Residential segregation based on taxable income in the metropolitan area of Athens

Conceptual approaches to income segregation

Income is a defining factor for the selection of a place of residence, while housing costs are strictly connected to the cost of neighbouring houses. Real estate agents, appraisers and buyers use the latest prices of comparable real estate in local housing markets to assess prices. In parallel, since mortgages are linked to income, the choices of residents are defined by their income. In general, these mechanisms set a budget constraint on people’s available choices of where to live, creating an income-based stratification of housing locations. The aforementioned income segregation -the unequal distribution of individuals and households into different spatial units based on income- is apparent in most cities and is characterised by three key parameters:

• First, the classification of residential areas on the basis of income can lead to the so-called prosperity and/or poverty separation. “Prosperity separation” indicates the uneven distribution of high income and low income households at a local scale and “poverty separation” is determined by the uneven distribution of low and non low-income families among spatial partitions.
• A second important aspect of income segregation is its connection to geographic areas demonstrating other forms of social segregation (e.g. racial). The neighbours that residents prefer contribute to hierarchies that are not identical, but complementary to those resulting from income inequality.
• A third parameter of income division is the geographic scale. Resident hierarchy based on income can be expressed mainly through large-scale hierarchy schemes of residential areas. That would be the case, for example, if all high-income families lived in the suburbs, and all low-income families in the city centre. Small-scale hierarchy schemes of residential areas would be the case if high and low-income residents were distributed in a chessboard pattern in the metropolis. In this scheme homogeneous wealthy neighbourhoods are located next to homogeneous poor neighbourhoods.

This division is a process with a high degree of heterogeneity. As for the analysis of spatial effects between socio-professional groups, there are two conceptual components of housing separation:

• The first component is spatial exposure or, in other words, spatial isolation. Spatial exposure assesses the extent to which people belonging to a particular group tend to remain isolated from (or mix with) members from other groups.
• The second component is spatial evenness or spatial clustering. Spatial evenness refers to the way in which members of a population group are allocated. It assesses how evenly they are distributed in space. In other words, it examines the degree to which members are placed near or away from one another.

Data and study area

The quantitative data used for our statistical analysis was drawn from the General Secretariat of Information Systems (GSIS) of the Ministry of Finance. It is based on tax returns of individuals per postal code and includes: the number of tax returns and declared income (at current values) of predetermined income brackets and broader professional categories. The timespan of the available data includes the time period between 2003 and 2013.

Table 1: 

At this point, four important observations should be made regarding data reliability and the validation of the explanatory power of the analysis. Recording reported income may not represent the levels of prosperity precisely, since it does not represent the (unknown) size of the shadow economy and tax evasion.

1. Even if one tried to depict the size of the shadow economy at a national level, it would be impossible to define the level of its spatial heterogeneity at a local level.
2. Over time, as shown below, declared income has significantly fluctuated -especially for certain professions- mainly due to changes in taxation (e.g. change of presumed income criteria, taxpayer obligation to submit a tax declaration form if they own real estate, even if they have zero income etc.). This resulted in abrupt fluctuations in the declared income distribution and changes in the calculation of average incomes.
3. The use of fixed income brackets at current values ​​can pose problems in long-term comparisons of income inequality, especially in times of intense economic change.

The study area covers the entire region of Attica and includes 508 entries corresponding to postal sectors, out of which 289 are spatially identifiable. The remaining entries refer to Post Office Boxes and Post Offices and therefore cannot be geographically located.

General overview

The Attica region accounts for over 42% of the reported income of individuals in the country, though with a steadily declining share for the period 2003-2013 (Figure 1). Accordingly, the average declared income (declared income divided by the number of tax returns) is constantly above 115% of the national average throughout the entire period. The development of total declared income both at country level and within Attica can roughly be broken down into three sub-periods (Figure 2.1 and 2.2):

a) Between 2003 and 2008 there was a steady upward trend that culminated in 2008 (18,900 per capita) and is linked to the general economic climate of the period.

b) Between 2008-2010 there was relative stagnation (for high income levels), which seems to not be aligned with the deteriorating socio-economic context, but is probably related to the change in taxation in 2009. Thus, shrinking available income per head was counterbalanced by the enlargement of the fiscal base.

c) From 2011 to 2013, however, the effect of the financial crisis became visible in the tax base too, which shrunk significantly up to 2013. Since then, the average declared income has been less than that of 2003. Average income breakdown is worth mentioning here, as there are significant changes in income distribution: up to 2005, there were extreme maximum and minimum values ​​that affected the reported average income (Figure 2.1 and 2.2).

Figure 1: Participation of the Region of Attica in the total number of tax returns and declared income throughout the country (2003-2013)

Source: GSIS 2004-2014

Figure 2.1: Evolution of declared income in 2005 prices for the Region of Attica and the entire country (2003-2013)

Source: GSIS 2004-2014

Figure 2.2: Breakdown of average declared income in 2005 prices for Attica (2003-2013)

Source: GSIS 2004-2014

As regards the allocation of tax returns and declared incomes to income brackets in the period 2003-2013 (Figure 3), there has been a gradual increase in the participation of higher income brackets compared to medium-low brackets through 2010. In 2011, this trend reversed, as the lower income bracket shrunk and low-middle brackets expanded, returning to the levels of the previous years. Over time, income inequality, as reflected in the Gini index (Figure 4.1), is associated with the increase in average declared income until 2010, the rapid declared income decline to 2003 levels thereafter and the income stabilisation during 2012-2013. Correspondingly, changes in polarisation (i.e. in the ratio of the two highest income brackets divided by the two lowest income brackets) shows the difference in the rate of change between higher and lower brackets both in the growth period and during the recession (Figure 4.2).

Figure 3: Allocation of tax returns and declared income into 10 income brackets for the Region of Attica and for the entire country (2003-2013)

Source: GSIS 2004-2014

Figure 4.1: Gini index (income inequality) for declared income in the Region of Attica (2003-2013)

Figure 4.2: Relative index of the rate of income inequality (high/low income brackets ratio) for declared incomes in the Region of Attica (2003-2013)

As regards the distribution in broader occupational categories, the main observations can be summarised in the following points (Figure 5.1): Salaried and Pensioners are the dominant source of taxable income. The widening of the tax base in the past three years corresponds to a slight increase in the participation of the Rentiers category in tax returns, however without a proportional increase in declared incomes. On the contrary, the weight of the Pensioners’ category gradually increased as of 2003 (tax return and income share). This also indicates structural unemployment problems both in Greece and in Attica in particular.

The percentage of declared income for Merchants/Artisans/Traders and Liberalss, who theoretically constitute the upper part of the socio-economic scale, remained stable . However, their contribution in tax revenue declined as a percentage of the total. All professional groups in Attica are clearly above the national average with respect to their average declared income. Reductions in income “shifted” after 2010, the year of the most significant cuts in wages and pensions, although the categories of Merchants/Artisans/Traders and Liberals seem to be affected more by the economic crisis (Figure 5.2).

Figure 5.1: Distribution of tax returns and declared income into broader income brackets for the Region of Attica and the entire country (2003-2013)

Source: GSIS 2004-2014

Figure 5.2: Average declared income of broader income brackets in 2005 prices for the Region of Attica and the entire country (2003-2013)

Source: GSIS 2004-2014

Spatial Inequality

Mapping declared income in the broader metropolitan area of ​​Athens for the year 2013 shows distinct patterns of high and low income concentrations (Map 1.1). For comparison purposes, we have used smaller size representations for the years 2003, 2008, 2010. Previous studies (Μαλούτας 2001; Καλογήρου2011) have evidenced the east-west residential division , which our analysis also shows, together with a division between centre and periphery with local differentiations between north-south, particularly in the eastern part of Attica. More specifically, higher ​​declared incomes are concentrated in the north-eastern part of the metropolitan agglomeration (Erithrea, Ekali, Nea Pendeli, Vrilissia), in the centre of the Municipality of Athens in areas around Lycabettus (Kolonaki, Mavili Square, Evangelismos), in the southern part (Voula, Vouliagmeni) and in an enclave that includes Psychiko and some zones of the Municipality of Filothei. Lower incomes cluster in the western part of the Municipality of Athens, in particular the area of Metaxourgio, a number of zones in the Municipality of Agios Ioannis Rendis and Tavros as well as parts of the north-western zone comprising Aspropyrgos, Fyli, Acharnes and Kamatero.. There is limited “mobility” of the spatial analysis units in the income hierarchy. Areas at the upper end of the distribution have a very high probability to stay there or move upward, while low-medium areas are likely to “move” either higher or lower (Table 2.1).

Map 1.1: Average reported income by Postal Code in the Athens metropolitan area for the years 2003, 2008, 2010 and 2013

Table 2.1: Probability of transition from the initial to the final distribution by deciles for reported average incomes inAthens metropolitan area, 2003-2013

To examine the “mobility” of a system, we use the Markov chain statistical method, which examines the possibility of variations in the distribution of values ​​from a specific (initial) state to a final state through a transition route with discrete steps. The (hypothetical) probability of a system’s distribution in the next step (and basically in all future steps) depends exclusively on the current state of the system and is not aggregated based on the state of the system in previous steps. In this particular case, the spatial distribution of the average declared income is considered to be a “system”, while the distribution of declared income into income brackets is the “condition”. Finally, annual recordings of the variable’s values are considered as the “steps”. For the purpose of this presentation, we used a statistical method that takes into account neighbouring values for each spatial unit (Rey 2001). Thus, the double entry table created also represents the probability of a value “moving” from one income bracket to any other, from the base year to the final year of the period reviewed. The values in the cells on the diagonal of the transition matrix describe the probability of an area to remain in the same income bracket from the beginning to the end of the period. High values ​​indicate stagnation in the spatio-temporal dynamic of the distribution of declared incomes. Values ​in the upper part of the matrix (defined by the diagonal) indicate movement towards a relatively lower income level and values ​in the lower part of the matrix indicate movement towards a higher level of income.