The research design and method

The research presented here forms part of a broader longitudinal study of 17 low-income households living in Ankara (Eroğlu, Reference Eroğlu2004). The study involved two visits to the field, one in April 2002 and the other in October 2002. Longitudinal design was chosen because it comprises a robust research base for identifying causal processes, and collecting high-quality data. The choice of design helped enhance data quality especially through allowing time for building rapport, which proved to be a gradual process. Field visits were set six months apart because, although there is no perfect time for longitudinal research, shorter studies are less susceptible to external shocks and sample attrition. This study was indeed able to avoid both risks. Besides, the six-month gap was first considered and then proved adequate for exploring household plans and whether they were realised.

The research setting was selected from a typical gecekondu settlement of Ankara which predominantly consisted of pre-1985 rural to urban migrants from surroundings of the city or other parts of Central Anatolia (Güvenç, Reference Güvenç and Yavuz2001). Gecekondu originally refers to squatter housing which was partly legalised in the 1980s. These areas make up a considerable part of the population in big cities of Turkey and tend to have a higher incidence of poverty (Bulutay, Reference Bulutay1998). The chosen settlement reflected this general tendency. There was also heterogeneity in the levels of poverty experienced in the settlement, given the occupational status of its inhabitants, which extended from public sector employees and seasonal labourers to the unemployed.

The sample included households with an average monthly incomeFootnote ² below US$370,Footnote ³ four members, nuclear structure, one child at or above the age of 15 and Alevi and Sünni religious backgrounds. Those who fell below the threshold correspond to the bottom 15 per cent of the Ankara population (Sönmez, Reference Sönmez2001). Further, the intended household size and structure reflect the typical characteristics of urban households in Turkey, including gecekondu dwellers, and the selected religious groups represent the two largest Islamic sects in Turkey, which differ in terms of their religious practices, political viewpoints and their approaches to women (Ayata, Reference Ayata1997; Shankland, Reference Shankland and Rasuly-Paleczek1996).

The sample was obtained through random selection by starting from an official database containing the intended household size, structure and age of children for two areas within the same gecekondu settlement: one predominantly Sünni and one predominantly Alevi. Random number tables were used to select from a sample frame, including 156 out of 2005 households, and then households with an income above US$370 were excluded during access negotiations.

Data were drawn from semi-structured interviews conducted separately with both partners. In all but one household they were interviewed on both occasions to cross-check their accounts. Qualitative interviewing was preferred to a household survey in order to generate detailed and valid data. The sample size was therefore small, requiring further research to test whether conclusions reached here are applicable to the wider population. Qualitative interviewing was also considered more appropriate for this part of the research than participant observation in that it made it possible to gather information about households’ financial motives and plans.

The generated data were analysed both qualitatively and quantitatively. Qualitative analyses sought to provide a detailed understanding of the gün features, member characteristics, risk-coping mechanisms, rationale for participation and the scope for asset accumulation. Quantitative analyses were performed to explore patterns of gün membership and its relationship to income. Due to the small sample size, these were based on descriptive statistics only, and used qualitative data and theoretical knowledge obtained through literature review as a basis for causal inferences made from cross-tabulations.

Understanding patterns of overall gün participation

In contrast to earlier studies where gün organisation was described as an activity of the middle class (Bellér-Hann, Reference Bellér-Hann and Rasuly-Paleczek1996; Khatib-Chahidi, Reference Khatib-Chahidi, Ardener and Burman1995), this research revealed considerable interest from low-income households. In April, 53 per cent (9 out of 17) reported participation in various güns. The total number of güns observed was ten, as one household took part in two. By October, the participation rates had fallen to 41 per cent (7 out of 17). The total number of güns recorded in October was nine, since this time there were two households with two separate memberships. Four of the güns recorded in April ended after the completion of the first round, and five new güns were formed by those with at least one membership in April, out of which one was completed before October. Hence, the total number of güns observed throughout the study was 15.

Compared to formal credit and savings institutions, güns seem more accessible to these households, as in April and/or October only 17 per cent (3 out of 17) had a bank loan and 12 per cent (2 out of 17) had savings in a deposit account, so there were in total five households with access to formal channels (29 per cent), four of which were also gün participants and had at least one regular earner. The only one without a regular income had a loan credited officially to a relative who had a secure job.

Despite the popularity of güns among the poor households, April figures indicate a greater tendency for those with a relatively higher income to have a membership. Table 1 shows that in April all higher-income households participated in at least one gün, whereas the participation rate was only 17 per cent (1 out of 6) for those with a lower income, which is expected given the former's greater capacity to save.

TABLE 1.

Household income categories by gün membership in April and October

Notes: *Income categories were formed by using cut-off points obtained through the SPSS frequency facility. These points correspond to 240 and 307 US dollars.

However, October results suggest a weaker relationship between membership and income size,Footnote ⁴ indicating a decline in the participation of higher-income households to 40 per cent (4 out of 10). How can we make sense of this decline? Between April and October, 71 per cent (12 out of 17) had experienced a considerable rise in their incomes due to this period being the peak work season. Nearly all non-member households with a relatively low or medium income in April had one or more seasonal earners, most of whom were main earners working in the construction sector for a daily wage. Despite an increase in their earnings over the six months, these households tended to avoid gün participation. There were two-member households with predominantly seasonal incomes, but these were able to participate only in güns involving either very small payments or a few members to allow quick completion of a round. The general tendency for non-participation may be attributed to income irregularity, which undermines the household capacity to keep up with long-term periodic payments. The following extracts from spousal interviews confirm this point.

Incidentally, the feeling of shame and embarrassment provoked by the thought of default appears to have affected decisions against participation. This is supported further by the excerpt below, taken from an interview with a female spouse:

Overall, three main conclusions can be drawn here. First, seasonality in the labour market is likely to reduce gün participation by creating income instability as well as reducing the amount of annual income available. It will be shown in the next section that seasonal work can also affect participation by undermining trust-building among work-mates, which is needed to compensate for the lack of a formal contract. Second, income stability may play as important a role as income size in determining gün enrolment. Third, poor households do not necessarily turn to gün-type informal credit and savings instruments because they lack access to formal opportunities. Their motives for participation will be explored later in detail.

Composition of gün membership

This section explores the composition of gün membership in terms of gender, the length of acquaintance and the nature of social contexts to which members belong. Thereby, it seeks to throw light on some of the risk-coping mechanisms used.

First of all, the majority of the observed güns were all-female. Only in 13 per cent (2 out of 15) did both spouses have a separate or a joint membership. Second, the güns generally consisted of members who had medium- to long-term acquaintance. This meant that members had formed an opinion of each other's level of trustworthiness and creditworthiness, as a strong trust basis is considered a requirement to ensure that members are committed to the payment schedule in the absence of a formal contract. Newcomers were also allowed entry through a reference from a trustworthy member, but this was quite a rare occurrence. Of the ten güns observed in April, for instance, two had one newcomer each; both were from the respondent households which participated in güns within adjacent neighbourhoods. Although newcomers were not well acquainted with other members, they were granted access through their member relative(s) who had long-term residence in the area.

Last, the güns contained members who belonged to relatively stable social contexts. In all but one of the güns observed, the neighbourhood was the social context that connected nearly all members. Most members were close neighbours of either the respondents or their relatives and had a long history of residence in the same gecekondu settlement. Their relationship was multi-stranded in nature; apart from being neighbours they were also relatives, hemşehris from the same home town, friends and/or affiliates from a particular religious circle. The Alevis and Sünnis tended not to mix. A much less common social context was the workplace. Only one of the observed güns comprised members who were regular employees of a textile factory, the implication being that for güns to be organised among work-mates, there had to be a certain degree of stability at the workplace. The following extract from an interview with the husband of a woman who was enrolled in this gün confirms this inference:

Two points can be made about the social context of membership. First, it seems that relatively stable social contexts are the more likely preference for membership. Second, the choice of social context appears to reflect the work situation of female gün members who represented the majority. In the sample, all but one female member were either housewives or had casual jobs that involved working in isolation (such as daily house cleaners or home-based pieceworkers). Hence, the social context from which güns were most likely to draw members remained the neighbourhood with its stable as well as convenient base for membership.

Overall, two criteria appear central to the selection of gün members: length of acquaintance and stability of the social context. Both help reduce the risk of default by strengthening the trust basis. The latter also allows a certain degree of surveillance over members’ movements, which pressurises them to maintain the payments as well as improves their ability to chase the defaulters. Therefore, the selection criteria can be viewed as one of the mechanisms used to cope with the risk of default.

Motives for gün participation

This section explores what motivates low-income households to become gün members. It starts by discussing why certain motives are less likely.

Firstly, despite women's disproportionately greater involvement in güns, the need to protect savings from men's conflicting financial demands is unlikely to be the key motive, as none of the memberships was concealed from husbands. This research did reveal secret kitties kept by women to this effect. However, these were separate from gün savings and were obtained from other sources. To illustrate, women cut back on claimed personal expenditure or inflated claimed household expenditure to generate funds for their kitties (Eroğlu, Reference Eroğlu2009).

Moreover, women are unlikely to have become members for purposes of leisure, as the social gathering aspect, which is a common feature of the güns organised by middle-class women (Bellér-Han, Reference Bellér-Hann and Rasuly-Paleczek1996; Khatib-Chahidi, Reference Khatib-Chahidi, Ardener and Burman1995), was missing from almost all of the güns observed. Women avoided such gatherings due to the monetary costs involved. The motive to obtain social support is also less likely given the length of acquaintance and the degree of connectedness among members.

Finally, as will be shown in detail later, gün savings were either kept for rainy days or used to buy household supplies in bulk or to service loans borrowed for investment puposes. The early purchase of a ‘lumpy’ consumption item, however, appears less likely as a primary motive mainly because the risk of being the last recipient is equalised across members through random assignment (see also Gugerty, Reference Gugerty2007). This risk seems rather to act as a deterrent as it means a delay in gaining access to funds when needed. This is evident from the excerpt below:

So going back to our original question: what was the key rationale for low-income households to join güns? One possible answer can be found within the following excerpt from an interview with a man who jointly participated with his wife in a gün that contained 11 members and required contributing a small gold coin every month.

Evidently, the failure to make gün payments was viewed as strongly as ‘being disgraced in public’. What seems implicit here is the concern for reputation and avoidance of shame and embarrassment one would feel in the case of default (see also above). From the strength of this perception, it appears that members were compelled to make regular savings for gün payments in order to preserve their reputation as a trust/creditworthy person and avoid feelings of shame and embarrassment. It also seems that such financial discipline may not be entirely self-imposed but emanate from community pressure. The excerpt below illustrates how the neighbourhood community could force members to remain faithful to a payment schedule:

In light of these considerations, gün organisations can be construed as a ‘forced savings mechanism’ providing the poor households with the discipline of saving a portion of income which would otherwise be easily spent on pressing consumption needs. Hence, one key motive for participation can be to achieve the discipline required to resist the consumption pressures obstructing the objective to save. As will be shown in detail later, the general tendency for households to have clear plans about their gün savings substantiates this argument.

Key features of the observed gün types

This study identified three gün types locally known as şeker günü (meaning sugar day), altın günü (meaning gold day) and dolar günü (meaning dollar day). This section explores their key features and working principles.

Gün types had the following features in common. First, all of them were based on a verbal contract. Second, none of them required a specific person to be assigned the role of organiser, as in roscas observed in other parts of the world (see Ardener, Reference Ardener1964). Third, the number of members was fixed for one round. Fourth, the order of rotations was determined by drawing lots. Fifth, gün payments or contributions at each rotation were the same for all members. Last, either the contributions or the lump sum was fixed to a medium whose value could be protected against inflation during the life of the gün.

The most distinctive difference among these güns was in the medium of the contributions or the lump sum, according to which they were labelled. There were also variations in terms of membership size and rotation frequency, but these features were not specific to any particular type (see Table 2).

TABLE 2.

Distribution and characteristics of the güns observed in April and October

Notes: * indicates the new güns that were formed after the April interviews. Also note that the tables exclude one şeker günü which was organised after April and completed before October.

† in determining the dollar equivalence of member contributions and the lump-sums, the study drew on April and October averages for the price of one quarter gold coin and dollar in the free market. The averages were obtained from the gold and dollar rates for the 1st, 15th and 30th of the given months, published respectively by Ankara Association of Jewellers and Altınkaynak (2002) Foreign Exchange Rates Archive.

The working principles of these güns are summarised below to examine the differences more closely. Starting with şeker günü, the lump sum here was in Turkish Lira but based on the wholesale price of 50 or 100 kg of sugar. At each rotation, the price was obtained from a certain dealer and divided by the number of members to determine the size of contributions. This allowed price increases to be accommodated, thereby giving members an equal chance to buy the same amount of sugar. In altın günü, the lump sum was composed of quarter gold coins. In every rotation, each member contributed a small gold coin, known as çeyrek (that is, quarter), to a fund which was then given to the member whose turn was due. Thereby, all members were able to acquire the same amount of coins, despite changes in the price of gold.

Finally, dolar günü was based on similar working principles as altın günü, the key difference being that the contributions were fixed to the US dollar (equivalent to $50 per member at each rotation). The lump sum was equal to the fixed amount multiplied by the number of members. This meant that members had the opportunity to obtain an equal amount of US dollars despite possible fluctuations in exchange rates.

The three gün types compared

This section examines the popularity and investment potential of the three gün types and compares their ability to cope with the risk of default, enable rapid access to funds, and prevent unequal distribution of (dis)advantages between members. Through this comparison, the pros and cons of the different gün types are highlighted.

Of the three types, şeker günü was most popular and dolar günü least. Şeker günü membership was evident in 53 per cent of all the güns observed (8 out of 15), whereas the respective figures for altın and dolar günü memberships were 40 per cent (6 out of 15) and 7 per cent (1 out of 15). Popularity seems inversely related to the size of contributions, which was smallest in şeker günü and greatest in dolar günü. From Table 2, it can be seen that the monthly contributions to the şeker günüs were less than half the size of those made to the others. It follows from this that in circumstances of poverty, the size of contributions can be critical to the popularity of a gün type since the smaller the size, the less the pressure on income and hence the lower the risk of default.

Güns’ popularity was thus closely linked with their ability to cope with the risk of default. Although there were no defaults, şeker günü members were better protected against this risk as the amount they had to contribute was very small. The fixed nature of the lump sum gave them the advantage of being able to keep the size of contributions small by increasing the number of members participating in a round. Members of altın and dolar günü were denied this advantage due to their contributions being fixed. Yet they had the flexibility to increase the fund size by admitting more participants at the start of a round. Şeker günü was less flexible as the need for sugar and/or other household supplies set the limits of the funds. Combined with the payment size, this explains why the fund sizes of altın and dolar günü tended to be larger than those of şeker günü (see Table 2). The opportunity to gain access to larger funds made the first two types more suitable for investment purposes.

The ability to allow urgent access to the funds was dependent on the order of rotations and rotation frequency. As mentioned earlier, in all güns the order by which members received the lump sum was determined by lots drawn before starting the rotations. This meant that the lump sum was not necessarily accessible by members when they needed it most. Nevertheless, those güns with more frequent rotations at least enabled quicker access. Table 2 demonstrates that all of the altın and dolar günüs rotated on a monthly basis, whereas some of the şeker günüs had twice-monthly rotations. Owing to the small size of contributions, members of şeker günü had the advantage of being able to organise more frequent rotations, and hence a greater capacity to respond to the urgent household needs, which required more cash than was normally available.

Finally, in comparing the ability of güns to prevent unequal distribution of (dis)advantages between members, it should be borne in mind that such inequalities are likely to occur where there is high inflation and/or an opportunity to obtain real returns from the lump sum. In an inflationary situation, members are faced with the risk of depreciation of the real value of the lump sum, which would disadvantage its subsequent recipients. In her study of altın günü, Khatib-Chahidi (Reference Khatib-Chahidi, Ardener and Burman1995) described the application of gold standard as a uniquely simple method of administration and organisation, which allows the members to combat inflation without having to use complicated compensatory mechanisms such as discounting and auctioning. This description can be extended to all three gün types as they simply apply the gold, dollar or sugar standard to protect the value of savings throughout rotations.

Şeker günü effectively counteracted inflation by accommodating changes in the wholesale price of sugar. Between April and October, the increase in the six-monthly rate of inflation was approximately 14 per cent (TÜİK, 2002a), whereas the price of sugar increased by 25 per cent: the cost of 100 kg sugar rose from 106 million TLFootnote ⁵ (US $80) to 138 million TL (US $82). The additional cost was equally distributed between members, resulting in an increase in their contributions. In this way, all members had an equal chance to buy the same amount of sugar despite changes in price. On the other hand, the effectiveness of altın and dolar günüs in offsetting inflation was dependent on whether the rates of real return from gold and US dollars were positive. The respective six-monthly rates for the period between April and October were 16.9 per cent and 11.8 per cent (TÜİK, 2002b),Footnote ⁶ indicating that altın and dolar günüs were also able to provide effective protection against inflation.

The above-inflation increases in the price of gold, sugar and US dollars indicate a possibility to make real returns from the funds during the life of the güns. However, the potential for real returns was subject to fluctuations in the commodity and money markets and tended to be limited. In particular, şeker günüs were likely to produce the lowest level of real returns as they had the smallest fund size. However, this put them in a better position to avoid inequalities likely to emerge in the distribution of real returns due to market fluctuations. Such inequalities might be more pronounced in altın and dolar günü whose members, due to the rotational allocation of funds, are faced with the risk of missing the opportunity to convert the funds into Turkish Lira when the gold or currency prices were more favourable. However, this risk was perceived as not significant enough to prevent participation.

Plans and reality: how were gün savings used?

Through an exploration of household plans, this section seeks to understand how gün savings were used, and to what extent they contributed to asset accumulation. The following analysis thus focuses on households with membership in April.

At this stage, nine households participated in güns, but as one household was a member of two separate güns, the observed total was ten. Three households, all of which were members of şeker günü, had received their lump sum payments prior to the interviews in April. Two of them used their gün savings to buy food and cleaning supplies in bulk (such as sugar, oil and washing up liquid), and carried on as members of the same gün whose first round continued throughout the fieldwork. The third used the lump sum to pay off a small portion of the loans borrowed over a period of unemployment for everyday needs. Given the size of the lump sum (53 m TL, or US $40 in April) which was equal to 3.6 per cent of the household's outstanding debts for that month, this household was unable to secure significant debt relief. Also the prospects for their new şeker günü membership were also low given the limited nature of the lump sum expected (69 m TL or US $41 in October).

On the other hand, seven households were due to receive their lump sum by October. Two of them were şeker günü members. Their plan was to use the lump sum for consumption purposes; for example, to buy food and cleaning supplies in bulk. Both households fully realised their plans, and, having obtained adequate supplies for the forthcoming winter, one quit after the first round ended, whereas the other embarked on a new şeker günü with fewer people due to further need for winter supplies.

The remaining five were either members of an altın or a dolar günü, with plans to use their lump sums for investment purposes. Two of them had the intention to keep the gün savings for ‘rainy days’. One was able to act as planned but had to quit after completing the first round since the coincidence of the second round with the school year made monthly savings no longer viable for this household. The other, however, abandoned its plan and used all the gün savings, according to the wife rather unwisely, to pay towards an old second-hand car. Despite some financial difficulty, this household managed to remain a member of the same gün whose first round continued throughout the fieldwork.

The other three households had plans to service part of the investment loans they borrowed from a bank or other people. One was planning to use the lump sum to finance the quarterly instalments of the title deed for the gecekondu they currently occupied. Separate membership of each spouse within the same altın günü had the potential to secure considerable debt relief, yielding funds that were nearly equal to two instalments per year (22 gold coins, equal to US $334 in April). However, the couple was able to pay off the June instalment only, as the rest was needed to cover the cost of their attendance at a relative's wedding and the following month's contribution to the new altın günü that was set up with ten members soon after the successful completion of the previous one. As her husband had recently been made redundant, only the female spouse took part in this gün, and managed the monthly savings with the help of their working son. As a result, the household's capacity to secure cumulative debt relief through gün membership was reduced by more than half.

The second household was intending to pay off part of the loans borrowed to set up a family bakery. The specific plan was to return the twisted golden bracelet they owed to a neighbour by topping up the lump sum with the female spouse's earnings from bootee sales. Although this plan did not fully materialise, the household was able to pay back US $180, corresponding to 6.7 per cent of their outstanding debts in April. The repayment was less than planned for because within the April–October period, when the effects of economic crisisFootnote ⁷ in 2001 had been still felt rather strongly, the female spouse was able to earn very little from bootee sales despite her large network of potential clientele. The female spouse was indeed fully aware that gün membership brought little debt relief. In actual fact, neither her earnings nor gün savings prevented the bakery from going bankrupt two months prior to the October interview. Despite the bankruptcy, the household managed to carry on in the same gün whose first round still continued, but their prospects for participation in future rounds were weakened dramatically.

The last household was planning to use the lump sum to pay off a portion of the debts accumulated through the purchase of a two bedroom flat. The lump sum (US $150) corresponded to 3.5 per cent of this household's outstanding debts for April. The expectation was to secure cumulative debt relief through several quick rounds between a few members. After the completion of the first round, it appeared difficult to realise this plan because, due to being composed of casual and seasonal labourers, this household lacked the regular and adequate income needed to make sizeable gün payments. The household managed to avoid default by becoming indebted to other gün members as well as using the first lump sum obtained to pay off the debts created in the process. The female spouse was able to continue with the second round but voiced her intention to quit after this round due to income constraints.

Overall, most households were able to use their gün savings in accordance with their consumption or investment plans. However, the level of success achieved in realising these plans varied greatly due to income constraints linked with precarious/seasonal work conditions, and the wider effects of the economic crisis. From the ways in which savings were used, it became apparent that the observed güns acted as a means of survival and/or investment. Şeker günüs appealed more to households with consumption plans, and were used to obtain extra cash required to buy certain household supplies in bulk. Except for the last recipient, all members had the opportunity to purchase these commodities earlier than they would if they had to save on their own. This gün type was much less attractive to households with an investment plan. Conversely, altın and dolar günüs were used either to make emergency savings or to service loans obtained for investment. However, their ability to promote asset formation was highly restricted for two reasons. First, the size of savings obtained through a single round was often very small, as compared with the amount of debts incurred on large-scale assets. Second, most members were unable to sustain membership beyond the first round due to the same constraints on income. Thus, for the majority, güns provided little scope for asset accumulation.