Political crisis and social trust: the case of Australia

ABSTRACT

Trust in Australian political institutions has declined markedly since 2007. Trust scholars and social capital theorists argue that institutional trust affects social trust, which implies that declining trust in political institutions should also reduce social trust. However, using data from the Australian Election Study, World Values Survey and HILDA panel study, I find no evidence of any aggregate-level decline in Australian social trust since 2007. While this could be a result of the ‘coup era’ in Australian politics from 2010 to 2018, a better explanation is that social trust remained high because permanent changes in trust occur primarily among younger people. A long-term decline through generational replacement therefore remains possible. This challenges both social capital theory and institutional theories of social trust, but would explain the experiences of other advanced democracies during episodes of political crisis, including New Zealand, Greece, Finland, Japan, Sweden, Spain and the United States.

KEYWORDS:

• Trust
• Australia
• generalised
• social
• institutional
• HILDA
• WVS
• AES

1. Introduction

Almost uniquely among advanced democracies, Australia has seen a substantial decline in political trust since 2007. This started to become apparent when trust in political institutions, previously cyclical, failed to rebound in its usual manner after the 2013 change of government (Dassonneville and McAllister 2021) and by 2019 it had reached its ‘lowest level ever recorded’ (Jiang 2022). Social capital theorists and other trust scholars have long suggested that institutional trust and social trust are linked (Hardin 1992; Kumlin and Rothstein 2005; Newton, Stolle, and Zmerli 2018; Tao et al. 2014) and a recent panel study reports that institutional trust affects social trust but not vice-versa (Dinesen et al. 2022). This implies that Australia's declining institutional trust should drag down social trust, which would have worrying consequences for democracy and economic growth (Putnam 1993; Zak and Knack 2001).

This paper examines whether Australia has experienced the predicted decline in social trust and is, to my knowledge, the first paper to examine how institutional trust shapes Australian social trust. I use data from the World Values Survey, Australian Election Study and HILDA panel survey to track changes in social and institutional trust from 1981 to 2022. The results are clear: while social and institutional trust are correlated at the individual level, there has been no aggregate-level decline across three different indicators of social trust. This contradicts social capital theory and institutionalist explanations for social trust (Dinesen et al. 2022; Hardin 1992; Newton and Zmerli 2011) but is consistent with the experiences of some other advanced democracies such as New Zealand, Sweden, Greece and Spain which, despite political crises and declining institutional trust, showed stable or even increasing social trust (Caïs, Torrente, and Bolancé 2021; Ervasti, Kouvo, and Venetoklis 2019; Newton 2006).

Why did Australian social trust fail to decline post-2007? Further investigation indicates that while trust in party-political institutions declined, this was not the case for apolitical institutions like the police or judiciary. This suggests that the institutional trust decline was not a broad secular trend but was instead possibly the result of factors specific to federal party politics such as the ‘coup era’ of 2010–2018. Given the theorised role of the judiciary, law enforcement and bureaucrats in translating political trust into social trust (Fairbrother et al. 2022; Freitag 2003; Kumlin and Rothstein 2005), perhaps the narrow, temporary decline in institutional trust seen in Australia was not sufficient to trigger a decline in social trust. Another explanation, which arguably fits the data better, is that permanent changes in social trust are limited mostly to younger people (Dawson 2019; Uslaner 2002). This is consistent with evidence that Australians who came of age during the institutional trust decline show decreases in social trust as well, and implies that country-level social trust could decrease in the future through a process of generational replacement. This paper contributes to our understanding of the determinants of Australian social trust – a topic which has received surprisingly little attention – as well as the impacts of political crises on social trust more generally.

2. Social and institutional trust in Australia

Social trust – which refers to citizens’ ‘horizontal’ trust towards each other – is important for a flourishing society. It assists economic development by reducing transaction costs and enabling collective action (Putnam 1993, 1995; Whiteley 2000; Zak and Knack 2001) and facilitates the formation of large corporations (Fukuyama 1995). It is also important for effective governance and the formation of the ‘civic culture’ which underpins democracy (Almond and Verba 1963; Putnam 1993). Social trust is also linked to higher vaccination rates and greater resilience to the COVID-19 pandemic (Lenton, Boulton, and Scheffer 2022; Mankell and Abdelzadeh 2023) – a vivid illustration of how trust generates collective benefit. Australia's social trust has traditionally been very high: in Wave 3 of the World Values Survey (1995–1999), only six countries out of 56 scored higher on a key indicator of social trust (Inglehart et al. 2022).¹

The possible determinants of social trust have therefore received a great deal of attention, ranging from country-level cultural factors (Bjørnskov 2007; Delhey and Newton 2005; Uslaner 2002) to individual life experiences (Glanville and Paxton 2007; Hardin 1992; Laurence 2015). It has also been frequently suggested that social trust is linked to other forms of trust, particularly one's trust in political institutions, which is a form of ‘vertical’ trust (Bäck and Kestilä 2009; Freitag 2003; Newton and Zmerli 2011). ‘Social capital theory’ contends that social trust and institutional trust are closely associated and mutually reinforcing – social trust facilitates institutional trust by making it easier to participate in civic affairs, while institutional trust, in turn, creates conditions where social trust can flourish (Putnam 2000; Newton and Zmerli 2011). This has received some degree of empirical support (Bargsted et al. 2023; Brehm and Rahn 1997; Jagodzinski and Manabe 2004; Newton and Zmerli 2011; Spadaro et al., 2020). Others have suggested that the direction of causality runs mainly from institutional to social trust, known as the ‘institutional’ view (Bargsted et al. 2023; Rothstein and Stolle 2008). Using instrumental variables, Tao et al. (2014) demonstrated that institutional trust affected social trust in China, while Dinesen et al. (2022) used Swedish panel data to show that institutional trust affected social trust but not vice-versa.

In both cases, the causal link from institutional trust to social trust is believed to occur through either of two broad channels. The first, which I label the ‘signalling’ channel, emphasises the role of political and institutional actors as a kind of heuristic which citizens use to make estimates about the trustworthiness of people in general. In this account, people draw inferences about others’ behaviour by observing the actions of bureaucrats and other public servants (Kumlin and Rothstein 2005, 349) – if they cannot be trusted, then why should people in general be trusted? Rothstein and Stolle (2008, 446) point out that the behaviour of police officers, bureaucrats and judges ‘function as important signals to citizens about the moral standard of the society in which they live’ and if they are perceived as corrupt then citizens will internalise a message that ‘in order to get ahead in life one must be engaged in various forms of corruption’. Accordingly, interactions with bureaucrats have been found to affect social trust (Fairbrother et al. 2022; Kumlin and Rothstein 2005). Secondly, institutional trust might affect social trust through the ‘enforcement’ channel, which draws attention to the role of strong institutions in fostering prosocial behaviour and punishing opportunistic or selfish individuals. As Freitag (2003) notes, ‘citizens will trust one another despite the temptation to freeload as long as they perceive that their political and societal institutions guarantee a credible environment where people can be confident that trusting will be requited and not exploited’. By enforcing contracts and punishing liars and cheaters, institutions create an environment where citizens feel comfortable trusting others (Hardin 1992, 161; Knack and Keefer 1997, 1279). This channel received empirical support from Spadaro et al. (2020) who found that the relationship between institutional and social trust is mediated by increased feelings of security. People who trust institutions will therefore trust other people in general.

We therefore have good reason to be concerned about Australian social trust. Trust in Australian political institutions has been in decline since 2007, and if there is a link between institutional and social trust this implies that Australian social trust should also have declined. Australian institutional trust was largely cyclical over the late twentieth century – supporters of the incumbent government tend to express higher trust in political institutions, so government turnover led to periodic shifts in trust levels (Bean 2001). But Dassonneville and McAllister (2021) noted that after a peak in 2007, political trust declined and failed to bounce back after the 2013 change of government. They concluded that trust in political institutions appeared to be trending downwards. And by 2019, institutional trust had reached its ‘lowest level ever recorded’ (Jiang 2022) with only 25% agreeing that people in government could ‘sometimes’ or ‘usually’ be trusted to do the right thing, down from 43% in 2007 (Cameron and McAllister 2022). The apparent crisis of confidence in Australia's political institutions triggered expressions of concern from both academics and the popular press (Evans 2019; Hutchens 2018).

This yields a clear prediction: Australian social trust should have declined alongside institutional trust since 2007. As trust in political institutions declined, Australians may have used that as a cue to revise downwards their trust towards other people in general. After all, if the very officials charged with overseeing the ‘public good’ cannot be trusted, what does that say about ordinary people (Kumlin and Rothstein 2005; Rothstein and Stolle 2008)? This is the ‘signalling’ channel. Alternatively, Australians may perceive that their untrustworthy political institutions are ineffective at sanctioning bad behaviour, that liars and cheats among the general population are therefore unlikely to face punishment for their misdeeds, and that as a result, they cannot afford to extend trust towards other people (Freitag 2003; Knack and Keefer 1997). This is the ‘enforcement’ channel. Regardless of which of these one finds most plausible, the empirical implications are summarised in the hypotheses below:

H₁: Australian social trust is correlated with institutional trust.

H₁ should be correct regardless of whether one prefers social capital theory or an ‘institutional’ link between the two forms of trust. A correlation between Australian social and institutional trust was previously identified by McAllister (2014) although his data only goes up to 2012, missing most of the institutional trust decline, and nor did it use HILDA or WVS data. A correlation between institutional and social trust of course does not prove that the former causes the latter, but it is a necessary condition and one which we would expect to observe if institutional trust exerts a causal impact on social trust. If this were the case, then we would also expect to observe the following:

H₂: Australia’s level of social trust should have declined alongside institutional trust since 2007.

All of the above notwithstanding, the experience of some other countries also gives us reasons to be sceptical that declining institutional trust pushes down social trust during episodes of political crisis. Newton (2006) examined the cases of Sweden, Finland, New Zealand and Japan, all of which underwent episodes of political crisis during the 1980s and early 1990s due mainly to economic problems and poor government performance. Trust in government institutions declined significantly in all four countries, but social trust changed very little. Caïs, Torrente, and Bolancé (2021) and Ervasti, Kouvo, and Venetoklis (2019) report that a similar phenomenon occurred in Spain and Greece in the wake of the 2008 economic crisis: institutional trust declined markedly, but social trust increased. To complicate matters further, Newton (2006) shows that institutional and social trust are nevertheless correlated at the country level. The relationship between institutional and social trust is clearly more complex than social capital theorists or ‘institutional’ scholars of trust might claim, and it is far from obvious whether Australia's declining institutional trust will have yielded a concomitant decline in social trust. This paper therefore makes two main contributions. Not only is it the first paper to test whether Australia's oft-mentioned decline in political trust has triggered a decline in social trust, but it also contributes to our understanding of the determinants of Australian social trust, a topic which has attracted surprisingly little attention. Some factors which have been found to affect Australian social trust include ethnolinguistic diversity (Leigh 2006; Mendolia, Tosh, and Yerokhin 2016), social network structure (Chong et al. 2011) sport participation (Brown, Hoye, and Nicholson 2014) and parental divorce (Viitanen 2014), but with the exceptions of Bean (2005) and Tranter and Skrbiš (2009), political factors have remained largely absent from the discussion.

3. Data

To test the hypotheses, I use three separate datasets: the Australian Election Study (AES) surveys from 1993 to 2022, World Values Survey (WVS) data for Australia from 1981 to 2018, and the Household, Income and Labour Dynamics in Australia (HILDA) panel studies from 2005 to 2022 (Inglehart et al. 2022; McAllister et al., 2022; Watson and Wooden 2012).² Only waves 5, 6, 8, 10, 11, 14, 18 and 22 of the HILDA studies are used because these are the ones with social trust data. Social trust is concerned with trust in ordinary people, such as neighbours or strangers on the street, whereas institutional trust refers to trust in institutions such as parliament, political parties, the press, the police or the judiciary. As Freitag (2003) notes, social trust belongs primarily to the private sphere while institutional trust belongs to the public sphere. It is also important to distinguish social trust from ‘particularised trust’ – the latter refers to trust towards specific people one knows personally, such as one's spouse, friends or family, whereas social trust is concerned with unspecified strangers (Freitag and Traunmüller 2009).

Across the three surveys, I use four different measures of social trust: generalised trust, people take advantage (reversed), neighbourhood trust and trust in strangers. Following Dassonneville and McAllister (2021) and Jiang (2022), institutional trust is operationalised in the AES with three items: satisfaction with democracy, belief that government can be trusted and belief that the government is run for the benefit of all.³ In the WVS, institutional trust is operationalised as the respondent's degree of ‘confidence’ in various institutions such as parliament, the government or political parties. Controls for age, sex, income, ethnicity, education, religiosity and state are also used in some regression models. A detailed description of the measures, coding and operationalisation of all variables is contained in Appendix A, and descriptive statistics are contained in Appendix B. In all the analyses, survey weights are applied wherever available.⁴ Consistent with Cameron and McAllister (2022) and Dassonneville and McAllister (2021), trust variables are recoded to binary in most cases.⁵

4. Results

To test H₁, I run a series of logistic regression models with social trust as the outcome and institutional trust as the predictors. I also control for a range of likely confounders, and survey weights are applied wherever possible. Each different type of social trust (generalised trust, trust in strangers and neighbourhood trust) is run in separate models, and to avoid multicollinearity only one form of institutional trust was included per model. Full regression tables are available in Appendix C. The ‘people take advantage’ measure of social trust was not included because this variable appears only in the HILDA dataset which does not include any institutional trust data. The results are displayed in Figure 1 and are consistent with H₁. Regardless of whether controls were used, generalised trust is associated with higher levels of all six measures of institutional trust. Confidence in parliament and government were linked to greater neighbourhood trust, and confidence in parliament, government and political parties were all associated with greater trust in strangers. It seems clear that trust in Australian political institutions is linked to greater social trust, and H₁ is therefore supported.

Figure 1. Relationship between social and institutional trust. Italicised variables from WVS, non-italicised from AES.

To test H₂, I examine how mean levels of social trust have changed over time. Has social trust declined alongside institutional trust? Before examining social trust, I first generate some plots in order to confirm that institutional trust has, in fact, declined. Although the decline in Australian institutional trust has been widely publicised by the likes of Dassonneville and McAllister (2021), this relied only on AES data. Is a decline also apparent in the Australian WVS data? The trends in institutional trust are shown in Figure 2 and also displayed numerically in Appendix D. The numbers refer to the aggregate level of trust in each year. Turning first to the left-hand panel, we can see that the measures of institutional trust from the AES show a clear decline as reported by Dassonneville and McAllister (2021) and Cameron and McAllister (2022). The proportion of people who believe the government can be trusted declined from 0.43 in 2007 to 0.30 in 2022, while the belief that government is run for the benefit of all declined from 0.62 to 0.46 and satisfaction with democracy fell from 0.86 to 0.70. Importantly, the WVS data shows a similar pattern, with confidence in government falling from 0.40 in 2005 to 0.30 in 2018, while confidence in parliament fell from 0.34 to 0.28 and political parties from 0.14 to 0.11. This is, to my knowledge, the first time that WVS data has been brought to bear on this question and confirms that Australian institutional trust has declined since the mid-2000s.

Figure 2. Mean Australian institutional trust over time.

With the decline in institutional trust now firmly established, we can turn to the question of social trust. Social trust levels over time are displayed in Figure 3 and also shown numerically in Appendix D. The results seem clear: there has been no discernible decline in Australian social trust, and this holds true across all three surveys and all four types of social trust. In the AES, generalised trust was 0.57 in 2007 and 0.56 in 2016. In the HILDA survey, the measures of social trust actually increased slightly over time: generalised trust went from 0.69 in 2006 to 0.70 in 2022, while neighbourhood trust went from 0.66 to 0.70 and people take advantage (reversed) rose from 0.66 to 0.67.⁶ In the WVS, neighbourhood trust went from 0.82 in 2005 to 0.80 in 2018, while generalised trust went from 0.46 to 0.49 and trust in strangers went from 0.49 to 0.48. Across all three surveys, some indicators of social trust increased slightly and others declined slightly, but the magnitude of these changes is small and recent levels of Australian social trust are almost identical to those in the mid-2000s. Evidently, social trust has not changed significantly since that time, which contradicts H₂.

Figure 3. Mean Australian social trust over time.

Overall, the results support H₁ but not H₂. Despite the evidence of an individual-level association between social and institutional trust, the aggregate-level decline in institutional trust has not yielded a similar decline in social trust. In the next section, I discuss some reasons why this might be the case.

5. Explaining Australian social trust

The above results are problematic for social capital theorists and ‘institutional’ scholars of trust. The fact that social and institutional trust are associated with each other (H₁) is consistent with these theories, but then how can we explain the fact that social trust failed to decline along with institutional trust (H₂)? I contend that there are two possibilities. First, the social capital/institutional theories could be correct, but the institutional trust decline was too narrow to trigger a decline in social trust. Second, institutional trust may not always exert a causal effect on social trust across all age groups, political contexts or levels of aggregation. I will discuss each of these possibilities in turn.

5.1. Was the institutional trust decline too narrow to affect social trust?

One possibility is that while institutional trust does generally affect social trust as the social capital and institutional theories suggest, Australia's institutional trust decline was too narrow in scope to yield an effect on social trust. Institutional and social capital theories of trust place particular focus on bureaucrats and civil servants (Kumlin and Rothstein 2005; Rothstein and Stolle 2008) and this is especially the case for the ‘enforcement’ mechanism where citizens decide to trust because they can rely on a well-functioning police and justice system to punish wrongdoers (Freitag 2003; Knack and Keefer 1997). And even in the ‘signalling’ mechanism, civil servants, police and judges are still highly visible agents of government and therefore can serve as potent ‘signals’ of the trustworthiness of the generalised other. But what if people observe misbehaviour among only a small segment of elected politicians, while bureaucrats, civil servants, police and judges continue to be perceived as clean? In this case, it would make sense that people continue to express high social trust – there would be little reason to have doubts about ‘enforcement’ and most government officials would still emit high trust ‘signals’.

Could this explain why Australian social trust remained high? After all, the decline in institutional trust coincided with an unusual period of turmoil in Australian federal politics. From 2010 to 2018, the country saw four sitting Prime Ministers deposed by their own parties, giving it the title of ‘coup capital of the democratic world’ (Cameron and McAllister 2024). In all four cases, a majority of voters disapproved of the coups and frequent changes of leadership became regarded as a national embarrassment (Cameron and McAllister 2024; Evans 2015). Perhaps the decline in trust in Australian political institutions merely reflects voters’ temporary discontent with the state of federal politics. The institutional trust indicators which declined in Figure 2 are all related to party politics – but if this was merely a symptom of discontent about the ‘coups’ in the federal Liberal and Labor parties, then we would not expect any decline in trust towards apolitical institutions like the police, civil servants or courts.

Evidence from the WVS appears to bear this out. In Figure 4, we can see that confidence in the police, civil services and the justice system has, if anything, then increased slightly since the mid-2000s. As with the previous figures, these trends are also recorded numerically in Appendix D. Clearly, the institutional trust decline evident in Figure 2 was limited to institutions closely associated with federal party politics – i.e. the government, parliament and political parties – and did not extend to the institutions that citizens rely on for ‘enforcement’ or as the source of many of their ‘signals’ about the trustworthiness of society. This is consistent with the idea that the institutional trust decline in Figure 2 and reported elsewhere by Dassonneville and McAllister (2021) and Jiang (2022) was merely a symptom of discontent with the ‘coup era’ of the 2010s. Perhaps most tellingly of all, by the time the ‘coup era’ was over in 2022, institutional trust somewhat rebounded (see AES data in Figure 2). If institutional trust continues to rebound in the next AES survey in 2025, this will provide even more compelling evidence that the initial decline was a temporary, narrow phenomenon related to the ‘coup era’ of the 2010s, and thus did not affect the institutions which citizens use as the basis of their judgements about social trust.

Figure 4. Mean Australian trust in police, courts and civil services.

However, this explanation – that the decline in institutional trust was too narrow to affect social trust – is also unsatisfying because several other countries experienced stable social trust levels despite ‘broad’ decreases in institutional trust which encompassed both party-political and apolitical institutions, including Greece (Ervasti, Kouvo, and Venetoklis 2019), the United States (Aassve et al. 2024) and Finland (Newton 2006). It seems that even a ‘broad’ decline is not sufficient to yield declining social trust. Moreover, Jiang (2022) found that approval of leadership changes did not explain Australian institutional trust, which is inconsistent with the idea that the decline was driven by dissatisfaction with repeated ‘coups’.

What about other factors, such as the Global Financial Crisis of 2008–2009? Ervasti, Kouvo, and Venetoklis (2019) contended that Greek social trust remained high in the wake of the GFC because the crisis caused people to ‘lean on each other when both political and impartial institutions fail’. This seems unlikely to explain the Australian results, however, because Australia avoided recession during the GFC and there was therefore little reason for Australians to ‘lean on’ each other. Moreover, New Zealand, Finland and Sweden saw instances of diverging social and institutional trust during the 1980s (Newton 2006), which suggests that this pattern is a more general trend and not the result of the GFC.

5.2. Do the social capital and institutional theories hold true for everyone?

It might therefore be preferable to consider an alternative explanation – that institutional trust may not exert a causal effect on social trust among adults. This is not as far-fetched as it might seem, even though H₁ is supported. A long line of trust scholarship contends that trust is largely fixed during people's adult lives, and adult experiences do not have long-lasting effects – including experiences with government officials or the political system. Instead, early-life social context and even genetic factors are the strongest influences on social trust, which is mostly set in childhood and changes little thereafter (Uslaner 2002). There is some evidence in support of this view (Dawson 2019; Stolle and Hooghe 2004; Wu 2020; Wu 2021). Rather than being the product of experiences, trust is instead shaped by one's parents, cultural background as well as other factors such as the degree of social inequality (Bjørnskov 2007; Rothstein and Uslaner 2005; Uslaner 2002). Other studies have pointed out that adult life experiences do have temporary effects on social trust, but these tend to dissipate quickly as people return to their baseline levels (Kumove 2024). There is even evidence that this occurs following interactions with bureaucrats in particular (Fairbrother et al. 2022). If this is true, then it would be no surprise that Australia's declining institutional trust failed to yield a decline in social trust. Social trust is simply not very amenable to change during people's adult life, and even if experiencing Prime Ministerial ‘coups’ or other political shenanigans might yield a temporary decrease in social trust through the ‘signalling’ or ‘enforcement’ mechanisms, this will likely have dissipated by the time the next AES or WVS survey is conducted. Institutional and social trust are correlated, as H₁ predicted, but this merely reflects the fact that they are both the result of one's upbringing: some people just have ‘trusting personalities’.⁷

This explanation also can account for the overseas results. As noted above, Japan, New Zealand, Sweden, Finland, Greece and Spain all experienced episodes of declining institutional trust yet social trust was largely unchanged (Caïs, Torrente, and Bolancé 2021; Ervasti, Kouvo, and Venetoklis 2019; Newton 2006). If the H₂ results were merely because Australia's institutional trust decline was too ‘narrow’ for reasons specific to the 2010–2018 ‘coup era’, then how do we explain these similar patterns overseas? ⁸ However, the idea that institutional trust may not affect country-level social trust would also explain the overseas results. Just like Australia, people in those countries mostly do not modify their social trust during adulthood, so even very negative experiences with the political system will have little immediate impact on aggregate social trust levels.

However, even if declining institutional trust does not usually yield social trust changes among adults, this does not preclude the possibility that Australia's institutional trust decline may yield declining social trust in the future through a process of cohort replacement. Young people who are raised during times of low institutional trust may grow up to show low social trust as well; this would be consistent with the contention of Uslaner (2002) and others that social trust is mainly shaped during childhood and adolescence. There is some evidence for this in Figure 5, which shows trends in Australian social trust for different birth cohorts. People born in the 1990s, whose adolescent years occurred during the institutional trust decline, show markedly different trends to earlier generations. Their generalised trust declined steeply from 2010, and from 2011 to 2016 they had the lowest score of any birth cohort – although it started to recover after 2014. Other cohorts do not show this kind of decline. The same is true for ‘people take advantage’, where the 1990s cohort initially showed the highest level of trust but, unlike all other cohorts, this declined rapidly during the 2010s. This is consistent with the idea that growing up during low institutional trust causes a scarring effect on social trust. If Australian institutional trust remains low, we might expect aggregate social trust levels to decrease in the future as older, high-trust cohorts are replaced by newer cohorts who were socialised to have lower trust levels. Cohort effects have already been shown to have contributed to the decline in social trust in the United States (Clark and Eisenstein 2013), although a full age-period-cohort analysis of Australian social trust will be necessary to determine whether the same process is likely to occur here.

Figure 5. Australian social trust by birth cohort.¹¹

The ‘cohort replacement’ explanation is also particularly attractive because it offers a way of reconciling Dinesen et al. (2022) and Tao et al. (2014) with the present findings. Those studies found an individual-level causal link from institutional to social trust. However, they did not disaggregate their results by age group, which leaves open the possibility that the link they identified operates mainly for adolescents and not adults. If so, it should only generate visible country-level impacts in the long term through cohort replacement. An individual-level link (Dinesen et al. 2022; Tao et al. 2014) could therefore coexist with country-level divergence in the short term. This suggests a potential refinement of the social capital and institutional theories where the causal link from institutional to social trust operates mainly for younger people. ⁹

6. Discussion and conclusion

This paper contributes to a growing body of evidence which shows that social and institutional trust diverge during episodes of political crisis. Social capital and institutional theories of trust predict that institutional trust affects social trust, and therefore Australia's declining institutional trust should cause a decline in social trust. However, while the results indicated that social and institutional trust are indeed correlated in Australia (H₁) there is no evidence of any decline across four different forms of social trust (H₂). One possibility is that Australia's institutional trust decline, which seems to have affected only institutions linked to party politics, was too ‘narrow’ to yield a decline in social trust, although this fails to explain why other countries such as Greece, Finland and the USA experienced ‘broad’ institutional trust declines and yet social trust still held steady (Aassve et al. 2024; Ervasti, Kouvo, and Venetoklis 2019; Newton 2006). Another possibility, which is consistent with a wide range of recent scholarship on social trust, is that the institutional and social capital theories do not apply to every person, or in every political context: institutional trust might only exert a persistent causal effect on social trust among adolescents, or perhaps the causal dynamics work differently during times of political crisis. If so, rapidly declining institutional trust may not yield a near-term decline in country-level social trust. This would explain not only the Australian results in this paper but also similar trends which have occurred in New Zealand, Sweden, Finland, Japan, Greece, Spain and the United States (Aassve et al. 2024; Caïs, Torrente, and Bolancé 2021; Ervasti, Kouvo, and Venetoklis 2019; Newton 2006).

The conclusion of this paper is therefore somewhat optimistic. Australian social trust is robust and is seemingly not at risk from declining institutional trust in the short term, although Figure 5 points to a possible long-term decline through cohort replacement. As of 2018, only ten countries (out of 76 total) in the WVS Round 7 survey showed higher levels of generalised trust than Australia (Inglehart et al. 2022) and HILDA data indicates that Australia does not seem to have experienced any decline in social trust since then, even during Covid-19 (see Figure 3).¹⁰ We should therefore pause before accepting claims recently promulgated by the media that Australian social cohesion has ‘plummeted’ (Murphy 2023). Those claims were based on the Scanlon Institute's Mapping Social Cohesion report, and were founded partly on the fact that Australians reported reduced financial security in 2023 (O’Donnell 2023). It is questionable, however, whether financial well-being can truly be considered a component of social cohesion (Schiefer and Van Der Noll 2017). In any case, the results in this paper indicate that Australians’ trust towards their neighbours, strangers and people in general remains very high by global standards, at least for now.

The paper also contributes to our understanding of social trust during episodes of political crisis. We can now add Australia to the growing list of countries which have experienced declining institutional trust yet no decline in social trust. It is becoming increasingly clear that two of the dominant explanations for social trust – social capital theory and institutional theory – may need to be refined to account for this phenomenon. Caïs, Torrente, and Bolancé (2021) suggest that the political crisis in Spain led to the development of new left-wing political movements which, although less trusting of institutions, helped to foster greater trust among ordinary people, and this is why social trust did not decline alongside institutional trust. While this may have been the case overseas, no left-wing movement along the lines of Movimiento 15M emerged in Australia, which means this explanation is of limited utility for explaining why social trust stays high. A more appropriate refinement of the social capital and institutional theories might be to suggest that institutional trust does affect social trust, as reported by Tao et al. (2014) and Dinesen et al. (2022), but mainly during childhood and adolescence. This would explain why country-level social trust fails to respond immediately to changes in institutional trust and would be consistent with recent studies showing that long-term changes in social trust are mostly due to early-life factors (Kumove 2024; Uslaner 2002; Wu 2020). It is also consistent with the birth-cohort trends in Figure 5.

Lastly, the paper contributes to the study of Australian social trust, a topic which has attracted surprisingly little attention from researchers. This appears to be the first study which addresses how Australian institutional trust is related to social trust. Although a handful of studies have tested various determinants of social trust in Australia (Brown, Hoye, and Nicholson 2014; Chong et al. 2011; Leigh 2006; Mendolia, Tosh, and Yerokhin 2016; Viitanen 2014), only two appear to have tested how political factors are linked to social trust (Bean 2005; Tranter and Skrbiš 2009) and these did not consider institutional trust or related attitudes. Future studies may wish to examine how Australian social trust is related to other political variables, such as rates of voting, protesting or party identification. Even if experiences with the political system do not affect social trust during adulthood, this does not preclude the possibility that certain types of political socialisation in early life may have an enduring effect on Australians’ social trust levels.

Acknowledgements

I am grateful to Daniel Casey, Ian McAllister and the attendees at the 2023 Australian Political Studies Association conference and 2024 POP workshop for their feedback, and to the two anonymous reviewers whose comments substantially improved the manuscript.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by funding from Norges Forskningsråd.

Notes on contributors

Michael Kumove

Michael Kumove is a Postdoctoral Fellow in political science at the Norwegian University of Science and Technology (NTNU). His research focuses on political behaviour and peace and conflict studies, examining questions related to social trust, ethnic conflict and intergroup attitudes.

Notes

1 This was generalised trust, which asks respondents whether ‘most people can be trusted, or if you can't be too careful when dealing with people’.

2 The AES, WVS and HILDA surveys were chosen because they offered a comprehensive range of different measures of social and institutional trust. Other surveys, such as the ANUpoll and Australian Survey of Social Attitudes (ASSA) also have some data for these variables, although ANUpoll's social trust data appears to be very limited and ASSA has only one type of social and institutional trust, both of which already appear in the AES.

3 Although only the second of these items explicitly mentions ‘trust’, all three were used as indicators of institutional trust by Dassonneville and McAllister (2021). The fact that all three exhibit very similar trends in Figure 2 further suggests that they all reflect some kind of underlying faith in the political system.

4 Pre-2010 rounds of the AES do not include weights, so these rounds were left unweighted.

5 The exception is the HILDA data which is measured on a 1–7 scale.

6 HILDA social trust data are on a 1–7 scale, but it was normalised in Figure 2 to make it visually consistent with the other surveys.

7 Another possibility is that institutional trust does not affect social trust, but social trust affects institutional trust. This view has attracted little support in modern trust research, although Rosenberg (1956) suggested that ‘the belief that political dishonesty is rife in a democracy may be based less upon actual knowledge of political corruption than upon the general conviction that nearly everyone is dishonest’. If true, this would explain why the two forms of trust were correlated in H₁ yet declining institutional trust failed to affect social trust in H₂. However, this theory is inconsistent with recent work which reports a causal link from institutional to social trust, including with experimental designs (Rothstein and Eek 2009; Tao et al. 2014; Dinesen et al. 2022).

8 After all, as noted above, some of these countries experienced ‘broad’ declines where trust in both party-political and impartial institutions declined (Ervasti, Kouvo, and Venetoklis 2019). Similarly, in the USA, the Covid-19 pandemic saw declines in trust towards both party-political institutions such as the federal government and the president as well as impartial institutions such as the Centers for Disease Control and the media. Once again, however, social trust held steady despite this ‘broad’ decline in institutional trust (Aassve et al. 2024).

9 Another refinement of the two theories could be that individual-level trust dynamics differ from the country-level shifts which happen during political crises. Perhaps during times of political crisis where others are seen lowering their institutional trust, many people reduce their own institutional trust because of these social cues, even though their underlying attitudes towards institutions relevant for ‘signalling’ and ‘enforcement’ remain largely unchanged. If so, then there would not be any subsequent effect on social trust. Conversely, during ‘normal’ non-crisis periods these social cues are absent and so reductions in institutional trust are more likely to reflect a genuine loss of faith in political institutions, consistent with the link identified by Dinesen et al. (2022). This would be a fruitful topic for future studies.

10 The ten countries in WVS Round 7 with higher generalised trust than Australia were: Austria, China, Denmark, Finland, Iceland, the Netherlands, New Zealand, Norway, Sweden and Switzerland.

11 Only HILDA data are used in this figure because the AES and WVS surveys do not have a sufficient range of years to identify trends for the 1990–1999 birth cohorts. HILDA also offers a much larger sample size.