2.1. Inequality Concepts and Assessment

2.1.3. Assessing Inequality and Its Consequences

In assessing inequality and its consequences, the emphasis is on differences in characteristics between and within alternative categorizations of population segments, as well as on trends in these differences and alternatives. Actuarial analysis is usually conducted on an ex-ante or expected basis, accompanied by documentation of the choices made. To this end, a series of questions should be addressed in the context and objectives of the analysis, as summarised in Figure 1.

Figure 1. Framework for inequality analysis.

These questions include, among others:

What are the objective(s) of the analysis?

• Who are the stakeholders and intended (and unintended) users?

• What is the inequality being studied, for example, of income, wealth, health, longevity, or happiness?

• Is it an assessment of a particular inequality, related inequities, or their causes, outcomes, or consequences?

• What will be the planned use of the analysis, e.g., as an input to the cost-benefit analysis of a public policy decision, the establishment of the price of a product, an assessment of the compensation structure of a financial institution, or the quality of a public health or educational system?

• What form will it take – analytical comparison, qualitative analysis, or both?

What characteristic(s) are being compared?

• What characteristics of individuals or groups are being compared? Multiple characteristics, such as social or economic characteristics, may be used, as well as those related to mortality and morbidity risks.

• Economic characteristics can include income (e.g., wages, self-employment profits, investment income, private or public social transfer benefits, public goods, and services), wealth, or consumption. They can be measured on a per-capita or per-household basis.

• Gross Domestic Product (GDP) per capita is often used as a proxy for income, as it can be readily available on a national level.

• In longitudinal analyses, adjustment may be appropriate, for example, for changes in inflation, relative currency values, consumption patterns, purchasing power, population dynamics, and demographic factors.

• Alternative definitions of characteristics may be better associated with inequality measures and trends (Rose, Reference Rose2018). For example, deducting taxes and transfer payments can considerably reduce reported income inequality; in some cases, the Gini coefficient (index) can decrease by up to a third when comparing income on a before and after-tax basis.

What are the categories or groupings of interest?

• Population segments can be based, for example, on age, sex, marital status, language, religion, range of income or wealth, race/ethnicity, type of location, extent of debt, disability, or employment status.

• Both income and wealth measure economic status, with income measured over a time period and wealth being a point-in-time measure.

• Some categories can be clearly defined (e.g., birth cohort), while others are fuzzy or continuous (e.g., economic status) or even ill-defined (e.g., ethnic group).

What are the aggregations or ranges of values of the categories over which the characteristic(s) are compared?

• Examples of units include income, age, and geographical ranges (e.g., global, national, or local).

• Level of aggregation can change the findings of an analysis, both within and across countries, because of differences in, for example, definitions and scope of measures, fixed costs, standards of living, tax systems, social benefits (e.g., access to or quality of healthcare or education), and labour market regulations.

What metric s are used?

• Several metrics or methodologies may be used to measure the extent of inequality. For example, the average value, the median value, or the Gini coefficient.

• Inequality of opportunity can be especially difficult to measure, partly because it represents potential rather than actual outcomes – thus, it may not be directly observable. Proxies are often used.

What data / information are available or obtainable?

• The availability, quality, reliability, relevance, and objective nature of the data or information used in the analysis are of central importance.

• Relevant and reliable data being accurate, clear, timely, consistent over time and space, credible, and with minimal response bias may be difficult to obtain. This may especially occur when dealing with individuals of extremely high or low income or wealth, and in lengthy or multinational studies.

• The most appropriate source(s) of data may depend upon the application. They can include, for example, tax records, surveys, or insurance records.

• In many cases, income or wealth may be underreported because of the use of offshore accounts, gig or informal work, subsidies, or the absence of accurate tax filings. Objectively derived data, such as that obtained through professional measurement or from birth certificates, are preferable to self-reported data, although the latter may be more readily available. If appropriate, estimates or adjustments may be necessary.

• Significant changes in conditions or the economic or social environment over the covered period(s), especially if volatility, asymmetry, or non-linearities are involved, can affect data or suggest needed adjustments.

What timeframe is involved?

• Populations, societies, and markets change. Although the past can often provide useful information, it may represent a biased indicator(s) of the future unless adjusted.

• The selected period studied depends on the objective(s) of the analysis, what is being studied, available and comparable reliable information, and the underlying drivers and trends.

• It is important, where practical, to use data from the same or similar periods. If cycles have occurred, experience for complete cycles should be included. Comparisons over more extended periods are often more appropriate than short-term comparisons, except when significant changes in drivers or conditions occur over a longer period.

• Careful consideration or adjustment may be needed if a discontinuity or change event has occurred, because conditions before and after it may not be comparable. For example, technological discontinuities (e.g., the introduction of the mobile phone, internet, armed conflict, or a pharmaceutical advance) can bring with them so much change that simple extrapolations from pre-discontinuity conditions may be inappropriate or misleading.

The perspectives of prospective stakeholders should be considered in determining prioritisation and tradeoffs among actions and their implementation. For example, should inequality be considered from the perspective of those in the bottom quantile or from society as a whole? Are differential subsidies or incentives justifiable between individuals or groups? For example, do they result in a greater reward for taking on greater risks that may benefit society as a whole?

To remain objective, multiple approaches, analytical methods, viewpoints, or scenarios should be considered to develop a clear, comprehensive, and objective assessment.

2.2. Metrics of Inequality, Modelling, Aggregation, and Grouping

It is essential to use relevant metrics and assess the quality and consistency of the underlying data used to analyse the extent of inequality and its consequences. This is especially true because much of the publicly available information regarding inequality can be misleading, mean different things to different people, or be derived from inconsistent sources. Further, applying multiple metrics or multiple data sets may help ensure an unbiased analysis.

Many of the metrics used are standard statistical and economic measures, while others are more inequality-specific. Several texts, including those by the Haughton and Khandkep (Reference Haughton and Khandkep2009), Cowell (Reference Cowell2011) and Cowell et al. (Reference Cowell, Karagiannaki and McKnight2017), provide detailed discussions of these metrics. Visual data comparisons can lead to enhanced insights, even when less statistically rigorous. Consistent with actuarial standards of practiceFootnote ⁴ , actuaries may find it useful to discuss and be transparent with non-actuaries about how these metrics, considerations, and practices are applied.

The following are commonly used metrics and techniques employed in the study of inequality:

• Statistical measures include range, average (statistical mean), median (the 50^th percentile value), mode (the most common value), percentile (a specified percentage of the probability distribution), standard deviation, coefficient of variation (a volatility measure), and regression coefficients.

• Both the range and shape of the distribution of the item measured can be important. For example, the distribution of incomes in a highly unequal society is quite different from that of a population with a solid middle class.

• Quantile (e.g., quartiles, quintiles, deciles) segments, the sorted values of the observed data into equal parts, can be usefully compared. A q-quantile divides the data into q parts. An inter-quantile range eliminates the top and bottom quantiles, focusing on the core of the distribution. The top quartile is the highest quantile of a distribution; for example, inequality can be expressed in terms of the percentage share of total income or wealth of the highest-income group.

• Ratios of values of one quantile relative to corresponding values of a different quantile. For example, the Palma ratio (Tutor2u, 2023) is the ratio of the top decile to the bottom four deciles.

According to Cowell (Reference Cowell2011), there are four desirable properties for an inequality measure:

• Transfers: A reduction in inequality occurs when income transfers from the rich to the poor (the rich become not-so-rich, and the poor become not-so-poor).

• Independence: Inequality does not change when everyone’s income is changed while leaving the individual’s share unchanged.

• Scale: Inequality should not depend on the size of the population; for example, when doubling the population, the inequality remains unchanged.

• Decomposability: Inequality remains the same within subgroups of the population.

Three ways to look at inequality are:

• The social welfare function (Haughton and Khandkep, Reference Haughton and Khandkep2009) considers how society treats and prioritises welfare, similar to social utility:

  (1) $$W\left( {{y_1},{y_2}, \cdots, {y_n}} \right) = \mathop \sum \nolimits_{i = 1}^n U\left( {{y_i}} \right)$$

  where U(y) is the utility of the welfare form (or the individual) or its aversion to inequality. A related approach uses, for example, a measure of relative wage inequality within a firm, computed as the ratio of the wage of the top-earning employee to the firm’s median wage. This has been used to assess the reasonableness of the wages of the highest-earning employees of a firm.

• Information theory builds measures of inequality by analogy between the probabilities of events (in the range of 0 to 1) and shares of income, used as weights in a valuation.

• The structural approach builds on axioms and principles essential in developing an index of inequality.

The Gini coefficient (Gini, Reference Gini1912) is a primary measure of the statistical dispersion of a characteristic such as income or wealth. It measures the concentration of a metric in a population commonly used to assess the extent of a population’s inequality.

• Its value ranges from 0.0 (perfect equality, e.g., where all population members have the same income) to 1.0 (perfect inequality, e.g., where one person earns all the income). A Lorenz curve can represent it, a graphical representation of the distribution of income (or wealth) within a population, where percentiles of the population are plotted against the cumulative income of those at or below that percentile. The more income (wealth) is equally distributed, the closer this curve is to the 45^o line. Conversely, the more unequal the income (wealth) distribution, the more concave it will be.

• However, a low Gini coefficient does not always indicate a low level of inequality. For example, the income Gini for Sweden is low (about 0.30 in 2021), while its wealth Gini is high (about 0.88 in 2021). In Sweden, considerable social welfare benefits are available, home ownership is relatively low, and earned income is heavily taxed. As a result, non-earned income (wealth in various forms) tends to be underestimated in income Gini, although it affects wealth Gini (Cowell, Reference Cowell2011; Cowell et al., Reference Cowell, Karagiannaki and McKnight2017).

Social inequality, as well as disease prevalence or life expectancy, is sometimes expressed in terms of its “relative” and “absolute” aspects. Absolute inequality is calculated by subtracting the value of one group from another. Relative inequality shows the proportional difference between the values of subgroups (Harper & Lynch, Reference Harper, Lynch, Oakes and Kaufman2006).

From information theory, entropy measures the degree of randomness in a set of information. Entropy means disorder and chaos. Several measures of entropy can be applied to inequality, for example:

• Theil’s index (Theil, Reference Theil1967) is a special case of the entropy function, designed for (economic) inequality purposes by measuring how far the observed entropy “distance” the population is from the lowest possible entropy (reached if all individuals have the same income).

• Mean logarithmic deviation index, the average of the log of each group’s average income divided by the group income, is another measure of income inequality that equals zero when all incomes are equal and increases as incomes become more unequal. It is used to decompose population inequality into inequality within and between groups, and is sensitive to changes in the lower tail of the distribution.

Actuaries often compare the mortality of population segments, for example, relative patterns of female-to-male mortality or the ratio of the age-adjusted mortality of subpopulations with different characteristics (e.g., medical conditions, income, or level of education). These comparisons can use a standard set of weights corresponding to a given population at a specified time with the same mix of characteristics, such as sex or smoking status. Similar measures can be applied to disability, morbidity, and longevity. Note that the inequality within a subgroup or population segment can also be significant—this may be gleaned through multivariate analysis or predictive analytical methods.

A model of inequality can be quite complex, involving several variables or a combination of variables. However, a concern with overly complex models is that they can over-fit the data; that is, they may not capture the most important relationships or patterns and may not be helpful for forecasting. A modeller needs to keep in mind the 80/20 Pareto rule (also known as the Pareto Principle (SixSigma.us, 2021)), which suggests that for many studies, roughly 80% of the effects can be identified or resolved by analysing 20% of the causes, or 80% of the modelling effort can be expended to explain 20% of the relevant findings.

Some mortality analyses have examined the extent of population heterogeneity by analysing the distribution and standard deviation of, for example, age at death. In some cases, actuaries can reduce the overall standard deviation in their models by recognising inequalities, additional variables, causes of death, or subgroups.

Using multiple/proxy measures, ranges, distribution shapes, clusters, groupings, time periods, or alternative scenarios can enhance many analyses. For example, while age-standardised mortality of the U.S. Hispanic population has generally been reported to be lower than the corresponding mortality of the U.S. non-Hispanic White population (referred to as the Hispanic mortality paradox), differences in smoking prevalence and mix of native-born and immigrant populations may explain a portion of this difference (Fenelon, Reference Fenelon2013). This example may enhance the understanding of the inequality within or between these groups by simultaneously comparing mortality outcomes while considering ethnic, smoking, and immigration characteristics.

Groups of individuals with one or more similar characteristics (e.g., income, age, sex, location, and country) might be combined with respect to these characteristics to increase the credibility of the data being assessed. In other words, aggregation and grouping may be appropriate or desirable. Predictive analytics can also be useful. Alternative groupings can be based on different characteristics. In any case, care is needed since the use of alternative groupings may affect the findings.

Illustrating possible misleading impacts of aggregation, in the first half of the twentieth century, different levels of inequality would have been found in certain European countries if the population from their colonies (with free movement of people) were combined with that of the colonising countries. Further, the range of average income between countries can be quite large, which can complicate comparisons both between and within multinational blocks, such as the Organization for Economic Co-operation and Development (OECD), the European Union (EU), and the Association of Southeast Asian Nations (ASEAN).

2.3. Drivers of Inequality

There are multiple drivers of inequality, the mix and importance of which can differ by country, region, population segment, and time. Their interactions can result in high and unstable correlations, as well as partial and reverse causations, which may change over time. This complexity and interactions can contribute to difficulty in determining the primary driver(s) and relations. For example, the causal relationships between income inequality, mortality, and self-rated health (Shimonovich et al., Reference Shimonovich, Pearce, Thomson, McCartney and Katikireddi2022) suggest the use of more complex indicators and the aggregation of several individual factors, which can be identified through multivariate analysis or predictive analytics.

The structure of the economy and the labour market can significantly impact inequality. Technological advancements enable shifts in economic status and job roles, propelling society through stages of agriculture, industry, service, and information. Each development stage alters the nature of employment, leading to shifts in the required skill mix and increased unemployment when mismatches occur. For example, the transition from agricultural to industrial jobs usually decreases the number of rural jobs and increases the number of generally higher-paid middle-class jobs, may result in a decrease in inequality.

The shift to service and information-based jobs in many countries has had several consequences, including an initial increase in lower-skilled jobs, sometimes followed by a decrease as these jobs are taken on by robotics and other technologies, and simultaneously fewer, but higher-skilled and compensated jobs. Similar developments are expected from the recent introduction of artificial intelligence, though it is too early to estimate its impact. These changes have, overall, increased inequality within the labour force.

Christine Lagarde, then head of the International Monetary Fund, indicated in 2015 that IMF research showed that less inequality is associated with greater macroeconomic stability and sustainable growth; if the income share of the poor and middle class is increased by 1 percentage point, GDP growth increases by as much as 0.38 percentage points over five years (Lagarde, Reference Lagarde2015).

Considering inequality within countries, Milanovic (Reference Milanovic2016) pointed out that there have been two basic views regarding why inequality decreased in the twentieth century:

• Structural, promoted by Kuznets (Reference Kuznets1955), including

  1. ○ economic forces, including urbanisation and industrialisation

  2. ○ expanded educational attainment

  3. ○ other public or community services and support

  4. ○ population aging increases demand for social retirement security and public healthcare, which can result in greater taxation or lower quality services

  5. ○ widespread subsidies and social assistance/welfare/care to help overcome the unaffordability of essential services

  6. ○ social cohesion that helps a society in times of war or depression.

• Political, espoused by Piketty (Reference Piketty and Goldhammer2013), includes wars (especially in Europe) requiring greater taxation, asset depletion in economic depressions (sharply reducing asset values and wealth), and pro-labour legislation.

The Kuznets (Reference Kuznets1955) hypothesis suggests that, on a macro level, as countries industrialise and incomes increase, their inequality will initially rise (due to the increasing productivity associated with more heterogeneous occupations) and then decrease (once traditional workers become a minority of the working population).

Milanovic (Reference Milanovic2016a and Reference Milanovic2016b) suggested that there have been two Kuznets waves, with the first occurring between about 1850 and 1912 and a second upswing since the 1980s, reflecting rapid technological changes in many higher-income countries. This transition has led to a shift from manufacturing jobs to those in service, information, and communications. Economic resources and consequential wealth have been increasingly concentrated in the hands of the owners of capital, while labour has tended, on average, to become relatively poorer, thereby increasing inequality.

Milanovic (Reference Milanovic2023) categorised global inequality into three eras:

1. (1) An increase between the industrial revolution in the early 19^th century to the middle of the 20^th century, driven by widening gaps between and within countries, especially between the industrialised world and the rest

2. (2) A further increase to the end of the century, characterised by a reduction in inequality within countries, which peaked around 1988 with the rise of a middle class and a welfare state in many countries

3. (3) Since around 2000, there has been a reduction in inequalities between countries due mainly to the economic rise of China, from a Gini coefficient of 0.694 in 1988 to 0.601 in 2018. Milanovic observed that historically, people of lower income in the rich world were better off than many in the rest of the world. In any case, vast income and wealth differences can hinder both economic progress and democracy.

At a macro-level, Milanovic (Reference Milanovic2016) suggested five factors that work toward further increases in inequality in the United States, which are likely to also apply to many other countries:

1. 1. Continued substitution of labour with capital, recently through the use of technologies such as artificial intelligence and robotics

2. 2. Increasing concentration of capital income

3. 3. Labour income and capital income increasingly captured by the same individuals

4. 4. Marriage/partnerships of highly skilled or knowledgeable individuals

5. 5. Concentration of income/wealth of individuals, reinforcing the political and economic power.

According to Bourguignon (Reference Bourguignon2017), globalisation (especially in supply chains) and skill-biased technological progress are responsible for an increase in the share of total income going to capital and for slow growth in the employment and wages of unskilled labour in developed countries (job migration, e.g., janitors, security support, and caregivers). However, in some countries, labour shortages in these areas have driven up immigration and wages and have reduced inequality.

There has been considerable debate regarding whether globalisation or technological advancements are more significant in driving inequality. Although the effect of these factors differs by industry sector and country, Bourguignon believes technological change has been more significant.

In some cases, conditions underlying one set of inequalities lead to a different set in subsequent generations. For example, the rich are often favoured because they receive both higher labour and capital income. As a result, inequality of wealth tends to increase and is usually greater than inequality of income. In contrast, if some drivers of inequality are eliminated, it may lead to a reduction in other drivers of inequality. For example, an emphasis on required or universal education in one generation can lead to greater financial success in the next one through intergenerational income elasticity, but limited education can lead to the opposite situation.

Drivers of inequality include several factors over which an individual has no or limited control, such as date of birth (age), sex at birth, household background, inherited genetic traits, racial/ethnic identity, and chance. In contrast, an individual can have limited or complete control over other relevant factors, including their attained level of education, willingness and ambition to work harder, longer hours, or over a longer period, preparedness to take risks or be innovative, lifestyle choices, and desire to save rather than consume.

Epidemics and pandemics usually increase inequality and can have a considerable effect on the number of people in poverty, disproportionately affecting households with lower incomes and small businesses. As Sanches-Paramo et al. (Reference Sanches-Paramo, Hill, Gerszon, Narayan and Yonzan2021) note, the COVID-19 pandemic “has translated into a sharp increase in global poverty. About 97 million more people are living on less than $1.90 a day because of the pandemic, increasing the global poverty rate from 7.8 to 9.1%; 163 million more are living on less than $5.50 a day.”

Other drivers of inequality include socioeconomic status, including racial and ethnic characteristics; immigrant status and source of immigration; marital status, income/wealth/inheritance, disability/health, educational attainment, employment status, and occupation; family status, structure, and stability; relationship between workers and firms; quality, accessibility, and cost of social, healthcare, and educational services, public policy decisions; and location.

2.4. Inequality and Poverty: Relative and Absolute Concepts

Arguably, poverty can represent a more severe economic and moral problem than inequality. Both, however, can raise questions concerning the fairness of the system people live in, such as whether economic unfairness resulting in low social mobility is more important than income and wealth inequality (i.e., are fair ex-ante opportunities more valuable than ex-post equality outcomes?). According to Andrews and Leigh (Reference Andrews and Leigh2009), “inequality may be more acceptable in a society with a high level of social mobility … [and] moving from rags to riches is harder in more unequal countries”.

Inequality, by definition, is a relative concept. If everyone in a population progresses or regresses consistently according to a specific metric, there is no change in inequality within that population. In contrast, other characteristics such as sex, geographical location, and health status are absolute characteristics, although their meaning or value can vary. There are also hybrid concepts such as poverty, which, although its definition can depend upon an absolute criterion (e.g., inability to attain a minimum standard of living for survival, such as a specified income or wealth for a household of four), also contains a relative aspect (e.g., when compared with the average standard of living in a community). In addition, it can change over time and location; for example, smartphone ownership may be viewed as a luxury in one community and a necessity in another.

Most people compare themselves to their peer group or neighbours. The poor in one country might be considered rich in another, although they may still feel poor. Poverty should thus be viewed in the context of the benchmark used for comparison.

Economic inequality and poverty are, of course, highly interconnected, representing different aspects of the same issue, as those in poverty are typically located in the lowest part of the income or wealth distribution. However, economic inequality can exist without poverty if everyone in a community lives above the defined poverty line. In contrast, if no one is living above this poverty line, everyone is in poverty. For example, the Gini coefficients for Bangladesh and Japan are fairly similar (0.33); however, the average annual income in Bangladesh is about US$2,900, compared to about US$39,000 in Japan.Footnote ⁵

Poverty is usually associated with a lack of social cohesion, poor health, lack of quality housing, inadequate public education, apprenticeship training and opportunity for the individual, expensive childcare, and high crime and rates of incarceration. Governments can take responsibility for helping to overcome these challenges by addressing essential functions such as healthcare, housing, education, public transportation, and ensuring personal safety. People with low incomes are also subject to greater financial instability, food insecurity, and mental stress, partly because of their limited contingency resources.

The doctrine of sufficiency (Frankfurt, Reference Frankfurt1987) suggests that what is important to the individual is not relative economic equality, but rather the extent to which an individual has “enough,” a non-comparative standard of acceptance of a person’s circumstances in the context of personal goals. In other words, everyone should have enough to satisfy their fundamental human needs, rather than comparing themselves to or competing with others. Still, the alternative comparative approach is used quite often, for example, by the OECD.

The comparative approach often quantifies and compares characteristics and the extent of poverty by using a deprivation index that aggregates several factors that, in combination, capture key characteristics of poverty in a population. There are several such measures, such as the United Nations’ Multidimensional Poverty Index (MPI) used by the United Nations Development Programme and the Oxford Poverty and Human Development Initiative 2019. The MPI is composed of ten indicators, weighted by health (1/3 for malnutrition and death of a child under 18 in the last five years), education (1/3 for years of schooling and school underattendance), and standard of living (1/3 for cooking fuel, sanitation, drinking water, electricity, housing, and home assets). Based on surveys conducted between 2011 and 2022 (UNDP, 2024) across 110 countries, about 1.1 billion people (18% of the global population of 6.2 billion) are multidimensionally poor.

There is a large variation in multidimensional poverty. Half of those who are multidimensionally poor are children under age 18, while a third are children under age 10. In addition, there is little or no association between economic inequality (measured by the Gini coefficient) and the corresponding MPI value. In the ten selected countries for which changes over time were analysed, deprivation declined faster among the poorest 40% of the population than among the total population.

Another composite index is the Human Development Index (HDI), which is the geometric mean of three factors:

1. (1) Health as assessed by life expectancy at birth

2. (2) Education as measured by the mean years of schooling for adults aged 25 years and older and expected years of education for children of school-entering age

3. (3) Standard of living as measured by gross national income (GNI) per capita. HDI uses the logarithm of income to reflect the diminishing importance of income with increasing GNI. It can be used to assess the overall socioeconomic status of a population and better understand trends in mortality and markets for products of financial institutions.

Currently, the most commonly used international threshold for extreme poverty is defined by the World Bank (2024) as income of less than US$2.15/day per person in 2025. The vast majority reside in Sub-Saharan Africa and South Asia. The World Bank reports that the global number of people in extreme poverty decreased substantially, from about 36% in 1990 to about 9.3% in 2018, and more than one billion people escaped extreme poverty due mainly to economic growth. Still, as shown in Baler et al. (Reference Baler, Kristensen and Davidson2021), while extreme poverty has declined in most of the world since 2000, it is projected to remain stable in the most fragile countries.

4.1. General Considerations

Policies and programmes aimed at reducing inequality need to recognise its causes and consider their impacts. A good example is the increase in the eligibility age for social security retirement programmes, which can disadvantage individuals in lower socioeconomic groups who need to work longer to maintain a decent standard of living, given their shorter life expectancy.

The desire for equal rights demonstrates some of the issues presented by efforts to reduce inequality. Eliminating gender inequality from actuarial considerations through the imposition of unisex rates for insurance, retirement benefits, and credit can be a two-edged sword. On the one hand, since females generally live longer, using a unisex premium structure – rather than sex-specific – increases the costs of life insurance policies for females relative to those for males, effectively subsidising one sex over the other. In contrast, the opposite situation arises for pension funding.

There is no single or simple solution to all the issues associated with inequality; changes can take a long time, if ever, to be effective. Nevertheless, approaches used have included right-to-work laws, minimum income, gender equality, accessible and quality healthcare, enhanced and effective universal education, and technology training, as well as accommodations for people with disabilities. Such programmes often result in higher economic growth, more well-paying jobs, investment in human capital, better infrastructure, transportation, healthcare, education, and reliable energy. At the same time, they may introduce new inequalities by subsidising some population groups at the cost of others.

Introducing inequality-reducing programmes can have unintended and undesirable consequences, such as the overuse of or excessive funds and subsidies, which can lead to an increase in unnecessary government jobs and projects, possibly supporting corruption and fraud. These programmes can contribute to increased unemployment, as employers seeking higher profits may substitute local employees with immigrants, adopt technological approaches, outsource, or employ other techniques. Programme desirability can depend on who is affected, the incentives and alternative jobs available or provided to them, and the effectiveness of the programmes. Their impact can also depend on whether a tight or loose labour market exists.

An inherent tension can exist between treating everyone equally and addressing everyone’s needs. Since addressing these inequality issues can involve a society’s fabric, culture, beliefs, and history, the proposed solutions are often politically or socially sensitive. Progress often results from public pressure or judicial rulings rather than natural economic forces.

These are complex issues, often involving multidimensional, multifaceted, multicultural, multiclass, and multigenerational considerations that complicate the design, planning, and running of such programmes.

One approach to mitigate undesirable consequences of inequality-reducing programmes is to use a multiple-tier system tailored to address the needs of those in each tier. To illustrate, a total financial support system to reduce risks of employment disruption could include:

• social insurance benefits provided through mandatory participation, possibly including governmental subsidies

• group insurance and retirement programmes supported or sponsored by an employer

• for those without a sponsoring employer, private sector insurance and savings programmes, partly subsidised by the government

• public safety net programmes, possibly involving income- or means-testing, used when the former tiers are deemed inadequate to satisfy the basic needs of some individuals.

Communities, charitable organisations, and voluntary groups often actively support efforts to reduce inequality and its consequences. Examples, sometimes tailored to individual circumstances, include strong support for the poor, disabled, and elderly by religious groups; charity-based and voluntary organisations that support the aged, disabled, and disenfranchised children, women, and families; and microinsurance and microfinance initiatives that involve the community in remunerating losses. In many countries, these efforts replace or supplement efforts otherwise expected of governments, although they may not be able to supply continuous, sustainable help to those in need.

Actuaries provide services to numerous organisations that help manage the adverse aspects of inequality. These organisations include insurance companies, retirement plan sponsors, providers of savings management for retirement and credit, safety net programmes, and government policymakers/regulators. Their products and programmes address the causes and consequences of inequality, as well as the potentially adverse financial impacts due to, and in turn affected by, contingent events or conditions such as death, longevity, ill health, accidents, property damage, and lawsuits. These consequences can have a significant impact on the financial security of individuals and the sustainability of organisations and governments. If not addressed, those involved would suffer, likely further exacerbating inequality.

The primary roles of actuaries regarding these programmes include:

1. (1) assessing their costs and benefits

2. (2) preparing analyses that can help decision-makers to assess whether programmes achieve their objectives and are sustainable in the long term.

This is achieved using actuarial tools such as predictive analytics, stochastic analysis, artificial intelligence, and scenario analysis, which recognise the range and distributions of expected outcomes.

To assess private and public sector products and programmes, actuaries gather information and data concerning relevant demographic, economic, financial, regulatory, and environmental phenomena and their impacts on these products and programmes. Often, they also consider societal and political conditions, events, and trend, as applicable. The data may be either public or proprietary, generated by a wide range of organisations. Scepticism regarding the data may be warranted, as future conditions and trends may not replicate those of the past, and as the data may be represented in different forms and formats and may be incompatible. In addition, data can be of low quality, incomplete, inconsistent, or irrelevant to the issues being assessed. In some cases, the experiences of different population segments can diverge, for example, due to inequality, especially when changes in these population segments are expected. Finally, relationships between data can be associative, rather than causative, with trends that differ between components. As a result, data should be decomposed where practical to distinguish between fluctuations, trends, cycles, and structural changes.

A fundamental characteristic of most population segments is that both their ex-ante and ex-post experiences are of interest. Factors that influence this experience can be either observable or unobservable, the latter possibly being due to inadequate data or information. Indeed, heterogeneity, uncertainty, and time values are key concepts that underpin the application of actuarial science. The disaggregation of data, combined with behavioural science, technologies such as predictive analytics and artificial intelligence, and actuarial science, can help address the causes and outcomes of the inequalities involved, expressed in terms of resulting costs, benefits, and population developments. Actuaries can also develop methods and measures to assess the dynamic effects and uncertainties associated with inequality, using pricing or costing models, risk adjustment, underwriting, reserving, and cost-benefit analysis.

A better understanding of the relevant costs, benefits, and risks of alternative programme designs can inform both private sector and public policy decision-makers who determine the effects and advisability of social security contributions, insurance prices, subsidies, and cross-subsidies.

4.2. Education-Based Solutions

Education, the foundational investment in human capital, is a critical factor in the long-term mitigation and reduction of inequality. The scope and quality of education significantly determine the future course of inequality and can have a profound impact on an individual’s life course, including their health, longevity, income, and wealth. An effective educational system can break the intergenerational transmission of poverty and influence and reduce inequality of opportunity. A generation that experiences greater convergence in educational attainment should ultimately help decrease the inequality of outcomes.

Achievement of higher educational attainment depends upon the availability, accessibility, affordability, quality, and relevance of educational resources at all levels, including preschool, primary, secondary, post-secondary college, university, post-university education, and trade training. Although benefit is often gained from higher educational attainment, there may also be a significant opportunity cost of attending secondary and post-secondary education (e.g., lost immediate wages, slower work advancement, cost of study materials, and reduced help with the family business), as well as other expenses (e.g., accommodations), which have a relatively greater financial impact on those whose families have a lower income. Even where education is free or largely subsidised, it may exacerbate inequalities, as all taxpayers contribute towards the cost of education, while those who attend are generally from better-off families,Footnote ⁹ especially at higher educational levels, where fees can be significant.

Colleges, universities, and public programmes can enhance the accessibility and affordability of post-secondary and trade training programmes through loans, grants (such as scholarships), part-time work (e.g., assistantships and internships), online and virtual courses and webinars, and other support programmes. Other subsidies can include forgiving student loan interest during or after the education period, providing loans at below-market interest rates, and repayment only when income exceeds a minimum threshold. Many of these programmes, which can differ significantly by country, target those from low-income or middle-class families (Ziderman, Reference Ziderman2005).

The size and market for student loans are large in some countries. According to the Education Data InitiativeFootnote ¹⁰ , total U.S. student loan debt has ballooned in recent years, reaching around US$1.8 trillion by about 43 million borrowers; many are afraid that their debt could hinder their financial future. Most of this debt was underwritten through federally provided student loans. Forgiveness remains a persistent political and economic issue.

Financial sustainability (i.e., systems that pay for themselves) is not usually the primary objective of education support and student loan systems, as they represent investments in human capital. For example, in the UK, more than 77% of those taking out student loans will have some, or even all, of the loans paid for by the government because many graduates will not earn enough to repay their loans on a timely basis (The Guardian, 2017). Only those with high incomes will eventually repay the full amount of loans. These systems thus redistribute from those with higher incomes to those with lower incomes.

Nevertheless, sustainability is vital to educational support and loan programmes. An actuarial valuation is a helpful tool to project future loans (e.g., the number of students receiving new loans, current loans, loans in repayment, and loans in default), repayment rates, revenues and costs of such programmes, their impact on future budgets, and in risk management.

For example, in Canada, such assistance is provided through the Canada Student Financial Assistance Program (CSFA Program). Currently, almost 50% of Canadian post-secondary students benefit from student loans and grants provided by the CSFA Program. Since 2000, this programme has been directly financed by the government. As a result, the Office of the Chief Actuary (2023) has a statutory responsibility to conduct triennial actuarial reviews (to provide an assessment of the current and previous twenty-five years’ cost of the CSFA Program). Projections of these costs are important for government budgeting and determining the cost to taxpayers. During the COVID-19 pandemic, the government introduced several relief measures to help the most vulnerable students. The programme actuaries worked closely with policymakers to quantify the financial impacts of these measures.

Another important educational tool for reducing inequality is financial education, both within and outside traditional schools. Long-term savings and investment management are essential for preparing for life cycle risks and retirement. Many people, however, have limited investment knowledge, aptitude, or patience to focus on their longer-term financial issues. They could benefit from relevant financial education.

Financial education programmes should be practical, entertaining, and focus on practical life skills, including financial decision-making and understanding the interaction of such decisions with relevant public and private programmes. Technology and social media can play a crucial role in ensuring the widespread availability of financial education. Regulators can promote such information, as is done by the Israel Capital, Savings and Insurance Authority, which supports such programmes through its website (https://www.boi.org.il – in Hebrew).

Actuaries can contribute to financial education efforts by participating in the development of educational materials for various interested entities, including insurance companies, pension funds, savings programmes, and social security systems.

4.3. Social Security Solutions

Social security can be provided by a system or programme that provides help and security against life cycle risks, such as unemployment, disability, prolonged sickness, aging, retirement, longevity, and death. It is typically provided through a long-term government programme, which can vary between countries and subpopulations in terms of benefits, types of beneficiaries, eligibility rules, quality, access, and financing. These are often enhanced by additional private programmes such as insurance and pension savings, as well as voluntary community organisations.

Social security programmes encompass both social insurance (e.g., providing healthcare, unemployment, and workers’ injury insurance, as well as retirement benefits) and social assistance/welfare/care programmes. Both types can be provided on a universal basis or include limitations and constraints such as means-testing or work/study requirements. Due to the broad scope of these various programmes, we focus on major governmental social security programmes.

Many social security systems provide long-term support, where funds are accumulated or budgeted over the working years of the participants and are paid as benefits when the designated needs arise. The manner in which funds are collected and maintained, benefits paid, and demographics of the affected subpopulations evolve, all affect the sustainability of these systems. Due to these characteristics, the financial soundness of social security systems is affected by the country’s economic and demographic development. In turn, the size of these systems and the affected populations make them a significant part of the economy. Further, by their nature, such systems, along with their scope, characteristics, and beneficiaries, are often at the core of liberal-conservative (or left-right) political discussions.

Social security can reduce inequality through two principal means:

• Providing universal benefits in the event of certain lifecycle risks by implicitly targeting those in social groups who are more likely to be impacted by these risks. For example, since the health status of those with low income may be worse, they may receive health benefits that are a relatively higher proportion of their incomeFootnote ¹¹ .

• Providing benefits and services that explicitly target disadvantaged groups. Such benefits may be tailored or targeted to those who would benefit most (e.g., through means-testingFootnote ¹² or minimum benefits).

Financing can be designed to reduce inequalities. For example, earnings up to a specific limit can be excluded from contribution requirements, or progressive general taxation can be used to finance benefits.

4.4. Healthcare Solutions

Issues associated with inequality relating to the provision of healthcare are similar to those of social security programmes and may depend on whether the country’s healthcare is provided through the public sector, private sector, individuals, or a hybrid approach. Additional issues specific to healthcare can arise, whether provided in facilities such as hospitals or a home setting. These issues (sometimes referred to as health inequities) are usually assessed in terms of access/availability, quality, timeliness of treatment, cost, and choice of provider(s). Affordability is a function of both access/availability and cost. It is difficult to optimise all these highly interdependent factors.

Public policy can help reduce health inequality. International health goals have been established by the WHO and the United Nations, first in 1985, and continued through the UN’s 2015 Sustainable Development GoalsFootnote ¹⁵ to be reached in 2030. These have met with uneven success but have focused attention on crucial health and inequality-related issues. Several European countries, including the U.K., Finland, and Lithuania, have introduced national targets to address socioeconomic inequalities (Europe Sustainable Development Report, 2022). The impact of policy on inequality, both in theoretical and empirical frameworks, shows that policy can affirmatively contribute to population health (Rose, Reference Rose1992). Providing better health conditions for lower socioeconomic groups is essential for reducing inequality in mortality and morbidity (De Gelder et al., Reference De Gelder, Menvielle, Costa, Kovacs, Martikainen, Strand and Mackenbach2017).

Access to and affordability of care are often based upon eligibility for (full or partial) reimbursement and the access/availability of appropriate healthcare infrastructure. Even in fully public programmes, the convenience of obtaining access to healthcare outlets and the accessibility of their services can differ by population segment, location, and income. A weak healthcare infrastructure can create healthcare inequalities. Urban and rural settings present additional challenges. In cities, higher-quality facilities and resources may only be accessible or affordable to those with higher incomes. In contrast, in rural areas, high-quality care may not be accessible to everyone.

Providing healthcare is becoming increasingly expensive almost everywhere, even where it is available, due to the costs of trained staff, diagnostic and treatment equipment, medical advancements, and pharmaceuticals. Cost, in turn, affects affordability, quality, and availability to different population segments. Inevitably, even in fully public programmes, evolution to multiple levels of care can occur, where those with higher incomes will be able to purchase some form of private insurance or other means of obtaining the highest possible level of care.

The primary purpose of providing healthcare is to add healthy years to life, not just more years. Thus, even though healthcare services are usually provided once an adverse condition has manifested itself, it may be more important to act in a preventative manner, including a nutritious diet, adequate physical activity, and avoidance of behaviours such as smoking. These healthy behaviours and habits are particularly difficult to maintain for those who cannot afford them. Better health outcomes will occur if these opportunities are incentivized for all or at least made convenient and affordable.

Healthcare inequality also stands in the way of accessing quality and cost-effective long-term care and services for those who are disabled or older. The ability of many older people and people with disabilities to afford to meet their needs with respect to performing their activities of daily living can be dramatically impaired. People with low incomes are more likely to need such services, while simultaneously lacking adequate resources due to a lower propensity and ability to save, as well as an inability to afford private insurance. Thus, public approaches may be needed, at least on a supplemental basis.

Actuaries can provide objective advice regarding these key healthcare factors, including interlinkages, cost estimates, unintended consequences, and input into many of the design decisions needed. Thousands of actuaries worldwide are involved in costing and analysing the demographics and distribution of healthcare costs. It is important that they consider the needs of individuals with low income or limited access to care in their evaluation and design of these systems and associated products.

4.5. Other Solutions

Many private, commercial, and voluntary insurance and savings programmes can mitigate the vulnerabilities and consequences of inequality. Since they are the subject of many papers and books, they will not be dealt with here other than to indicate that, in many cases, they can be provided through financial institutions, governments, communities, and joint private/public programmes. In some cases, solutions to address inequality can be sought through public and private partnerships or the operation of private sector products, developed and maintained under the oversight of applicable regulators.

Although adverse outcomes or consequences of certain conditions or events (e.g., death, longevity, disability, ill health, property damage, poor credit, or lawsuits) can be addressed by private or social insurance or accumulated individual wealth, many who most need such protection cannot afford it. In these cases, public-sector approaches may be desirable.

Contributions by actuaries to these solutions can include such functions as:

• Providing input to and analysis of the development of default options tailored for low- and middle-income employees, participants, and their dependents and survivors. Actuarial experience with defined benefit pension plans can provide guidance regarding setting targets aimed at reducing inequality, as well as developing algorithms to determine contribution rates and benefits.

• Assessing the effectiveness of intermediaries and providers in terms of cost and quality of services provided.

• Helping design effective and efficient microinsurance programmes.

• Helping devise programmes to convert home equity into lifetime income solutions, including through the use of lifetime annuities.

• Assisting decision-makers to understand the uncertainties involved under a range of alternative scenarios.

• Designing risk pools that are actuarially sound, practical, and affordable for lower and middle-income individuals.