Social protection and food insecurity

The USA has a long history of running food-specific social security programmes. Originally tied to agriculture surpluses and then enlarged and reformed in repose to a strong anti-hunger lobby⁽Reference Power, Little and Collins⁸⁸⁾, the SNAP is a programme targeted towards ensuring that households do not go without food. It is a means-tested entitlement programme and eligibility criteria require households have no more than US$2250 in saving and assets, a gross monthly income at or below 130 % of the poverty line (about $26 000 per year for a three-person family), and a net income equal or less than the poverty line (about $20 400 annually)⁽⁸²⁾. Given the highly selective criteria to qualify for the programme, many studies have observed high rates of food insecurity among recipients⁽Reference Wilde⁸⁹⁾. This creates a methodological challenge when evaluating programme effectiveness. To address it, selection bias must be accounted for in analyses aiming to identify the effect of programme participation on food insecurity. In general, studies that have accounted for these methodological issues have found that participation in SNAP is associated with a significant decline in food insecurity⁽Reference Bartfeld and Dunifon^45, Reference Ratcliffe, McKernan and Zhang^90–Reference Nord⁹³⁾. One recent example is an analysis of data from a novel survey of SNAP participants by Mabli and Ohls⁽Reference Mabli and Ohls⁹⁴⁾. Here, the authors compared the rates of food insecurity among new entrants into the programme with those who have been receiving SNAP for 6 months or more, as well as examining food insecurity outcomes in a longitudinal sample. They found that food insecurity was significantly reduced by about four percentage points in the cross-sectional sample and 11·1 percentage points in the longitudinal sample. Severity of food insecurity was also considerably reduced⁽Reference Mabli and Ohls⁹⁴⁾. However, although studies have shown a significant decline in food insecurity prevalence, even after intervention, rates of food insecurity among SNAP recipients remain high. This suggests the level of SNAP support and potentially the programme's administration⁽Reference Edwards, Heflin and Mueser⁹⁵⁾ are not sufficient to prevent food insecurity among recipients. Others have argued that the paternalistic nature of SNAP reduces its effectiveness and do not allow households to maximise their own utility⁽Reference Power, Little and Collins⁸⁸⁾.

Given that income and other measures of financial resources are most consistently associated with food insecurity, a key question is whether income transfers reduce food insecurity. Most high-income countries provide social security for households and children with low incomes in the form of cash transfers, including means-tested income support, tax credits and universal benefits, such as child benefits. A handful of studies have evaluated how these programmes impact food insecurity, also employing econometric techniques.

In Canada, a difference-in-difference approach was used to examine the effect of an introduction of the 2006 Universal Child Care Benefit, a monthly income supplement of $100 for each child under age 6 years across Canada⁽Reference Ionescu-Ittu, Glymour and Kaufman⁹⁶⁾. Comparing eligible families to those without eligible children, the authors estimated that introduction of the child benefit reduced food insecurity by 2·4 percentage points, or about 25 %. The effect was stronger when the analytic sample was restricted to households with incomes below median income. Also from Canada, Milligan and Stabile examined how generosity of child benefits related to a number of child outcomes, including parents’ reports of their child ever experiencing hunger⁽Reference Milligan and Stabile⁹⁷⁾. Using a stimulated benefits approach, whereby variation in benefit eligibility across provinces and over time was used as an instrument for actual benefits a family was eligible for over 1994–2005, they found that there was a significant decrease in the probability of children experiencing hunger as the value of benefits increased. Other studies have evaluated food insecurity rates among households receiving social assistance over periods of investment in cash transfer programmes and other poverty reduction interventions, finding significant reductions in food insecurity for these households during these periods, also suggesting food insecurity is responsive to expanded social security programmes in Canada⁽Reference Li, Dachner and Tarasuk^98, Reference Loopstra, Dachner and Tarasuk⁹⁹⁾.

Natural variation in welfare policies has also been used to evaluate the relationship between non-food safety net programmes and food insecurity across states in the USA. For example, Schmidt et al. examined the individual and combined effects of food-related (e.g. SNAP and National School Lunch Programme), non-food social security programmes (e.g. cash welfare programmes that support low-income families, disabled people and low-income working families) and public health insurance, on food insecurity among lone-parent families⁽Reference Schmidt, Shore-Sheppard and Watson¹⁰⁰⁾. They also used simulation eligibility instrumental variable approach, examining how the average programme generosity in each state-year for each demographic group related to household food insecurity. This overcomes the selection effects of there being more families eligible for benefits and more food insecurity during difficult economic times. They found that increasing a family's combined income and food entitlements by US$1000 reduced food insecurity by 1·1 percentage points. They also examined the independent effects of food-based, cash-based and public health insurance programmes, and found that there was no evidence that food-based programmes were more effective than cash-based programmes; both were associated with a decline in food insecurity. The results were inconclusive for public health insurance.

Another element of many countries’ social safety nets are state pensions. In Canada, the old age security pension is a universal provided to all senior citizens aged 65 years and older, and guaranteed income supplement is an additional cash transfer for low-income seniors. Combined, these cash transfers are significantly more generous than support for working-age low-income people who are out of work. By exploiting differences in eligibility between cash transfers before and after pension age, McIntyre et al. showed how comparable individuals have significantly lower food insecurity once they are of pension age and eligible for state pensions⁽Reference McIntyre, Dutton and Kwok¹⁰¹⁾.

In light of evidence that unaffordable housing and health conditions are associated with food insecurity, the provision of universal health care and subsidised housing or housing benefits may also impact food insecurity. To the author's knowledge, there are few studies that have focused on these, likely due to the lack of data and difficulty evaluating policies that vary cross-nationally rather than within countries. One example is a recent study of social protection spending across Europe, which broadly captures the cumulative generosity of interventions that provide cash benefits and in-kind transfers, including housing subsidies. Using EU Statistics on Income and Living Conditions data, which include a measure of self-reported inability to purchase meat, chicken, or fish (or vegetarian equivalent) every other day as a proxy for food insecurity, Loopstra et al. found that during the Great Recession in Europe, in countries where there were low levels of per capita investment in social protection spending, increasing unemployment and declining wages were associated with rising food insecurity, whereas this was not the case where social protection spending was high⁽Reference Loopstra, Reeves and McKee¹⁰²⁾. Spending on housing, in particular, had a strong buffering effect on the relationship between rising unemployment and food insecurity. In contrast, many studies exploring food insecurity and social housing have found high rates of food insecurity among household living in social housing, which likely reflects the selection of vulnerable households into this type of housing⁽Reference Kirkpatrick and Tarasuk^68, Reference Fafard St-Germain and Tarasuk¹⁰³⁾. Studies exploring the expansion or reduction of the provision of affordable housing on food insecurity would enable examination of the effects of this type of social intervention on food insecurity.

In summary, investment in a variety of social protection interventions has been shown to reduce food insecurity. Given that these policies are most often implemented at the population level, the effects of such policies translate into large reductions in the number of households experiencing food insecurity and potentially reduce spill-over costs of food insecurity to other areas of state spending (i.e. healthcare)⁽Reference Tarasuk, Cheng and de Oliveira¹⁰⁴⁾.

Food banks (or food pantries) and food insecurity

Given the evidence that investment in social security and expanded welfare entitlements can protect households from food insecurity, periods of welfare retrenchment have often been characterised by the rise of charitable food provisioning in the form of food banks as communities have sought ways to respond to growing hunger in their midst⁽Reference Poppendieck^105–Reference Lambie-Mumford¹⁰⁸⁾.

A food bank (called a food pantry in the USA) is a place where individuals can go to receive groceries free of charge. There is an important variation in food bank operations across and within countries, but common features of food banks in Canada, the USA and Australia have been reviewed in order to evaluate their effectiveness in addressing food insecurity⁽Reference Simmet, Depa and Tinnemann¹⁰⁹⁾. Common features include a reliance on donated food, either from the community or from surpluses donated by the food industry; a reliance on volunteers, although many food banks employ staff as well; and their extra-governmental nature, meaning they are not directly run by government, although they may receive governmental in-kind support or grant funding.

Food banks aim to prevent people from going hungry; however, studies evaluating food banks across different dimensions suggest they are largely unable to meet this aim. Firstly, the levels of need observed among food bank users suggest that they experience high levels of severe food insecurity, which is often recurrent in the context of food bank use⁽Reference Loopstra and Lalor^62, Reference Loopstra and Tarasuk⁶³⁾. An early study of women using food banks in Toronto by Tarasuk and Beaton⁽Reference Tarasuk and Beaton⁶⁵⁾ observed that despite regular food bank use, there was a high prevalence of severe food insecurity among study participants (70 %). Over a 30  d period, experiences of hunger were reported by 57 % of participants, even in the context of food bank use. So while in some studies respondents have emphasised that the food provided to them from food banks is an important supplement to otherwise depleted or non-existent food supplies⁽Reference Hamelin, Beaudry and Habicht^5, Reference Baskin, Guarisco and Koleszar-Gree^110–Reference Williams, Watt and Amero¹¹²⁾, the amount and regularity of help provided from food banks may be insufficient to ensure that households do not go without food.

To the author's knowledge, intervention trials evaluating before and after effects of receiving help from food banks have not been conducted, and few studies have followed people using food banks forward from the point of food bank use⁽Reference Loopstra and Tarasuk⁶³⁾. One recent example by Roncarolo et al. ⁽Reference Roncarolo, Bisset and Potvin¹¹³⁾ examined change in food insecurity over 1 year in a cohort of people who started using food banks within 6 months of recruitment. They reported that after 1 year, food insecurity was significantly reduced and attributed this effect due to the food bank intervention. However, attributing the observed decline in food insecurity to food bank use could have been erroneous for several reasons. The study lacked a true baseline measure (i.e. before receiving help from food banks) and how often households were receiving this kind of assistance over the year of follow-up was not taken into account. There was a high attrition rate (more than 50 %), and omitted variable bias is likely, as employment increased in the sample but was not accounted for in statistical models. Income was only adjusted for using a categorical variable, which was unable to capture changes in income in the magnitude of $5000 or less. There was also no control group, so it is impossible to know if the change observed was in accordance with a time trend in food insecurity in this community over this period. Lastly, there are selection effects that must be accounted for. Households that experienced serious declines in their circumstances may be harder to re-contact, thus biasing the follow-up sample towards those that did better over the year. This work highlights the need for well-designed studies to examine the impact of food bank use on food insecurity. While food banks have proliferated in many high-income countries⁽Reference Riches and Silvasti¹¹⁴⁾, most recently in the UK⁽Reference Lambie-Mumford¹⁰⁸⁾, they have done so in the absence of well-designed evaluations of their effectiveness.

Studies exploring the nature of food bank operations and quality of food distributed highlight why they may have limited effectiveness. For example, an in-depth study of food banks operating in Toronto showed that the functioning of food banks was highly contingent upon donations, resulting in assistance that was highly variable and unresponsive to users’ needs⁽Reference Tarasuk and Eakin¹¹⁵⁾. Food banks were open for only a limited number of days and times and had to ration the amount they gave away to make food supplies stretch farther, thereby further disassociating client need from what was received⁽Reference Tarasuk and Eakin¹¹⁵⁾. Studies across Canadian cities have also documented serious limitations in the amount of food and quality of food distributed, including foods that were past best before dates, highly processed and superfluous, and quantities that would only provide for 1–3 d worth at most⁽Reference Jacobs Starkey^116–Reference Teron and Tarasuk¹²⁰⁾. Other concerns about the limited nature of food banks for vulnerable populations have come from the medical community. Based on a survey of food pantry programmes operating in the New York City⁽Reference Gany, Bari and Crist¹²¹⁾, it was found that programmes were inaccessible due to few operating hours and entry requirements, and that the lack of variety and quantity of foods provided likely meant the programmes would not be able to meet the needs of medically underserved patients.

Reports on the nutritional inadequacy of foods provided from food banks are common⁽Reference Simmet, Depa and Tinnemann¹⁰⁹⁾. A recent review of studies that have examined the nutritional quality of foods provided through food banks. It found that while there was wide variability in the nature and amount of food provided, in general, studies reported insufficient amounts of fruit and vegetables and milk products, and insufficient amounts of calcium, vitamin A and vitamin C⁽Reference Simmet, Depa and Tinnemann¹⁰⁹⁾.

Importantly, studies among food bank users have documented feelings of shame about having to use food banks⁽Reference Tarasuk and Beaton^65, Reference Hobbs, MacEachern and McIvor^122, Reference Garthwaite¹²³⁾. These findings raise questions about whether food banks can address food insecurity, as receiving help from a food bank means acquiring food in a socially unacceptable way⁽Reference Lambie-Mumford¹⁰⁸⁾. Stigma and shame may explain why there is evidence that food banks are only used by less than one-quarter of the food-insecure population in Canada⁽Reference Loopstra and Tarasuk¹²⁴⁾. Examination of reasons why households do not use charitable food assistance, even in the context of severe food insecurity, suggest that food banks are considered an intervention of last resort and that households endure going without food rather than turning to charity⁽Reference Loopstra and Tarasuk⁶³⁾.

Cumulatively, research on food banks from high-income countries suggests that charitable food provisioning through food banks is inherently limited in its ability to meet the needs of individuals and households experiencing food insecurity, and at a population level, has done little to impact the widespread problem.

Community food programmes

In light of the limitations of charitable food provisioning through food banks, organisations have often sought alternative ways to provide no-cost or low-cost food to people and/or enhance the types of services offered at food banks. Sometimes referred to as community food security activities, these have been positioned as sustainable alternatives to charitable hunger programmes such as food banks, which have been viewed as ‘treatment’ and ‘social welfare’⁽Reference Hamm and Bellows¹²⁵⁾. These initiatives can take different forms. Some food banks in the UK and Canada offer additional programmes (e.g. www.trusselltrust.org/what-we-do/more-than-food/) or operate as a part of a community food centre (e.g. cfccanada.ca/what_is_a_community_food_centre), where people can receive additional services such as benefits counselling with a caseworker, debt counselling, cooking classes or fuel vouchers, or participate in community kitchens or community gardens. Another model is the community shop, which through a membership structure (possibly operated through volunteering time or a low membership fee), members make their own choices from the donated or surplus foodstuffs available.

Martin et al. ⁽Reference Martin, Wu and Wolff¹²⁶⁾ examined a combination of these services in a randomised intervention design, where clients using a regular food bank were compared with food bank clients who were offered a Freshplace membership. The Freshplace intervention allowed members to choose their own food, much of which was fresh and perishable, and included motivational interviews with a project manager, and additional services and referrals. The authors reported that, at some but not all points over the follow-up period, the odds of severe food insecurity among Freshplace compared with those receiving help from a traditional food bank were significantly reduced, while self-sufficiency scores and fruit and vegetable score were significantly increased. Given the combined nature of the intervention, it was not clear what aspect of the intervention was most effective, and drop-out rates were not reported. Participants were also allowed to switch their intervention group, which means randomisation was not perfectly achieved, potentially biasing results. However, the results suggest that one-on-one counselling, greater food choice and membership-based models may be more effective at meeting the needs of food-insecure people than traditional food banks.

Community kitchens are another form of community food programme. Programme participants prepare large amounts of food together and take home the meals prepared, and simultaneously often are taught budgeting and cooking skills⁽Reference Tarasuk¹²⁷⁾. In a review of research on community kitchens across high-income countries, Iacovou et al. ⁽Reference Iacovou, Pattieson and Truby¹²⁸⁾ found that a variety of methods have been employed to investigate their effectiveness. These included participant observation, questionnaires (cross-sectional and pre/post), interviews and focus groups. Most studies were small and descriptive. The main themes that emerged across the studies reviewed were an increase in reported intake of nutritious foods and increased healthy food access, increased self-reliance and engagement with social services, improved social skills and enhanced social support, and increased skills, confidence and enjoyment from cooking⁽Reference Iacovou, Pattieson and Truby¹²⁸⁾. However, in an evaluation of the experience of people using a community kitchen programme in Canada, Engler-Stringer and Berenbaum noted the tenuous nature of any benefits provided to programme users since programme availability was subject to holiday schedules, funding constraint and time allotment of staff⁽Reference Engler-Stringer and Berenbaum¹²⁹⁾.

Largely absent from the evaluations of these types of programmes are objective, pre- and post-test evaluations of how programmes relate to household food insecurity, although even this type of evaluation can be biased to retain participants who derive the most benefit if people who discontinue use are not followed. There are few studies that have used a randomised trial design and followed participants over a longer period of time. One exception is a recent evaluation of the SNAP-ED programme in the USA, which offers a nutrition, budgeting and lifestyle intervention education programme⁽Reference Rivera, Maulding and Abbott¹³⁰⁾. This study showed that in the 10 weeks after baseline, the reduction in food insecurity was the same for both the control and intervention groups, which may have been due to the financial compensation received by both groups for participating. After 1 year, food insecurity scores were significantly lower in the intervention group, though this was driven by a greater rate of rebound to baseline scores for the control group rather than a further reduction in food insecurity for the treatment group. This suggests that the programmes could buffer families from a decline in food insecurity in the face of economic shock or other drivers, rather than cause a decline in food insecurity. In the intervention group, only 41 % were fully food secure 1 year after the intervention, suggesting its effects were not powerful enough to move all households into food security. The dropout rate was also 43 %, which may mean those who were deriving the most benefit from the programme were more likely to stay enrolled in it.

Community food programmes also often focus on improving food availability and improving food skills. For example, reflecting concern about food deserts, some programmes focus on providing alternative places for people in low-income areas to buy or acquire food by establishing mobile, low-cost fruit and vegetable markets, fruit and vegetable box drop-off programmes, or community or allotment garden spaces for people to grow food⁽¹³¹⁾. Funding and provision of these programmes can be through third-sector organisations, government (particularly local governments if public health is devolved), or both.

These programmes have also largely not been evaluated in randomised intervention designs. An examination of the impact of a subsidised fruit and vegetable box programme (e.g. good food box programme) compared the food insecurity and dietary quality of individuals participating in the programme to those not participating and those who had dropped out of the programme⁽Reference Miewald, Holben and Hall¹³²⁾. It was found that those enrolled in the programme had a lower prevalence of food insecurity than those not participating at baseline, and there was a non-significant increase in food insecurity among those who discontinued the use of the programme over the follow-up period was reported⁽Reference Miewald, Holben and Hall¹³²⁾. Rather than highlight the effectiveness of programme use for reducing food insecurity, these findings may indicate that those with deteriorating circumstances were not able to remain in the programme. Further, over the 8-month follow-up period, the prevalence of food insecurity among the forty-six individuals that remained in the programme and completed both questionnaires was unchanged⁽Reference Miewald, Holben and Hall¹³²⁾.

This raises the problem of reach, relevant to many community food programmes as well as food banks. While programme benefits can be documented among participants, how programmes impact the prevalence and incidence of food insecurity in communities is often not discussed. Even if a large effect size is observed, if only a small subset of the food-insecure population is ever inclined or able to maintain sustained participation in these programmes, a wide-scale reduction in food insecurity at the population will not be achieved. When the use of community kitchens was explored in a Toronto-based sample of low-income families with children, only a very small proportion of families participated in these initiatives, and many reported not participating in them because they did not feel in need of such a programme⁽Reference Loopstra and Tarasuk¹³³⁾. In addition, children, employment, disability, social interaction disorders and lack of transportation may all reduce the ability of households vulnerable to food insecurity to participate in these programmes.

Researchers have also raised concerns about community food programmes being positioned as initiatives to address food insecurity because they focus on skills, behaviour and physical food access, which have not been empirically documented as drivers of household food insecurity. Focus in these areas may serve to draw attention away from systemic drivers of insufficient finances for food⁽Reference Tarasuk^127, Reference Allen^134–Reference Johnston and Baker¹³⁷⁾.