washers, refrigerators, air conditioning, motor cars, mobile phones, PCs and Internet connections (if they had PCs) (see Table 8).

TABLE 8. 
Household classes and material resources (in percentages)
Poor Non-empAff Non-empPoor Ed Non-empAff Ed Non-empPoor WCAff WCPoor MCAff MCTotal
Not enough money for food and clothes 89 66 82 61 77 52 60 39 68 
DVD 27 39 22 44 40 51 40 56 41 
Auto washer 12 34 26 44 22 38 46 65 31 
Fridge 58 86 65 85 70 89 81 94 76 
Air cond 14 23 
Mob phone 71 83 71 93 85 93 93 98 85 
Car 12 21 13 35 23 26 36 40 24 
PC 10 16 31 18 29 46 15 
If have PC, internet access 31 53 77 87 50 63 60 76 65 
Possessions score 2.7 3.8 3.1 4.6 3.5 4.4 4.5 5.5  
Savings 16 14 20 22 12 
Debts 60 45 54 26 58 37 41 32 49 
2008 c/f 2007          
Better off 10 24 27 15 28 17 33 19 
Worse off 38 26 36 23 36 25 29 21 31 
2009 c/f 2008          
Better off 16 25 11 22 11 24 14 
Worse off 43 35 34 22 41 29 32 22 35 
          
Rungs on ladder 
1 & 2 42 19 35 21 27 12 17 20 
3, 4, 5 55 73 61 59 68 72 70 71 67 
6–10 20 16 13 28 13 
          
Relative to other households 
Very poor, poor 51 25 52 22 33 20 21 12 31 
Fair 46 66 44 66 62 63 66 65 60 
Good, very good 12 17 13 22 10 
          
Min monthly income for normal life 
500 39 20 42 14 24 15 21 
750 30 36 28 26 32 20 34 14 27 
1250 20 32 19 36 29 38 25 30 29 
Over 1250 11 12 11 24 15 34 27 52 23 
N 950 295 88 71 1412 846 263 554 4479 
Poor Non-empAff Non-empPoor Ed Non-empAff Ed Non-empPoor WCAff WCPoor MCAff MCTotal
Not enough money for food and clothes 89 66 82 61 77 52 60 39 68 
DVD 27 39 22 44 40 51 40 56 41 
Auto washer 12 34 26 44 22 38 46 65 31 
Fridge 58 86 65 85 70 89 81 94 76 
Air cond 14 23 
Mob phone 71 83 71 93 85 93 93 98 85 
Car 12 21 13 35 23 26 36 40 24 
PC 10 16 31 18 29 46 15 
If have PC, internet access 31 53 77 87 50 63 60 76 65 
Possessions score 2.7 3.8 3.1 4.6 3.5 4.4 4.5 5.5  
Savings 16 14 20 22 12 
Debts 60 45 54 26 58 37 41 32 49 
2008 c/f 2007          
Better off 10 24 27 15 28 17 33 19 
Worse off 38 26 36 23 36 25 29 21 31 
2009 c/f 2008          
Better off 16 25 11 22 11 24 14 
Worse off 43 35 34 22 41 29 32 22 35 
          
Rungs on ladder 
1 & 2 42 19 35 21 27 12 17 20 
3, 4, 5 55 73 61 59 68 72 70 71 67 
6–10 20 16 13 28 13 
          
Relative to other households 
Very poor, poor 51 25 52 22 33 20 21 12 31 
Fair 46 66 44 66 62 63 66 65 60 
Good, very good 12 17 13 22 10 
          
Min monthly income for normal life 
500 39 20 42 14 24 15 21 
750 30 36 28 26 32 20 34 14 27 
1250 20 32 19 36 29 38 25 30 29 
Over 1250 11 12 11 24 15 34 27 52 23 
N 950 295 88 71 1412 846 263 554 4479 

Relatively prosperous households were less likely to report having debts, more likely to have savings, and less likely to report that they did not have enough money for food and clothes. The more prosperous households tended to feel, and were more likely than poorer households to feel, that their standards of living were improving year by year, and (if employed) that they were doing ‘fair’ or ‘well’ compared with other households. The better-off households were more likely to place themselves above the bottom, though rarely higher than the lower-middle rungs on a 10-point socio-economic ladder, and felt that it was necessary to have higher total incomes in order to live a normal life than members of poorer households considered necessary.

The prosperity of households was related to their possessions, and also to their likelihood of having access to potentially useful forms of social capital: other family members who worked in the public sector or for international organisations (Table 10), and relatives and friends who were living and/or working abroad (see Table 9). Members of relatively prosperous households were also the most likely to feel that they could rely on help from both within and beyond their own families if difficulties arose (see Table 10).

TABLE 9. 
Household classes and international contacts (in percentages)
Poor Non-empAff Non-empPoor Ed Non-empAff Ed Non-empPoor WCAff WCPoor MCAff MCTotal
Relative currently abroad 37 55 33 51 43 48 45 56 45 
Friends abroad 16 23 24 31 22 27 29 41 25 
International contacts score 0.6 0.9 0.6 1.0 0.8 0.9 0.8 1.1  
N 950 295 88 71 1412 846 263 554 4479 
Poor Non-empAff Non-empPoor Ed Non-empAff Ed Non-empPoor WCAff WCPoor MCAff MCTotal
Relative currently abroad 37 55 33 51 43 48 45 56 45 
Friends abroad 16 23 24 31 22 27 29 41 25 
International contacts score 0.6 0.9 0.6 1.0 0.8 0.9 0.8 1.1  
N 950 295 88 71 1412 846 263 554 4479 
TABLE 10. 
Household classes and social support (in percentages)
Poor Non-empAff Non-empPoor Ed Non-empAff Ed Non-empPoor WCAff WCPoor MCAff MCTotal
Member with gov, pol, inter job 16 31 17 23 26 36 27 47 29 
If ill, HH member would help 42 52 38 63 48 55 53 71 51 
Can borrow cash from outside HH 48 55 51 64 52 64 61 73 57 
Social support scores 0.8 1.0 0.8 1.2 0.9 1.1 1.1 1.3  
N 950 295 88 71 1412 846 263 554 4479 
Poor Non-empAff Non-empPoor Ed Non-empAff Ed Non-empPoor WCAff WCPoor MCAff MCTotal
Member with gov, pol, inter job 16 31 17 23 26 36 27 47 29 
If ill, HH member would help 42 52 38 63 48 55 53 71 51 
Can borrow cash from outside HH 48 55 51 64 52 64 61 73 57 
Social support scores 0.8 1.0 0.8 1.2 0.9 1.1 1.1 1.3  
N 950 295 88 71 1412 846 263 554 4479 

Individual class was related to an entirely different set of differences, which were driven by different processes. Class-related differences among individuals lay primarily in their attitudes and features of their individual lifestyles (how they chose to spend their own money), and education rather than whether or not a person was employed and his or her personal income was responsible for most of these differences. The highly educated respondents were the most likely to read newspapers regularly, to use the Internet, to travel abroad, and they were the most interested in politics (see Table 11)

TABLE 11. 
Individual classes and lifestyles (in percentages)
Ed, empl, poorEd, empl, afflEmp, poorEmp, afflEd, poorEd, afflPoorAffl
Read papers at least once a week 59 66 31 39 47 65 23 19 
Use internet: ever 47 63 16 30 35 42 12 16 
Travelled abroad in last five years 23 45 17 28 20 31 13 15 
Agree or completely agree that TV journalists serve my interests 24 24 22 24 25 20 20 18 
         
Interested or very interested in: 
Foreign policy 64 67 49 55 61 50 37 31 
Domestic policy 69 75 53 63 66 62 44 40 
Local policy 67 63 56 59 63 54 47 42 
Interest in politics (high=interested) 2.9 3.0 2.7 2.8 2.8 2.8 2.5 2.4 
N 380 284 1164 285 514 26 2799 68 
Ed, empl, poorEd, empl, afflEmp, poorEmp, afflEd, poorEd, afflPoorAffl
Read papers at least once a week 59 66 31 39 47 65 23 19 
Use internet: ever 47 63 16 30 35 42 12 16 
Travelled abroad in last five years 23 45 17 28 20 31 13 15 
Agree or completely agree that TV journalists serve my interests 24 24 22 24 25 20 20 18 
         
Interested or very interested in: 
Foreign policy 64 67 49 55 61 50 37 31 
Domestic policy 69 75 53 63 66 62 44 40 
Local policy 67 63 56 59 63 54 47 42 
Interest in politics (high=interested) 2.9 3.0 2.7 2.8 2.8 2.8 2.5 2.4 
N 380 284 1164 285 514 26 2799 68 

The DIS included a variety of questions on trust in institutions. The most trusted institutions in the countries were religious institutions, the national armies, and the presidents. The respondents who expressed the highest levels of trust in the armies and presidents were those without higher education (see Table 12).

TABLE 12. 
Individual class and trust in institutions (in percentages)
Ed, empl, poorEd, empl, afflEmp, poorEmp, afflEd, poorEd, afflPoorAfflAll
Fully trust: 
Army 47 42 59 50 45 46 53 65 52 
Banks 14 12 19 21 15 16 17 16 17 
Education system 20 12 19 16 11 16 18 15 17 
Health system 14 10 18 15 11 12 15 12 15 
Court system 11 11 12 10 
NGOs 10 12 11 11 10 
Parliament 12 12 15 19 14 16 12 
PM and ministers 13 10 15 18 28 16 21 15 
President 29 32 34 43 22 40 35 57 34 
Police 15 13 18 13 14 12 16 21 16 
Media 11 14 11 10 10 
Local gov 10 11 19 13 12 15 15 15 
Religious institutions 58 53 59 54 63 56 56 45 57 
Ombudsman 12 12 19 15 12 14 15 12 15 
Mean trust score 3.3 3.3 3.4 3.5 3.2 3.2 3.4 3.4  
N 380 284 1164 285 514 26 2799 68 5520 
Ed, empl, poorEd, empl, afflEmp, poorEmp, afflEd, poorEd, afflPoorAfflAll
Fully trust: 
Army 47 42 59 50 45 46 53 65 52 
Banks 14 12 19 21 15 16 17 16 17 
Education system 20 12 19 16 11 16 18 15 17 
Health system 14 10 18 15 11 12 15 12 15 
Court system 11 11 12 10 
NGOs 10 12 11 11 10 
Parliament 12 12 15 19 14 16 12 
PM and ministers 13 10 15 18 28 16 21 15 
President 29 32 34 43 22 40 35 57 34 
Police 15 13 18 13 14 12 16 21 16 
Media 11 14 11 10 10 
Local gov 10 11 19 13 12 15 15 15 
Religious institutions 58 53 59 54 63 56 56 45 57 
Ombudsman 12 12 19 15 12 14 15 12 15 
Mean trust score 3.3 3.3 3.4 3.5 3.2 3.2 3.4 3.4  
N 380 284 1164 285 514 26 2799 68 5520 

When presented with the alternatives, the less educated were the more likely to opt for ‘governments should take care of the people, like a parent’. Those with higher education were more likely to opt for ‘the people should control the government’ (Table 13).

TABLE 13. 
Individual class and the role of governments (in percentages)
Ed, empl, poorEd, empl, afflEmp, poorEmp, afflEd, poorEd, afflPoorAfflAll
Role of gov 
Take care of people like a parent 52 44 63 59 48 44 66 70 61 
People should control the government 41 50 33 36 47 48 31 24 35 
N 380 284 1164 285 514 26 2799 68 5520 
Ed, empl, poorEd, empl, afflEmp, poorEmp, afflEd, poorEd, afflPoorAfflAll
Role of gov 
Take care of people like a parent 52 44 63 59 48 44 66 70 61 
People should control the government 41 50 33 36 47 48 31 24 35 
N 380 284 1164 285 514 26 2799 68 5520 

The above social and cultural features must be regarded as part of what the relevant classes ‘are’ rather than effects or correlates of class. It is such features that account for the ability of some classes, in particular places and at particular times, to become political actors, and to make a real difference – that is, to have significant societal effects.

The main class-related differences among households in the South Caucasus are in living standards, and these differences are gradational. There are clusters of households in terms of the education of their members, their chances of being in employment and their earnings, but the main outcome at the household level is variations in living standards which position households on a continuum along which there are no clear breaks. Class divisions in the South Caucasus are created by the differences among individuals, which are driven mainly by education, contrasting the university-educated (typically employed, in the public sector, in better-paid jobs), who themselves are concentrated within middle-class households, with the rest. Individuals in corresponding class households, or households where all adult members have the relevant class characteristics, can be regarded as comprising the core units in a class, as in the list of ‘composite’ classes (combining individual and household features) in Table 14. Such individuals and households are at the centre of class clusters around which there are individuals in households which pull the individuals into the upper or lower fringes of their classes, as indicated by the individuals’ positions (see Table 14). Alternatively, it may be said that some individuals pull their households up or down into the class fringes.

TABLE 14. 
Class comparisons, all countries
IndividualsHouseholdsComposite
Ed Emp Aff Aff MC 12 Core MC 10 
Ed Emp Poor Poor MC Fringe MC 
Emp Aff Aff WC 19 Upper WC 
Emp Poor 21 Poor WC 32 Core WC 26 
Ed Aff Ed Aff Under Lower WC 
Ed Poor Ed Poor Under   
Aff Aff Under Fringe Lower 27 
Poor 51 Poor Under 21 Core Lower 29 
IndividualsHouseholdsComposite
Ed Emp Aff Aff MC 12 Core MC 10 
Ed Emp Poor Poor MC Fringe MC 
Emp Aff Aff WC 19 Upper WC 
Emp Poor 21 Poor WC 32 Core WC 26 
Ed Aff Ed Aff Under Lower WC 
Ed Poor Ed Poor Under   
Aff Aff Under Fringe Lower 27 
Poor 51 Poor Under 21 Core Lower 29 

One outcome in the South Caucasus countries is distinctive middle classes amounting to around 20% of the populations in the capital cities, and around 10% elsewhere. Contrary to suggestions made elsewhere and referring to the situations in other ex-Soviet countries in the 1990s (for example, Ilyin 1998), we find that by 2009 new middle classes really had been formed in the South Caucasus, and that these classes were not mirages. The middle classes in the South Caucasus are well-educated, more likely than the members of other classes to be in employment, typically in the public sector, and they tend to be the highest earners. Their households are well-equipped with durables, and we know from previous research that these middle classes supply most visitors to the theatre, the cinema, galleries and museums, restaurants and suchlike (see Roberts et al. 2009b). This makes the middle classes high profile – visible to one another, and to members of other classes. The middle classes are well-connected in numerous ways – to other people who are in employment, in good jobs and to people who are abroad. They read newspapers regularly, use the Internet, travel outside their own countries and are more interested in public affairs than the members of other classes.

In the evidence from the DIS it is impossible to detect a clear socio-cultural break between the working classes and the lower classes. Nor is there a clear demographic break: this is because most non-employed individuals are in households where someone is employed (an employed husband with a non-employed wife, for example). This prevents those concerned forming distinct lower or under-classes (as also found in East-Central Europe by Domanski 2002). Employed non-graduates, typically with lower-paid jobs, and their households are probably best regarded as simply less deprived than the rest of the lower classes. In the lower classes most people of working age are not in regular employment. Those with jobs tend to earn less than the middle classes. A minority have regular jobs in the public sector or with private businesses, but employment outside the middle classes is more likely to mean self-employment in agriculture or fishing, trading, in construction or transport (taxi-driving, for example). Members of the lower classes are unlikely to use the Internet or read newspapers regularly, or travel outside their own countries. We are suggesting that the countries do not possess distinct under-classes or, therefore, distinct working classes. Other investigators have noted that throughout Eastern Europe the working classes have become ideologically and politically invisible (Simonchuk 2005; Stenning 2005). We find that in the South Caucasus working classes cannot be identified even as demographic formations. Such classes would have existed under communism, but have been destroyed by the closure of factories and the break-up of state and collective farms.

We do not intend to attribute to the classes that we have identified interests that they themselves might not recognise, and we acknowledge that in identifying class cultures we are limited to the evidence collected in the DIS. If more or different questions had been asked, a different picture may have emerged. That said, our evidence does not indicate a relationship of conflict between the middle and lower classes.

Most lower-class individuals and families did not feel especially deprived. They were more likely to perceive themselves as situated on the lower-middle than on the very bottom rungs of their countries’ socio-economic ladders, and they were more likely to describe their circumstances as ‘fair’ than ‘poor’ or ‘very poor’. They probably saw themselves as sharing with most other families the elements of prosperity and the hardships that were normal in their countries where roughly two-thirds of the DIS respondents said that their households did not have enough money to buy all the food and clothing that they needed, and more felt (in 2009) that they were becoming worse off rather than better off from year to year. Nevertheless, roughly 80% of all households owned their own dwellings, colour TV was virtually universal, and 85% had mobile phones. Longer-term, most expected their situations to improve. Three-quarters expected their own children to be better off when they reached the respondents’ ages.

Lower-class respondents were less likely than middle-class respondents to do so, but most of the former declared that they were ‘very interested’ or ‘quite interested’ in politics. Trust in institutions was low in all classes. As mentioned above, the most fully trusted institutions were religious institutions (57%), national armies (52%), and presidents (34%). The lower classes expressed the highest levels of trust in the armies and presidents, and preferred governments ‘to take care of’ rather than ‘to be controlled by’ the people. We can infer, though this was not asked in the DIS, that for the lower classes ‘taking care’ meant creating jobs and providing education and health services as well as maintaining external and internal security. If so, no parts of this agenda clashed with typical middle-class concerns and aspirations.

Most members of the lower classes did not feel especially deprived, and few members of the middle classes felt particularly privileged. Two-thirds described their circumstances as simply ‘fair’ compared with other households. A similar proportion placed themselves on rungs three, four or five – lower-middle on a 10-point socio-economic scale. Middle-class respondents felt that they needed more money than lower-class respondents found necessary to lead a normal life. The circumstances of the majority of middle-class individuals and households appeared to fall well short of their own aspirations. Over a half of the higher-education graduates in Georgia, around 40% in Armenia, and around 30% in Azerbaijan (which has a lower participation rate in higher education than the other two countries) were not in employment at the time of the DIS. Among those who were in employment, a bare majority in Azerbaijan, but only minorities of higher-education graduates in Armenia and Georgia, were earning at least $250 per month after taxes. It seems highly likely that the middle classes will want their governments to create more commensurate jobs for well-qualified young people, and to raise the relevant (public sector) salaries. Most middle-class respondents in the DIS were not satisfied with conditions in their countries, or with the direction of change. Less than a half believed that people like themselves were treated fairly by their governments, or that politics was going in the right direction. Most had additional complaints about politics in their countries. The typical lower-class grievance was that the governments were insufficiently or unsuccessfully paternal. The typical middle-class complaint was that politics was insufficiently democratic – responsive to the people's wishes.

Neither the middle nor the lower classes appeared to blame the other for their own discontents. All the anecdotal and research-based evidence indicates that the main targets of middle-class and lower-class grievances are ‘the rich’ and ‘politicians’ as a career group, with whom the rich are believed to be thoroughly implicated. The rich were indistinguishable within the DIS but we do not doubt their existence (see Eyal et al. 1998; Hoffman 2002). Class cultures form over generations rather than years or even decades. In the longer term, class-based political movements may be formed in the South Caucasus. If so, in our view, at present the most likely outcome will amount to a reversion to pre-communist politics – the people versus the new (rich) aristocracies and their (political) accomplices. However, the post-communist context is different. There is no utopian vision, let alone a theory about how it could be made a reality.

Abbott
,
P.
,
1987
.
Women's social class identification: Does husband's occupation make a difference?
,
Sociology
21
(
1987
), pp.
91
103
.
2003
.
Blossfeld
,
H-P.
, and
Timm
,
A.
, ed.
Who Marries Whom? Educational Systems and Marriage Markets in Modern Societies.
Dordrech
:
Klewer
;
2003
.
Bogomolova
,
T.
,
1998
.
Income mobility in Russia in the mid-1990s
, paper presented at the International Sociological Association Congress, Montreal, Canada, July.
Bridger
,
S.
,
Kay
,
R.
, and
Pinnick
,
K.
,
1996
.
No More Heroines? Russia, Women and the Market.
London
:
Routledge
;
1996
.
Britten
,
N.
, and
Heath
,
A.
,
1983
. “Women, men and class analysis”. In:
Garmarnikow
,
E.
,
Morgan
,
D.
,
Purvis
,
J.
, and
Taylorson
,
D.
, eds.
Gender, Class and Work.
London
:
Heinemann
;
1983
. pp.
79
96
.
2005
.
Data Initiative Survey 2005
, Baku/Tbilisi/Yerevan.
Charles
,
N.
,
1990
.
Women and class: A problematic relationship
,
Sociological Review
38
(
1990
), pp.
43
89
.
Clarke
,
S.
,
2000
.
The closure of the Russian labour market
,
European Societies
2
(
2000
), pp.
483
504
.
Domanski
,
H.
,
2000
.
On the Verge of Convergence: Stratification in Eastern Europe.
Budapest
:
Central European University Press
;
2000
.
Domanski
,
H.
,
2002
.
Underclass and social stratification in post-communist societies
,
Sisyphus (Social Studies)
16
(
2002
), pp.
109
21
.
Erikson
,
R.
, and
Goldthorpe
,
J. H.
,
1988
.
Women at class crossroads: A critical note
,
Sociology
22
(
1988
), pp.
545
53
.
Evans
,
G.
, and
Mills
,
C.
,
1999
.
Are there classes in post-communist societies?
,
Sociology
33
(
1999
), pp.
23
46
.
Eyal
,
G.
,
Szelenyi
,
I.
, and
Townsley
,
E.
,
1998
.
Making Capitalism Without Capitalists: The New Ruling Elites in Eastern Europe.
New York
:
Verso Books
;
1998
.
Falkingham
,
J.
,
2005
.
The end of the rollercoaster? Growth, inequality and poverty in Central Asia and the Caucasus
,
Social Policy and Administration
39
(
2005
), pp.
340
60
.
Ganzeboom
,
H. B. G.
,
De Graaf
,
P. M.
, and
Treiman
,
D. J.
,
1992
.
A standard international socio-economic index of occupational status
,
Social Science Research
21
(
1992
), pp.
1
50
.
Goldthorpe
,
J. H.
,
1983
.
Women and class analysis
,
Sociology
17
(
1983
), pp.
465
88
.
Hoffman
,
D. F.
,
2002
.
The Oligarchs: Wealth and Power in the New Russia.
Oxford
:
Public Affairs Limited
;
2002
.
Hoffmann
,
E.
,
1999
.
International Statistical Comparisons of Occupational and Social Structures: Problems, Possibilities and the Role of ISCO-88.
Geneva
:
International Labour Office
;
1999
.
Ilyin
,
V.
,
1998
. “The new middle strata in modern Russia”. In:
Kivinen
,
M.
, ed.
The Kalamari Union: Middle Class in East and West.
Aldershot
:
Ashgate
;
1998
. pp.
118
29
.
Khmelko
,
V.
,
2002
.
Macrosocial change in Ukraine: The years of independence
,
Sisyphus
16
(
2002
), pp.
125
35
.
Kutsenko
,
O.
,
2002
.
Dynamics of class formation process in Ukrainian society
,
Sisyphus
16
(
2002
), pp.
137
50
.
Leiulfsrud
,
H.
, and
Woodward
,
A.
,
1987
.
Women at class crossroads: Repudiating conventional theories of family class
,
Sociology
21
(
1987
), pp.
393
412
.
Lockwood
,
D.
,
1958
.
The Blackcoated Worker.
London
:
Allen and Unwin
;
1958
.
Oksamytna
,
S.
, and
Khemlko
,
V.
,
2004
.
Social exclusion in Ukraine in the initial stages of the restoring capitalism
,
Ukrainian Sociological Review
4
(
2004
), pp.
179
92
.
Piirainen
,
T.
,
1998
. “From status to class: The emergence of a class society in Russia”. In:
Kivinen
,
M.
, ed.
The Kalamari Union: Middle Class in East and West.
Aldershot
:
Ashgate
;
1998
. pp.
314
41
.
Prandy
,
K.
,
1998
.
Deconstructing classes: Critical comments on the revised social classification
,
Work, Employment and Society
12
(
1998
), pp.
743
53
.
Predborska
,
I.
,
2005
.
The social position of young women in present-day Ukraine
’,
Journal of Youth Studies
8
(
2005
), pp.
349
65
.
Roberts
,
K.
, and
Pollock
,
G.
,
2009
.
New class divisions in the new market economies: Evidence from the careers of young adults in post-Soviet Armenia, Azerbaijan and Georgia
,
Journal of Youth Studies
12
(
2009
), pp.
579
96
.
Roberts
,
K.
,
Pollock
,
G.
,
Manasyan
,
H.
, and
Tholen
,
J.
,
2008a
.
School-to-work transitions after two decades of post-communism: What's new?
,
Eurasian Journal of Business and Economics
1
(
2
) (
2008a
), pp.
103
29
.
Roberts
,
K.
,
Pollock
,
G.
,
Rustamova
,
S.
,
Mammadova
,
Z.
, and
Tholen
,
J.
,
2009a
.
Young adults’ family and housing life-stage transitions during post-communist transitions in the South Caucasus
,
Journal of Youth Studies
12
(
2009a
), pp.
151
66
.
Roberts
,
K.
,
Pollock
,
G.
,
Tholen
,
J.
, and
Tarkhnishvili
,
L.
,
2009b
.
Youth leisure careers during post-communist transitions in the South Caucasus
,
Leisure Studies
28
(
2009b
), pp.
261
77
.
Roberts
,
K.
,
Tholen
,
J.
,
Huseynadze
,
D.
,
Ibrahimov
,
A.
, and
Pollock
,
G.
,
2010
.
Graduate careers during the post-communist transition in the South Caucasus
,
Educational Research and Reviews
53
(
3
) (
2010
), pp.
108
18
.
Roberts
,
K.
,
Tholen
,
J.
,
Khachatryan
,
G.
,
Velidze
,
R.
, and
Pollock
,
G.
,
2008b
.
Transitions to adulthood in rural villages during the transition from communism in the South Caucasus
,
Social Problems
4
(
2008b
), pp.
117
40
.
Rose
,
D.
,
1998
.
Once more unto the breach: In defence of class analysis once again
,
Work, Employment and Society
12
(
1998
), pp.
755
67
.
Rose
,
D.
, and
Harrison
,
E.
,
2007
.
The European Socio-Economic Classification: A new social class scheme for comparative European research
,
European Societies
9
(
2007
), pp.
459
90
.
Simonchuk
,
E.
,
2004
.
The working class in Ukraine: Chronicle of losses
,
Ukrainian Sociological Review
5
(
2004
), pp.
155
78
.
Stanworth
,
M.
,
1984
.
Women at class crossroads: A reply to Goldthorpe
,
Sociology
18
(
1984
), pp.
159
70
.
Stenning
,
A.
,
2005
.
Where is the post-socialist working class? Working class lives in the spaces of (post)-socialism
,
Sociology
39
(
2005
), pp.
983
99
.
Tilkidjiev
,
N.
,
1996
. “Social stratification in post-communist Bulgaria”. In:
Coen-Huther
,
J.
, ed.
Bulgaria at the Crossroads.
New York
:
Nova Science Publishers
;
1996
.
Trapido
,
D.
,
2007
.
Gendered transition: Post-Soviet trends in gender wage inequality among young and full-time workers
,
European Sociological Review
23
(
2007
), pp.
223
37
.
Treiman
,
D. J.
,
1995
.
Social Stratification in Eastern Europe After 1989.
Los Angeles
:
University of Los Angeles Press
;
1995
.
Yakubovich
,
V.
, and
Kozina
,
I.
,
2000
.
The changing significance of ties: An exploration of hiring practices in the Russian transitional labour market
,
International Sociology
15
(
2000
), pp.
479
500
.

Ken Roberts is Professor of Sociology at the University of Liverpool. Since the early 1990s he has coordinated a series of studies in former communist countries on the impact of the changes among their young people. His books include Youth in Transition in Eastern Europe and the West (2009) and Class in Contemporary Britain (2011).

Gary Pollock is Head of the Department of Sociology at Manchester Metropolitan University. His research interests include the transitions of young people from school to work, and the application of sequence analysis to youth life course development.

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) license, which permits unrestricted use, distribution, and reproduction in any medium, provided the use is non-commercial and the original work is properly cited. For a full description of the license, please visit https://creativecommons.org/licenses/by-nc/4.0/legalcode.