How Can a Company Assess Social Needs to Reduce Poverty among Its Workers? The Case of the Export Banana Industries

Abstract

(1) Background: In Southern countries, companies that employ a large number of workers are concerned with poverty reduction, which is the priority of their CSR policy. The banana sector has made great efforts to calculate a “living wage”, but this is not sufficient. How do we assess a decent living standard in Southern countries, where basic social needs are rarely covered by the public authorities? (2) Methods: the selection and analysis of the living standard assessment methods are based on a systematic review of the literature. (3) Results: We selected 427 references in total and classified them into 10 method families. Among them, we have not found a ready-to-use method. The paper therefore suggests an innovative methodology (CSDA) inspired by Townsend’s non-monetary deprivation score and by the importance of social norms and ties. Instead of using a list of goods to define a decent living standard, the new method relies on access to services proven to be important for public health by covering social needs. Moreover, this method is easy to implement for the private sector, highlights the priority social needs for workers on which the company can act with its CSR policy, and allows for cross-country comparison. (4) Conclusions: This last point is crucial for the producers’ commercial negotiations with large-scale distribution and also to prevent the flight of workers to other activities or competitors. With regard to poverty alleviation, it will depend on how the company uses the results of the CSDA assessment.

1. Introduction

The export agricultural industries represent a large proportion of world trade flows from developing countries (referred to as “South/Southern” in this paper) to Western countries. These industries are currently undergoing major transformations, such as unprecedented climate conditions, emerging consumer demands, or compliance with changing regulatory standards. These transformations relate to respect for the environment, but also—and increasingly—respect for the people working in these industries.

Consumers’ social concerns seem to be settling on the concept of “a decent living standard”. When it comes to the private agricultural industries, this is a major challenge when there is a value chain linking workers (smallholders, agricultural labourers, etc.) with Northern consumers who are the recipients of the products and increasingly demanding fair treatment of the former.

1.1. Addressing the Challenges of the Export Banana Industry

The case of the export dessert banana industry is representative of these current transformations [1]. Due to the new consumer concerns, retailers are requesting the producers (in this case, the plantations) to consider fair distribution of added value to the workers as a priority of their own Corporate Social Responsibility (CSR) policy. The idea is to take them out of poverty, since they are regarded as the losers in the battle for distribution of added value [1,2] and as the most vulnerable players in the industry [3]. To keep their “licence to operate”, producers need to show that the banana plantation workers or banana smallholders, and their families, have a “decent living standard” and that the company contributes to it. This first challenge is coupled with another concern for producers and importers. The latter also require expenditure on workers (hospitals, schools built, etc.), as part of their CSR policy, to be taken into account to justify a higher price during commercial negotiations. Indeed, the dessert banana trade takes place in a highly strategic arena. Bananas are exported to Europe or the USA from origins with widely differing socio-economic contexts (Latin America, Africa, French West Indies, etc.), and are in fierce competition on the same markets. During sale price negotiations between the producers and traders, or between exporters and the supermarket sector circuits, arguments for additional social costs are raised. Providing proof justifying their additional supports is an argument in the sale price negotiations. Finally, a third challenge is increasingly emerging: the necessity to increase the attractiveness of the agricultural sector, in order to keep workers in the company. Indeed, at least in places where rival plantations are in close proximity to each other, they increasingly compete for trained farm workers. The problem of excessive employee turnover in the South is becoming increasingly acute and is affecting the profitability of the plantations. Dissatisfied employees tend to stay in their jobs for a short time and, in general, to engage in independent farming as soon as possible.

To address these three challenges, evaluations have been attempted by the World Banana Forum (WBF), which is a formal inter-professional group bringing together dessert banana operators (producers, labourer unions, exporters, traders, ripeners, transporters, supermarket sector retailers, certifiers, NGOs, government representatives, development agencies) and agricultural research (Cirad). All studies on a “decent living standard” initiated or commissioned by the WBF are in fact studies on a living wage only, inspired by British precursor works. A living wage is defined by the GLWC (Global Living Wage Coalition) as follows: “Remuneration received for a standard work week by a worker in a particular [time and] place sufficient to afford a decent standard of living for the worker and her or his family. Elements of a decent standard of living include food, water, housing, education, healthcare, transport, clothing and other essential needs including provision for unexpected events.” [4].

The review conducted in [5] for the WBF recommended the CREA method. CREA categorises the remuneration received into four levels, the highest of which is regarded as decent. In this case, decent remuneration is able to cover the basic needs (food, clothing, housing, energy, travel, healthcare, and access to primary education), participation in social activities, and enables small savings to be made. The approach puts together a basket of goods and services, monetarily evaluates its contents, and then compares this level to the income actually received. Using this approach, the consultancy firm Ergon Associates (London) produced “wage scales”, which were applied by the CIMS (Centro de inteligencia sobre mercados sostenibles, Costa Rica) to create an initial group of living wage indicators (made comparable by applying purchasing power parity coefficients). These were completed between 2013 and 2015 for eight exporter countries (five in Latin America, the Dominican Republic, Cameroon, and Ghana). The data came from official statistics, without hearing from agricultural workers. A specific Ecuadorian study was conducted in early 2012 by INCAE (an Ecuadorian business school) among workers on six types of farms.

In 2016, the WBF teamed up with Fairtrade International (FI), a main member of the Global Living Wage Coalition (GLWC), to launch the Living Wage Advocacy Initiative (LIWIN) pilot project, co-financed by the IDH—the Sustainable Trade Initiative, and other public or private entities [6]. Its objective is to create a national consensus on markers for a subsistence wage, using the methodology developed by [7]. These markers can be used as a basis for future wage negotiations and minimum wage setting interventions in each country’s banana sector. This budget method is based on calculating costs for a reference-sized family in the context of the region, to cover food and housing needs, as well as other so-called “non-food and non-housing” needs (health coverage, education, domestic appliances, clothing, hygiene, mobile phone, leisure, and culture) [4]. The costs of food and housing needs are determined from mainly primary data, while the so-called “non-food and non-housing” needs are estimated using secondary data on regular expenditure habits [4]. This project was implemented in two banana-producing countries, Ecuador and Ghana. The living wage levels obtained provide a concrete basis for negotiations aimed at filling the gap between actual wages and the calculated living wage. The initiative sets out to be multipartite, involving both banana workers and producers but also distributors and governments of exporter and importer countries [6]. The report on the pilot project conducted in Ghana was published in February 2017 and suggests a wage scale including the World Bank extreme and absolute poverty thresholds, national poverty thresholds, and national wage thresholds in agriculture (and more particularly in the export banana sector), as well as the living wage calculated using the Anker and Anker methodology [7].

Outside of the WBF framework, the GLWI has also conducted reports on other export banana-producing countries, by determining a national living wage threshold for rural areas, using the same method, but without specifically including the export banana sector (e.g., Cameroon in 2020 and the Dominican Republic in 2013). Similarly, a living wage threshold was determined for Costa Rican banana-producing provinces [8] as part of a project known as “Next steps in sustainability-measuring impact and testing living wage”. This project, initiated in 2016, co-financed and supported by Fyffes, IDH, and other private players, was implemented by the Rainforest Alliance. A second study in Belize should also lead to a living wage threshold being set.

1.2. The Methodological Challenge

As explained in Section 1.1., the methods used are monetary and simulate budgets, based on the ones historically designed for Western countries (see Appendix A). This raises the following methodological difficulties highlighted by Brown et al. [9].

First, one needs to correctly determine nutritional standards, housing type, expense types, savings, and allowances in case of the unexpected. Moran explained that estimating family size and the number of workers it includes is an arbitrary exercise [10]. Sometimes, it is impossible to estimate a wage monetarily. Van Rijn et al. used five proxies (savings level, poverty level, security of access to food, property, satisfaction with lifestyle) as an approach to the living wage, as they were not able to measure it directly [11]. All these difficulties are present in the work undertaken by the WBF.

Then, the budget methods assume that the goods and services in the basket have a price (set by comparing supply against demand), i.e., that there is a market for these goods and services. Goods and services assumed to be freely supplied by the state (e.g., public transport) are removed from the basket. Yet, when public services are malfunctioning or incomplete, certain goods and services do not have a market or are even not accessible locally or inaccessible to certain parts of the population (e.g., school for girls). However, the shortage of public services varies greatly according to the areas explored by the WBF.

We can also lament that the works initiated by the WBF rarely hear from smallholders and agricultural labourers. This omission wrongly suggests that the basic services are the same everywhere, regardless of the context, but also that aspirations in terms of services are the same everywhere. Yet, it is not possible to guess arbitrarily—or according to international standards—what the local priorities are, since there is a wide variety of concepts as to what has value [12]. The diversity of these contexts leads to services to the population in varied forms. Family ties, the traditional mechanisms of social support, can effectively replace the more conventional forms of state support [13].

However, it is with regard to comparisons between countries that the criticisms have been the most serious. Yet, due to the competing multi-player composition of the WBF, any study on living wages is examined there in terms of comparison of competing origins. At best, the comparisons take into account differences in purchasing power parity, when they should also take into account the different macro-economic conditions [10]. The monetarisation of the living standard gives the illusion of a comparison but can only really compare living standards for two countries with similar socio-economic contexts. The monetary units erase differences of nature between the objects, hence the fact that a wage cannot allow comparisons, as it is not known whether it reflects otherwise equal situations in terms of access to services. We can note that even for comparing European countries, INSEE warns that harmonised price indices are not based on the same basket of goods [14]. This would be misleading, since two different European populations do not consume the same goods.

When it comes to developing countries, some authors reject any possibility of drawing up robust comparisons, since the cultural differences combining to establish variable local consumption levels are exacerbated [10]. Srinivasan (2001, cited in [9]) calls into question the pertinence of attempts to calculate a decent wage in a poor country. He writes that focusing attention on the relatively well-off employees of multinationals distracts us from the real causes of poverty, yet this dimension of comparability (and so price competition) of studies conducted in various countries is fundamental for WBF members, as we have seen. In order to induce a positive change in the workers’ living standard, and even simply to determine it, one has to know how to assess it.

1.3. The Research Questions

The limits that we have just mentioned illustrate the gaps to deal with in order to conduct a satisfactory living standard assessment in the South. As a member of the WBF, Cirad was asked to propose another method to assess the living standard of workers in the South. The first research question is therefore: “Are there any methods to assess a decent living standard for workers, which enable comparisons between the different contexts in the South?”

If the answer is no, a second question would then arise: “Can we construct an assessment method of a decent living standard for workers, enabling comparisons between the different contexts in the South?”

The objective of this article is to contribute to both questions. Section 2 develops the methodology for seeking and analysing methods able to assess a decent living standard for Southern workers. Section 3 sets out and discusses the results of this research. Section 4 proposes a new assessment method for a decent living standard in the South, known as “the consensual services deprivation approach” (CSDA). This has been built by combining proven methods identified in the literature review. Section 5 presents the discussion and Section 6 the conclusion.

2. Materials and Methods

2.1. Selection of Assessment Methods

The selection and analysis of the assessment methods of decent living standard are based on a systematic review of the literature. The objective is to answer the research question “Are there any methods to assess a decent living standard for workers, which enable comparisons between the different contexts in the South?”

The research question was converted into main concepts (in this case: “assessment” and “living standard”) and then into keywords, which were enriched by synonyms, analogues, and thematic exploration [15]. The main keywords used (including appropriate variations and permutations) here were combinations of the following terms:

• For the concept of assessment: assess/evaluate/measure/estimate/calculate/appraisal/determine/set/achieve and their derivatives.
• For the concept of living standard: living standard/living wage/decent income/decent salary/minimum wage/minimum income standard/poverty line/decent living standards/decent living conditions/standard of living/sustainable livelihood.

The queries were run identically for all the selected databases: Google Scholar (search engine specialised in academic research, which provides access to grey literature), EconLit (database specialised in human and social sciences), and Scopus (general scientific database). They combined all of the above keywords in English and were restricted from the 1990s to the present. In EconLit and Scopus, the keywords were searched for in the title, abstract, or keywords, while in Google Scholar only in the titles.

After a quick first read-through, the corpus of references found was labelled and sorted in order to restrict the number of documents [16]. The criterion governing inclusion was presentation or use of methods to assess the living standard of individuals or households, directly or indirectly. Consequently, we would not retain methods assessing this concept solely on the scale of a bigger entity (as in [17] for the Earth as a whole). Double counts, as well as references whose readable and available content was insufficient for understanding, were deleted. Some documents that were already known or had been identified by snowballing were added. A list of selected references was created [16].

2.2. Document Analysis Methodology

From this first selection, we drew up a list of living standard assessment methods, which were analysed in view of the objects assessed and the assessment process. This work brought together papers that were similar on these points, thereby forming groups of works which were close together in terms of object and process. In this way, step by step, we discerned method families, which are presented in Section 3.1.

In order to answer the first research question, we referred to the criteria derived from the methodological gaps revealed in Section 1.2 and reiterated below. For a method to be able to assess a decent living standard in the South, we believe it must comply with the following four conditions:

• Condition 1: The method approaches the living standard of an individual or household directly or indirectly. Through construction, the selection of documents stemming from the bibliographic search always complies with this criterion.
• Condition 2: It is suited to situations where there is no price defined for certain basic services for populations, and it can work without assuming the existence of a monetarised market for the service. Accordingly, all the method families employing monetarisation only (income, wages, etc.) assume the existence of markets for everything and do not comply with this criterion.
• Condition 3: it takes into account representations from the people concerned (without projecting the prejudices of other societies onto the field), by asking the opinion of the main stakeholders as to the main object, representing the living standard undergoing assessment.
• Condition 4: It enables meaningful and complete comparisons between countries and sectors, including when the socio-economic contexts are very different. Comparing wages, income, or baskets of goods between countries makes some sense but gives only a little information on the comparative living standards of the two populations.

All the methods included in the same family provide an identical response when we subject them to the 4 conditions above. That is why the methods will be analysed by family, rather than by reference selected.

3. Results

Appendix B presents the number of results obtained for the various queries, as well as the number of documents selected. Figure 1 presents the number of references selected by source type (databases and “snowballing”).

3.1. Result: Method Families

The following sections (Section 3.1.1 and Section 3.1.2) present the method families. Appendix C presents a summary table.

3.1.1. Absolute Methods

The feature shared by these methods is that the object assessed as an approach to living standard is understood independently from the living standard in the society or the division of this living standard in society.

• Monetary budget methods: These methods look at the income required to cover the expenditure relating to obtaining a predefined basket of goods and services. The precursor of these methods is Morris [18] in the United Kingdom. Estimated expenditure was still monetary, as in the Living Wage Calculator [19], applied in the USA, or by Anker [7] and Anker and Anker [20], who study the wages of agricultural workers. Haveman and Wolff add asset valuation to the monetary elements to calculate the poverty line [21]. In some methods, such as [22], monetary items are converted to time in order to calculate the Freely Disposable Time (FDT) of a household. The poverty line stands for FDT = 0. The representation of stakeholders is not directly surveyed and comparisons between countries are incomplete due to monetarisation.
• Budget methods based on food/energy: These approaches were completed by [18] and are reduced to estimating coverage of food or energy needs. The most well-known, the Food Energy Intake method (FEI), considers total expenditure (food and non-food) covering the recommended calorie intake. Here, the possible absence of monetarised trade for certain goods and services is no longer a limit. Hence, they have been applied mainly in Southern countries, e.g., by the Indonesian Central Bureau of Statistics (BPS) from 1987 to 2002 [23] but also in Kenya [24], in Mozambique [25], or in India [26]. Dietary habits are taken into account before calculating the cost of calories or energy required via national surveys or other databases, but the representation of stakeholders is not directly incorporated.
• Ratio methods: These approaches are guided by the comment by Engel [27], who showed that the proportion of food expenditure in income falls as households become richer. There are numerous applications of this, such as the food-share method [28], which calculates income from the proportion of food expenditure, or the calculation of income enabling a good food expenditure ratio (low income cut-offs [29]), as well as the application of the LES model (Linear Expenditure System) and enlarged LES model, which use sampling surveys for the same purpose, as in [30]. These methods deduce total income by evaluating income dedicated to food, thereby avoiding the need for the presence of monetarised trade. In this case, the representation of stakeholders is a secondary consideration. Comparison between countries is meaningful but only provides information on food.
• Physiological deprivation score methods: These approaches are used to establish a score that reflects the state of health. The Body Mass Index method [31] is based on universal biological indicators (height, weight, size, etc.) to establish whether individuals are in good health and by extension whether they have a good living standard. These methods can thus be applied even in the absence of monetarised trade. Biological indicators are objective and do not incorporate the representation of individuals. Since at no time either expenditure, costs, or income are used, comparisons between countries are meaningful in terms of physical health. However, living conditions which influence the state of health remain unknown.
• Absolute deprivation score methods: These are methods that assess multi-dimensional poverty using a deprivation score, calculated from a panel of non-monetary indicators. These indicators are the same whatever the context, as for the Multidimensional Poverty Index (MPI) [32] initiated by the work of Alkire and Foster [33]. The use of non-monetary indicators avoids the issue of the non-existence of monetarised trade. In general, secondary data are used, and no stakeholder surveys are required. These methods enable only incomplete comparisons between countries.

3.1.2. Relative Methods

These methods are based on the hypothesis that the object assessed as an approach to living standard relates to the “normal” living standard in the surrounding society.

6.

Relative budget methods: These methods compare the income level to the cost of a basket of goods and services. Yet, this basket is identified according to the one owned by a certain proportion of households in the society or of what is regarded as a necessity by a certain proportion of households (Rowntree approach cited in [34,35]). So, unlike monetary budget methods, they do incorporate representations from the households. However, these methods are dependent on the presence of monetarised trade and do not therefore allow meaningful comparisons between countries and regions.

7.

Relative budget methods based on food: These methods described in the works of Ravallion [36,37] and Ravallion and Bidani [38] are derived from relative budget methods and ratio methods. On the one hand, they estimate food expenditure according to local costs of a basket of food satisfying the minimum nutritive energy needs (and belonging to people whose consumption expenditure is below a predetermined amount), while adhering to local traditions. On the other hand, they estimate two non-food expenditure thresholds using an equation from Engel. These methods are applicable even in the absence of monetarised trade. They provide incomplete comparisons of living standards. Surveys to incorporate representation of households are not required.

8.

Income or consumption proportion methods: Poverty is seen as having an income below a certain proportion of the average or median income (or consumption expenditure) of the population in question. This is the case with the Fraction of Median Income Approach from an OECD study (1976) cited in [39] or in [40,41]. These methods do not work in the absence of monetarised trade and do not take into account representation of stakeholders.

9.

Deprivation score methods: Poverty is here understood as multi-dimensional, and living standard is assessed according to non-monetary indicators, which are counted to establish a score, as in the Townsend’s Deprivation score [42] or in the works of Alkire and Foster [33]. Moreover, there are the cases in livelihood approaches, such as applications of the Sustainable Livelihood Framework and the associated tools “Participatory Rural Appraisal” or “Participatory Poverty Assessment” [43,44,45] or “Sustainable Livelihood Security Index” [46]. Livelihood approaches are inspired by Sen [47,48] to suggest operationalisation of the capability approach. The use of non-monetary indicators enables application of these methods even in the absence of monetarised trade. The stakeholders are surveyed to determine the deprivation level. However, they are not systematically asked for prioritising and choosing the non-monetary indicators. Comparisons between countries are meaningful only between the same indicators, but they are promising.

10.

Consensual methods: The stakeholders themselves define the goods making up the sufficient basket [49,50] or directly the minimum income that they regard as sufficient [51] or rate their current income level ([52]; The Leyden Poverty Line method in [53]; The Subjective Poverty Line method by [54]; Center for Social Policy Poverty Line by [55]). The method of Dubnoff [56], which is about asking prospects to assess the income level of a hypothetical family, belongs to this group. The main characteristic of this family is that representation of the stakeholders is put at the centre of the assessment. These methods are not intended to enable comparisons between countries, but it seems possible to improve them in order to achieve this objective (see Section 4).

3.2. Conclusion about the Analysis of the Methods under the Four Criteria

Table 1. Features of the method families.

Method Family	Objects Assessed as an Approach to Living Standard (Condition 1)	Works in the Absence of Monetarised Trade? (Condition 2)	Representations of Stakeholders? (Condition 3)	Meaningful Comparison of Living Standard? (Condition 4)
Absolute Methods
Monetary budget methods	Basket of goods and wages/income	No	No	No
Budget methods based on food/energy	Cost of minimum calories/energy required	Yes	No	No
Ratio methods	Deduction of total revenue by evaluation of income dedicated to food	Yes	No	Yes, for food
Physiological deprivation score methods	Score reflecting state of health	Yes	No	Yes, but incomplete
Absolute deprivation score methods	Score reflecting living standard	Yes	No	Yes, but incomplete
Relative methods
Relative budget methods	Basket of goods and income	No	Yes	No
Relative budget methods based on food	Social standard in calories, and estimated non-food expenditure, as per Engel	No	No	Yes, for food
Income or consumption proportion methods	Fraction of income or expenditure of society	No	No	No
Deprivation score methods	Score reflecting living standard	Yes	Yes, but incomplete	Yes, but incomplete
Consensual methods	Goods or income indicated by the stakeholders	Yes/No	Yes	No

recaps how the various families meet the four conditions set in advance for selecting a method. None of the absolute methods integrate the representation of the stakeholders, and they are not appropriate for comparisons between different socio-economic contexts.

So, we have to look elsewhere. Are the relative methods suitable? Of the relative methods, some (relative budget methods, deprivation score methods, and consensual methods) are associated with representations from the stakeholders. Of these, relative budget methods are still dependent on the existence of a market, and consensual methods are not intended for the comparison of living standards between two sites.

For this reason, only the deprivation score methods family remains in contention. As a result, the deprivation score methods family seems most appropriate for our objectives. However, none of the methods included in this family are ready for the use we want to make of them. Indeed, in the method of Alkire and Foster [33] and for the Sustainable Livelihood Framework [44], indicators are relative to the standard of living of the society but are generally chosen by the investigator, making the representations of the respondents incomplete. Important elements of the standard of living could therefore be omitted. In addition, for both methods, comparisons are incomplete as they are meaningful only between the same non-monetary indicators and as methods are not designed for it. Townsend’s method is promising but is based on a closed questionnaire that always includes the same non-monetary indicators, regardless of location [42]. Hence, it enables complete comparisons between sites, but the representations of the respondents about their own standard of living are incomplete.

In conclusion, there are promising parts of methods, rooted in the relative methods, but we have not found a ready-to-use method suitable for assessing a decent living standard in the South. Thus, in response to the second research question, “can we construct an assessment method of a decent living standard suitable for the South?” we propose a new method (dubbed the consensual services deprivation approach or CSDA), a combination inspired by both the deprivation score methods and consensual methods, as explained below.

4. Proposing a New Method: The Consensual Services Deprivation Method (CSDA)

This Chapter presents the different concepts in use within the CSDA method, without taking into account the chronological order in which the ideas actually arose. It discusses the concepts of “social services” (Section 4.1), what are the “decent” and “normal living standards” (Section 4.2), and the “comparison of gaps” (Section 4.3). Section 4.4 provides a guide to the practical implementation of the method in the field, as actually tested in July 2021 and spring 2022 in Africa.

4.1. Focusing on “Social Services” Deprivation

The CSDA approach is a modest extension of the works of Sen [57], who observed that during a famine the poor did not have access to food, although it was abundant. Access to goods and services counts, not just the very existence of these goods and services.

As to the nature of what causes a shortfall, we opted to look at “social services”. These are services relating to harmonious development of the individual (health, education, recreation, and retirement). Rostila et al. demonstrated an astonishing point about these [58]. After observing the known correlation between “growing inequalities” and “self-declared health deterioration”, these researchers showed that this statistical correlation disappeared if the level of municipal expenditure on social services was integrated. Along the same lines, Beckfield and Bambra attribute the excessive mortality rate in the USA to the lack of a Welfare State and assert that Americans would live on average 3.77 years more if social expenditure in the USA was only equal to the average expenditure in OECD countries [59].

These works suggest that the adverse effects of inequalities on health can be offset by sufficient access to social services and that the availability of these services makes a strong contribution to people’s health [58]. Yet, health is a universal and indisputable component of wellbeing [60]. Therefore, the CSDA method will focus on access to social services.

4.2. Toward a New Definition of a Decent Living Standard

The new definition of a decent living standard seemingly picks up on the definition from Morris [18] of the “standard for healthy living”. However, the two approaches to health are different. While Morris inferred good health from individual consumption habits (healthy diet, sports activities, etc.) and according to scientific standards [18], we construct good health through sufficient access to the social services deemed normal in a given society. In this sense, our approach is similar to Townsend [42] and Platt [61], who used representative indicators of access to socially prescribed activities. Here, a decent living standard is one providing access to a decent state of health, itself achieved and maintained through sufficient and constant access to social services deemed normal in a given society at a certain time. This choice raises three remarks: (1) for many historical and other reasons, access that is considered normal (albeit very low) in society A might be considered highly inadequate in society B; (2) as people watch television and become more and more aware of how other people live, the level considered normal becomes more and more demanding over time; (3) it has been recognised that access to acquired benefits leads to a “Ratchet Effect” so that demands for access to services increase simply because of their improvement [62]. The “Ratchet Effect” is a mechanical analogy in economics that refers to the escalation of production, price, or wage that tends to self-perpetuate and resist falling back. For example, people are happy when their salary increases, but after a few weeks this increase seems normal; hence, they will tend to ask for more. They are influenced by the previous highest level, which makes it difficult to reverse the change.

Moreover, it is necessary to take into account that the priority social services from one social group may be different to those prioritised for another, which led Alkire and Foster [33] to consider different indicators for assessing living standards in different social groups. Hence, in the CSDA approach, only priority social services for workers in the local area are determined and studied. These priority social services are identified by surveys of stakeholders who are able to comment on the determinants of workers’ living standards. We will use proxies representing certain social services deemed important [11,63].

Some authors have tried to implement the relative deprivation approach, acknowledging that it is very difficult to quantify the extent of the relative deprivation [64]. Fortunately, implementing the CSDA method does not require a measurement of the extent of relative deprivation but an assessment of household access level to social services “consensually deemed to be normal” (see next section for the explanation of “normal”).

4.3. Assessment of the Access Level “Consensually Deemed to Be Normal”

The quality of access to a service (i) is judged by the concerned households surveyed [65], (ii) relates to a given group, and (iii) is evaluated by comparison between what is deemed normal by the group at the time of the interviews and the average access level achieved in the households surveyed. As a consequence, the households are not only asked to rate their own access to the service, as in [66], but to say what would be the normal service level in their society, i.e., for a hypothetical average family, as in [56]. When households were asked to rate the living standard of hypothetical families randomly assigned to them, Dubnoff observed that the judgements made were highly consensual, which seems to confirm that there is a shared social standard underlying these judgements [56].

To compare the living standard achieved in two different socio-economic contexts, at a certain time, we compare service by service, without seeking to aggregate the various services, as illustrated in Figure 2. As for comparison between countries or sites, it is based on a reference common to all sites, namely the “level consensually deemed normal”, whereas its content usually differs depending on the site (see Figure 2).

According to the sites, the service may take different forms and be linked to different situations (e.g., for the “health” service, the issue on site A could be combating malaria and on site B combating HIV). Nevertheless, the “level consensually deemed normal” can be the same on both sites A and B (e.g., “Having free access to primary care combating the principal disease”) or can be different (e.g., “Having free access to primary care combating the principal disease” for A and “Having free access to hospital care” for B). In both cases, the fact that a household does not have access to the consensually deemed normal service level defines a “gap”. Between sites, the gaps (between the service level actually achieved and the local “consensually deemed normal service level”) are compared for interpretation.

What matters here are the people’s representations rather than objective aspects (such as an indicator reporting the number of cases of malaria). It is true that a shortcut is made by equating a decent standard of living with access that is considered “normal” by those concerned. In fact, for CSDA, what is decent is what is deemed “normal” for a given society at a certain time. For another society, it might not be the same. This bias is deliberate, since we consider the notion of normal access to service to be completely subjective and socially constructed. In CSDA, we choose to focus on respecting peoples’ representations (what is important to them and what is satisfactory for them).

4.4. In Practice: Step-by-Step Implementation of CSDA

The CSDA method has been successfully tested on a first case study in a Southern country in Africa, where public services are faulty. The assessment was conducted within a collaboration between researchers and a company (which provided field access for the researchers), so called research action [67]. This is why we cannot reveal the detailed results which belong to the company.

• Selection of the important social services for workers

As explained in Section 4.2., the researchers first must select the social services deemed important by the workers. These are identified by surveys of stakeholders who are able to comment on the determinants of workers’ living standards. The researchers used the typology of stakeholders of Mitchell, Agle, and Wood to embrace the full spectrum of viewpoints [68]. Twenty-four stakeholders were interviewed, enabling us to identify eight social services (occurrences in the interviews). Finally, with the participation of the company, five important social services from the stakeholders’ answers were selected for the next step. The company chose the important social services that could be improved through CSR actions.

• Interviews with households

Then, open interviews with households enabled us to determine the level consensually deemed normal locally for each social service and the current level each household reaches. The households surveyed constituted selected samples [69]. It is not realistic to contemplate statistical validity, which would entail interviewing hundreds of households with identical questionnaires. The other way of obtaining robust results is to make a “selected sample” from as diverse a set of individuals as possible. If despite this heterogeneity some consistent responses are revealed, the results may then be generalised [69].

We chose from the same region: 12 households, some members of which are employees in or suppliers to this value chain; 12 households situated in the immediate surroundings, which do not have any members working directly in the value chain but which benefit from certain services provided by the company, due to their proximity; and finally, 12 control households situated in a remote different region. The number of households interviewed can be increased for the sake of completeness, if necessary. In the three groups, the interviews were conducted in the same period at home with one or more persons (including the head of the household), as per the semi-directive focused interview method [69]. We asked, with their consent, as per the European GDPR law:

• To assess the access level consensually deemed normal “what do local people think is the current correct level for such or such a service?”. An example of a question applied to schooling access would be: “For access to schooling, what do your neighbours think in general is the correct level that children in the village should be able to achieve?”. By definition, this assessment is consensual if we have chosen the members of the groups correctly. This was the case in the African case study.
• What they think of the access level of their own household (much less, less, better than, or equal to the consensus level) for each service and how to improve it to reach the level consensually deemed normal.

Thanks to close collaboration with the company, 59 households were surveyed at home, alone with the investigator. The households interviewed lived in two regions, where the predominant ethnic group is the same. They constituted three groups of selected samples [69], as described in the previous paragraph: 22 households in the first group and 19 in the second group, both groups in the same region, and 18 in the third group in another region. For each region, despite the heterogeneity of the households’ characteristics and situations, they agreed on the same level consensually deemed normal locally for each social service.

• Identification of gaps

For each service, the researchers created decision trees in order to create histograms. There are examples of a decision tree in Figure 3 and of a histogram in Figure 4.

To represent the gaps, the researchers broke down the level consensually deemed normal of each service into different levels, which are not hierarchical in relation to each other (Figure 3). For each level, there are at least two possible modalities to access the service: the most “satisfactory” modality, which is assigned a value of one point, and the “unsatisfactory” modality, which is assigned a value of zero. If there are more modalities within a given level, the one point is divided. For example, in Figure 3, there are three modalities for level 2. Hence, the value of “modality 1” is one; the value of “modality 2” is 0.5, and the value of “modality 3” is zero. For each household, the researchers calculated the total value by adding up the values obtained for each level. For example, a household has access to modality 1 for level 1 (value = 1), modality 2 for level 2 (value = 0.5), and modality 2 for level 3 (0 point); the total value for this household is 1.5 out of a maximum possible total value of three points (which is the value for the level consensually deemed normal).

In Figure 4, each bar in the histogram represents the total value obtained by each surveyed household; 12 households for each region are represented as an example. The y-axis of Figure 4 is the total value obtained calculated by adding up the values obtained for each level in the decision tree (Figure 3). For the previous example, the y-axis value of the household is 1.5 points.

Then, the researchers identified the gap between the level consensually deemed normal and the actual levels reached by the interviewed households (Figure 4). These gaps were then discussed with the company (i.e., which levels are unsatisfactory and why). Aspects of services that cannot be represented by a budget in monetary terms have been proven to be key for the service quality (overcrowded classrooms, lack of medicines in the dispensary, availability of transport schedule, etc.). These aspects are taken into account by CSDA and are managed by the company. The two regions showed a slightly different “level consensually deemed normal” service by service. The regions could be compared on the basis of the model shown in Figure 4.

5. Discussion

Traditional methods for assessing living standards have been criticised for failing to meet one or more of the three conditions outlined above. What about the CSDA method? The first case study highlighted that it could work even if the services are not available on a monetarised market (Condition 2). Participation of the workers concerned is requested via the household open interviews at home (Condition 3). As for comparison between countries or sites, it is based on the reference common to all sites, that is, the “level consensually deemed normal” (Condition 4), while its nature was slightly different depending on the site in the African case study. The originality of CSDA lies in combining the multidimensional and non-monetary poverty approach of Townsend [42], the inclusion of social norms of Dubnoff through the representation of people [56], and the importance of social services for health [58], while allowing comparisons between countries or sites.

As social requirements are increasing, there is a growing awareness of the idea that organisations making profit from resources of the country must redistribute it locally and are responsible for the living standard of the workers and their families [70]. This phenomenon is even more accentuated for companies operating in the Southern countries where state institutions are often malfunctioning [71]. Hence, the CSDA method is implemented with and for the value chain actors, giving the opportunity to the company to address their responsibility by limiting their negative social impacts and focusing their CSR actions on the real social service needs of households.

• Limits during the implementation

However, some companies are already performing concrete actions (e.g., housing labourers’ families), and could be tempted to exaggerate them through investigating exclusively the services on which they are already acting. We must be sure to select services representing a fairly varied panel to cover a large part of a family’s living needs. It is also possible that some serious issues might be covered up, which we will discover when conducting the surveys.

The interviewers will at the outset be seen as envoys of company Y, since it is Y that will introduce them. So, they can receive complaints or even grievances. It is up to the interviewers to have the skill to pull the conversation out of these ruts.

As not all the interviews are conducted on the same day, it is possible that the first households interviewed will interact with households interviewed later. We therefore cannot exclude a homogenisation effect on opinions. However, the sample selected contains households as different as possible. The research evolves with each new case [72]. The representations of the households may also evolve through local discussions, outside of any interview.

• Limits in the outputs of the methodology

As we do not calculate a monetarised basket of goods, there is no possible offsetting between the level of two items. Although, to compare different sites, ideally, we would have to be able to investigate the same services. Yet, since people’s concerns differ according to the context, at least some of the services investigated can differ from one site to another. Thus, unlike Townsend’s method, which always assesses the same non-monetary indicators, the CSDA method does not enable a complete comparison because it puts the representation of the people concerned at the heart of the method, in a way that no deprivation score method carries out.

As we compare the gaps, company A in the region with the lower access level consensually deemed normal, in our performative point of view (for example, free access to primary school), might appear to be in a “better” situation than company B in another region (for example, free access to secondary school) if the compared gap is lower. Company A might hide behind such results to avoid increasing salaries or only for favourably comparing themselves with company B in a competitive arena. Moreover, in the short term, company A might be tempted to slow down its efforts in alleviating poverty because the gap for households is smaller than in another region. However, in the longer term, the company will be forced to continue to improve social services access. Indeed, the access level consensually deemed normal will naturally evolve as access to social services increases, as described by the “Ratchet effect” [62].

In summary, we tried to anticipate and address the main practical drawbacks of CSDA before its implementation. The first case study lived up to its promise. However, it is likely that new field tests could reveal further necessary adjustments.

6. Conclusions

The approaches to a decent living standard were initially derived from those for poverty, since enjoying a decent living standard can be redefined as not being “poor”. There are two main families of scientific approaches to poverty, based on opposing theoretical premises.

The first approach assumes that, in general, individuals seek to maximise their function of utility. Therein, poverty is defined exclusively according to an income level. Regardless of whether their neighbour is rich, or they live in a failed state, individual poverty is seen as having insufficient monetary resources [73]. So, individual poverty can be defined as “an income level” below a set threshold in a given country, a level that can change over time. Wellbeing is therefore achieved through rising above the poverty threshold. The second approach emphasises the importance of comparison with the corresponding social group, as soon as the living standard exceeds the threshold of biological survival. It asserts that being poor is falling short of the decent living standard, defined by the corresponding social group [42]. Social wellbeing depends on maintaining satisfactory social relationships with their social group. This is effective when the individuals and the household “maintain their rank”, according to this locally defined social consensus, in terms of ownership of socially significant goods and access to social services.

CSDA postulates the existence and the importance of such local comparisons of the living standard in a given society. It is also based on the idea that access to social services is important for wellbeing or that reducing inequality of access to services is favourable for health. It can be applied to a wide range of research projects and can address a wide variety of objects: relative poverty, inequality of all types, access to services, role of the Welfare State in health, etc. The CSDA method is a modest illustration of the possibility of an alternative approach to following the theory of utility only.

CSDA can be an asset in alleviating poverty by focusing companies’ efforts on what really matters to families. Better meeting the real social service needs of families will also lead to minimising the costly turnover of workers. The results of the studies conducted by CSDA will also serve as arguments to discuss the additional social costs and ultimately to justify higher selling prices for bananas produced under these better social conditions, in a highly competitive arena. We hope it is not the sole motivation for the company to move to improve the social services it delivers.

Today, the company supporting the assessments made on the first case study is currently considering which concrete CSR actions it can put in place to fill the identified gaps. Indeed, the company is aware of how improving social services delivered to the population around the plantation improves its own competitiveness. Nevertheless, the final consequences of using CSDA depend on how the company will harness the results of the assessment.

Through construction, it is probable that CSDA would be suitable for assessing the living standard of workers’ households in other export industries and in other socio-economic contexts in the South. CSDA might also be a precious tool for other actors (such as certifiers or governmental agencies) to monitor the appropriateness of development actions implemented by the companies.