Introduction

Statistics show that about 14.5% of the population in 121 countries is multi-dimensionally poor, with South Asia and sub-Saharan Africa having the greatest proportions of 17.3% and 51.8%, respectively (World Bank, 2024). According to the United Nations Development Programme (2023), out of about 6.1 billion people in more than 100 countries, 1.1 billion (18%) live in acute multi-dimensional poverty, and sub-Saharan Africa is home to more than half of this population. The majority of the poor (84%) reside in rural areas, and half are children less than 18 years old (ibid.). Although income or consumption spending remains relevant in the definition and measurement of poverty, it is not the only manifestation of the phenomenon as implied by the welfarist approach (D’Attoma & Matteucci, 2023; Eshetu et al., 2022; Zulkifli & Abidin, 2023). The non-welfarist paradigm asserts that aspects such as poor health, lack of education, precarious housing, poor sanitation, unsafe drinking water, job insecurity, over-indebtedness, social exclusion, and lack of freedom or powerlessness (in various forms) are also crucial (Amornbunchornvej et al., 2023; Bourguignon & Chakravarty, 2003; Chamboko et al., 2017; Ntsalaze & Ikhide, 2018; Shah & Debnath, 2022; Zulkifli & Abidin, 2023). People’s capabilities to function in society are the basis for considering non-monetary variables in the definition and measurement of poverty (Sen, 1999).

Eradicating multi-dimensional poverty requires an understanding of its nature, measurement, effects, and causes (along with the strategies that have already been adopted to reduce it). Bibliometric analyses and systematic literature reviews are among the avenues through which this knowledge is gathered. Literature on multi-dimensional poverty is vast, but there is a paucity of systematic reviews of what has been investigated so far. A pioneering systematic literature review of this concept was done recently (D’Attoma & Matteucci, 2023), but it focused on the definition and measurement of multi-dimensional poverty and the dimensions (criteria) used in empirical studies. Moreover, their review only covered the years up to 2019 and excluded most citation-based analyses. In a related systematic literature review of poverty, Zulkifli and Abidin (2023) focused on the dimensions of poverty based on 88 publications. However, the search string applied excluded the key concept of “multi-dimensional poverty”, and the publication dates of the consulted research articles were not explicitly stated.

Therefore, this study explores the literature on multi-dimensional poverty published before March 2024. A bibliometric analysis is conducted to present the knowledge map of the concept, and this is the first study to undertake such an analysis of multi-dimensional poverty. Based on the results of the bibliometric analysis, a systematic literature review was performed. While such hybrid reviews have been implemented in other areas (Goyal & Kumar, 2021; Nisa & Chalid, 2022; Yao et al., 2023; Z. Zou et al., 2023), a combined analysis of multi-dimensional poverty does not exist. A hybrid review provides richer contributions and clarifications than a single approach (Klarin, 2024). Based on bibliometric analysis, the evolution of multi-dimensional poverty research and the themes investigated are identified. The content analysis conducted after the bibliometric analysis enables the researcher to identify underexplored themes and new methods, thus unearthing gaps for future research (Liao et al., 2023). Thus, a combined review facilitates quantitative and qualitative insights, thus enhancing methodological rigour and ultimately producing robust findings (Marzi et al., 2024).

Moreover, the focus of existing systematic literature reviews on multi-dimensional poverty (D’Attoma & Matteucci, 2023; Zulkifli & Abidin, 2023) has been narrowed to its measurement. These systematic reviews have been based on the traditional method, an approach prone to researcher bias and with limited capability to analyse a large volume of publications. The use of bibliometric software in this study eliminates researcher bias and enables the analysis of a large number of publications (Mhlanga & Dzingirai, 2024; Umeokafor et al., 2022). This provides a holistic overview of the intellectual structure of multi-dimensional poverty.

Thus, the contribution of this study was to offer a comprehensive understanding of multi-dimensional poverty research based on a hybrid review. This study is the first to conduct a bibliometric analysis and content co-occurrence literature review of multi-dimensional poverty using computer software (VOSviewer). A total of 1513 articles were analysed, and the results showed that multi-dimensional poverty research rapidly increased during the last decade; more than 90% of the journal articles were published between 2013 and 2024. The content co-occurrence literature review showed that previous research focused on the measurement, causes, and effects of multi-dimensional poverty and the strategies that have been adopted to alleviate the problem. Generally, longitudinal investigations of multi-dimensional poverty to better understand the problem over time are scant, hence the need for such research. Moreover, research can also focus on the effects of recent technological innovations (for example, the Internet of Things) on multi-dimensional poverty.

The study aims to highlight the state of research on multi-dimensional poverty based on the following research questions:

  1. 1.

    What are the trends of peer-reviewed journal articles on multi-dimensional poverty?

  2. 2.

    What is the distribution of publications on multi-dimensional poverty in terms of journals and countries?

  3. 3.

    Which keywords have been widely used in peer-reviewed journal articles on multi-dimensional poverty?

  4. 4.

    Which multi-dimensional poverty themes have been widely investigated?

  5. 5.

    What gaps exist in multi-dimensional poverty research?

To answer these research questions, the rest of the study is organised as follows: the next section provides a description of the methodology of the study, followed by a presentation of the results of the bibliometric analysis. Thereafter, a content co-occurrence analysis is undertaken to identify the specific issues investigated by different publications on multi-dimensional poverty. Based on the identified themes, the policy implications and directions for future research are proposed. Finally, the conclusions and limitations are provided.

Methodology

Data sources, search terms, and inclusion criteria

The Elsevier Scopus database was searched on 08 March 2024 using the following search string: (“multi-dimensional poverty” OR “multi-dimensional poverty” OR “multi-dimensional well-being” OR “multi-dimensional well-being” OR “multi-dimensional well-being” OR “multi-dimensional well-being” OR “multi-dimensional inequality” OR “multi-dimensional inequality” OR “multi-dimensional deprivation” OR “multi-dimensional deprivation”). Since multi-dimensional poverty is an issue largely investigated in the humanities and social sciences, the search was performed using the Scopus database. Generally, Scopus is among the databases with comprehensive coverage of social sciences and humanities publications from reputable publishers—it is multidisciplinary (Martins et al., 2022; Nisa & Chalid, 2022). Moreover, the Scopus database is rigorously designed to include publications from credible sources, and de-duplication of the downloaded list is not required (AlNemer, 2023; Klarin, 2024; Mushtaq et al., 2023). Using a single database, such as Scopus, also avoids the problems associated with merging multiple databases (AlNemer, 2023). The search was restricted to peer-reviewed journal articles published in English (both final and in press) until March 2024, and it was based on article title, abstract, and keywords.

Bibliometric analysis

Bibliometric analysis can be used to evaluate scientific literature of a certain field or concept. It is a quantitative method that involves the use of computer software such as VOSviewer to analyse and visualise data (Moral-Muñoz et al., 2020). Through bibliometric analysis, researchers can easily assess research impact in terms of citations and identify research trends and occurrences of keywords (Ciolomic et al., 2024; Mhlanga & Dzingirai, 2024). By visualising the quantitative indicators generated from bibliometric analysis, academics and researchers can easily understand the knowledge domain of a particular concept.

VOSviewer enables the “visualisation of similarities” of the research output of different authors on a map such that closely related items plot closer to each other (Goyal & Kumar, 2021; Van Eck & Waltman, 2010). Co-occurrence network analysis and co-citation analysis are important features of VOSviewer. These culminate in highly correlated publications falling into common research clusters, paving the way for a systematic literature review based on thematic content analysis.

Content co-occurrence literature review

Based on the downloaded Scopus database and the results of the bibliometric analysis, content analysis was conducted. Content analysis is one of the procedures for deriving meaning from relevant publications and has been widely applied in systematic literature reviews (Bhandari, 2023; Goel et al., 2021; Goyal & Kumar, 2021; Khirfan et al., 2020; Seuring & Gold, 2012; Z. Zou et al., 2023). This allows the researcher to scrutinise the subject matter of each publication, enabling a deep understanding of the specific themes of the concept under investigation (Khirfan et al., 2020). Thematic content analysis undertaken after a bibliometric analysis can be based on the results of a bibliometric coupling of documents (Bhandari, 2023; Chrysikopoulos et al., 2024; Das, 2024; Yao et al., 2023), citation analysis (Ahmad et al., 2020), co-citation analysis (Ciolomic et al., 2024; Goyal & Kumar, 2021), or keywords co-occurrence analysis (Goel et al., 2021; Samanta & Aithal, 2024; Z. Zou et al., 2023). Undertaking content analysis based on the output of a bibliometric coupling of documents, citation analysis, or co-citation analysis might not be desirable because priority is given to the most impactful publications. Consequently, fairly new and relatively old publications with few citations are likely to be excluded, resulting in a biased review.

Klarin (2024) emphasises the use of an approach that addresses the limitations of the aforementioned techniques, the content co-occurrence or co-word analysis. With the content co-occurrence approach, the actual content of publications rather than citations is prioritised. The terms (words and phrases) that frequently occur together in publications are extracted and mapped so that closely related terms are categorised in a common cluster. Moreover, with the co-occurrence of terms, whether an article has been cited or not is immaterial; hence, the content analysis yields robust and reliable results (ibid.). In this study, content analysis was based on the clusters generated from the co-occurrence of terms in VOSviewer.

Results

To view the distribution of publications by country and journals, the search results from the Scopus database were exported to VOSviewer. Using VOSviewer, co-occurrence of keywords and content co-occurrence analysis were also undertaken. The results of the bibliometric and content analysis are presented in this section.

Trends in journal publications

From the 1527 journal articles downloaded from the Scopus database, 14 irrelevant publications were removed after scrutinising the abstracts, leaving 1513 articles for the analysis; the articles eliminated were largely published in mathematical journals and focused on the mathematical concept of multi-dimensional inequalities. Figure 1 illustrates the annual research output trend. The number of journal publications on multi-dimensional poverty has grown rapidly since 2013. Before 2013, fewer than 50 articles were published each year worldwide. On average, there were 64 publications per annum during the period 2013–2017. From 2018 onwards, at least 100 research articles were published each year. In fact, 69% of the journal articles considered in the bibliometric analysis were published between January 2018 and March 2024, indicating the growing interest in multi-dimensional poverty research.

Fig. 1
figure 1

Annual scientific production.

Full size image

Bibliographic coupling of journals and countries

With a bibliographic coupling of sources (journals), the size of the circles depicts the intensity of relevant publications for a particular journal; generally, the larger the circle, the higher the number of relevant publications. The colours of the circles indicate journal clusters, implying that sources with the same colour belong to a common cluster. The thickness of the lines connecting the journals is also important because they reveal the degree of correlation among the sources. Journals that are highly correlated in terms of common citations are linked with thicker lines, whereas the reverse is true for thin lines.

The bibliographic coupling of journals that published research on multi-dimensional poverty is illustrated in Fig. 2. A minimum of 10 publications from a source (regardless of the number of citations) was set, and 20 journals met the criteria. As shown in Fig. 2, the top 5 publishers on multi-dimensional poverty are Social Indicators Research (n = 173), World Development (n = 42), Child Indicators (n = 37), PLoS One (n = 29), and Journal of Economic Inequality (n = 25).

Fig. 2
figure 2

Bibliographic coupling by source.

Full size image

Concerning the publications and citations by country, the size of the circles and the thickness of the lines represent the countries with the highest number of publications and co-citations, respectively. The bibliographic coupling of countries in this study was determined by setting a minimum of 25 publications per country. Of the 137 countries, 22 met the criteria (see Fig. 3). It can be observed that the United States, the United Kingdom, China, India, Italy, Australia, Germany, Spain, South Africa, and Pakistan are the top ten countries with publications on multi-dimensional poverty. Generally, high-income countries dominate in terms of research in this area. The distribution of the number of publications by country is summarised in Table 1.

Fig. 3
figure 3

Bibliographic coupling by country.

Full size image
Table 1 Top ten countries by publications.
Full size table

Keywords co-occurrence network analysis

Lastly, the co-occurrence of keywords (both author and index keywords) was analysed, and the lowest number of appearances per keyword was set at 70. Of the 5604 available keywords, 20 met the criteria, and the results are shown in Fig. 4. As Martins et al. (2022) highlight, the thicker the line linking two words, the stronger the degree of co-occurrence between them. Also, the closer the nodes are to each other, the greater is the strength of the two-node association (ibid.).

Fig. 4
figure 4

Co-occurrence of keywords.

Full size image

As depicted in Fig. 4, the keywords with the highest frequencies include “poverty” (525 occurrences), “multi-dimensional poverty” (511 occurrences), “human” (225 occurrences), “humans” (165 occurrences) “poverty alleviation” (154 occurrences), “female” (144 occurrences), “male” (126 occurrences), “adult” (105 occurrences), and “poverty determinant” (103 occurrences). Generally, keywords falling in the same cluster correspond to a common research theme, and the density of the keywords enables the investigator to easily identify research hotspots. From Fig. 4, three clusters (red, green, and blue) can be observed. 13 words appear in the red cluster, and the prominent ones are multi-dimensional poverty, poverty, multi-dimensional poverty index, poverty alleviation, and poverty determinant. The green cluster is composed of 6 words (human, humans, male, female, adult, child), while the blue cluster has one word (income). Although content analysis can be based on keyword co-occurrence analysis results (Samanta & Aithal, 2024; Z. Zou et al., 2023), this study relies on co-word co-occurrence analysis. The relative merits of the approach are highlighted in the section “Content co-occurrence literature review”.

Content analysis

Following bibliometric analysis, content analysis is usually undertaken to identify knowledge gaps and guide future studies. The content analysis in this study was based on the results of a content co-occurrence analysis of terms in the titles and abstracts of research articles. For citation normalisation, VOSviewer’s default option (association strength) was maintained, and a minimum of 40 occurrences for each term was specified (implying that the term should appear in at least 40 publications from the downloaded Scopus database). Four research clusters emerged from the analysis (see Fig. 5). The sizes of the circles reflect the frequencies of the terms in the topic areas of the research articles; the larger the circle, the higher the occurrence. For example, the terms “impact”, “effect”, and “model” are more frequent in the red cluster, while the green cluster is dominated by words such as “measure”, “measurement”, and “framework”. The overlay visualisation (Fig. 6) shows that recent multi-dimensional poverty research has focused on issues such as “strategies”, “impact”, “effect”, “covid”, “poverty alleviation”, and “poverty reduction”. The most frequent and relevant terms from the co-word analysis are also summarised in Table 2.

Fig. 5
figure 5

Content co-occurrence.

Full size image
Fig. 6
figure 6

Content co-occurrence (overlay visualisation).

Full size image
Table 2 Key themes on multi-dimensional poverty research.
Full size table

Relevant publications for content analysis were subsequently identified and classified among the four clusters. A document was considered for classification if at least three relevant cluster terms were contained in its topic areas (title, abstract, and author keywords). Based on the frequency of relevant terms in the topic areas, a document was considered to belong to a specific cluster. The publications that met the stipulated criteria were distributed as follows: red (221 publications), green (267 publications), blue (218 publications), and yellow (133 publications). Thereafter, using the Microsoft Excel RAND function, 10-15 research articles were randomly selected from each cluster for content analysis, and the findings are summarised in the following sub-sections.

Green cluster: measuring multi-dimensional poverty

As highlighted in the literature review section of this paper, the debate on the best measure of multi-dimensional poverty remains unresolved. A measure that has gained wide international recognition is the Multi-dimensional Poverty Index (MPI); an approach that emphasises three dimensions (education, health, and standards of living) and was propounded by Alkire and Foster (2011). The approach resembles the Human Development Index (HDI) and the UNDP has since adopted it in several countries. To facilitate the speedy and comprehensive analysis of MPI data, a specialised package (mpitb) was recently incorporated into the Stata econometrics software (Suppa, 2023).

“The package mpitb comprises several subcommands to facilitate specification, estimation, and analyses of MPIs and supports the popular Alkire-Foster framework to multi-dimensional poverty measurement.” (Suppa, 2023, p. 1).

Concerning the dimensions for measuring poverty, advocates of the MPI recognise that variations in the prevailing conditions across countries, provinces, or districts may warrant the development of suitable indices, but such endeavours should be anchored on expert opinion and policy relevance, among other considerations (Amornbunchornvej et al., 2023; Ntsalaze & Ikhide, 2018).

Since the calculation of the MPI partly involves the aggregation of binary indicators assigned 0 and 1 values depending on the severity of an issue, an improvement or deterioration of a particular poverty issue undoubtedly changes the overall index and may also influence other dimensions that make up the MPI. It follows that instituting measures to alleviate a certain dimension may enhance or worsen another aspect of poverty. For the development of more effective strategies to reduce poverty, Amornbunchornvej et al. (2023) highlight the need to know the causal connections among the MPI indicators; hence, they propose a framework for inferring such links, the R package “BiCausality”. Besides the MPI, the World Bank’s Multi-dimensional Poverty Measure (MPM) also incorporates exciting dimensions of multi-dimensional poverty (Alkire et al., 2023). Alkire et al. (2023) propose an adjustment to the MPI to show moderate rather than acute multi-dimensional poverty, that is, the Moderate Multi-dimensional Poverty Index (MMPI). The new measure is expected to unmask certain aspects of multi-dimensional poverty.

However, empirical studies have largely adopted the Foster-Alkire counting approach to measure multi-dimensional poverty. The method was applied to identify relevant dimensions for measuring water and sanitation poverty among households in Manhiça, Mozambique (Giné‐Garriga & Pérez‐Foguet, 2019) and to assess the importance of livestock and land ownership in the MPI for households from a drought-prone Ethiopian rural region (Abeje et al., 2020). The approach was also used to evaluate the relevance of household farm and non-farm income on the MPI of a farming community in the Philippines (Cerio et al., 2019). Ferreira and Lugo (2013) argue that the measurement of multi-dimensional poverty should go beyond Alkire-Foster scalar indices and the dashboard approach of Ravallion (Ravallion, 2011) and suggest the use of stochastic dominance analysis, Venn diagrams, and copular functions.

Based on Demographic and Health Survey (DHS) data for Zimbabwe, Tanzania, Malawi, and Mozambique, Berenger (2019) used four counting-based poverty approaches (the Alkire-Foster method and three approaches sensitive to inequality) to measure multi-dimensional deprivations in education, health, and standard of living. Using Consumer Expenditure Survey data for Brazil, Tavares and Betti (2024) embraced the Alkire-Foster method and a fuzzy approach to explore gender differences in multi-dimensional poverty. Ntsalaze and Ikhide (2018) applied multiple correspondence analyses to the National Income Dynamics Data to assess the significance of incorporating over-indebtedness and lack of employment among the traditional Alkire-Foster variables for the measurement of multi-dimensional poverty in South Africa.

Yellow cluster: determinants of multi-dimensional poverty

Regarding the determinants of multi-dimensional poverty, researchers analysed primary data gathered through household surveys (Dika et al., 2021; Eshetu et al., 2022; Mandal et al., 2024; Mishra et al., 2022; Shah & Debnath, 2022), while others relied on information from secondary sources (Agyire-Tettey et al., 2021; Altamirano Montoya & Teixeira, 2017; Battiston et al., 2013; Cheng et al., 2021; Cowling et al., 2014; Padda & Hameed, 2018; Pradhan & Pradhan, 2023; Pradhan et al., 2023; Satapathy et al., 2023; Shakoor & Shah, 2022; Zeeshan et al., 2022; W. Zou et al., 2023). Moreover, the focus of these studies was on the determinants multi-dimensional poverty among rural households (Dika et al., 2021; Eshetu et al., 2022; Mandal et al., 2024; Padda & Hameed, 2018; Shakoor & Shah, 2022; Sinha et al., 2022; Zeeshan et al., 2022) or at the national level (Agyire-Tettey et al., 2021; Altamirano Montoya & Teixeira, 2017; Cowling et al., 2014; W. Zou et al., 2023). However, other studies have specifically investigated multi-dimensional child poverty (Agyire-Tettey et al., 2021; Mandal et al., 2024; Pradhan & Pradhan, 2023; Pradhan et al., 2023).

Eshetu et al. (2022) established that cooking fuel, housing, electricity, land ownership, and tropical livestock units were major contributors to rural multi-dimensional poverty in Ethiopia. Research on the determinants of multi-dimensional poverty in rural Pakistan (Shakoor & Shah, 2022) revealed that 75% of overall poverty emanated from the dimensions of health and living standards. Living standards also significantly contributed to early childhood poverty in Ghana (Agyire-Tettey et al., 2021). An Indian study on early childhood multi-dimensional poverty found that

“…of the 15 indicators considered, availability of TV/Radio, followed by immunisation, sanitation, cooking fuel, housing condition, and stunting, are the leading contributors with a combined contribution of more than 55%.” (Pradhan et al., 2023, p. 1).

In general, most studies relied on the Alkire-Foster methodology. The dimensions considered varied among the research papers, and in some cases, more than 10 aspects were investigated in the same study. Generally, living standards or conditions such as inadequate sanitation facilities, poor housing, unsafe drinking water, and lack of clean cooking fuel significantly impact multi-dimensional poverty. This was largely the case regardless of whether the study focused on households at the national level, rural areas, or children. Several studies also found that the head of a household’s education level, income, and job insecurity influenced multi-dimensional poverty. Households with more educated heads and higher incomes were less deprived, and job insecurity increased deprivation. In addition, maternal education reduced the likelihood of multi-dimensional child poverty. Health indicators such as child mortality and malnutrition have also been widely researched and have emerged as significant contributors to multi-dimensional well-being.

Particularly in rural areas, the size of cultivated land or land ownership proved to be an important factor; limited access to land increased multi-dimensional poverty. Also, being rural was generally associated with increased deprivation, underscoring the need for policies to eradicate poverty in rural communities. Other dimensions investigated among the sampled research articles but not very prominent across different studies include the availability of social security benefits, good road networks, access to markets, climate change, pollution, amount of credit taken, birth order, dependency ratio, household size, and household head’s age, gender, and occupation.

Blue cluster: effects of multi-dimensional poverty

Multi-dimensional poverty has multi-faceted effects such as reduced labour productivity and economic growth, physical and mental health deterioration, and social problems (crime, substance abuse, school drop-outs). From the sampled research articles, most studies focused on the health effects of multi-dimensional poverty. Díaz et al. (2022) assert that in comparison to material poverty, deprivations linked to work, medical insurance, and schooling have a greater likelihood of causing mental health problems during adolescence. A Colombian study found that the risk of mental health challenges was 50% higher among multi-dimensionally deprived households than in non-poor (ibid). A strong positive association between multi-dimensional poverty and dementia was established among South African adults of Soweto (Trani et al., 2022).

Studies have also found associations between multi-dimensional poverty and non-communicable diseases (NCDs). The odds of hypertension were higher among South Africans living in more deprived wards of a district in KwaZulu-Natal (Madela et al., 2023). In Tanzania, stroke survival was found to be higher among individuals from less deprived households than among those from more deprived (Jørgensen et al., 2023). A high incidence of multi-dimensional poverty has also been linked to mortality due to rabies (Taylor et al., 2023) and COVID-19 (Mendoza Cardozo et al., 2023).

“…our study revealed that the CMPI (Colombian Multi-dimensional Poverty Index) was associated with increased mortality in hospitalised COVID-19 patients in Colombia. Patients with a higher CMPI index had a higher likelihood of death compared with those of the lowest poverty level.” (Mendoza Cardozo et al., 2023, p. 4).

Besides the health effects highlighted above, multi-dimensional poverty influences people’s behaviour and happiness. Research among employees of an insurance firm in America established that improved multi-dimensional well-being increased job satisfaction and reduced work distraction (Fung et al., 2024). Multi-dimensional poverty increased the odds of being unhappy among individuals in China, Japan, and Korea (Nozaki & Oshio, 2016). A study on the association between multi-dimensional poverty and antibiotic misuse among patients in Kenya, Tanzania, and Uganda showed that poverty does not necessarily cause deviant behaviour (Green et al., 2023). The study found that misuse of antibiotics (self-medication and skipping doses or not completing a course) was prevalent among the least deprived than among the more impoverished.

Red cluster: fighting multi-dimensional poverty

Studies have been conducted to evaluate the effectiveness of measures aimed at alleviating multi-dimensional poverty in different countries. Ownership of resources positively impacted agricultural productivity among small farmers in Pakistan, ultimately reducing poverty incidence (Abrar ul Haq et al., 2021). In Ethiopia, the adoption of improved rice varieties positively impacted farmers’ yields, thereby reducing multi-dimensional poverty (Assaye et al., 2022). Chowdhury and Mukhopadhaya (2012) found that in comparison to financing granted by non-governmental organisations (NGOs), micro-finance extended by the government was more effective in improving the well-being of rural households in Bangladesh. Rural land consolidation (infrastructural improvements, conservation programmes, financing) improved the welfare of poor rural households in China (Cheng et al., 2021). Although financial support is necessary to fight poverty, the risk of over-indebtedness cannot be ignored. Using data for South Africa, Ntsalaze and Ikhide (2017) established that debt service-to-income levels beyond 42.5% caused a deterioration rather than an improvement in household welfare.

In Ethiopia, improved resilience capacity (access to basic services, access to social safety nets, and related measures) caused a decline in multi-dimensional poverty (Haile et al., 2021). Zakat (charity-giving to the less privileged) reduced multi-dimensional poverty in Pakistan as it enabled the beneficiaries to meet some of their necessities in life (Aziz et al., 2020). Using data from 110 countries, Santos et al. (2019) found that economic growth, exports, control of corruption, and an increased share of industry and services reduced multi-dimensional poverty. In Thailand, the growth of industrial parks has alleviated poverty in surrounding areas, including at the village level (Hutasavi & Chen, 2022). Based on the China Family Panel Studies data, Liu et al. (2021) confirmed the poverty-alleviation effects of digital information technology use, especially in urban areas. Other Chinese studies also found a reduction in multi-dimensional poverty as a result of smartphone usage (Liang et al., 2024) and digital financial inclusion (proxied by the number of electronic accounts, number of users, transactions per capita, and transaction costs) (Wang et al., 2024).

Implications

This analysis has implications for stakeholders. This review showed that the literature on multi-dimensional poverty has focused on its measurement, causes, effects, and strategies to eradicate the problem. The Alkire-Foster methodology has been a cornerstone of multi-dimensional poverty research, with dimensions such as health, income, education, and living standards prominent in different countries. Lack of resources, such as land and finance, also intensified poverty.

In developing countries, agriculture is an avenue through which rural households can escape poverty. To help rural households fight multi-dimensional poverty, governments can institute measures to promote access to agricultural land and inputs. However, diversification of smallholder agriculture rather than land expansion holds greater potential in reducing multi-dimensional poverty among rural households (Fan & Cho, 2021). Measures such as improving access to basic services like health and education, market access, diversification of livelihoods, infrastructural improvements, industrialisation, and social safety nets are also ideal for alleviating multi-dimensional poverty.

In this Fourth Industrial Revolution (4IR) era, the role of information and communication technologies (ICTs) in combating multi-dimensional poverty is paramount. Hence, partnerships with the private sector to improve ICT infrastructure are crucial. For development practitioners, financial support for specific projects and donations can improve households’ probability of moving out of poverty. Equally important is the need for donors to align their interventions with those of the governments of recipient nations (Larrú & González, 2021). Above all, local challenges are most effectively tackled through local solutions; hence, interventions to fight multi-dimensional poverty should be connected to the resources, heritage, and human capital available in communities being assisted. In short, the effectiveness of interventions could be context-specific. Measures that work in certain regions of the world may not be effective in other countries, and critical evaluation of alternatives is necessary before implementation.

Directions for future research

Longitudinal and time series studies on multi-dimensional poverty are generally scant, primarily because of data limitations. Most of the studies reviewed in this research were based on survey data collected by researchers or sourced from national statistics agencies in various countries. A better understanding of multi-dimensional poverty over time is crucial and can be the focus of future studies. For example, short panels based on multi-dimensional poverty survey data from different countries can be utilised to investigate the socioeconomic, demographic, and environmental determinants of multi-dimensional poverty in developing countries. The content analysis on the determinants of multi-dimensional poverty showed that there is room for more research on the role of factors such as road networks, access to markets, climate change, pollution, amount of credit taken, and social protection.

Other empirical studies have underscored the importance of ICTs in alleviating multi-dimensional poverty in China (Liang et al., 2024; Liu et al., 2021; Wang et al., 2024). Although the lack of digital infrastructure has inhibited the wide and rapid adoption of 4IR technology in most developing countries, multi-dimensional poverty alleviation research in this area remains important. Thus, rather than limiting research on the effects of ICTs such as computers, mobile phones, and the Internet on multi-dimensional poverty, researchers can assess the impact of technological innovations in a broader sense. In addition, instead of relying on traditional surveys to collect data for multi-dimensional poverty analysis, researchers can use artificial intelligence (for example, drones) and gather information from otherwise difficult-to-reach areas.

The overlay visualisation and content analysis revealed that research on the effects of multi-dimensional poverty is fairly new and topical. While it is recognised that multi-dimensional poverty has multi-faceted effects, the interest of most studies has been on health, leaving economic and social effects underexplored. For example, more studies on the impact of multi-dimensional poverty on worker morale and productivity can be conducted in both developing and developed countries. Multi-dimensional poverty also has implications for crime rates, substance abuse, and related social behaviour, and future research can delve deeper into these areas. Borrowing from the findings of a study on the effect of multi-dimensional poverty on antibiotic misuse (Green et al., 2023), mixed findings on the association between multi-dimensional poverty and other social behaviours are likely to be obtained. Additionally, micro and macro studies on the nexus between multi-dimensional poverty and corruption may yield interesting findings.

Conclusions and limitations

This research provides a comprehensive overview of the intellectual structure of multi-dimensional poverty. To achieve this, a bibliometric analysis and content co-occurrence literature review on multi-dimensional poverty was undertaken based on research articles downloaded from the Scopus database and published in English until March 2024. While related systematic literature reviews on multi-dimensional poverty exist (D’Attoma & Matteucci, 2023; Zulkifli & Abidin, 2023), these have been restricted to the measurement of multi-dimensional poverty, justifying the need for a study that provides a holistic overview of what has been investigated so far. In addition, the previous studies relied on traditional literature review, hence the risk of researcher bias cannot be ruled out. Additionally, traditional literature reviews are not suited for analysing a large volume of publications, a limitation addressed using computer software like VOSviewer.

Through bibliometric analysis, insights into the most impactful publications, journals with the most publications on multi-dimensional poverty, and leading themes of multi-dimensional poverty research were provided. A co-word analysis revealed four research themes: “measuring multi-dimensional poverty”, “determinants of multi-dimensional poverty”, “effects of multi-dimensional poverty”, and “fighting multi-dimensional poverty”. The content analysis performed thereafter highlighted areas for future research, including conducting longitudinal investigations of multi-dimensional poverty to better understand the problem over time, analysing the effects of recent technological innovations on multi-dimensional poverty, and further exploring the economic and social effects of multi-dimensional poverty.

This study relied on the Scopus database, and relevant studies indexed in the Web of Science, Dimensions, and Google Scholar could have been omitted. Thus, future multi-dimensional poverty research can use these alternative databases. Moreover, instead of using VOSviewer software for analysis, future research can use alternative programmes, and different multi-dimensional poverty research clusters might be generated. Additionally, rather than restricting the analysis to documents published only in English, it is worthwhile to incorporate multi-dimensional poverty studies written in other languages. Finally, although the Microsoft RAND function was useful in ensuring randomness in the selection of research articles for content analysis, this was achieved at the cost of excluding some of the most cited papers. Thus, future studies can focus on a specific multi-dimensional poverty theme and adopt other criteria to identify publications for content analysis.