Only 40% Trust Their Government — The Social-Capital Ledger

Trust is the least visible line on a national balance sheet and one of the most consequential. The Nobel economist Kenneth Arrow argued in 1972 that virtually every commercial transaction contains an element of trust, and that much of the world’s economic backwardness could be explained by the lack of mutual confidence [53]. Half a century later, the measurement has caught up with the intuition. The OECD’s Survey on Drivers of Trust in Public Institutions, whose third round was published in June 2026 from fieldwork in late 2025, found that 40% of respondents across 33 member countries and 5 accession candidates had high or moderately high trust in their national government, while 43% had low or no trust ✓ Established [1]. The previous round, fielded in late 2023 among about 60,000 people in 30 countries, produced almost the same split: 39% against 44% [5]. The OECD summarised the 2026 result in five words — trust steadies, but people still do not feel heard [1]. Stability at this level is not reassurance. It means that in the average rich democracy, more citizens distrust their government than trust it, and that has now held across three survey rounds.

The averages hide the real story, which is dispersion. In the 2023 round, trust in national government ranged from 61.9% in Switzerland and 47.6% in Norway to 30.4% in Chile, 28.7% in Latvia and 26.7% in the United Kingdom, against an OECD average of 39.3% ✓ Established [6]. The OECD’s Government at a Glance 2025 reported that in a third of countries surveyed trust in national government sat below 35%, and that across institutions the courts were trusted by 54% of people, the civil service and local government by 45%, and parliaments by only 37% [7]. The United States, which does not appear in the OECD tables, is measured by Pew Research Center: in 2025 only 17% of Americans said they trusted the federal government to do what is right just about always or most of the time, down from 22% in 2024 and from 77% in 1964 ✓ Established [10]. Gallup’s World Poll found that in 2023 confidence in the national government in both the United States and the United Kingdom stood at 30%, tied for lowest in the G7 — two countries that had been highest in the group when the poll began [11].

Trust in government and trust in other people are related but distinct quantities, and confusing them distorts the analysis. Generalised social trust — the share of people answering that most people can be trusted in the World Values Survey and its European counterpart — shows the starkest international spread. Our World in Data, compiling those surveys, reports that more than 60% of respondents in Norway and Sweden say most people can be trusted, while in Colombia, Brazil and Peru fewer than 10% do ✓ Established [12]. It also documents a strong positive relationship: countries with higher self-reported trust are countries with higher economic activity [12]. Institutional trust is more volatile. Norway’s trust in national government fell from 64% in 2021 to 48% in 2023 before recovering to 57% in 2025 [2]. Denmark, the country that has topped Transparency International’s Corruption Perceptions Index for eight consecutive years with a 2025 score of 89, records trust in its national government of 44% — almost exactly the OECD average [13] [6]. The Nordic advantage is overwhelming on social trust and modest on trust in politicians.

Behavioural measures broadly confirm that the survey numbers capture something real. In a field experiment published in Science in 2019, researchers turned in 17,303 apparently lost wallets at banks, hotels, police stations and public offices in 355 cities across 40 countries and recorded whether the recipients contacted the owner [52]. Wallets were returned 40% of the time when they held no money and 51% of the time when they held money — a finding that contradicted the predictions of most economists surveyed, and that showed large differences between countries ◈ Strong Evidence [52]. A separate line of laboratory research, beginning with Glaeser, Laibson, Scheinkman and Soutter in 2000, found that the standard survey question predicts how trustworthy a person behaves more reliably than how trusting they behave [31]. That matters for interpretation. When a population reports that most people can be trusted, it is partly reporting on the behaviour it observes around it. Survey trust is, to a significant degree, a measurement of the surrounding society’s trustworthiness.

The European data tell a similar story of stable averages and sharp national swings. The Spring 2025 Standard Eurobarometer found trust in the European Union at 52%, its highest level since 2007, while trust in national governments across the bloc stood at 36% and in national parliaments at 37% ✓ Established [14]. A year earlier, the Spring 2024 wave had recorded falls in national-government trust of 13 points in the Netherlands and 11 points in Finland, and only 25% trust among Europeans who described themselves as working class [56]. Commercial surveys use different methods but point in the same direction. The 2026 Edelman Trust Barometer, based on 33,938 online interviews in 28 countries in late 2025, found that 70% of respondents were unwilling or hesitant to trust people with different values, rising to 90% in Japan and 81% in Germany [8]. Edelman’s chief executive, Richard Edelman, described the trajectory as a descent from fear to polarisation to grievance and now to insularity [8] [9].

These numbers frame the question this report addresses. Trust is not a fixed national character, since several countries have gained or lost 15 to 20 points in a few years. Nor is it random, since the same countries cluster at the top and bottom of every measure decade after decade. The central analytical task is to separate the slow-moving components of trust — the historical and cultural inheritance that economists have traced back centuries — from the fast-moving components that respond to how institutions actually behave ◈ Strong Evidence [21] [25]. The policy question follows directly. If trust is mostly inherited, low-trust societies are trapped and reform must wait for generations. If it is substantially produced by institutions, it can be rebuilt — though not by communication campaigns, only by changes in the conduct of the institutions themselves. The weight of evidence reviewed in the sections that follow points towards the second answer, with important qualifications.

The OECD trust framework, refined over three survey rounds, separates the drivers of trust into competence — whether institutions are responsive and reliable — and values — whether they are open, fair and act with integrity [5]. Its regression analysis, repeated across countries, consistently finds that the values side carries more weight than headline economic performance. In the 2024 results, 69% of people who felt they had a say in what government does reported high or moderately high trust in national government, against 22% of those who felt they did not ✓ Established [5]. In the 2026 results the gap between the two groups reached 47 points across the OECD, and 48 points by the calculation of the Consultation Institute, up from 44 points in 2021 [1] [4]. Strikingly, 68% of respondents believed that voting gives them some influence over government, but only 31% believed that people like them have a say in what government does [1]. Elections confer formal voice. Most citizens do not experience that as being heard.

Fairness and integrity form the second cluster. The OECD survey asks whether respondents expect a public employee to refuse a bribe, whether they expect a high-level official to refuse a well-paid private sector job in exchange for a political favour, and whether they expect to be treated equally regardless of income, gender or background [5]. Expectations of integrity and fair treatment are among the strongest predictors of trust in every round ◈ Strong Evidence [5] [7]. This finding aligns with two decades of academic work. Bo Rothstein and Jan Teorell defined quality of government as impartiality in the exercise of public authority — the principle that officials do not take into consideration anything about a citizen that is not stipulated in advance by law or policy [18]. Impartiality, in this framework, is the institutional property that generates trust. Rothstein and Eric Uslaner traced a causal chain in which economic inequality fosters corruption, corruption depresses trust, and low trust sustains the inequality — what they called an inequality trap [17].

Reliability is the third pillar, and it is where the pandemic offered a natural test. The Lancet study by Thomas Bollyky and colleagues, covering 177 countries between January 2020 and September 2021, found that measures of trust in government and interpersonal trust were associated with lower infection rates and higher vaccine coverage, while standard pandemic-preparedness indices were not ◈ Strong Evidence [15]. Bollyky told the Associated Press that the team found no links between COVID outcomes and democracy, populism, government effectiveness, universal health care, pandemic preparedness metrics, economic inequality or trust in science [16]. As reported by the Associated Press, the authors estimated that if every country had the interpersonal trust of Denmark, global infections would have been about 40% lower — roughly 440 million fewer — over the 21-month period, and 13% lower with Danish levels of trust in government [16]. In Denmark, the HOPE project led by Michael Bang Petersen, which conducted more than 400,000 interviews in 8 democracies, found that about 90% of Danes trusted the health system and that trust in the authorities’ handling of the pandemic was the best predictor of vaccine acceptance [42].

Economic shocks damage trust, but mostly through the channel of perceived unfairness. Yann Algan, Sergei Guriev, Elias Papaioannou and Evgenia Passari, analysing European regions after the 2008 crisis, found that unemployment across the European Union rose from 7% in 2007 to 11% in 2013, and that a 5-point rise in regional unemployment was associated with a 1.5-point fall in political trust ◈ Strong Evidence [30]. Using the pre-crisis share of construction as an instrument, they estimated that a 1-point rise in unemployment raised the anti-establishment vote by 2 to 3.9 percentage points [30]. The effect on interpersonal trust was far smaller than the effect on trust in political institutions, with a standardised coefficient of about -0.065 against about -0.15 [30]. Their summary was blunt: crisis-driven economic insecurity is a substantial driver of populism and political distrust [30]. The crisis did not make Europeans distrust each other. It made them distrust the institutions they held responsible.

Distribution matters as much as the level of hardship. The 2026 OECD results found trust lower among women, younger adults and people with less education, and noted that the gap by education had widened substantially since 2021 [1]. In the United Kingdom the pattern has reversed: by 2025 graduates trusted government less than non-graduates, with an education gap of 31 points, while the British voice gap was only 8 points — a sign that distrust there has spread beyond the groups that traditionally feel excluded [4]. The 2025 Edelman Trust Barometer recorded that 61% of respondents across 28 countries held a moderate or high sense of grievance against business, government and the wealthy, and that the share fearing government leaders deliberately mislead them had risen from 58% in 2021 to 69% [9]. Grievance, in Edelman’s framing, is the bridge between economic insecurity and political distrust ◈ Strong Evidence [9].

Taken together, the drivers point to a consistent hierarchy. Trust follows the experience of being treated fairly and heard by institutions that behave predictably, and it falls when citizens perceive that rules bend for the connected, that hardship is distributed unequally, or that their participation changes nothing. Economic growth helps at the margin, but the OECD data show that countries with similar growth rates differ widely in trust, and that the variables most strongly associated with trust are procedural rather than material ◈ Strong Evidence [5] [7]. The British philosopher Onora O’Neill has argued that the aim should not be more trust as such but more trust in the trustworthy and less in the untrustworthy [33]. The survey evidence fits that framing. Where institutions are impartial and responsive, trust is high and largely deserved. Where they are not, low trust is a rational response rather than a cultural pathology.

The Nordic pattern is the reference point for every theory of trust. Kenneth Newton and Jan Delhey, analysing 60 nations in 2005, found that high-trust societies share five characteristics — ethnic homogeneity, a Protestant tradition, good government, wealth and income equality — and that these traits cluster most strongly in the Nordic countries ◈ Strong Evidence [29]. Importantly, the same associations held more weakly across the other 55 nations, suggesting that the Nordic countries are an extreme case of a global pattern rather than an exception to it [29]. Transparency International’s 2025 index places Denmark first for the eighth consecutive year with a score of 89 and Finland second with 88, while the global average fell to 42 and 122 of 182 countries scored below 50 [13]. Only 5 countries scored above 80, down from 12 a decade earlier [13]. The Nordic countries thus combine the world’s highest social trust with its lowest perceived corruption — a coincidence that most researchers now treat as cause and effect.

The key question is the direction of causation. If Nordic trust were simply a cultural trait, it would produce good institutions; if institutions produce trust, the arrow runs the other way. Kim Mannemar Sønderskov and Peter Thisted Dinesen addressed this directly using two Danish panel surveys covering 18 years, tracking the same individuals over time with fixed-effects and cross-lagged models [25]. They found strong evidence that trust in state institutions causes social trust, and only limited evidence for the reverse direction ◈ Strong Evidence [25]. Dinesen’s earlier study of immigrants to Western Europe found that both the trust culture of the origin country and the institutional quality of the destination country were highly significant predictors of an immigrant’s trust — evidence that trust has cultural roots but responds to the institutional environment in which a person lives [26]. People who move from a low-trust to a high-trust society become more trusting. That is difficult to explain if trust is a fixed inheritance.

The mechanism most associated with Nordic trust is the universal welfare state. Staffan Kumlin and Bo Rothstein, using Swedish survey data, found that contact with means-tested, selective welfare programmes was associated with lower social trust, while contact with universal programmes was not ◈ Strong Evidence [20]. The logic is procedural. Means-tested programmes require caseworkers to make discretionary judgements about who deserves help, which invites suspicion of cheating by claimants and of arbitrariness by officials. Universal programmes — child benefits, health care and education available to all on the same terms — remove that discretion and with it a major source of mutual suspicion [20] [17]. Rothstein and Uslaner argued that universal programmes are the main route out of the inequality trap, because they reduce inequality while treating citizens as equals [17]. The Nordic welfare state is therefore not only a redistributive system. It is a trust-generating machine, because almost every citizen experiences the state as impartial in routine encounters.

The historical record supports the institutional account, with caveats. Alina Mungiu-Pippidi’s study of how societies develop control of corruption traces Danish rule-of-law building to the absolute monarchy established in 1660, and finds that most of today’s least-corrupt countries made their transition during the 19th century ◈ Strong Evidence [41]. Rothstein has argued that Sweden moved from a corrupt, patronage-based administration to an impartial one through a concentrated burst of reform in the mid-19th century — a big bang [19]. Anders Sundell, in a 2016 study for the Swedish government’s Expert Group for Aid Studies, disputes the speed of that transition, finding that Swedish change was better characterised as gradual and incremental, with informal fees to officials persisting until salaries rose and income tax was introduced [40]. Both accounts agree on the essential point: Nordic trust was built, over generations, by the deliberate construction of impartial administration. It was not an original condition of Scandinavian society.

The Nordic balance sheet is not immune to depreciation. Norway’s trust in national government fell 16 points between 2021 and 2023, from 64% to 48%, before recovering to 57% in 2025 ✓ Established [2]. Finland recorded 47% in the 2023 OECD round, and the Spring 2024 Eurobarometer then registered an 11-point fall in Finnish trust in national government [6] [56]. Denmark’s OECD figure declined from 49% in 2021 to 44% in both 2023 and 2025 [60]. These movements are large in absolute terms, but they occur around a high baseline of social trust that has barely moved: Nordic citizens can lose confidence in a particular government without losing confidence in each other or in the courts and civil service [12] [7]. That distinction — between trust in incumbents and trust in the system — is what separates a temporary dip from a structural crisis. The Nordic systems absorb political disappointment because the administrative layer below the politicians remains impartial.

Two complications deserve attention. First, the Nordic countries are not the only high-trust democracies in institutional terms: Switzerland, with 61.9% trust in national government in 2023, and Luxembourg, at 56%, outrank all four Nordic states on that measure [6] [7]. Switzerland’s direct democracy, with frequent referendums, offers a different route to the same outcome — voice rather than universalism. Second, a meta-analysis of 1,001 estimates from 87 studies by Dinesen, Merlin Schaeffer and Sønderskov found a statistically significant negative association between ethnic diversity and social trust, strongest for trust in neighbours and when diversity is measured locally ⚖ Contested [28]. That finding is often cited to argue that Nordic trust depended on homogeneity and will erode with immigration. The same body of research, however, shows that the effect is modest relative to institutional quality, and that immigrants in high-trust countries become more trusting over time [26] [28].

Bo Rothstein calls the low-trust condition a social trap: a situation in which people fail to cooperate for mutual gain because they do not trust others to cooperate too ◈ Strong Evidence [19]. If most officials take bribes, the honest official gains nothing by refusing and the honest citizen loses by not paying. If most taxpayers evade, compliance is a sucker’s game. Each actor’s distrust is individually rational and collectively ruinous, and because it is rational it does not dissolve when the original cause disappears [19]. Edward Banfield’s ethnography of a village in southern Italy in 1955 described the same equilibrium as the inability of villagers to act together for their common good or for any end transcending the immediate, material interest of the nuclear family [59]. Robert Putnam’s study of the Italian regional governments created in 1970 found that their performance two decades later tracked civic traditions reaching back centuries: northern regions with dense civic associations governed well, southern regions without them governed poorly [23].

Economists have since measured how long such traps persist. Luigi Guiso, Paola Sapienza and Luigi Zingales found that Italian cities that had been self-governing communes in the Middle Ages have more civic capital today, and that the effect increases with the length and intensity of independence ◈ Strong Evidence [24]. Their estimate is that at least 50% of the North-South gap in Italian social capital is due to the lack of a free city-state experience in the South [24]. Nathan Nunn and Leonard Wantchekon, using the 2005 Afrobarometer survey of 21,702 respondents in 17 countries, found that individuals from ethnic groups more heavily raided during the Atlantic and Indian Ocean slave trades express less trust today in relatives, neighbours, members of their own ethnic group and local government [22]. Using historical distance from the coast as an instrument, they concluded that most of the effect operates through factors internal to the individual — cultural norms, beliefs and values transmitted across generations — rather than solely through the quality of present-day institutions [22].

The economic cost of these traps is large. Yann Algan and Pierre Cahuc measured the trust that descendants of immigrants to the United States inherited from their ancestors’ countries of origin, using the General Social Survey from 1977 to 2004 across 24 origin countries, and used that inherited trust to estimate the effect of trust on growth in the origin countries themselves ◈ Strong Evidence [21]. They found that inherited trust explained more than a third of cross-country differences in income per head. Their counterfactual estimates were striking: income per head in 2000 would have been 17% higher in Italy, 11% higher in France, 7% higher in Germany and 6% higher in the United Kingdom if those countries had the inherited trust of Sweden — and 69% higher in Russia, 59% higher in Mexico and 546% higher in Africa [21]. These figures carry uncertainty, and a 2017 replication questioned parts of the identification strategy [57]. But even heavily discounted, they imply that trust is among the largest determinants of national prosperity.

Latin America illustrates the trap at regional scale. Our World in Data reports that fewer than 10% of respondents in Colombia, Brazil and Peru say most people can be trusted, among the lowest figures recorded anywhere ✓ Established [12]. Latinobarómetro’s 2023 report found that support for democracy across the region had fallen to 48%, from 63% in 2010, that only 28% were satisfied with how democracy works, and that 17% would accept an authoritarian government in some circumstances [47]. Support recovered to 52% in the 2024 survey [48]. In the Middle East, Lebanon offers an extreme case: Arab Barometer data analysed by the Economic Research Forum show that by 2024 fewer than 10% of Lebanese trusted the government or parliament, while trust in the army exceeded 85% [49]. When the civilian state is distrusted and only the security forces command confidence, the incentives for democratic reform weaken further. Low trust in this configuration is not merely a symptom of poor governance. It narrows the political space in which better governance could be built.

Traps also form within wealthy democracies. Greece offers the clearest European case: Eurobarometer data show that trust in national government exceeded 50% in 2004 and averaged 44% before 2008, then collapsed to 8% in autumn 2011 and 7% in 2012 during the sovereign debt crisis ◈ Strong Evidence [45]. It has not exceeded 37% since [45]. The pattern matches the finding by Algan and colleagues that crisis-driven unemployment damages trust in political institutions far more than interpersonal trust [30]. A society can therefore enter a low-trust equilibrium in its politics while retaining reasonable trust in neighbours, and the political component alone is enough to fuel anti-establishment voting [30]. France provides a milder variant. The CEVIPOF political trust barometer for 2025 found that only 26% of French respondents trusted politics, against 47% in Germany and 39% in Italy, with trust in the government at 17% and in the National Assembly at 20% — but trust in mayors at 61% [46].

The French mayors figure points to the most important qualification in the low-trust literature. Distrust is rarely uniform across institutions. Courts, local government and the civil service generally command more trust than parliaments and national governments: across the OECD in 2023, courts were trusted by 54% and parliaments by 37% ✓ Established [7]. The institutions citizens encounter directly and experience as rule-bound tend to be trusted more than those they see through partisan media. This is consistent with the institutional account. It also identifies where repair can begin. A society cannot change its history, but it can change the institutions through which most citizens meet the state — the police officer, the tax office, the school, the registry — and the evidence from the next sections suggests that changes at that level move measured trust faster than any political reform. The trap is persistent, but its walls are made of expectations, and expectations respond to evidence.

The United States offers the longest continuous series and the steepest decline among rich democracies. Pew Research Center records that about 73% of Americans trusted the federal government to do what is right always or most of the time in 1958 and 77% in 1964; the figure fell to about a quarter by 1980, has not exceeded 30% since 2007, and stood at 17% in 2025 ✓ Established [10]. Gallup’s measure of confidence across nine long-tracked institutions averaged 28% in 2025 against 48% in 1979 [55]. In April 2024 Gallup noted that 2022 marked the first time that the United States and the United Kingdom sat at the bottom of G7 rankings for confidence in national government, having both been highest at the outset of the World Poll nearly two decades earlier [11]. The American decline has continued across administrations of both parties. That pattern is difficult to attribute to any single government’s conduct and suggests structural drivers — polarisation, inequality and a fragmented information environment — rather than a cyclical response to performance.

The United Kingdom has followed a similar path more recently and more abruptly. The British Social Attitudes survey, published by the National Centre for Social Research in June 2024, found that 45% of Britons almost never trust governments of any party to place the needs of the nation above the interests of their own party, up 22 points since 2020, and that 58% almost never trust politicians to tell the truth when in a tight corner, up 19 points ✓ Established [44]. 79% said the system of governing Britain needed improvement [44]. In the OECD’s 2023 round, British trust in national government was 26.7%, the lowest of the countries in the Irish statistics office’s comparison [6]. The 2026 results show a further shift in composition: graduates now trust government less than non-graduates [4]. In a pattern that reverses most international findings, distrust in Britain has migrated from the economically excluded to the educated middle class.

Japan and Korea show that trust can rise quickly. Japan’s trust in national government stood at 24% in the 2021 OECD survey and reached 46% in 2025, above the OECD average of 40% ✓ Established [3]. The increase came despite the Liberal Democratic Party slush-fund scandal of 2023, which pushed Prime Minister Fumio Kishida’s approval rating as low as 17%, the lowest for any prime minister since the party returned to power in 2012 [50]. The OECD data do not isolate a single cause, and the gain coincided with leadership changes and a different political cycle. Korea’s rise was similarly sharp: according to Financial News, Korea recorded 51.03% trust in the 2026 results, up from 37.15% in the 2024 survey, placing it sixth in the OECD [43]. Korea’s Minister of the Interior, Yoon Ho-jung, attributed the gain to efforts to normalise state affairs quickly and strengthen communication with the public [43].

Estonia complicates the assumption that efficient public services automatically produce trust. The country is known internationally for its digital public services, yet in the 2024 OECD results only 38% of Estonians trusted their national government, down 9 points since 2021 ✓ Established [58]. The same survey found trust in the police at 74%, in the courts at 62%, and in other people at 60% [58]. Estonia therefore resembles the Nordic pattern in miniature: high trust in impartial administrative institutions and in fellow citizens, combined with volatile trust in elected governments. Efficient services build trust in the services. They do not by themselves protect governments from the political consequences of inflation, tax increases or contested policy choices. The Estonian case is a warning against technocratic optimism — the belief that better delivery will repair political distrust on its own.

Greece and Lebanon show how quickly trust can collapse when institutions fail visibly and the costs fall unevenly. In Greece, trust in national government fell from above 50% in 2004 to 7% in 2012, and has not recovered beyond 37% ◈ Strong Evidence [45]. In Lebanon, Arab Barometer surveys show trust in government and parliament below 10% by 2024, after a banking collapse that wiped out the savings of ordinary depositors [49]. The collapse in both countries followed a pattern consistent with the academic evidence: an economic shock that was experienced as unfair, with costs borne by those who had not caused it, damaged trust in political institutions far more than trust in neighbours or in institutions such as the army [30] [49]. Recovery has been slow in both cases. Lost trust appears to be regained more slowly than it is lost, because citizens need repeated evidence of changed behaviour before revising expectations formed during a crisis.

Switzerland and Norway, at the other end of the distribution, illustrate two routes to sustained high trust. Switzerland recorded 61.9% trust in national government in 2023 and 61.57% in the 2026 results, the highest in the OECD [6] [43]. Its system combines federalism, consensus government and frequent direct democracy, giving citizens formal voice on specific decisions several times a year. Norway relies instead on universal welfare institutions and an impartial civil service, and its trust in government has proved resilient: after falling to 48% in 2023 it recovered to 57% in 2025 [2]. Both systems give citizens repeated evidence that institutions are fair and responsive — one through participation, the other through impartial delivery. The comparison suggests that there is more than one route to high trust, but that all routes pass through visible fairness and voice ◈ Strong Evidence [1] [17].

Georgia’s reform is the most frequently cited modern case. After the 2003 Rose Revolution, the new government dismissed about 15,000 traffic police officers — roughly 85% of the force — in a single day in July 2004, and in total about 30,000 officers left the police service ✓ Established [36]. The force was rebuilt as a smaller, better-paid patrol police with new uniforms, new vehicles and new recruitment standards; the ratio of police to citizens moved from 1 per 78 to 1 per 214 by 2006 [36]. The World Bank’s 2012 study of the reforms documented parallel changes in tax administration, customs, electricity supply, public registries and university admissions [34]. The results were measurable. Georgia’s rank in the Corruption Perceptions Index improved from 130th of 158 countries in 2005 to 64th in 2011, and its World Bank Doing Business rank from 112th to 16th [35]. By 2009 an International Republican Institute poll ranked the police the third most popular institution in the country [36].

Survey data on police trust confirm the transformation, while also showing its limits. The Caucasus Barometer, run by the Caucasus Research Resource Centers, recorded trust in the police at 67% in 2011, 50% in 2012 and 58% in 2013 ✓ Established [37]. Trust was higher in rural areas than in the capital — 67% against 49% in 2013 — a pattern consistent with the view that citizens judge institutions by their own encounters with them [37]. The fall after 2011 is a reminder that trust built through institutional change remains contingent on continued conduct, and that it is re-earned in every encounter [37]. The World Bank framed the lesson as a challenge to fatalism: its report argued that corruption is sometimes seen as endemic, a product of local culture, and inevitable — and that Georgia’s experience showed otherwise [35]. The decisive element was not a communication campaign. It was the visible disappearance of the roadside bribe, an experience nearly every citizen could verify directly.

Hong Kong and Singapore show the same principle operating over a longer period. In June 1973, Chief Superintendent Peter Godber fled Hong Kong while under investigation, with assets of about HK$4.3 million, provoking public outrage ✓ Established [38]. Governor Murray MacLehose responded by creating the Independent Commission Against Corruption in February 1974, explaining that the public would have more confidence in a unit that is entirely independent and separated from any department [38]. Godber was extradited and tried in 1975, and the commission went on to arrest large numbers of serving police officers in its first years [38]. Singapore’s Corrupt Practices Investigation Bureau followed a similar model of independent enforcement; in Transparency International’s 2024 index Singapore scored 84 and ranked third of 180 countries, first in the Asia-Pacific region [39]. In both city-states, the credibility of the reform depended on prosecuting powerful insiders early, demonstrating that the rules now applied to everyone.

Bogotá offers a different model, focused on civic norms rather than enforcement. Antanas Mockus, elected mayor in 1995, launched a citizenship-culture programme that used symbolic interventions — including mimes who mocked traffic violators at intersections — alongside conventional policing [51]. During the following years homicides fell from 88 to 22 per 100,000 residents by 2003, annual traffic deaths fell from about 1,300 to about 600, and water consumption fell by about 40% during a shortage ◈ Strong Evidence [51]. Some 63,000 residents voluntarily paid an extra 10% in tax [51]. Attributing these changes to the programme alone is difficult, since policing and economic conditions also changed. But the Bogotá case illustrates a mechanism the trust literature predicts: when citizens see others complying with shared rules, their own expectations of reciprocity rise, and the social trap begins to unwind [19].

The common features of successful rebuilding cases are consistent with the academic evidence. First, change was visible at the point where citizens meet the state — the traffic stop, the tax office, the registry [34] [37]. Second, reforms targeted the powerful early, signalling that impartiality was real [38]. Third, the institutions were made structurally independent of the political actors they might otherwise serve [38] [39]. Fourth, success was measured by experience rather than announcement ◈ Strong Evidence [1] [5]. The historical debate about Sweden — whether its transition was a big bang or a gradual evolution — matters less for policy than this shared lesson: trust followed when citizens could observe, repeatedly and directly, that institutions behaved impartially [19] [40]. The risks that accompany such reforms are equally consistent across cases, and they determine whether a reform produces durable trust or a temporary gain.

The rebuilding evidence also clarifies what does not work. No case in the literature shows trust rising durably because a government declared its trustworthiness or ran a campaign to promote confidence. The OECD’s 2024 findings, which link trust most strongly to voice, integrity and fair treatment, imply that trust-building is a by-product of institutional conduct rather than a goal that can be pursued directly ◈ Strong Evidence [5]. Onora O’Neill’s argument that the aim should be more trust in the trustworthy, not more trust as such, is the operational principle behind every successful case [33]. The question for low-trust societies is therefore not how to persuade citizens to trust, but how to make institutions demonstrably trustworthy at the points where citizens encounter them. The answer in Georgia, Hong Kong and Singapore was enforcement against insiders and structural independence. In the Nordic countries it was universal, impartial delivery over generations.

The cultural account rests on strong evidence. Algan and Cahuc found that Americans inherit trust from their ancestors’ countries of origin, and that this inherited trust predicts economic outcomes in those origin countries [21]. Eric Uslaner found that Americans whose grandparents came from high-trust countries such as the Nordic states or Britain are more trusting, and that ethnic heritage appeared to matter more than present surroundings ◈ Strong Evidence [27]. Nunn and Wantchekon traced African mistrust to the slave trades of previous centuries and found that most of the effect operates through internal norms and beliefs [22]. Guiso, Sapienza and Zingales linked Italian civic capital to medieval self-government [24]. If these findings capture the main channel, trust is a slow variable that policy can influence only over generations. Low-trust societies would be trapped by their histories, and high-trust societies would enjoy a durable inheritance that reasonably good governance can preserve.

The institutional account rests on equally strong evidence of a different kind. Sønderskov and Dinesen showed, within individuals over 18 years, that changes in institutional trust preceded changes in social trust [25]. Dinesen found that immigrants’ trust responds to the institutions of their destination as well as the culture of their origin [26]. Kumlin and Rothstein linked the design of welfare programmes to the trust of those who use them [20]. The rebuilding cases — Georgia, Hong Kong, Singapore — show measured confidence in specific institutions changing within a decade ◈ Strong Evidence [35] [38] [39]. The two accounts are not strictly incompatible. Cultural transmission explains why trust levels are persistent; institutional effects explain how they can change. The policy-relevant finding is that institutions are the only part of the system that governments can reform directly, and that their effects compound across generations once established.

A second debate concerns whether falling trust is a problem at all. Pippa Norris, in her 2022 book In Praise of Skepticism, drew on survey data from more than 100 societies to compare citizens’ trust with the actual competence, honesty and impartiality of their governments, and argued that the risks of excessive, credulous trust are commonly underestimated ⚖ Contested [32]. On this view, the relevant measure is not the level of trust but its alignment with trustworthiness. Low trust in a corrupt government is a sign of a functioning citizenry; high trust in an untrustworthy one is dangerous. O’Neill has argued that the supposed crisis of trust may be more a matter of what people tell pollsters than of any active refusal of trust in daily life [33]. Citizens who tell pollsters they distrust doctors or civil servants continue to use their services. The behavioural evidence from lost wallets and pandemic compliance suggests that survey trust does correlate with behaviour, but the correlation is imperfect [52] [15].

A third debate concerns measurement across political systems. The Edelman Trust Barometer consistently ranks China, the United Arab Emirates, India and Indonesia as the most trusting societies, with Trust Index scores of 73 to 80 in 2026, while Japan, France, Germany and the United Kingdom score between 38 and 44 [8]. Critics, including a Radio Free Asia fact-check, argue that respondents in autocracies overstate their trust out of caution, and note that some governments ranked highly have had commercial relationships with Edelman ⚖ Contested [54]. The OECD survey, limited to democracies and members and accession candidates, uses different methods and gives a different picture, with Japan at 46% in 2025 rather than at the bottom [3]. A fourth debate concerns diversity: the meta-analysis of 87 studies finds a modest negative association between ethnic diversity and social trust, but Rothstein and Uslaner argue that inequality and corruption matter more [28] [17].

The weight of current evidence supports a synthesis. Cultural transmission is real and explains persistence: trust levels are sticky because expectations are passed from parents to children and reinforced by everyday experience [21] [22]. Institutional effects are also real and explain change: when institutions behave differently for long enough, expectations update, first among those who encounter them directly and then across the society [25] [26] ◈ Strong Evidence. The historical cases fit this pattern. Italian civic capital persisted because institutional conditions in the South remained less favourable for centuries [24]; Georgian police trust rose because the institution itself was replaced [36]. Culture, in this synthesis, is accumulated institutional experience. It changes slowly because experience accumulates slowly — not because it is immune to change.

The synthesis has a practical implication for how trust should be measured and discussed. Headline figures on trust in government, which respond to scandals, elections and economic cycles, are poor guides to the underlying condition of a society’s social capital. Measures of trust in courts, police, civil service and fellow citizens move more slowly and better capture the institutional foundation [7] [12]. A country can lose trust in a particular government while retaining trust in its institutions, as Norway did after 2021, or can lose both, as Greece and Lebanon did [2] [45] [49]. Distinguishing the two is essential for diagnosis. The first is a political problem that elections can resolve. The second is a structural problem that requires institutional reform — and, on the evidence, years of consistent conduct before measured trust recovers ◈ Strong Evidence [37] [45].

The social-capital ledger can now be read with some confidence. On the asset side, high-trust countries hold an accumulated stock of institutional experience: generations of impartial administration, universal services and low corruption, which have produced citizens who expect fair treatment and behave accordingly [13] [17] [25]. On the liability side, low-trust countries carry a stock of rational suspicion inherited from predatory institutions, sometimes centuries old [22] [24]. The OECD survey shows that the flows into and out of these stocks are driven mainly by procedural experience — voice, fairness, integrity — rather than by growth or delivery alone ◈ Strong Evidence [1] [5]. The 47-point gap in trust between those who feel heard and those who do not is the largest single finding in the data [1]. It identifies the most important deficit in rich democracies: not competence, but the sense that participation changes outcomes.

The economic stakes are substantial. The inherited-trust estimates of Algan and Cahuc, even if discounted for methodological uncertainty, imply income differences of 6% to 17% among major European economies attributable to trust alone [21] [57]. The pandemic evidence suggests that trust affected infection rates more than formal preparedness plans did [15] [16]. The populism evidence shows that crisis-driven distrust translates into anti-establishment voting, with consequences for policy stability and institutional quality ◈ Strong Evidence [30]. Trust is therefore not a soft variable to be managed by communication departments. It is a determinant of economic performance, public-health capacity and political stability — and because it is built slowly and lost quickly, it is a variable on which short-term political incentives systematically underinvest.

The policy implications follow directly from the drivers. First, institutions should be designed to minimise discretion at the point of contact, through universal rather than means-tested programmes where feasible and transparent rules where discretion is unavoidable [20] [18]. Second, anti-corruption enforcement must be structurally independent and must reach the powerful early, as in Hong Kong and Singapore [38] [39]. Third, participation must be consequential: the OECD data show that formal voting rights do not close the voice gap, and that citizens need evidence that their input changes decisions ✓ Established [1] [4]. Fourth, costs in crises must be distributed in ways citizens recognise as fair, since unfairness rather than hardship drives the collapse of political trust [30] [45]. Fifth, success should be measured by trust in specific institutions and by behavioural indicators, not by approval ratings for governments [7] [52].

Several uncertainties remain. The relative weight of culture and institutions is not settled, and the inheritance estimates are sensitive to methodological choices [57] [27]. Cross-national surveys may not capture trust equivalently across political systems, as the debate over autocratic bias in commercial surveys shows [54]. The reasons for the sharp rises in Japanese and Korean trust between 2021 and 2025 are not yet established, and could reflect political cycles rather than institutional change ⚖ Contested [3] [43]. The effect of ethnic diversity is real but modest, and its interaction with institutional quality is not fully understood [28]. And the argument by Norris and O’Neill — that low trust in untrustworthy institutions is healthy — implies that some measured declines may represent improved citizen judgement rather than deterioration [32] [33]. These uncertainties counsel humility in interpreting short-term movements, but they do not alter the central finding.

The comparison of countries suggests a final lesson about time. The Nordic stock of trust was built over more than a century of institutional reform, and it absorbs political shocks that would destabilise lower-trust societies [41] [2]. Georgia built measurable trust in its police within a decade, but saw part of it erode when political conditions changed [37]. Greece lost most of its political trust within 4 years and has recovered only partly in more than a decade since [45]. Trust is accumulated slowly and spent quickly. For governments, this means that the trust they inherit is mostly the product of their predecessors’ conduct, and the trust they bequeath will be the product of theirs. For citizens, it means that distrust is often a reasonable reading of the evidence, and that the appropriate demand is not for reassurance but for institutions that deserve confidence.

The social-capital ledger thus has an answer to the question in its title. Some countries trust their institutions because their institutions have, over long periods, treated citizens impartially, listened to them and refrained from predation — and because that experience has been transmitted across generations as expectation [17] [25] [21]. Others do not trust their institutions because their experience has been different, and their distrust is a rational adaptation that persists until conduct changes. Trust can be rebuilt; Georgia, Hong Kong, Singapore and, more tentatively, Japan and Korea show that measured trust responds to visible institutional change ◈ Strong Evidence [35] [38] [39] [3] [43]. But it is rebuilt only by trustworthiness demonstrated repeatedly where citizens meet the state. The 40% who trust their government across the OECD, and the 43% who do not, are not expressing national character. They are reporting what their institutions have taught them.