By Ugo Gentilini, Senior Economist, Margaret Grosh, Senior Advisor, and Michal Rutkowski, Senior Director, Social Protection and Jobs Global Practice, The World Bank

---

Check out the upcoming international conference Together to achieve Universal Social Protection by 2030 for more on this topic

---

The notion that social protection is “universal” rests on two elements, namely that “everyone” is “covered.”

In many cases, the debate revolves around the “everyone” aspect – that is, the rationale and modalities to cover all members of society and not just some. Yet, this assumes clarity on the meaning of “coverage.” This is a big assumption.

In health insurance, for example, the goal is to provide coverage to all, so that in the event people fall sick, they get health services. For contributory pensions, unemployment or disability insurance programmes, coverage is used in an analogous way.

In most periods, people covered by such insurances will benefit from a guarantee or a promise of help when needed, but not necessarily from a payout. In pensions, people are covered for many years before they receive a payment; and many may never be unemployed and hence receive a payout for such a shock.

For social assistance, instead, coverage is often interpreted as receiving an actual transfer. This is quite a difference and a critical issue to clarify.

Such a difference in definition has implications for universal social protection in three ways.

First, if a country has a guaranteed minimum income (GMI) programme, which provides cash when incomes fall below a threshold, the social insurance interpretation would be that – like in the case of unemployment – everyone is covered independently of the event occurring (i.e., income falling). Thus, coverage would be many times greater than the actual benefit roster. A GMI is universal in insurance terms, but it is targeted from a social assistance standpoint.

Second, part of the social assistance community refers to programmes for the elderly or children as universal. But what is really meant is that eligibility for such programmes entail no requirement other than age. Hence, a “universal social pension”, for example, is again universal from an insurance perspective, but it is targeted from a social assistance viewpoint.

Third, coverage can refer to a gross or net benefit. Take the quintessential universal programme defined in social assistance terms – i.e., universal basic income (UBI). While everyone receives a payout, those benefits must be financed. Net payments may be lower than gross payments, and some people will pay more than the benefits they receive as an UBI payout. Thus, some people are nominally beneficiaries, but de facto financiers. The programme, therefore, stops being universal in social assistance terms.

So where do we go from here? We offer four considerations.

First, targeting and universality are often portrayed as opposing forces. This does not have to be the case. People at the bottom of the welfare distribution require special and urgent support, and the overall benefit structure of universal social protection systems should account for that. Expressed within a vision of universality based on inclusive needs and the gentle ‘tapering’ of interventions, targeting means prioritisation, not exclusion. This is what progressive universalism, espoused by our recent World Development Report 2019 on the changing nature of work, is about.

Second, we need to weigh the trade-offs between coverage, adequacy, affordability and incentives to work. Choices around these interconnected parameters are hard to balance in any country, but especially in lower-income contexts. Those tensions could be relaxed, but not eliminated, by allocating more resources to social assistance compared to current low levels of spending (about 1.5% of GDP), raising revenues through more effective and fair taxation, and directing more resources where needed the most.

Third, such considerations should be interpreted at the system-level. Most current discussions centre on important but micro-issues – e.g., how to define and measure needs, what targeting method to use, which programme to select. But both assistance and insurance approaches encompass a wide range of interventions, and hence a multitude of specific choices. We need to look at the entire set of benefits and taxes across the welfare distribution to understand how the overall system affects the population.

Fourth, coverage is only a means to an end. We should be driven by clear, relevant and measurable outcomes and leverage coverage to achieve them. The education sector, for instance, is increasingly framing its goals around children’s learning performance, and not coverage of pupils. We should think of ways to translate social protection coverage into similarly compelling metrics.

Ultimately, for universal social protection (USP) to be achieved by 2030, we need a steep acceleration in progress – including via key partnerships around USP2030 and SDG 1.3 and through clear thinking about concepts and trade-offs.