Multiparty democracy, social cohesion, and human development in Sub-Saharan Africa: A panel data analysis

The regime type and poverty theories

Regime type and poverty theories are founded on the presumption that a close relationship exists between the mode or type of a country’s government and the standard of living of its people. Moreover, these theories imply that democracy has the potential to improve the living standards of the lowest income groups, while nondemocracy lowers the standards (Acemoglu, Naidu, Restrepo, & Robinson, 2019; Altman & Castiglioni, 2009; Besley & Kudamatsu, 2006; Blaydes & Kayser, 2011; Deacon, 2009; Eterovic & Sweet, 2014; Gerring et al., 2015; Gerring, Thacker, & Alfaro, 2012; Haggard & Kaufman, 2008; Hanson, 2015; Kudamatsu, 2012; McGuire, 2013; Miller, 2015; Sen, 1999; Sen, 1981).

Three theories are at the center of the regime type and poverty model (Ross, 2006), two of which are associated with the works of Sen (1999, 1981) on the causes of famine and poverty (Ross, 2006). In his works, Sen first argued that democracy enables the poor to inflict sanctions on governments that enable famines to occur, through electoral processes, and that political leaders would act judiciously to prevent famines and avoid sanctions (Ross, 2006). Secondly, Sen argued that democracies are relatively better than nondemocracies in terms of their capacity to disseminate information from remote and poor areas to the central administration due to the freedom of the press. Thus, according to him, even when leaders of a democratic and a nondemocratic government are equally devoted to preventing the occurrence of famine, democratic leaders are more likely to know precisely when action is required (Ross, 2006).

The third theory suggests that a democratic system would make more expenses in terms of welfare benefits for the poor, as compared to nondemocratic regimes. Hence, several scholars opine that democracies expend more on services and public goods as the electoral process compels them to, while non-democratic governments face no such constraint (Ross, 2006). A priori, there are good indications that these theories are at least partially correct - democratic regimes appear to be more concerned about social goods and services than non-democratic governments.

Indeed, historical studies have shown a partial correlation between suffrage extension and the size of government welfare spending both in the US (Gouveia & Masia, 1998) and more generally in the Western and Latin American countries (Kristov, Lindert, & McClelland, 1992; Lindert, 1994). In addition, a study of 44 African countries by Stasavage (2005) has outlined a strong substantiation that democracy has augmented government spending on education, while a series of analyses of Latin American countries has found that democracy is highly correlated with higher spending on health, education, and social security (Avelino, Brown, & Hunter, 2005; Brown & Hunter, 2004; Kaufman & Segura-Ubiergo, 2001). However, even if democratic governments increase social expenditure, a posteriori, they do not improve infant or child mortality rates, which are considered by many to be the most accurate and comprehensive indicators of social welfare among the poor (Ross, 2006).

Thus, the literature on democracy and development appears to demonstrate conflicting views regarding the plausible effects of multiparty politics on the well-being of people. Until recently, the most commonly held view in this context has been that multiparty democracy ameliorates human development (Gerring et al., 2012; 2015). This view has been observed in various scholarly works (Altman & Castiglioni, 2009; Besley & Kudamatsu, 2006; Blaydes & Kayser, 2011; Brown & Mobarak, 2009; Deacon, 2009; Eterovic & Sweet, 2014; Haggard & Kaufman, 2008; Hanson, 2015; Kudamatsu, 2012; McGuire, 2013; and Miller, 2015) for instance.

Lately, however, the commonly held view that the well-being of individuals is influenced by democracy has been fiercely challenged and debunked, with numerous quantitative and qualitative studies conducted to corroborate the absence of a tangible association between regime type and diverse aspects or measures of human development (Gerring et al., 2012; 2015; McGuire, 2006; Ross, 2006; Shandra, Nobles, London, & Williamson, 2004). Further, given recent factual analysis, some of the arguments in favor of the above opinion are questionable. While most people generally believe that democracy would result in higher social expenditures, which, in turn, would improve the welfare of the poor, it has been noted that there is little to no relationship between public spending and human development beyond the context of the OECD countries (Filmer & Pritchett, 1999; McGuire, 2006).

In bids to reconcile these two views and to probe deeper into the relationship between democracy and human development, Gerring et al. (2012) have introduced the possibility that the developmental effects of democracy may be longer-term, characterized by a distal rather than proximal causal relationship. Hence, these authors opine that empirical works must test the relationship between the two variables while accounting for the time lag of the model. Owing to further research and a series of statistical checks with the infant mortality rate (IMR) as the main measure of human development, in conjunction with two hypotheses (the first, replicating the traditional causal model that links IMR to a country’s achievement or level of democracy in the previous year; and the second, measuring democracy with a “stock” indicator that captures a country’s regime history from 1900 to the observation year), Gerring et al. (2012) have found that the contemporary level of a country’s democracy has a weak association with enhanced human development, while a country’s historical experience with democracy has a strong and robust influence on human development. Hence, the researchers concluded that democracy does ameliorate human development only in cases where it is considered a historical or “stock” phenomenon. In simple terms, this implies that if a democratic system of governance is upheld over a longer period, its net effect will be resolute for the welfare of its citizens. Where one stands today depends essentially upon where one has been (Gerring et al., 2012).

In the context of the present study, the measure of human development extends beyond the infant mortality rate to relate to the quality of life, improvement in personal income, environmental conditions, and healthcare and education attainment (Seers, 1979). Hence, it is difficult to suggest that sustained or meaningful human development has occurred in Sub-Saharan Africa. Thus far, multiparty democracy has not eradicated poverty or hunger in the region, which are incompatible with any form of development. The measures of human development as well as the other variables used in the present study can be found under Underlying data (Diori & NaRanong, 2022).

Social cohesion theory

In his work entitled “The quest for good governance and development in Sub-Saharan Africa,” Chabal (2002) quoted the NEPAD (New Partnership for African Development), declaring that peace, quality governance, socio-political stability, institutional consolidation, and the rule of law are necessary for investment and economic growth. Moreover, NEPAD is of the opinion that the aforementioned factors necessary for growth can only be derived from democracy, which has long been considered by Western development agencies and donors as the main precondition for the “takeoff” of any nation intending to transition from a state of underdevelopment to development.

However, such a perception of democracy and its impact on development has been met with strong criticism. Today, it is an open secret that democracy in SSA has neither brought about social cohesion and political stability, nor better institutions and economic growth, which are normally considered the prerequisites for any development. This idea was supported by Easterly et al. (2006), who provided substantial proof to show how social equity and group cohesion can lead to better institutions and economic growth. For these researchers, strengthening cohesion within society by creating and maintaining quality institutions that benefit all members and lower socio-economic cleavages is imperative for countries struggling with growth (Easterly et al., 2006).

Social cohesion within a country is critical for establishing the trust and civic participation required to enact specific reforms as citizens must accept the short-term depravations that naturally arise from reform implementations before the situation ameliorates in due course (Easterly et al., 2006).

This study drew on Easterly et al.’s theory of social cohesion, institutions, and growth (2006). Social cohesion theory was deemed suitable for this study as it emphasizes the reasons for which multiparty democracy has rarely fostered human development in the SSA region. Hence, the search for tangible effects of multiparty democracy and social cohesion on the development of the SSA region occurred in line with the theoretical lens offered by Easterly et al. (2006) as well as the regime type and poverty theories. The same methodological lenses were utilized to address the research hypotheses, which consisted of testing the short-run and long-term upshots of multiparty democracy and social cohesion on the well-being of the people in SSA.

As for the literature on social cohesion and human development, it has not been prolific regarding the relationship between the two concepts. Few theoretical and empirical attempts have been made to establish a linkage, both direct and indirect, between the two concepts. In a study that aimed to explore the potential effects of social cohesion on human development using state legitimacy as a mediator between the two variables, Seyoum (2020) finds that social cohesion has a direct and an indirect effect on the well-being of people, i.e., human development. Using data from 180 countries and the state fragility index as a proxy of social cohesion, the latter study identifies several aspects of a socially cohesive state with a significant influence on poverty alleviation and sustained development (Seyoum, 2020). However, decreased or absent social cohesion was found to generate a lack of successful institutions, which, in failed states, are characterized by political and economic rules and regulations that frequently promote indecency and undermine the fair distribution of public wealth (Easterly et al., 2006; Seyoum, 2020). Although resources may proliferate in a failed state, they are seldom used to address public poverty or improve the health system. Hence, the consequence of state fragility would eventually be felt in education (e.g., poor school systems and facilities, the lack of quality and quantity of teaching materials), healthcare (poor healthcare systems and facilities, and inadequate treatment), and income distribution (inequality and discrimination in GDP per capita, unequal access to state resources). The findings consider a tangible relationship between the inadequate provision of goods and public services with state failure and fragility; hence, Amate-Fortes, Guarnido-Rueda and Molina-Morales (2017) conclude that the incapacity of the state to provide for security, necessary goods, and services also has a negative effect on economic growth, and consequently, on human development.

A similar study by Rotberg (2010) finds that in instances where a state fails, its population disperses, human capital is drained, and the total production and per capita income start to dwindle. State failure results in governments failing to account for their citizens and, at times, being unwilling to formulate policies to reduce poverty and advance human development. In short, failed states suffer from a dearth of good governance, ranging from the establishment of credible institutions to the consolidation of democratic achievements, in addition to a dearth of rule of law and equity.

Furthermore, a 2018 OECD report has found that state fragility is mainly caused by a lack of cohesion within a society as well as its fractionalization into groups with opposing identities and daily struggles. In the absence of social cohesion, even the elites, who are to unite such groups, tend to be factionalized, regarding the state as a means of personal profit rather than a politically organized unit that may be developed for common interests.

Inequalities and ethnic fractionalization are current challenges that Sub-Saharan Africa faces due to the dreadful disparities between the rich and the poor. Ethnic fractionalizations and various types of inequality operate as channels through which considerable social unrest can erupt (Easterly et al., 2006).

In summary, social cohesion is proven to be indispensable for SSA countries to transit to human development. Its absence may cause dire social pathologies that may not only have an effect on economic issues but also endanger the subsistence and viability of society at large (Durkheim, 1897). Nowadays, it is unanimously agreed that human development not only depends on economic issues but also on numerous social consensuses to be met.

Study design

The present study uses a panel data technique to determine the effect of multiparty democracy and social cohesion on human development in SSA. Panel data analyses are correlational research strategies entailing a methodical examination of possible relations between a set of variables. They are utilized to determine if variations in an outcome or criterion variable are related to possible changes in one or more predictor variables. This is exactly the case of this study where efforts were made to determine whether the variations in human development measures are attributed to the regime type (multiparty democracy) in use in the SSA region, or to the presence or absence of social cohesion, while controlling for other possible determinants. The panel data approach was selected in this study to address the research hypotheses, developed as follows:

In hypothesis 1, a distinction is made between the long- and short-term effects of multiparty democracy on human development. This hypothesis is consistent with the literature on democracy and development, where an immediate relationship and a log-run effect between the two variables are often observed. This may be viewed through the lens of “supply” (post-electoral government provisions and stimulus packages) or “demand” (advances in human development as demanded by voters). In either observation, multiparty democracy may have a short-term relationship or effect on human development.

Alternatively, with regard to the distal effect of multiparty democracy on human development, this effect may manifest as an increase in economic growth, which eventually leads to higher per capita income, higher achievement in health, education, etc. In this optic, it is possible to affirm that the effects of multiparty democracy on human development are not immediate; rather, they present themselves in the longer run. Countries must enjoy economic growth first before their populations receive the benefits attached to the same. This long-run effect is also echoed in H1.

In hypothesis 2, however, the empirical evidence offered by extant literature indicates a long-term relationship between social or community cohesion and human development. This is observed via the attainment of quality institutions first, where the latter induces economic growth, subsequently bringing about an improvement in human development (see Easterly et al., Social Cohesion, Institutions, and Growth theory - section 2).

Similarly to H1, hypothesis 3 exhibits a twofold relationship; one that is immediate and the other long term, in relation to the interaction effect between multiparty democracy and social cohesion on human development. These relations are observed through “supply” and “demand”, as explained in hypothesis 1. The assumption of all of the hypotheses suggests a positive, rather than negative relationship, between multiparty democracy, social cohesion, and human development. This is in line with most theories and empirical research findings in the literature on these relations.

Sampling

The population of this study comprises the member states of the SSA region that have experienced democracy for a long period of time. There are 46 countries in total in SSA according to the UNDP. The sample (a non-probability convenience one of 35 Francophones, Anglophones, and Portuguese language speaking countries) was made up of all of the countries in SSA that have had a minimum of 25 years of democratic experience. As for the sampling frame, it constitutes the Eastern, Western, Central, and Southern African countries, and is representative of the four sub-regions of the main Sub-Sahara region (see Table 1 below).

Data collection

Secondary data about four dependent variables (human development index, infant mortality, life expectancy, and basic drinking water), five independent variables of interest (multiparty democracy 1 and 2, political rights, civil liberties, and social cohesion), and four control variables (GDP per capita, institutional quality, population growth, and political stability), were employed in the present study to test the research hypotheses. These data, which consisted of socio-economic development and human capital indicators of 35 SSA countries for the years 1995-2019 (see Diori & NaRanong, 2022), were compiled from diverse sources such as the 2020 United Nations Development report (UNDP, 2020), for data pertaining to human development; the 2021 World Bank World Development Indicators (World Bank, World Development Indicators, 2021), for data on infant mortality, life expectancy, and basic drinking water; the Freedom House annual survey of civil liberties and political rights (Freedom House, 2021); and Marshall and Gurr (2018) for data on democracy. The data for social cohesion was sourced from the State Fragility Index and Matrix (Marshall & Elzinga-Marshall, 2020) while data on governance and political stability was sourced from Kaufmann and Kraay (2020). Other data regarding population growth was sourced from the United Nations 2019 World Population Prospects (UNDESA, 2020); the data for economic growth was sourced from the 2020 World Bank national account databank (World Bank National Accounts Data, 2020); while data for per capita GDP was sourced from the Maddison Project Database (Bolt & van Zanden, 2020). This approach is consistent with the literature, ensuring that findings were of the same standard as previous empirical studies on multiparty democracy and human development. As far as the time span of the collected data is concerned, one may note that the year 1995 represents three to four years after a full democracy was established in most SSA countries that were sampled in this study. The year 2019 represents the latest year data were available for most or all countries in the sample, and all variables. The full dataset, the data tables, and the link to the data sources can be found under Underlying data (Diori & NaRanong, 2022).

Data analysis

Panel data was used in this investigation to address the research purpose, consisting in finding out whether multiparty democracy and social cohesion have some effects on human development. Three different techniques were utilized to ensure that the data met all of the assumptions of the regression analysis. First, measures of central tendency were used—mean, standard deviation, minimum, and maximum—in order to probe the normality and linearity of the dataset (Table 2). Secondly, correlation matrix (Table 3) was used to check whether the variables were strongly correlated. Thirdly, Breush-Pagan/Cook-Weisberg and Wooldridge tests (Table 4) were carried out to look at issues concerning heteroscedasticity and serial correlation.

In order to control for the specific characteristics, two estimation techniques were used: panel fixed-effects (FE), and an instrumental variable in the system generalized methods of moments (system-GMM) estimation. The aforementioned panel data estimations were most appropriate for this study for several reasons. First, because they helped to explain how multiparty democracy and the social cohesion variables within each Sub-Saharan country affect human development over time, and by so doing, they allow for the heterogeneity and individuality of each country with respect to the relationship between the dependent and independent variables. Second, with the combination of cross-section and time series observations, these panel data estimations allowed for additional information, greater variability, less collinearity, greater degrees of freedom, and thus more efficiency. Third, they permitted the detection and measurement of the effects of unobserved country specific effects and therefore reduced the bias in the estimated coefficients.

It is worth noting however that, while the fixed-effect technique was used to control for unobserved country-specific heterogeneities such as structural characteristics, historical differences and time-invariant shocks, the system GMM was used to deal with endogeneity issues and to check for the robustness of the findings. System GMM was chosen and utilized in this study because of its use of internal instruments as parts of the independent variables (Arellano & Bond, 1991).

Finally, note also that, data about some of the variables used in the analysis were incomplete for some countries and time periods, making this panel data unbalanced. The problem was addressed using the Stata 15 software which handled the missing data issue automatically.

Model specification

The specification of the statistical models within the study was based on the review of the associated theories. Four different measures of multiparty democracy were included in the basic models, with two in the dynamic models. Of the four measures of democracy, three were sourced from the most popular institution, Freedom House, while one was drawn from Polity 5. As previously indicated, this approach remains consistent with the existing literature, as several scholars have utilized at least two of the same measures for the variable of democracy (see Gerring et al., 2012, 2015). Hence, multiparty democracy 1 (the Freedom House measure of democracy) is termed Democ1, while multiparty democracy 2 (the Polity 5 measure of democracy) has been termed Democ2. Democ1 was reduced to its component levels with two more measures and variables having been added to the analysis - political rights (PR) and civil liberties (CL). The choice of the fifth independent variable, social cohesion (SC), was also consistent with the literature (see Amate-Fortes et al., 2017; Easterly et al., 2006; OECD, 2018; Rotberg, 2010; Seyoum, 2020). The dependent variables in this study were the human development index (HDI), the variable of interest, and three of its proxies: infant mortality (IMR), life expectancy (LEXP), and basic drinking water (BASW). With the exception of the fourth variable, basic drinking water (BASW), which was created by the authors of this study, infant mortality, life expectancy, and human development index are considered prominent in the context of human development research (see Altman & Castiglioni, 2009; Besley & Kudamatsu, 2006; Blaydes & Kayser, 2011; Deacon, 2009; Hanson, 2015; Gerring et al., 2012, 2015; Kudamatsu, 2012; McGuire, 2006, 2013; Miller, 2015; Ross, 2006; Shandra et al., 2004). The interaction between multiparty democracy 1 and social cohesion was termed Democ1SC, while Democ2SC referred to the interaction between multiparty democracy 2 and social cohesion. Finally, the decision was made to control for institutional quality (IQ), per capita GDP (GDPPC), population growth (PGR), and political stability (PS) to curb their influence on the dependent variable, which may have caused a bias in the statistical results.

The basic models are specified as follows:

For infant mortality (IMR,it), life expectancy (LXP,it), and basic drinking water (BASW,it), which represent the other dependent variables, the same equation was used along with the independent variables.

The dynamic panel models are specified as follows:

For infant mortality (IMR,it), life expectancy (LXP,it), and basic drinking water (BASW,it), which represent the other dependent variables, the same equation and independent variables were used.

Descriptive statistics

The descriptive statistics in Table 2 show the potential presence of outliers for variables such as infant mortality: mean (65.389) and standard deviation (27.56); life expectancy: mean (56.217) and dispersion of (6.934); and basic drinking water: mean (63.065) and low dispersion of (15.223). The data indicated that certain countries perform better than others with regard to particular variables. However, owing to the large data observation (822), the outliers did not significantly violate the normality assumption.

Matrix of correlation

Table 3 illustrates that the correlation between multiparty democracy 1 and political rights on one side, and the correlation between multiparty democracy 1 and civil liberties on the other side, are very high, with respective values of 0.97 and 0.96. To resolve this issue of multicollinearity, the concerned variables were separated in the estimation models.

Heteroscedasticity and serial correlation tests

In Table 4, the results of the analyses revealed the absence of heteroscedasticity. As for the Wooldridge test, the results indicated the presence of serial correlation, which was rectified over the course of the analysis.

Effect of multiparty democracy on human development

Four varying measures were used to test the hypothesis across the fixed-effects estimation: multiparty democracy 1 or Democ1, multiparty democracy 2 or Democ2, political rights, and civil liberties. The dependent variables were the human development index, the variable of interest, and three of its proxies: infant mortality, life expectancy, and basic drinking water. Other variables were added to the model as control variables in accordance with the existing literature.

From models (1) to (3) in Table 5, results showed that both multiparty democracy 1 and 2 (Democ1 and Democ2) appeared related to human development. While multiparty democracy 1 was shown to exert a negative impact on human development at 5%, multiparty democracy 2 appeared to be positively related to human development at a 1% error level. The data indicate that a 5% change in the value of multiparty democracy1 is likely to be associated with -0.0147 points worth of decrease in the human development index, while a single unit change in the value of multiparty democracy 2 will be associated with 0.0110 points worth of increase in the human development index. The immediate implications of these findings corroborated Gerring et al.’s (2012) study, wherein the contemporary level of a country’s democracy was found to have only a weak association with enhanced human development; however, its historical experience with democracy has a strong influence on human development. Given the periodicity of the data that began in 1995, i.e., only 3 to 4 years after the democratization of most countries in SSA, in addition to the fact that multiparty democracy is relatively new in Africa, one may expect it to have little or no effect on the well-being of the population in SSA.

With regard to the control variables included in the models, the findings indicated that GDP per capita was the only variable associated with human development. This variable was shown to have a positive impact on human development in models (2) and (3). Bear in mind that the focus of the argumentation of the pro-democracy theories remains on the consolidation of per capita GDP through redistribution. For many scholars, including Aristotle, Madison, and some present-day political economists, it is undeniable that democracy serves as a mechanism for redistribution, and consequently, for human development (Meltzer & Richard, 1981). Hence, these pro-democracy scholars would opine that the advent of democracy would result in higher government social spending, which, in turn, would improve the welfare of the poor and influence human development. Today, this point is contradicted by a mass of existing research findings (see Gerring, et al., 2012; McGuire, 2006; Ross, 2006; Shandra et al., 2004).

From models (1) to (3) in Table 6 below, the findings showed that multiparty democracy 2 is negatively associated with infant mortality at the 0.01 level. The forecast of the results suggested that for a unit of change in the value of multiparty democracy 2, a -3.622-point decrease is likely to occur in the struggle toward improvement in infant mortality. As explained in the previous table (Table 4.4), these findings are in line with the current literature. Owing to in-depth exploration and a series of regression checks with the infant mortality rate as the core measure of human development alongside two hypotheses (the first, replicating the traditional causal model that links IMR to a country’s achievement or level of democracy in the previous year; and the second, measuring democracy with a “stock” indicator that captures the regime history of a country from 1900 to the observation year), Gerring et al. (2012) finds that multiparty democracy has a long-term effect on human development characterized by a distal rather than proximal causal relationship. According to these authors, multiparty democracy ameliorates human development only in case it is considered a historical or “stock” phenomenon. This implies that if a democratic system of governance is upheld over a longer period of time, its net effect will be resolute for the welfare of its citizens (Gerring et al., 2012). However, this is not the case with SSA Africa, which has a relatively new democracy, which is a form of government that has not been upheld for long in most countries.

The findings further showed that GDP per capita may dampen progress toward the amelioration of the infant mortality rate. The DGPPC exerts a negative impact on infant mortality at the 0.05 level, which may be explained by the lack of sufficient healthcare systems and services throughout SSA. Moreover, many countries in SSA are imbued with unrelenting discrepancies in terms of wealth and rampant poverty. To improve the infant mortality rate in SSA, rather than solely enhancing personal income, governments must implement better public policies that could lead to improvements in environmental conditions and healthcare.

From models (1) to (3) in Table 7, the results showed that multiparty democracy 2 is the only variable that was significant out of all independent variables. Democ2 was significant at the 0.01 level, which implies that a unit change in its value is likely to be associated with a unit increase in life expectancy. This result reinforces the prodemocracy theories, which emphasize that popular participation in government empowers ordinary citizens and, consequently, leads governments to become more accountable for their interests (Gerring et al., 2012, 2015; Ross, 2006). The proponents of these theories strongly believe that democracy would lead to higher government social spending, which, in turn, would improve the welfare of the poor and lead to the amelioration in social welfare as measured by mortality, literacy, and other human development outcomes (Gerring et al., 2012, 2015). However, given the limited resources in most SSA countries and the lack of quality institutions, unrelenting discrepancies are observed in terms of wealth and rampant poverty, which may impact life expectancy at birth.

From models (1) to (3) in Table 8, the results demonstrated that none of the four measures of multiparty democracy were related to the measures of human development, which contrasts the prodemocracy theories. However, they corroborate the findings of recent works on the relationship between democracy and human development. Numerous studies with large samples for quantitative and qualitative inquiries have been conducted to corroborate the absence of tangible relations between regime type and diverse aspects or measures of human development (Gerring et al., 2012, 2015; McGuire, 2006; Ross, 2006; Shandra et al., 2004). These studies have fiercely challenged and debunked the commonly held view that democracy influences the well-being of individuals. Beyond the context of the OECD countries, no substantiation of tangible relationships between public spending and human development has been observed (Filmer & Pritchett, 1999; McGuire, 2006).

Regarding the control variables that were included in the models, GDP per capita appeared to have a positive effect on basic drinking water, while institutional quality demonstrated a negative effect on the same. The low magnitude of the relationship between per capita GDP and basic drinking water is atypical in the Sub-Saharan African region, where a lack of quality institutions hinders public policy implementations and the potential for an impact of economic growth on human development.

Overall, the findings showed that the four measures of multiparty democracy had mixed effects on the measures of human development. Multiparty democracy 1 appeared to have a negative relation with the human development index, also carrying an insignificant impact on infant mortality, life expectancy, and basic drinking water. Multiparty democracy 2 was shown to entertain two positive relations, one with the human development index and another with life expectancy. Furthermore, negative and insignificant associations were observed with infant mortality and basic drinking water, respectively. Nevertheless, these mixed findings lend support to part of research hypothesis 1 and the recent literature (Gerring et al., 2012, 2015; Ross, 2006), indicating certain long-term positive effects of multiparty democracy on human development. According to the recent literature, multiparty democracy has little to no effect on human development, as characterized in these results by a negative and a low magnitude positive relationship between the two variables (Gerring et al., 2012, 2015; Ross, 2006).

Effect of social cohesion on human development

In order to test this hypothesis across the estimations, one measure of social cohesion (SC) was used, with four varying measures of human development as the dependent variables: the human development index, infant mortality, life expectancy, and basic drinking water. Other variables were added to the model as control variables in accordance with the existing literature: per capita GDP, institutional quality, population growth, and political stability.

From models (1) to (4) in Table 9 below, the results showed that social cohesion has a positive effect on infant mortality at the 0.01 level, while it corresponds with a negative impact on the remaining three dependent variables: human development index, life expectancy, and basic drinking water at the same level (0.01). To forecast these results, a unit change in the value of social cohesion will be associated with a 4.164-point increase in infant mortality rate amelioration, and a -0.0132, -1.210, and -1.554 decrease in the human development index, life expectance, and basic drinking water services, respectively. This is consistent with the current literature, which makes frequent usage of infant mortality as a measure of human development, insisting on the presence of a long-term relationship between the two variables.

Referring to the control variables that were included in the models (per capita GDP, institutional quality, political stability, and population growth rate), only two appeared to have impacts on the dependent variables. Hence, per capita GDP seemed to dampen progress in infant mortality improvement at the 0.01 level while it showed a positive impact on basic drinking water at the same error level. If we forecast the results of the effect of GDPPC on infant mortality, a 1% change will appear in the value of per capita GDP, which will be associated with a decrease of -0.000951 points toward improvement in the infant mortality rate. Furthermore, a change of 10% in the value of per capita GDP will be associated with a 5.98e-06-points increase in the human development; a 1% change in the same variable is likely to be associated with a 0.000526-point increase in the addition of clean water. These results demonstrate that GDPPC has no positive effect on the accomplishment of a good infant mortality rate in the context of SSA, rather, it shows little effect on the improvement of the human development index and basic drinking water. One reason to believe in this conclusion is that, according to the raw data, the country in SSA with the highest GDPPC, Equatorial Guinea (with $47,562), is not among the best performers as far as infant mortality rate is concerned. The results further showed that political stability appeared to dampen two measures of human development (HDI and BASW) while it showed no effect on infant mortality and life expectation.

Overall, the results have shown that social cohesion was related to human development via infant mortality, as it appeared to have a positive impact at the 0.01 error level. This supports the hypothesis that social cohesion exerts a positive influence on human development. Yet, social cohesion also shows a negative but significant association with other measures of human development (HDI, LEXP, BASW), leading to some mixed findings and support for the research hypothesis, current literature, and the supporting theory, which indicated some long-term positive effects of social cohesion on human development.

Effect of the interaction between multiparty democracy and social cohesion on human development

To test this hypothesis across the estimation, SC, which represented a measure of social cohesion was used along with two varying measures of multiparty democracy: multiparty democracy 1 and multiparty democracy 2. However, due to the high correlation between the two measures of multiparty democracy, they were used in separate models. The dependent variables were human development, infant mortality, life expectancy, and basic drinking water. Other variables were added to the model as control variables in accordance with the existing literature.

The results from models (1) to (4) in Table 10 showed that multiparty democracy 1 had no effect whatsoever on any measure of human development. In contrast, social cohesion had a significant positive relation with infant mortality, although it showed a negative impact on the human development index, life expectancy, and basic drinking water at the 0.01 and 0.1 levels, respectively. These results may be forecasted by hinting that a unit change in the value of social cohesion is associated with a 4.297-point increase in progress toward improvement in infant mortality. Moreover, a unit change in the value of social cohesion is likely to be associated with a -0.0198, -1.567, and -1.671-point decrease in progress toward ameliorating the human development index, life expectancy, and basic drinking water, respectively, in the Sub-Saharan African region. Thus, social cohesion, in this study, improves only the infant mortality rate. Surprisingly, however, the interactive effect of multiparty democracy and social cohesion showed a positive association with the human development index and basic drinking water. It is also worth noting that multiparty democracy 1 failed to show any positive relationship with the human development index or basic drinking water on an individual level. However, when multiparty democracy 1 interacted with social cohesion, the sign of the coefficient of these variables changed, showing a positive association with the human development index and basic drinking water. These findings suggested that despite the level of significance, the interaction between multiparty democracy 1 and social cohesion has the potential to improve human development conditions within a country. The interaction term between multiparty democracy and social cohesion was significant at the 1% level.

Regarding the control variables included in the models (per capita GDP, institutional quality, political stability, and population growth rate), GDP per capita appeared to have a positive effect on the human development index and basic drinking water, while it demonstrated a negative impact on infant mortality. Meanwhile, political stability showed a negative effect on the human development index and basic drinking water in models (1) and (4). Lack of political stability, when coupled with a lack of quality institutions, uncontrolled population growth, and inadequate income distribution could only hinder progress toward sustained human development. A closer look at the raw data indicated that the best-performing countries in terms of human development in the Sub-Saharan African region had four important factors in common - a controlled population growth rate, better institutions, better per capita share among their population, and above all, a stable political environment.

From models (1) to (4) in Table 11, the results demonstrate that multiparty democracy 2 (Democ2) is related to only one measure of human development - basic drinking water. On the contrary, social cohesion was positively related to infant mortality, albeit posing a negative impact on the human development index, life expectancy, and basic drinking water at a 0.01 error level. These results are forecasted to hint that a unit change in the value of social cohesion is associated with a 4.566-point increase in the amelioration of infant mortality rate. Furthermore, a unit change in the value of social cohesion is likely to be associated with a -0.0117, -1.015, and -2.115-point decrease in progress toward amelioration of the human development index, life expectancy, and basic drinking water, respectively, in the Sub-Saharan African region. Thus, social cohesion only improves the infant mortality rate. Note that multiparty democracy 2 had no significant positive relation with any measure of human development. Yet, when it interacted with social cohesion, the effects of the interaction term have become significant with both infant mortality (negatively) and basic drinking water (positively). These findings suggested that regardless of the level of significance, the interaction of multiparty democracy 2 with social cohesion may potentially improve human development conditions within a country. In other words, social cohesion is at its best when associated with multiparty democracy. The interaction term between multiparty democracy 2 and social cohesion was negative and significant at the 0.05 level in model 2 despite social cohesion being positively associated with infant mortality, and positively significant at the 0.1 level with basic drinking water. These results suggested a distal, rather than a proximal, effect of the interaction of both variables in the context of human development.

Overall, the results have shown that the two measures of interaction between multiparty democracy and social cohesion (Democ1SC and Democ2SC) had positive effects on two measures of human development (HDI and BASW), with a negative impact on infant mortality and no significant association with life expectancy. Further, with regard to the magnitude of the relationship, one could only conclude that the interaction between multiparty democracy and social cohesion has a positive long-term effect on human development. These mixed findings lend support to the third research hypothesis (H3), which assumes that the interaction between multiparty democracy and social cohesion has a positive effect on human development.

Instrumental variable approach and robustness checks

Note that fixed effects estimations may have certain disadvantages that derive from the presence of endogeneity and omitted variable prejudice, which has been addressed in this study using an instrumental variable (IV) approach in the system generalized methods of moments.

The results from the system GMM estimation in Table 12, specifically columns (1) to (5), indicated that the lag of the human development index was statistically significant (0.997, 1.001, 0.992, 0.996, 0.999) at the 1% level. This further implies that human development is strongly persistent in SSA. Considering the choice of one lag length, the insignificant specification test results of the AR (2) (0.790, 0.835, 0.945, 0.882, and 0.528) revealed that the system GMM model did not suffer from second-order serial correlation and that the null hypothesis could be accepted. In addition, the results of the Hansen test (0.167, 0.144, 0.266, 0.185, 0.230) have shown that the instruments used were not over-identified, suggesting that they were valid or not correlated with the error term. The results of the system GMM were, to a lesser extent, consistent with the fixed effects estimation albeit with certain minor differences. The coefficients of many of the variables were not significant compared to the fixed effects. In columns (2, 3, and 5), for instance, social cohesion was shown to have no influence on human development, which is unlike the case of fixed effects estimation, where it exerted a negative impact on three measures of human development (HDI, LEXP, BASW) and a positive impact on the fourth (IMR). Multiparty democracy 2 was also not significant as it was in the fixed effects estimation, just as in the case of its interaction with social cohesion, which was not only insignificant but also illustrated a negative coefficient. In column (3), however, apart from the error level, the statistical direction of multiparty democracy 1 and its interaction with social cohesion did not change, remaining negative and significant for the former and positively significant for the latter at the 1% error level. The positive sign of the interaction between multiparty democracy 1 and social cohesion (Democ1SC) suggested that social cohesion has the potential for dampening the negative effect of multiparty democracy on human development.

The results from the system GMM estimation in Table 13 below, specifically columns (1) to (5), indicated that the lag of infant mortality was statistically significant (0.970, 0.970, 0.965, 0.967, 0.981) at the 1% level. This suggests that infant mortality is strongly persistent in SSA. Considering the choice of one lag length, the insignificant specification test results of the AR (2) (0.528, 0.531, 0.450, 0.554, and 0.417) revealed that the system GMM model did not suffer from second-order serial correlation and that the null hypothesis could be accepted. Furthermore, the results of the Hansen test (0.230, 0.219, 0.662, 0.112, and 0.262) showed that the instruments used were not over-identified; hence, they were valid or not correlated with the error term. The results of the system GMM were, to a lesser extent, consistent with the fixed effects estimation, albeit with several differences. The coefficients of many of the variables were significant after controlling for endogeneity issues, unlike in the results of the fixed effects estimation. In column (3), for instance, the coefficient of multiparty democracy 1, which has never shown a positive significance in the fixed effects estimation, has shown a positive value in the GMM estimation, although the statistical direction of its interaction with social cohesion (Democ1SC) is unchanged. In columns (3) and (5), social cohesion was shown to have a positive and a negative impact on infant mortality, unlike in the fixed effects estimation, where it has always been positive. Hence, in column (3), social cohesion has retained its positive effect on infant mortality while the magnitude of the relationship has shrunk from 4.164 at the 1% error level to 1.054 at the 5% level. In column (5), the sign of the coefficient of the variable has become negative at the 10% level. Note that social cohesion has always shown a positive significant relation with infant mortality in the fixed effects estimation. Moreover, in column (5), the sign of the coefficient of multiparty democracy 2 was shown to be negative and significant, similar to its state in the fixed effects estimation. However, its interaction with social cohesion has become positive at the 10% error level, indicating that the negative effect of multiparty democracy 2 on infant mortality has been dampened by its interaction with social cohesion.

From the results of the system GMM estimation method in Table 14, one may note how columns (1) to (5) indicated that the lag of life expectancy was statistically significant (-0.200, -0.114, -0.179, -0.204, -0.182) at the 1% level, suggesting that life expectancy is weakly persistent in SSA. The insignificant specification test results of the AR (2) (0.122, 0.101, 0.012, 0.100, and 0.185) revealed that the system GMM model did not suffer from second-order serial correlation and that the null hypothesis could be accepted. Furthermore, the results of the Hansen test (0.936, 0.936, 0.936, 0.936, 0.937) show that the instruments used were not over-identified; hence, they were valid or not correlated with the error term. The result of the system GMM estimation was, to a lesser extent, consistent with the fixed effects method, albeit with several major differences. The coefficients of many of the variables, which were insignificant in the fixed effects estimation, were found to be significant in the GMM estimation. Hence, in column (1), the sign of the coefficient of multiparty democracy 1 is significant and positive at the 10% level, unlike in the fixed effects estimation, where it appeared to be non-significant and negative. In column (2), the sign of the coefficient of social cohesion remains the same. In column (3), the sign of the coefficient of multiparty democracy 1, which was positive in column 1, was found to be negative and significant at the 1% level, in contrast with the data indicated in the fixed effects estimation results. Further, the interaction between multiparty democracy 1 and social cohesion (Democ1SC), which was insignificant in the fixed effects estimation, has become significant and positive in the GMM estimation (0.314) at the 1% level. In column (5), the coefficient of the social cohesion variable has become positive and significant at the 10% level (0.238), in contrast with the fixed effects estimation results, where it has never been positive. Furthermore, in the same column, the sign of multiparty democracy 2 has remained positive and significant, as in the fixed effects estimation, at the 1% level in the GMM estimation (2.708), whereas its interaction with social cohesion, which was insignificant and negative in the fixed effects estimation, has become significant and negative in the GMM model at the 1% level. This suggests that the combined effects of the two variables had a negative effect on life expectancy.

The results of the system GMM estimation method in Table 15, columns (1) to (5) indicated that the lag of basic drinking water was statistically significant (0.993, 0.994, 0.986, 0.997, 0.982) at the 1% level, suggesting that basic drinking water is strongly persistent in SSA. The insignificant specification test results of the AR (2) (0.081, 0.084, 0.097, 0.087, and 0.092) revealed that the system GMM model did not suffer from second-order serial correlation and that the null hypothesis could not be rejected. In addition, the results of the Hansen test (0.771, 0.713, 0.754, 0.634, 0.821) show that the instruments used were not over-identified; hence, they were valid or not correlated with the error terms. The results of the system GMM estimation were, to some extent, totally inconsistent with the fixed effects estimation, with several differences. The coefficients of many of the variables of interest were not significant, relative to the fixed effects. In columns (2, 3, and 5), for instance, social cohesion was shown to have no impact on basic drinking water unlike in the fixed effects estimation where it had a negative but significant effect on all measures of BASW. Moreover, the coefficient of the interaction between multiparty democracy 1 and 2 with social cohesion (Democ1SC and Democ2SC), which were significant and positive in the fixed effects estimation at 1% and 10%, respectively, have become insignificant in the system GMM estimation. Moreover, the coefficients of the multiparty democracy 1 and 2 variables have remained insignificant in the system GMM estimation.

Overall, after controlling for endogeneity issues, the results of the system GMM appeared to confirm parts of the research hypotheses, which imply a long-term relationship between multiparty democracy and human development, despite the hybrid character of the findings. Two measures of multiparty democracy, namely Democ1 and democ2, are shown to be positively correlated with two of the four measures of human development (IMR, and LEXP), with a relatively low magnitude association. This suggests a distal, rather than proximal, relationship. Additionally, social cohesion was positively correlated with two measures of human development (IMR and LEXP), much like its interaction with multiparty democracy 1 and 2 (Democ1SC and Democ2SC), which were also positively correlated with the human development index and infant mortality rate, respectively. In sum, these positive significant relationships contrasted with the negative significant and non-significant relations that these variables demonstrated with other measures of human development. Nonetheless, these mixed findings do lend support to the recent literature and hypothesis, which suggests that multiparty democracy has little to no effect on human development, or implies a distal relationship with human development (Gerring et al., 2012; 2015; Ross, 2006).

Underlying data

Dataverse: Data on the effects of multiparty democracy and social cohesion on human development in Sub-Saharan Africa, https://doi.org/10.7910/DVN/GPTLSJ (Diori & NaRanong, 2022).

This project contains the following underlying data:

Data are available under the terms of the Creative Commons Zero “No rights reserved” data waiver (CC0 1.0 Public domain dedication).