Financial instruments of the poor: initial findings from the South African Financial Diaries study

Abstract

A new data set called the South African Financial Diaries has been produced, based on a sample of 166 households, drawn from three different areas in South Africa – Langa, Lugangeni and Diepsloot. The selected households represent a range of dwelling types and wealth categories. A unique methodology was used to create a year-long daily data set of every income, expense and financial transaction used by every one of these households. Within this sample, households used, on average, 17 different financial instruments over the course of the study year. A composite household portfolio, based on all 166 households, would have an average of four savings instruments, two insurance instruments and 11 credit instruments. The same composite household portfolio would have about 30 per cent formal instruments and 70 per cent informal instruments. Interestingly, it was found that rural households use as many financial instruments as urban ones.

1. Introduction and review of the literature

There is a strange irony when thinking about financial management of poor households. One thinks that by the very nature of not having money, the poor cannot possibly manage the limited resources they do have. However, empirical facts do not support this assumption. In financial diaries surveys in both Bangladesh and India (see Rutherford, 2002; Ruthven, 2002; Ruthven & Kumar, 2002) it was found that the poor tended to manage their money through a variety of financial instruments. In South Africa, the same is found to be true.

The environment within which households operate in low-income countries makes the process of financial decision making quite different for those in developed countries. The life cycle hypothesis (Modigliani, 1970), for example, states that earnings are less than consumption after retirement and exceed consumption during the middle years of earning. Rational people should base their consumption decisions on expected lifetime income rather than current income. However, many researchers (Ando et al., 1992; Deaton, 1993; Spio & Groenewald, 1996) reject this hypothesis in most low-income countries. Savings seems, instead, to be precautionary and held for insurance reasons.

Another classic economic theory, the permanent income hypothesis (Friedman, 1957), proposes that rational individuals will try to smooth consumption if income is disrupted. Therefore, transitory income shocks should have no significant effect on consumption. Permanent income shocks (such as suffering a major disability) will, however, translate into lasting changes in consumption. This theory works reasonably well in developed countries where mechanisms such as insurance and credit can be used to effectively smooth income streams with little disruption in consumption. However, in low-income countries existing mechanisms do not always work well, and households may be forced to cope following a shock by drawing down savings, selling assets, working longer hours and doing without key services such as health and education or without key goods such as certain foods.

Based on data generated by a financial diaries method, Rutherford (2002) tracked household financial flows over the course of a year in Bangladesh and confirmed that households manage their portfolio of cash assets actively with a broad range of instruments. Moreover, different levels of poverty do not mean different levels of active management. Taken altogether, financial flows in poor areas are substantial, but mostly small per transaction.

Ruthven (2002) used the same financial diary methodology in India and her results echo much of what was found by Rutherford (2002) in Bangladesh. The results confirmed that significant sums of money were raised for lifecycle costs. Although they were poor, people allocated disproportionately high amounts to issues related to health. House construction was also extremely important. The results also confirmed that the most widely and frequently used financial device was family and reciprocal contacts. The transactions are small but interest free. Leaning on friends and neighbours is a regular strategy to cover deficits and bridge cash flow. Lastly, they confirmed that slightly different portfolios of financial devices are used by households of differing levels of wealth or livelihood, although all levels of wealth used financial devices. Most respondents were saving by hiding money at home, giving interest-free loans or putting money into a bank savings account. Most were borrowing by taking an interest-free loan, a wage advance or a private loan with interest.

2. Response to the literature and research questions

The approach used by Rutherford (2002) and Ruthven (2002) in Bangladesh and India was shown to provide helpful data on the use of a wide range of financial instruments by low-income households, suggesting that households are managing their money actively in an attempt to smooth consumption in some way. The approach measured how successfully financial devices are used to accomplish this. In South Africa, a perspective on the financial management of poor households will be useful to form the context of debates about how well financial instruments are serving the needs of the poor. The purpose of the South African Financial Diaries data set is therefore to fill in some of the missing background to the financial lives of South Africa's poor households.

This paper provides the background to the South African Financial Diaries data set. In Section 3, the sample and sample selection are discussed. The methodology is outlined in Section 4. Section 5 provides some of the initial observations of the data set. It indicates the numerous financial instruments households use and outlines some of the differences between households in rural versus urban areas and households of differing relative wealth. Section 6 concludes, and highlights some of the further discussions to which the Financial Diaries data set could contribute.

3. The Financial Diaries sample

To draw the sample for the South African Financial Diaries, the same method was used as in Rutherford (2002) and Ruthven (2002): a participatory wealth ranking (PWR). In South Africa, this method is used by the Small Enterprise Foundation (SEF), a prominent NGO microlender based in the rural Limpopo Province. Research has shown the PWR process to robustly identify poor households within selected villages and neighbourhoods. Simanowitz (1999) compared the PWR method to the Visual Indicator of Poverty (VIP) and found that the VIP test was seen to be at best 70 per cent consistent with the PWR tests. At times one-third of the list of households that were defined as the poorest by the VIP test were actually some of the richest according to the PWR. The PWR method was also implicitly assessed in van der Ruit et al. (2001) by being compared to the Principle Components Analysis (PCA) used by the Consultative Group to Assist the Poor (CGAP) as a means of assessing client poverty. They found that three-quarters of those defined as poor by the PCA were also defined as poor by the PWR. This study closely followed the SEF manual to conduct wealth rankings on which to select the sample for the South African Financial Diaries data set. After the year-long survey, it was found that the sampling method resulted in a wide variety of household incomes and sources of funds in the three Financial Diaries samples.

Monthly income per household

Three sites were selected for the Financial Diaries sample. Langa, the urban sample, is close to Cape Town in the Western Cape; Diepsloot, the peri-urban sample, lies between Johannesburg and Pretoria in Gauteng province; and Lugangeni is a small village outside Mount Frere in the Eastern Cape. As Figure 1 shows, most of Financial Diaries households have monthly incomes below R5000 per month, but the income distributions for each area differ widely. In Langa and in Diepsloot, three-quarters of the households have monthly incomes between R1000 and R5000, while only 15 per cent have monthly incomes below R1000. In rural Lugangeni, on the other hand, nearly 45 per cent of the households have incomes below R1000 per month. This is even more striking when one considers that this definition of income takes into account all non-financial cash flows that a household might receive, including remittances from relatives living elsewhere, business profits, rental income, regular and casual wages, grants and pension income.

Figure 1: Average monthly household income distribution (percentage of households in each area)

Monthly income per household member

This picture becomes more pronounced when one looks at monthly income per household member, as shown in Figure 2. The United Nations Millennium Development Goals (MDGs) suggest that it is key to assess those living below the threshold of $1 per person per day. The fluctuating rand/dollar exchange rate makes it difficult to determine a clear lower boundary in rand for the Financial Diaries households. We chose to set our lowest bracket at R200 per month or less, which is roughly $1 per day at an average exchange rate of R6.40 per US$.

Figure 2: Average monthly income per household member (percentage of households in each area)

Not only do rural households receive less income as a whole, they also tend to have more members. Whereas the average household size in the urban and peri-urban samples is about three, in the rural sample it is just over five. As a result, a far larger proportion of households in the rural sample of Lugengeni (25 per cent) fall within an income per member range of less than R200 per month than in the urban areas (0 per cent and 4 per cent, respectively, in Langa and Diepsloot).

Sources of income

Another key difference between the rural and urban samples of the Financial Diaries is the source of income (as shown in Figure 3). On average, a rural household relies on government grants for 48 per cent of their household income. Another 19 per cent comes from remittances from relatives, while 15 per cent comes from regular jobs. In Diepsloot, on the other hand, 60 per cent of household income, on average, comes from regular jobs, while only 5 per cent of income is from grants. In Langa, income from regular jobs accounts for 55 per cent of the average household income, while grants account for 15 per cent of average monthly income. Self-employment is only 10 per cent of household income in the urban and peri-urban areas, while in Lugangeni self-employment income registers a meagre 3 per cent of household income.

Figure 3: Sources of income (percentage of average monthly household income)

4. Diary methodology

This study adopts the same diary methodology that was used in Rutherford (2002) and Ruthven (2002) in Bangladesh and India, but attempts to build on the information gathered in the first studies to enhance the quantitative output in the South African Financial Diaries study. The sample was also expanded from 42 households to 166 households, and whereas the financial diaries studies in India and Bangladesh were mostly unstructured interviews and open-ended discussions, the South African Financial Diaries uses a combination of closed- and open-ended questionnaires. The initial three questionnaires are structured and gather information on household demographics, physical assets, typical income and expenditure patterns, historical and current employment, and lastly current and previous use of financial instruments. There are roughly 28 predefined financial instruments, each with its own questionnaires to record different aspects of the instrument. Each existing financial instrument receives its own financial device code against which cash flows are captured in the future. The initial interviews not only allow the household to become more comfortable with the fieldworkers but are also used to create an initial balance sheet position, as well as a typical monthly cash flow statement.

Diary questionnaires

[Examples of the exhibits discussed in the following section may be downloaded from the Financial Diaries website (www.financialdiaries.com) under the Methodology section.]

The households were interviewed every other week for a year, capturing every cash flow coming into and going out of the household, including income, expenditure, changes in physical assets, servicing financial instruments and initiating financial instruments. To facilitate the collection of data, data from the initial three questionnaires was used to produce a diaries questionnaire specific to each household. This was used to prompt memory, aid data collection, save the respondents time and encourage patience. Data was inputted by fieldworkers each week for the previous week's interviews and new diary questionnaires were generated. The diary process was enabled by a specially conceived and built Access Database and the consistent weekly capture of data.

Each week the respondents were also asked if they had done anything new in the past two weeks, for example opened a new bank account, joined a new stokvel (savings club), or stopped a financial device. Each new financial device was captured on a specific form and cash flows generated by that device were captured thereafter. If a financial device was closed, a ‘close’ form was captured and cash flows on that device were not captured thereafter. They were also asked if a major event had occurred, if a person had joined or left the household, if a new or casual job had been taken on or abandoned or if a new physical asset has been bought, sold or stolen. Each week the fieldworker also filled in a journal in which he or she noted various observations, events or comments made by the respondent that had not been captured elsewhere in the diary questionnaire.

Ongoing data checks

A key feature of the database and the ongoing diary system is the ease with which the data can be checked. Two key checks were built into the system. The first is a report that shows the project coordinator which households have been interviewed in two-week periods for the last six weeks and which have not. This report was run every week and followed up in a conference call with all six fieldworkers. The second is a report that tracks the sources and uses of funds in and out of the household from the time of the most recent interview to the time of the previous interview. Allowing for cash in hand in both interviews, the sources of funds (income, taking a loan, drawing money from a bank account, etc.) should equal the uses of funds (paying for food, putting money under the mattress, repaying a loan, etc.). Any household that had an excess or deficit of sources versus uses of funds greater than R200 was examined individually to detect where the imbalance was coming from. If the imbalance was coming from an input error, the data was corrected. If it seemed to come from a lack of reporting, the fieldworker was informed and told which details to follow up in the next interview. In this way, the data were continuously checked to ensure quality.

5. Initial results from the Financial Diaries data set

The poor hold a portfolio of diverse financial instruments (Figure 4)

Over the course of the study year, it was found that households would use an average of 17 different financial instruments. Some financial instruments, such as a bank account, would ‘stay open’ all year, while others, such as borrowing between neighbours, would open and close within days. Although credit is the type of financial instrument most often associated with poor households, we found that most households used a variety of insurance, credit and savings instruments. The poor do not tend to use only one type of financial instrument – they manage a portfolio. Most households have at least one credit, insurance and savings instrument. They do not only borrow, and they do not only save.

Figure 4: Average number of financial instruments used during the year (per household)

Rural households tend to use as many financial instruments as urban households (Figure 5)

One might assume that, because they are further from formal financial services, the rural poor might use fewer financial instruments. However, it was found that rural households tended to use as many financial instruments as urban households. They also tend, however, to use more informal than formal instruments. One reason for this is that they use not just one burial society but several, and they have very active lives lending to and borrowing from each other.

Figure 5: Average number of financial instruments used during the study year (per household)

Most transactions go through formal financial instruments (Figure 6)

All households push and pull a significant amount of money through financial instruments over the course of a month. We capture this activity by turnover. Turnover is calculated by adding the inflows into a financial instrument to the outflows of a financial instrument, over a particular period. In Figure 6, we chose to show August as a typical month of financial instrument turnover in various types of financial instruments.

Figure 6: Monthly turnover for types of financial instruments (rands)

Overall, the average household in the sample in a typical month will transact about R3000 through financial instruments. Not surprisingly, these transactions primarily happen in transactions-based formal financial instruments such as bank accounts. A household getting a payment through the bank – for instance, a regular wage, pension or grant – will receive this money into the instrument plus take the same amount out in a typical month.

Langa: giving credit is the most frequently used financial instrument (Figure 7)

In Langa, the most frequently used financial instrument is credit, obtained from a small business. This is not because there are a large number of businesses in Langa, but it may be because the businesses in Langa are more inclined to provide the customers with credit lines. A financial instrument is created every time a new customer takes credit and households can have up to 15 credit accounts running at the same time. Next on the list is one-on-one borrowing and lending. The average household in the Langa sample will have borrowed at least twice during the study year from neighbours, friends or family and lent an equal number of times.

Figure 7: Langa: average use of financial instruments (average = total number/52 households)

Lugangeni: burial societies are frequently used (Figure 8)

As in Langa, households in Lugangeni have frequent borrowing and lending patterns among themselves, but unlike Langa they are likely to have at least one informal burial plan, as well as one formal funeral policy. They are also more likely than households in Langa and Diepsloot to keep savings in their house. Although more informal financial instruments tend to be used most frequently, this does not mean that households do not have bank accounts. The average household in Lugangeni will have at least one bank account.

Figure 8: Lugangeni: average use of financial instruments (average = total number/60 households)

Diepsloot: bank accounts are more frequently used (Figure 9)

Although very few people have lived in Diepsloot for more than two years, financial ties in the form of lending and borrowing are as common as in the more established areas of Langa and Lugangeni. In Diepsloot, however, households use bank accounts more frequently than the other two areas. The average household will have at least one bank account. Despite this frequent use of banks, however, most households will also save in-house.

Figure 9: Diepsloot: average use of financial instruments (average = total number/54 households)

6. Urban versus rural case studies

The Financial Diaries data set provides a depth of information about households, although it cannot be used as a nationally representative survey. The following two case studies show that, contrary to expectations, some poor rural households can in fact use more financial instruments than a better-off urban household.

7. Conclusion and further research

The South African Financial Diaries project suggests several conclusions about the financial lives of the poor. Poor households have much more complex financial lives, using many more financial instruments, than may have been expected. Using a variety of sources of funds, households managed their money among a breadth of informal and formal instruments, and among savings, insurance and credit instruments. The depth of the Financial Diaries data provides a considerable resource for further research. Research areas to be probed include the extent of indebtedness within the sample, the link between employment and financial service access, methods of savings, the financial impact of shocks and the extent and financing of health expenditures.

Additional information

Notes on contributors

Daryl Collins

Research Associate, Southern Africa Labour and Development Research Unit, University of Cape Town. The author would like to acknowledge funding for the Financial Diaries project from the Ford Foundation, FinMark Trust and the Micro Finance Regulatory Council.