Topic 54 of 64· Wellbeing, equity and societal indicators

Social capital metrics

Social capital metrics quantify the networks, trust, and civic participation that let communities and institutions function efficiently — the "connective tissue" that has no line on any balance sheet but that visibly collapses cost and friction when present, and visibly raises it when absent. The modern framing comes from Robert Putnam's "Bowling Alone" (2000), which distinguished bonding capital (ties within a similar group) from bridging capital (ties across different groups); the UK's Office for National Statistics has since built a standing indicator set to track it nationally.

Why it matters

Putnam's central empirical claim — documented through declining US civic-association membership, church attendance, and union participation across the late twentieth century — was that social capital predicts outcomes conventional economics struggles to explain: lower crime, better child welfare, more effective local government, faster economic recovery after shocks. Bonding capital (strong ties within a tight-knit group) is good for mutual support but can calcify into insularity; bridging capital (weaker ties across different groups) is what typically correlates with access to opportunity, information flow, and institutional trust. The ONS took this seriously enough to build a national indicator framework — its "Social Capital in the UK" series (https://www.ons.gov.uk/peoplepopulationandcommunity/wellbeing/bulletins/socialcapitalintheuk/latest) tracks four pillars: personal relationships, social network support, civic engagement, and trust and cooperative norms, each built from established survey questions (Community Life Survey, Understanding Society). For public-sector digital services, social capital is doubly relevant: it is both an outcome some programmes are trying to build (community resilience funding, social prescribing) and an input that determines how well a service will actually be adopted — a service rolled into a high-trust, well-networked community will spread through word of mouth in a way an identical service in a low-trust area will not.

The maths

ONS four-pillar framework (indicators, illustrative):

Personal relationships:      % with someone to rely on in a crisis
Social network support:      % who could borrow money from friends/family if needed
Civic engagement:            % who volunteered or took civic action in past 12 months
Trust and cooperative norms: % agreeing "most people can be trusted"

No single ONS composite score is published — the pillars are reported
separately, deliberately, because aggregating them into one index would
hide which specific pillar is weak.

Putnam's bonding/bridging split (framework, not a formula):
  bonding capital ≈ density of ties within a homogeneous group
  bridging capital ≈ frequency/strength of ties across distinct groups

Worked example

Neighbourhood social capital snapshot: a Community Life Survey-style poll of a local area finds 78% have someone to rely on in a crisis (personal relationships), 61% could borrow money if needed (network support), 24% volunteered in the past year (civic engagement), and 41% agree "most people can be trusted" (trust and norms) — versus national averages of roughly 85%, 70%, 30%, and 45% respectively (illustrative, calibrate against the current ONS bulletin). The area under-indexes on every pillar but most sharply on trust (41% vs. 45% national, a 4-point gap) and civic engagement (24% vs. 30%, a 6-point gap) — flagging civic engagement, not trust, as the largest relative deficit worth targeted investment (a community grants programme, say) rather than a generic "build trust" initiative.

Bonding vs. bridging, service design: a jobs programme in a tight-knit community finds referrals travel fast within the community (high bonding capital: word spreads within days) but the programme struggles to reach residents outside that network (low bridging capital: uptake outside the core community is near zero after months). The fix implied is not "more marketing" but deliberately building bridging ties — partnering with organisations that sit outside the existing network, since bonding capital alone cannot solve a bridging-capital problem.

Software engineering connection

  • Digital platforms that route mutual aid, volunteering, or community grants (a "local connector" service, for instance) are literally building bridging-capital infrastructure; their success metric should be network diversity of connections made, not just transaction count — see government-as-a-platform for the broader pattern of infrastructure that others build value on top of.
  • Where a programme's theory of change explicitly targets social capital as an outcome (a community resilience fund, a social prescribing service), its theory of change and logic model should name the specific pillar (trust, civic engagement, network support) it expects to move, rather than an undifferentiated "build community" outcome that can't be measured against the ONS baseline.
  • Social capital indicators are a useful equity lens alongside the Index of Multiple Deprivation: an area can be income-deprived but socially rich, or vice versa, and the two point to very different interventions.

Pitfalls

  • Collapsing the four ONS pillars into one composite score — ONS deliberately does not do this; a single number hides which specific pillar is driving a low reading, and averaging masks a community that is high-trust but civically disengaged versus one that is the reverse.
  • Assuming social capital is always good — dense bonding capital in an insular group can actively resist outside institutions (including government services); Putnam's own analysis treats bonding and bridging as different goods with different, sometimes conflicting, effects.
  • Using survey-based social capital measures as a real-time operational metric — the underlying surveys (Community Life Survey, Understanding Society) run annually or less often; treat social capital data as a slow-moving contextual indicator, not something a service dashboard can update weekly.

Sources