The economics of trust: what happens to a society when the thing that makes markets work starts to disappear

Enforcing a commercial contract through the courts takes 485 days in Denmark and 801 days in Brazil. Both countries have functioning legal systems, commercial courts, and constitutional protections for property rights. The formal institutional infrastructure is, for these purposes, roughly comparable. The outcome is not.

Those 316 days don’t measure corruption or catastrophic dysfunction. They measure substitution: the additional formal machinery that has to run when the informal expectation of honest dealing is thin. Every step that trust would have made unnecessary — the verification, the due diligence, the escalation to counsel before a contract is signed, the insurance against counterparty default — adds time and cost. The court is not a failure of the system. It is the system, doing what systems must when the social substrate beneath them can’t be relied upon.

But even these numbers capture only the visible tip. They count disputes that reached a court. They say nothing about what happens upstream: the risk premiums baked into every transaction in a low-trust environment, the compliance infrastructure that exists to verify what professional norms would once have governed, the layers of contractual protection that businesses in high-trust economies never need. Those costs appear nowhere as a “trust surcharge.” In Denmark, they largely don’t arise — not because they’ve been counted and found small, but because the social condition that would generate them doesn’t exist.

We have built extraordinarily precise instruments for measuring the formal overhead of commercial life. We have built nothing for measuring the thing that determines whether that overhead is necessary. And the thing itself is running down.

What economists left off the balance sheet

The idea that trust is economically significant isn’t novel. Kenneth Arrow stated it plainly in 1974, in “The Limits of Organization”: “virtually every commercial transaction has within it an element of trust.” This wasn’t an observation about human nature. It was a point about market function — about how buyers and sellers, unable to verify everything, still transact millions of times daily without paralysis or litigation. They extend trust, and the extension is cheap. The alternative is not.

Oliver Williamson spent two decades working out the economic logic of that alternative. Transaction cost economics — developed through the 1970s and 1980s, summarised in “The Economic Institutions of Capitalism” (1985) — showed that governance structures, contract design, and vertical integration decisions are partly responses to how much informal trust is available. Where trust is high, simpler structures work. Where it’s low, firms invest in monitoring, verification, and formal enforcement. All of which produce overhead without producing anything else.

What neither Arrow nor Williamson could easily do was aggregate. Individual transactions carry a trust premium — or not. But how do you measure it at the level of a national economy? How do you compare Denmark and Brazil not just on contract enforcement days but on the totality of trust-related friction? The answer, for decades, was: you can’t, not cleanly, and so macroeconomic growth models — capital accumulation, labour supply, technological change, institutional quality — largely treated the social substrate beneath all those inputs as a fixed given. It isn’t.

The empirical breakthrough came from Stephen Knack and Philip Keefer’s 1997 study in the Quarterly Journal of Economics. Working across 29 market economies using the World Values Survey’s generalised trust measure, they found that average trust levels correlate strongly with GDP per capita and investment rates. A ten percentage point increase in trust was associated with a growth increase of four-fifths of a percentage point. Not trivial, in a field where fractions of a percentage point compound dramatically over decades. Crucially, they found that measures of group membership — the standard proxy for civic engagement — were not independently associated with better economic performance. What mattered was the generalised measure: whether respondents believed that most people could be trusted — meaning people outside their network, people they didn’t know.

Francis Fukuyama drew the organisational implication in “Trust: The Social Virtues and the Creation of Prosperity” (1995). High-trust societies — Germany, Japan — could build large, professionally managed enterprises because the cooperation problem was solved informally. Low-trust societies ended up constrained to smaller, family-based structures, because formal verification was too expensive to scale. This isn’t cultural determinism dressed up as economics; it’s an argument about what kinds of productive organisation are possible under different social conditions.

Robert Putnam’s contribution, in the 1995 “Bowling Alone” essay in the Journal of Democracy and in the 2000 book, was to show where generalised trust comes from — dense networks of civic association — and to introduce a distinction that is vital and largely ignored in policy conversations: bonding social capital versus bridging social capital. Bonding capital is trust within a group. Bridging capital is trust across groups, between strangers, between people with no prior relationship. The economic literature that Knack and Keefer confirmed is about bridging. Bonding capital, on its own, doesn’t do it.

The academic literature has had the mechanism for thirty years. The empirical support was solid by 1997. And yet: no finance ministry models the fiscal implications of the generalised trust trend. No central bank includes social capital in its growth projections. The variable that determines whether all the formal infrastructure of commerce functions efficiently sits unmeasured and unmanaged on the balance sheet of every economy in the world.

The measurement problem

The World Values Survey "most people can be trusted" question has attracted two distinct methodological challenges. Johnson and Mislin (2012), in Economics Letters, examined whether the WVS measure correlates with experimentally measured trust — trust game behaviour where respondents make real decisions about entrusting money to strangers — and found that it does, giving the measure reasonable external validity. Sturgis and Smith (2010), in the International Journal of Public Opinion Research, identified a different problem: when respondents are asked about "most people," a significant proportion answer about people they know personally rather than strangers in general, which measures something else entirely. Neither critique destroys the measure — the Knack and Keefer growth correlations survive various robustness checks — but both are reasons to treat point estimates with care. What survives scrutiny is the directional finding: across multiple measurement approaches, higher generalised trust correlates with better economic performance.

The longitudinal picture — decline, divergence, and chronic deficit

The trust decline is not one story. Three different configurations of trust failure are playing out across the countries most frequently cited together, and treating them as parallel cases obscures more than it reveals. The economic damage looks different in each. So does the path — if there is one — out.

The United States is the active decline case. The General Social Survey — the most methodologically consistent longitudinal instrument in American social research — shows approximately 46% of respondents answering “most people can be trusted” in 1972, falling to approximately 32% by 2018. The World Values Survey corroborates, showing roughly a seven percentage point decline between 1981 and 2017. But the most striking data point predates both instruments: Putnam, writing in the Journal of Democracy in 1995, found that 58% of Americans reported trusting most people in 1960, declining to 37% by 1993. Thirty-three years, twenty-one percentage points, no social media in sight. Facebook didn’t exist. Twitter didn’t exist. The decline was already structurally embedded before the technologies routinely blamed for it were invented.

That point is left here without elaboration. The full argument about causation belongs later. But it does real work simply by existing.

The United Kingdom is not a decline case — at least not in the direction the political conversation implies. The Policy Institute at King’s College London, in its 2023 “The State of Social Trust” report, found that 46% of Britons agreed that “most people can be trusted” in 2022, a significant increase from 29% in 1999, which was the lowest recorded figure. UK interpersonal generalised trust has risen substantially over two decades and now ranks among the highest in the world.

What has fallen is institutional confidence. Trust in Parliament, in government, in political leadership — the kind measured by the Edelman Trust Barometer and the focus of most political commentary — has moved in the opposite direction: parliamentary confidence fell from 32% in 2018 to 23% by 2022, a sharp drop over the period that included the Brexit parliamentary crisis and the Covid lockdown conduct revelations. These are not the same construct. They can diverge. In the UK, they have diverged sharply, moving in opposite directions within the same society over roughly the same period.

Brazil is a third thing altogether. World Values Survey data shows Brazilian generalised trust below 10% consistently across multiple survey waves spanning decades. Not a recent decline from a previously higher state. A chronic deficit — a stable, multi-generational low-trust configuration that was never otherwise. The OECD’s 2023 assessment of Brazilian public institutions documented the persistent substitution burden and endemic compliance costs that result from operating an economy in these conditions across generations.

The Edelman Trust Barometer for 2026 supplies a current global reading: 70% of people worldwide are unwilling or hesitant to trust someone with different values, different approaches to social issues, or different information sources. That’s not a 2026 phenomenon. In the US case, it reflects a trajectory that began in 1960.

Interpersonal vs institutional trust

The distinction between interpersonal generalised trust ("most people can be trusted") and institutional confidence (trust in Parliament, government, courts, media) matters more than the popular conversation acknowledges. The economic literature — Knack and Keefer, Zak and Knack — works primarily with interpersonal generalised trust, because that's what correlates with investment and growth. Institutional confidence is what Edelman and most political pollsters primarily measure. The two are related but distinct constructs that can move in opposite directions, as the UK case demonstrates. The practical implication is an early-warning dynamic: institutional confidence collapse can precede interpersonal trust erosion by years. By the time interpersonal trust falls and the economic costs become clearly visible in the data, the underlying process may be well advanced.

The price of distrust — what institutional substitution costs

The substitution mechanism is not complicated. When informal trust is insufficient to underwrite a transaction, formal institutions step in. Legal contracts replace implicit agreements. Compliance departments replace professional norms. Monitoring infrastructure replaces assumed good faith. Insurance replaces expected reciprocity. Auditing replaces reputation.

None of these are free. None produce anything except the managed absence of a problem that trust would have prevented for nothing.

The deadweight character of these costs is what makes them economically distinctive. Legal fees for a contract that wouldn’t exist in a high-trust environment produce no output. Compliance staff verifying supplier integrity produce no output. The insurance premium covering counterparty default in an environment where default is a known risk produces no output. Aggregate these across an economy and you have a structural tax on all economic activity — invisible in productivity statistics because it’s spread across millions of transactions, absent from GDP because the measure has no category for costs created by the absence of something.

It’s there. In a chronically low-trust economy, it’s large.

The precise scale is difficult to quantify, and the formal data is instructive precisely because of what it fails to show. The World Bank’s Doing Business dataset records the formal cost of contract enforcement in both countries: Denmark at 23.3% of the claim value, Brazil at approximately 22%. Almost identical. What is not identical — not even close — is the time: 485 days versus 801 days, a differential that represents capital tied up, commercial decisions deferred, relationships that never stabilised into productive exchange. And even this gap captures only disputes that materialised and reached a court. What happens before any dispute: the elevated due diligence, the additional contractual protections, the counterparty risk premiums priced into every transaction in an environment where bad faith is a calculable possibility, the deals that never reached the table because the informal expectation of honest dealing wasn’t there to underwrite them. That overhead appears in no formal dataset. The formal costs are a floor, not a ceiling.

Zak and Knack’s 2001 analysis in the Economic Journal found that countries with trust one standard deviation below the mean show substantially lower investment-to-GDP ratios. Trust deficits don’t just add costs — they reduce the fundamental willingness to commit capital at all. The Knack and Keefer cross-country data makes the same point from the investment side: lower-trust countries consistently show lower investment rates, which compounds into lower capital stock, which compounds into lower long-run growth, all without any of the underlying mechanism showing up as a line item.

Robert Putnam’s “Making Democracy Work” (1993) remains the most compelling worked example. Comparing Italian regional governments — equivalent formal structures, comparable resources, similar constitutional frameworks — Putnam found dramatically better institutional performance and economic outcomes in northern regions than southern ones. The north had denser civic networks, higher generalised trust. The south didn’t. The formal institutions cost the same in both places. They worked radically differently, because the social substrate beneath them was radically different. Enforcement in the south required more monitoring, more legal overhead, more formal substitution for the informal cooperation that the north managed without thinking about it.

The intellectual genealogy goes back to Edward Banfield’s “The Moral Basis of a Backward Society” (1958) and his concept of “amoral familism” — high family loyalty, minimal civic trust — which Banfield saw as the structural condition of southern Italian backwardness. Putnam updated and quantified what Banfield had intuited. Same mechanism, four decades later, still explaining the same differential.

Daron Acemoglu and James Robinson, in “Why Nations Fail” (2012), don’t use trust as their central variable, but the mechanism they describe — extractive institutions emerging and persisting where citizens can’t coordinate to resist elite capture — is the political economy form of low generalised trust.

Where generalised trust is high, citizens can coordinate around shared expectations and hold institutions accountable. Where it’s low, coordination is costly and unreliable, elite capture becomes easier, and extractive institutions become self-sustaining. The causal relationship runs both ways: extractive institutions erode trust further, which degrades the conditions for inclusive governance, which makes trust repair more difficult. The loop closes.

Measuring what's missing

The World Bank's Doing Business project — now succeeded by Business Ready — provides the most systematic cross-country data on formal contract enforcement costs: time in days, cost as a percentage of the claim value, and an index of procedural quality. These figures capture what is visible after a dispute materialises. What they cannot capture is the informal overhead of operating in a low-trust environment before any dispute: the additional due diligence, the higher counterparty risk premiums, the contracts that were more elaborate than they needed to be, the transactions that never happened. The academic literature consistently identifies this hidden portion as substantially larger than the formal overhead, which is itself already significant. Any quantitative comparison of high- and low-trust economies using the available formal data will therefore understate the actual gap. The costs in the Doing Business figures are a floor, not a ceiling.

The poison pill — why high group cohesion without generalised trust makes things worse

The institutional substitution argument, left there, produces an obvious conclusion: rebuild trust, and the deadweight costs fall. Strengthen communities, restore civic bonds, encourage local association, recreate the social fabric. This response is not only predictable but widely proposed. It is also wrong, for a specific reason that the Putnam literature makes clear and that policy discussions almost never engage with.

Putnam’s bonding/bridging distinction is where it goes wrong. Bonding social capital connects similar people — family, religious community, ethnic group, neighbourhood. Bridging social capital connects unlike people: strangers, members of different groups, people with no prior relationship. The economic literature on trust is about bridging — the willingness to extend cooperation to the unknown other, the stranger across a marketplace. Knack and Keefer’s finding is specific on this: group membership measures — the density of formal civic ties — were not independently associated with better economic performance. The “most people can be trusted” measure was. Bonding capital, by itself, doesn’t produce generalised trust. Under certain conditions, it prevents it.

The mechanism that operates when in-group trust is high and generalised trust is low is not a mild form of trust failure. It is the social structure of corruption.

Rothstein and Uslaner’s “All for All: Equality, Corruption, and Social Trust,” published in World Politics in 2005, works through the logic: inequality generates particularised trust — strong bonds within close networks — and distrust of outsiders. The conditions this creates are precisely those under which corruption functions efficiently. A corrupt relationship requires some minimum trust between its parties, supplied by in-group bonds, while lacking the generalised trust that would generate accountability pressure from outside. Christian Bjørnskov’s 2007 analysis in Public Choice provided empirical support for the causal pathway: inequality produces low generalised trust combined with high in-group trust, and this combination produces corruption.

Italy again, but from a different angle. The Mafia is not a trust failure in the conventional sense. It is a trust structure that works with extraordinary precision — within the network, contracts are enforced with remarkable reliability, the free-rider problem is solved through credible threat, and internal cooperation is high. What it lacks is generalised trust. The high-trust network extracts from the low-trust general environment around it, and the extraction is made possible precisely because the wider society cannot coordinate resistance. Maximum internal trust, minimum external trust, severe economic dysfunction. The bonding/bridging distinction in its most concrete form.

Easterly and Levine’s “Africa’s Growth Tragedy” (1997), in the Quarterly Journal of Economics, found that ethnic fragmentation correlates with lower growth partly through a related mechanism: high within-group trust combined with active cross-group distrust produces the factionalism that diverts economic activity from production toward rent-seeking and redistribution. The causal claims in this literature are contested — ethnic diversity and growth interact through many channels — but as pattern evidence, the finding is consistent with what the bonding/bridging logic predicts.

The policy implication is uncomfortable. Strengthening group identity, rebuilding community bonds, encouraging in-group solidarity as a response to generalised trust decline does not automatically produce the kind of trust that matters economically. Under certain conditions, it produces the opposite: high-trust factions in a low-trust general environment. That combination — strong internal bonds, weak external ones — is the precise social configuration from which rent-seeking and institutional capture grow. What matters for economic cooperation is precisely the trust that is hardest to generate: the willingness to deal honestly with people you do not know, have no prior relationship with, and cannot sanction through social pressure. Building communities gives you more bonding capital. Bridging capital comes only incidentally, under specific civic conditions, and not reliably.

The asymmetry — why the optimistic scenario requires what democratic politics can’t efficiently supply

Paul Slovic, writing in Risk Analysis in 1993, identified the asymmetry at the heart of any trust repair discussion. Trust is typically created slowly, through accumulated consistent experience. It can be destroyed by a single mishap. Trustworthiness requires a large number of confirming instances to establish; a relatively small number of relevant negative instances to disconfirm. Bad news about an institution travels faster and lands heavier than good news. Negative trust events have outsized and durable effects.

The 2008 financial crisis destroyed trust in financial institutions across the developed world in months; some of that damage, measured in public attitudes toward banks and regulators, persists nearly two decades on. The Catholic Church abuse scandal in countries where the Church was a primary trust institution — Ireland, Poland, the United States — produced trust collapses that formal apologies and institutional reform have not reversed, and probably cannot reverse on any timescale relevant to policy. The UK government’s documented conduct during Covid lockdowns — parties while the population complied with rules that carried legal penalties — was a trust event with a specific character. It didn’t introduce a new suspicion. It confirmed an existing one: that the rules apply differently to those who make them.

That confirmation is not a one-cycle event. It updates a generalised model of how institutions behave, and each subsequent instance — each further case where the rules turn out not to apply to those who make them — reinforces the prior rather than requiring fresh evidence. A model updated once in this direction requires substantially more contrary evidence to move back than it took to move forward. The Slovic asymmetry, applied at the level of institutional credibility, compounds.

The dominant policy response misidentifies the mechanism. The framing is almost always communicative: improve transparency, issue corrections, counter disinformation, communicate performance more effectively. Martinangeli, Povitkina, Jagers, and Rothstein’s 2024 experimental study in the American Journal of Political Science demonstrated that it is institutional quality itself — specifically, the ability to prevent corrupt behaviour — that drives generalised trust, not communication about that quality. The researchers exposed participants to institutions of different quality, operationalised as their ability to prevent corrupt behaviour, and measured trust using a trust game. Institutional quality moved trust. Messaging didn’t. You cannot communicate your way out of a credibility gap. The update mechanism is behavioural, not informational.

The time dimension is where democratic politics runs into its structural problem. Rebuilding trust requires sustained, consistent institutional performance at timescales that substantially exceed electoral cycles. A democracy holds elections every four or five years. The incentive structure of electoral competition is designed to produce results on that timescale — policies whose benefits are visible before the next election, reforms that show effects within a parliamentary term. The investments trust repair requires produce benefits long after the politicians who made them have left office. This isn’t a failure of political will. It is an inherent feature of electoral incentives operating against trust dynamics that move at entirely different speeds.

How much faster trust declines than it can be rebuilt is not precisely measurable from the existing research. The asymmetry is well-documented in the directional sense; specific reconstruction timescales are rarely estimated in the academic literature. What the rate of US decline, combined with the Slovic asymmetry, implies as an order of magnitude is generational — not one or two election cycles, but sustained consistent institutional performance over twenty or thirty years. No electoral system is designed to deliver that. None has delivered it.

Social media and the amplification question

The US trust decline from 58% in 1960 to 37% in 1993 predates the internet entirely. Any account that centres on social media cannot explain three decades of erosion that happened before those technologies existed. This doesn't mean social media has no effect. The mechanism it most plausibly operates through is amplification: negative trust events that previously required newspaper and broadcast cycles to propagate now reach global audiences within hours, meaning the Slovic asymmetry — negative events outsized impact — operates at dramatically higher velocity and scale. Social media didn't cause the trust deficit. It accelerated the rate at which each confirming instance of institutional unreliability reaches every possible observer. Treating it as the root cause, and anti-disinformation policy as the primary intervention, addresses the amplifier while leaving the underlying institutional dynamics untouched.

The prognosis — what the evidence says about societies that can’t arrest the slide

Not collapse. The word is too dramatic, which makes it easy to dismiss. What the evidence supports is less cinematic and harder to argue against: persistent, compounding underperformance, operating through self-reinforcing dynamics that the political framing of the problem cannot address — not because politicians are incompetent, but because they’re solving a different problem.

The self-reinforcing loop is documented in the social capital literature. Uslaner’s “The Moral Foundations of Trust” (2002) and the Rothstein and Uslaner analysis identify the sequence: low trust produces higher institutional substitution costs, which produce worse economic outcomes, which produce greater inequality, which erodes generalised trust further. At some point — the research is honest about not knowing exactly where — this becomes self-sustaining rather than a decline to a new equilibrium. The feedback is clear. The threshold is not. What the literature can say is that no society that has allowed generalised trust to fall substantially has subsequently shown significant spontaneous recovery without the kind of sustained institutional performance that the political economy analysis says is structurally difficult to deliver.

The Acemoglu and Robinson mechanism — developed primarily for developing countries and historical cases — applies to high-income democracies only as an interpretive extension of their framework, not as a finding they themselves make. But the mechanism they describe is not obviously limited to developing countries. When generalised trust falls below the level needed to sustain inclusive institutions, the political conditions for extractive capture improve. Elite groups can, through ordinary political economy — lobbying, regulatory capture, coalition management — install governance structures that serve their interests more effectively precisely when citizens cannot coordinate resistance. Effective coordination requires the generalised trust that is failing. Whether the US and UK are tracing the early stages of this pattern is a question the evidence doesn’t definitively answer. Neither does it rule it out.

What the comparative growth data does show clearly is that the high-trust societies — Denmark, Norway, Sweden, Finland — have maintained or improved their positions in productivity, innovation, and per capita income relative to their factor endowments for decades. Generalised trust scores in these countries consistently rank among the highest in the world, and they have largely avoided the institutional confidence collapses that have characterised the US, UK, and Brazil over the same period. The Nordic premium isn’t explained by natural resources or demographics alone. The trust differential explains a significant portion of it. If the US and UK trajectories continue, the economic literature implies persistent higher transaction costs, lower investment ratios, lower institutional quality, and diminished capacity for the productive cooperation that trust makes cheap and substitution makes expensive.

Not collapse. Chronic drag.

The kind that is felt as stagnation, attributed to globalisation or automation or political dysfunction, and never accurately diagnosed.

The public debate about trust decline is framed almost entirely in political and social terms — polarisation, disinformation, democratic backsliding, leadership failure — and the responses follow suit: messaging, coalition-building, transparency initiatives, counter-disinformation campaigns. Almost none of it engages with the economic cost, the asymmetry of repair, or the structural mismatch between electoral timescales and trust dynamics. The tools are calibrated to electoral timescales. The process they’re aimed at operates on generational ones. The substitution costs compound while the conversation addresses something else.

The gap between what the evidence shows and what the political conversation acknowledges isn’t incidental. It’s a symptom of the same problem it describes.

Every piece of physical infrastructure has a replacement cost, a depreciation schedule, a maintenance budget. Roads, power grids, broadband networks appear on national accounts because, at some point, someone worked out that infrastructure unmaintained fails, and that failure costs more than maintenance. The accounting framework was built to capture what can be seen.

Social trust is infrastructure in every economically meaningful sense. It determines whether the formal infrastructure of commerce functions efficiently, whether transactions clear at normal cost, whether institutions perform their functions without expending most of their energy on enforcement. It appears on no balance sheet. No finance ministry has a line item for the fiscal implications of the GSS trend. No central bank includes the trust trajectory in its productivity projections. The GDP accounting frameworks developed across the twentieth century are extraordinarily sophisticated instruments. They are mute about this.

We argued, across decades, about interest rates and fiscal multipliers and capital buffers and current account imbalances. We built institutions to manage those things — central banks, fiscal councils, international bodies with mandates and metrics and accountability mechanisms. We left the thing that determines whether any of those instruments actually work unmeasured, and consequently unmanaged.

If trust is infrastructure, and infrastructure that goes unrecognised and unmaintained eventually fails, what do we call the moment when the failure becomes visible?

Not a crisis. Crises have sharp edges, defined events, a before and an after. What the evidence describes is something slower: a gradual thickening of friction in every commercial interaction, a rising overhead on all economic cooperation, a progressive narrowing of the cooperative range. It doesn’t announce itself. It shows up as stagnation, attributed to other things, managed with tools designed for other problems.

English doesn’t have a good word for that. And the absence of the word is part of the problem.

Esclusione di responsabilità di Gen AI

Alcuni contenuti di questa pagina sono stati generati e/o modificati con l'aiuto di un'intelligenza artificiale generativa.

Media

Aedrian Salazar – Pexels

Key Sources and References

Arrow, Kenneth J. The Limits of Organization. W. W. Norton, 1974.

Williamson, Oliver E. The Economic Institutions of Capitalism: Firms, Markets, Relational Contracting. Free Press, 1985.

Knack, Stephen, and Philip Keefer. “Does Social Capital Have an Economic Payoff? A Cross-Country Investigation.” Quarterly Journal of Economics, 112(4):1251-1288, 1997.

Fukuyama, Francis. Trust: The Social Virtues and the Creation of Prosperity. Free Press, 1995.

Putnam, Robert D. “Bowling Alone: America’s Declining Social Capital.” Journal of Democracy, 6(1):65-78, 1995.

Putnam, Robert D. Bowling Alone: The Collapse and Revival of American Community. Simon & Schuster, 2000.

Putnam, Robert D. Making Democracy Work: Civic Traditions in Modern Italy. Princeton University Press, 1993.

Banfield, Edward C. The Moral Basis of a Backward Society. Free Press, 1958.

Zak, Paul J., and Stephen Knack. “Trust and Growth.” Economic Journal, 111(470):295-321, 2001.

Johnson, Noel D., and Alexandra Mislin. “How Much Should We Trust the World Values Survey Trust Question?” Economics Letters, 116(2):210-212, 2012.

Sturgis, Patrick, and Patten Smith. “Assessing the Validity of Generalized Trust Questions: What Kind of Trust Are We Measuring?” International Journal of Public Opinion Research, 22(1):74-92, 2010.

Slovic, Paul. “Perceived Risk, Trust, and Democracy.” Risk Analysis, 13(6):675-682, 1993.

Rothstein, Bo, and Eric M. Uslaner. “All for All: Equality, Corruption, and Social Trust.” World Politics, 58(1):41-72, 2005.

Uslaner, Eric M. The Moral Foundations of Trust. Cambridge University Press, 2002.

Bjørnskov, Christian. “Determinants of Generalised Trust: A Cross-Country Comparison.” Public Choice, 130(1):1-21, 2007.

Martinangeli, Andrea F.M., Marina Povitkina, Sverker C. Jagers, and Bo Rothstein. “Institutional Quality Causes Generalized Trust: Experimental Evidence on Trusting Under the Shadow of Doubt.” American Journal of Political Science, 68(3):972-987, 2024.

Acemoglu, Daron, and James A. Robinson. Why Nations Fail: The Origins of Power, Prosperity, and Poverty. Crown Publishers, 2012.

Easterly, William, and Ross Levine. “Africa’s Growth Tragedy: Policies and Ethnic Divisions.” Quarterly Journal of Economics, 112(4):1203-1250, 1997.

World Bank. Doing Business 2020: Comparing Business Regulation in 190 Economies. World Bank Group, 2020. archive.doingbusiness.org

Policy Institute at King’s College London. The State of Social Trust: How the UK Compares Internationally. King’s College London, June 2023. kcl.ac.uk/policy-institute/assets/the-state-of-social-trust.pdf

Edelman. 2026 Edelman Trust Barometer Global Report. Edelman, January 2026. edelman.com/trust/2026/trust-barometer

OECD. Drivers of Trust in Public Institutions in Brazil. OECD Publishing, 2023.

NORC at the University of Chicago. General Social Survey. gss.norc.org

World Values Survey Association. World Values Survey, Waves 1–7 (1981–2022). worldvaluessurvey.org