Can We Trust the “Society of Distrust”?

The global crisis, both financial and economic, that broke out in the spring of 2007 has placed the themes of confidence and trust at the centre of world public debate: trust in interbank relations, confidence of households and firms in the future, trust of markets in the signature of the public authority. [1] From this verbal profusion has emerged two series of general questions: the importance of confidence and trust, which appears to be considerable in “market democracies” [2], is manifested by their disappearance; in addition, it seems difficult to determine exactly what is included in those notions, whose malleable usage appears to mask great confusion. Trust and confidence distinguish themselves by their absence and remain elusive when they exist.

Contemporary research on trust in Europe has focused on the case of Italy for diverse reasons, the most important of which probably being the strength of regional contrasts. The issue of trust in the French society, even though it was briefly mentioned by Putnam (1993) [3], had not been the object of a true study until Algan and Cahuc’s work, La Société de défiance: comment le modèle social s’autodétruit. Hence, this work is the first attempt to cast light on the question and, in that capacity, its ambition deserves to be praised, the subject being of importance. However, Algan and Cahuc’s study suffers from important theoretical and empirical defects that compromise its scientific significance. To begin with, we will rapidly call to mind the argument of this concise text, which is freely available in its electronic version [4], before examining its analytical framework and the methodology the authors employ.

The Meaning of the Words: Lack of Trust Is Not Distrust

Let us start from the observation posited by the authors. The French society is characterised by an exceptionally weak level of “mutual trust” and “civism”, trust being measured principally by the rate of positive responses [5] to the question of generalized trust in the World Values Survey (WVS). “France […] is ranked 58^th among 82 countries, only surpassed by states that are much poorer or that have experienced armed conflict”, the authors write. They nevertheless describe this undeniable reality in an incomplete manner: to be sure, France is situated at a weak average level of generalized confidence, but it is to be found in reality near the world average in 2000, the year the authors take as a reference, from which it is only separated by a difference of 25% (28.4% on the average for sixty-four countries in the fourth wave of the WVS, as opposed to 21.4% for France).

Yet, the central problem of the study, which appears from the first pages onward, is not there. Since their analyses, hypotheses and recommendations critically depend on it, we would have wished that the authors had begun, at the opening of their demonstration, by two preliminary steps: the first aiming to warn the reader of the theoretical fragility of the concept of generalized trust; the second, forewarning him as to forming excessively indulgent interpretations of results that are heavily dependent on the subjectivity – variable over time and space – of individuals polled only about their opinion on the subject of a question whose very preciseness and signification themselves are controversial and contested.

Now, the way in which the authors treat the question of generalized trust is much more problematic than an absence of warning. Throughout the study an erroneous French translation of the question in the WVS is repeated: “En règle générale, pensez-vous qu’il est possible de faire confiance aux autres ou que l’on est [sic] jamais assez méfiant?” The question asked by the WVS [6] in English is the following: “Generally speaking, would you say that most people can be trusted, or that you can’t be too careful dealing with people?” A possible translation might have been: “D’une manière générale, diriez-vous que l’on peut faire confiance à la plupart des gens ou que l’on n’est jamais trop prudent dans ses rapports avec autrui ?” The authors’ translation is therefore truncated. Moreover, it is, in addition, incorrect.

The key to the error in the French translation is the term “careful”, which cannot reasonably be translated as “méfiant”. “Prudent”, “avisé”, “réfléchi”, even “reserve” would have been acceptable possibilities. However, “méfiant” [mistrustful] is obviously an error in translation. [7] The official French translation proposed in the questionnaire of the WVS, which is to say, in reality, the question to which surveyed Frenchmen responded, was logically the following: “D’une manière générale, diriez-vous qu’on peut faire confiance à la plupart des gens ou qu’on n’est jamais assez prudent quand on a affaire aux autres?” Yet, this mistake between “prudent” and “méfiant” permeates the whole study. It can be found again in the form of the observation that “Frenchmen are more mistrustful, on the average, than most inhabitants of other developed countries”. Since this statement is based on a translation error, it is itself marred by error, as is each usage the authors believe they can make of the words “méfiant” [mistrustful], “défiant” [distrustful], “méfiance” [mistrust] or “defiance” [distrust], beginning with the title of their work – except by straining the interpretation of the opinions expressed beyond the sense of the words used by the respondents.

However, it seems that we are dealing here with more than a simple error in translation: it is a conceptual error. As numerous authors emphasize, and notably those that have participated in the Russell Sage Foundation’s project on trust, the contrary of trust is not distrust or mistrust; it is the absence of trust. And, supposing that generalized trust is a robust concept and the WVS can truly measure it (two controversial conditions), the interpretation corresponding to the observation of a weak rate of positive response to the WVS’s question of generalized trust is that the society under study is characterized by an absence of declared trust, or of a greater prudence. France may thus be, eventually, a “society of prudence”, not a “society of distrust”, which is appreciably different. In any case, nothing in the data bearing on the trust of the French presented by the authors supports the finding of “distrust”.

In addition, and still on the theme of the analytical framework and of conceptual rigour, generalized trust is not to be confused with trust in institutions, yet the authors constantly bring the two together to form the block of “mutual distrust”, to which they also aggregate toward the end of the study distrust directed at the market or competition. “We are dealing with a global phenomenon”, they write, but this phenomenon of generalized social distrust has never been detected in the literature until now, and practically speaking, nuances do happen to exist that would be interesting to describe and analyse in the answers the French give to value surveys bringing the various forms of trust into play. Furthermore, while trust in institutions, Parliament, the judicial system and labour unions is in fact summoned to appear in the introduction, it then takes second place to bivariate correlations mostly exploiting generalized trust. Generalized trust is thus called upon to represent all types of trust, and notably trust in institutions, while precisely it is theoretically and empirically distinguishable from it.

Civism and Trust: Distinguishing the Concepts

Let us suppose nevertheless, so as to understand the authors correctly, that the weak rate of positive response to the question of generalized trust in the WVS survey in France, which is a fact, is the sign of “mutual distrust” among the French, where would it come from? The posited relation between (generalized) trust and civism seems clear in the eyes of Algan and Cahuc: “In all logic, available information indicates therefore that the French distrust each other more because they respect the rules of life in society less than the inhabitants of other wealthy countries”. The explanation selected for justifying the weakness of generalized trust in France is thus “incivism”: the French are said to be less civic, hence less trusting, in the sense of generalized trust. This relation is, however, rendered nebulous from the beginning by formulations that are, actually, prudent: “The observed distrust in France is probably the quid pro quo of behaviour that is truly less civic than in other countries”. Should “quid pro quo” be understood in the sense of “explanatory variable”?

Let us admit this imprecision to get to the heart of the matter: the theoretic framework built for the study. In the literature bearing on generalized trust, the civic explanation is the oldest and the closest to the work of Putnam (1993), but it is also the least empirically certain, not to mention the theoretical criticism addressed to it, and first of all by the supporters of the theory of generalized trust. A rapid examination of the multivariate correlations of the responses in the European Social Survey [8] with respect to generalized trust and civism indicate that civism, at least in the minds of the respondents, plays an insignificant role in comparison with their level of trust. [9] Moreover, no global empirical study exists to date, which is to say one that tests all the hypotheses in presence, of generalized trust that confers importance or significance to the civic variables. In spite of this, the authors affirm, relying on a single bivariate correlation (Chart 8), that “people are all the more mistrustful with respect to others that they live in a country in which their fellow citizens display a lesser civic sense” (we find again, here as elsewhere, the error in interpretation of the term “careful”) and that “trust is very strongly correlated with public-spiritedness”. This observation, which relies on very partial data, cannot be understood as implying that civism (or incivism) plays a significant role in the level of generalized trust, in France or elsewhere.

It may well be that the French are less civic than others, and the authors relate anecdotal evidence of this, but, given the data presented and the current state of the literature, nothing permits the affirmation that they are less trustful due to this fact. Yet, “distrust” and “incivism” mixed together are made responsible for very great evils: “Distrust and incivism in France do not constitute a new deal. They have persisted for at least two decades. As an example, the deficits of trust and civism have in fact accompanied the mediocre performance of the French economy over this period”. The authors imply here that there is a link between the French economy’s “mediocre performance” and the “deficits in confidence and civism” that they believe they have identified, but they do not propose any empirical method to test their hypothesis and compare it with alternative explanations, a hypothesis that remains for this reason at the stage of pure speculation. Even the observation of the French economy’s “mediocre performance” “for the last two decades” is not at all self-evident.

The History of Trust’s Variations: the Uncertainty of Measurement

Let us come now to what constitutes the greatest innovation of the study: the historical analysis of French confidence throughout the XXth century. The authors present their thesis without the conditional verb tense: “Confidence seemed much stronger before the Second World War; it has in all likelihood deteriorated since then. The defeat of France and the schism of the Collaboration had undoubtedly favoured this evolution. However, it was especially the establishment of a corporatist and statist social model that undermined trust: by establishing statutory inequalities, the French State contributed to the erosion of solidarity and mutual trust”.

From an historiographical perspective [10], as Delalande remarked, Algan and Cahuc’s thesis conflicts with the memory of the Liberation as a “moment of reconstruction of the society” by depicting it, on the contrary, as an “episode of destruction of the allegedly ‘harmonious’ social relations during the inter-war period”. It is indeed altogether surprising to be invited to imagine the establishment of the Social Security laws – more inspired in fact, contrary to what the authors assert, by the National Council of the Resistance (CNR) than by the Vichy regime – as more harmful to the mutual trust among Frenchmen than the economic and political stagnation of the 1930s on the one hand, and on the other, of the military defeat and German occupation, marked by the persecution and deportation of a part of the population and by violent and profound civil divisions. And nevertheless, the authors affirm that the Liberation and the establishment of the Social Security laws supposedly have “especially” undermined trust among the French – the “defeat of France” and the “Collaboration” having only “favoured this evolution”. The historical reversal is staggering. Should we not rather concur with Delalande when he writes: “By the consensual character they assumed, the reforms of the Liberation (nationalisations, Social Security, works councils, etc.) enabled France to ‘remake society’?”

However, let us once more admit the argument. How do the authors justify their historiographical audacity? To do so, they used a property Putnam already identified, and which seems to be robust in the long term: the relative stability over time of the levels of trust that is hypothetically due to their cultural determinants and to their profound historical roots. Without this hypothesis, indeed, the study of the levels of generalized trust over time in the case of France would stop at the point of departure of the first wave of the World Values Survey (WVS), which is to say in the beginning of the 1980s. However, the American General Social Survey (GSS) makes it possible to go farther back in time: to the end of the 1950s, which enables the authors to put forward an interpretation of the findings of this latter survey, which authorizes them to formulate conjectures about the level of generalized trust of the French since the 1930s.

Let’s be very clear straight away that before the 1980s, no survey data on the generalized trust of the French are available, nor any data on the generalized trust of Americans before the 1950s. Let us also specify that the authors employ the GSS data for the sole period 1977-2004, but that, according to them, “it is possible to reconstruct the evolution of social attitudes in France thanks to the evolution of immigrants to the United States who came from different countries”. “If we find that the descendents of French immigrants in the beginning of the XXth century have a level of mutual trust that is higher than the descendents of immigrants coming from other countries in the same period,” they write, “it is in all likelihood that the French immigrants who arrived in the beginning of the XXth century were more trustful than those from other countries”. The method, baptised “epidemiological approach” to generalized trust, seems at first sight very improbable: it consists of inferring from the level of generalized trust of Americans from the end of the 1970s the level of generalized trust of Frenchmen of the 1930s for the purpose of drawing historical conclusions about the evolution of trust in France in the course of the XXth century. The double interference of time and space when dealing with such uncertain data does not seem to discourage the authors – much to the contrary.

They make clear moreover, while on the subject of the GSS, that “this survey presents the advantage of asking exactly the same questions about civism and mutual trust as those we have used up to now” [the data from the WVS]. Yet, this is not the case for what concerns generalized trust, the question in the GSS being formulated differently from the question in the last wave of the WVS [11] (the same approximations is repeated by the authors in the other articles developing their “epidemiological approach” to civism or generalized trust [12]). The nuance may seem minor and even negligible, but since data dealt with here are opinions tinged with the subjectivity of the person polled, “you need to be very careful in dealing with people” and “you can’t be too careful dealing with people” do not objectively have the same meaning – neither in English, nor in French, nor, it may be supposed, in the different languages [13] of the countries where the WVS is carried out, [14] which means that a difference in formulation of the question asked is added to all the “noise” already identified.

After having arrived at the conclusion of an historical evolution of mutual trust in France during the XXth century, the authors ask themselves: “To when exactly does this reversal of attitudes date back?” Yet the matter becomes more complicated methodologically since the “General Social Survey does not indicate the year of arrival of the parents of second-generation Americans”. Not a problem for Algan and Cahuc: “It is possible to identify periods of rupture in the evolution of social attitudes by relating them to historical events”, for example, to the “Second World War”. Yet nothing specifies the historical basis of the authors’ choices, which at this stage become purely arbitrary. “To corroborate this hypothesis”, the authors compare “the social attitudes of second-generation Americans born before the Second World War and born after the war”, and affirm thereby being able to “reconstitute a large part of the mutual trust that prevailed in the country of origin before and after the war, and which was transmitted by the parents in function of their wave of immigration”. The pivotal moment in the analysis is therefore the “Second World War”, without semblance of further detail.

The authors then unveil their fundamental finding: “A comparative study of this sort brings to light a gripping difference between the two waves of immigration, before and after the war”. Surely enough, “all else held constant, Americans whose parents immigrated before the war had an 8% higher probability of trusting others when they originated from France rather than Sweden. The image is completely reversed for the Americans whose parents emigrated after the war, the level of mutual trust for Americans of French origin being 13% lower than the level of Americans of Swedish origin”.

Institutional or Cultural Analysis of Trust?

“The lack of trust and of civism of the French is thus not immutable”, conclude Algan and Cahuc, who undertake in the remainder of the work to support their theory by the analysis of “certain characteristics of French institutions put in place after the war”, which “in all likelihood have contributed to the emergence and perpetuation of that state of fact”. Here, the authors have opted for an interpretation that owes nothing to their economic analysis or to their statistical apparatus, but everything to their intuition or their beliefs, none of the abundant historical research on the period of the 1930s or on the Second World War being mobilized to make the Liberation or the enactment of Social Security laws the moment of rupture for the French, in preference to any other.

The finding presented as the one on which rests the whole edifice the authors elaborate in their attempt to reconstruct the historical evolution of generalized trust in French over the course of the XXth century is thus formulated in the following manner: a differential of 21% [15] between the two different groups of Americans at the end of the 1970s, whose parents had arrived respectively from France and Sweden before and after the war, in the probability of responding positively to the question: “Generally speaking, would you say that most people can be trusted, or that you can’t be too careful dealing with people?” All the rest of the authors’ argumentation flows from this result, since it is in its name that they undertake in the remainder of the study to comprehend the “rupture” in trust that dates, according to them, from the Liberation (the choice of that date, which they endeavour to justify a posteriori being purely discretionary at the moment of the study where it is formulated, except if the logical order of the demonstration were to be inversed). To say that the solidity of their hypothesis is not quite assured is an understatement.

This does not signify for as much that the idea of intergenerational transmission of social capital, and more specifically of trust, and still more precisely of generalized trust, has no foundation, if only one adheres to the cultural approach of generalized trust. The central idea of this approach is that generalized trust is a stable value over time, varying little. Eric Uslaner affirms that “when immigrants from a trusting country come to their new homes, they will carry on their cultural traditions of trust rather than simply ‘adapt’ to the new realities of their adopted environment”. [16] However, Uslaner’s ambition is much more modest that Algan and Cahuc’s, even if, as they do, he attempts to determine “Which matters more: Whether your ancestors came from a trusting society or whether you live among people who are likely to be trusting?”

He indeed stops before the obstacle of the unavailability of data: “It would be nice if we could match the levels of trust in the home countries when grandparents immigrated to the United States with contemporary estimates of how trusting people are in Sweden, Italy, Germany, or Latin America. But we can’t. There were no public opinion surveys in the 1890s or 1920s, so there is no firm way to establish a direct link between grandparents’ homeland experiences and their successors’ beliefs in the United States”. And it is he the most ardent supporter and best representative of the cultural approach to generalized trust who expresses himself in this way. What Uslaner compares, and what at first sight seems comparable, is the influence of the ethnic origin and the influence of the social milieu in which the immigrants are immersed on their arrival. Under no circumstances does he try his hand at inferring socio-historical conclusions over a century of history in the new Americans’ countries of origin. Yet, one of Uslaner’s results does concern us here: according to him, examination of the GSS data shows that, contrary to other ethnic origins, no significant effects are attached to Russian, Italian or French origins. Otherwise stated, according to his econometric analysis, the GSS data do not permit the detection of the least effect on the level of generalized trust of ethnic origin for the Americans of French ancestry, which weakens the thesis of Algan and Cahuc still somewhat more.

In addition to the great uncertainty surrounding the authors’ empirical methods for estimating the evolution of generalized trust declared by the French over time, their argumentation conceals an internal contradiction: why, if trust among the French is fundamentally stable, would it have abruptly plummeted for afterwards to stabilise again during five decades at a weak level? Two approaches to generalized trust are amalgamated here: the cultural approach (the transmission of trust is through the family, and institutions only play a limited, even nonexistent role) and the institutional approach (the public authority in the broad sense has the means to cause variations in generalized trust, by either increasing or decreasing it). This theoretical mixture has an ad hoc appearance.

An historical parallel does exist in the literature that may perhaps have served as a reference for the authors: the case of the Soviet Union. We know nothing of the level of generalized trust in various countries of Central and Eastern Europe prior to their annexation by the USSR, but we do know that the level of generalized trust in these countries is still particularly weak in our time and that some authors relate this situation to the socially destructive and persistent effects from the era of the communist regime. But precisely, how can one relate France’s Trente Glorieuses (1954-1975, the height of the “society of distrust” according to the authors since afterwards, in the 1980s, 1990s and 2000s, the French social model has become more hybrid by integrating “beveridgian” elements)? How can the economic and social performance of France be explained, for example, in comparison to countries of the Nordic model during this period, when the French model was even more corporatist and statist than it is today? Here, the explanation the authors give becomes even more uncertain: the Trente Glorieuses allegedly “eclipsed” the social and economic cost of the French distrust, a cost that supposedly only became evident from the middle of the 1970s onward. The last-ditch recourse to the received wisdom of the end of the “Fordist model” does not seem very convincing and appears above all quite distant from the scheme of the authors’ analysis.

Likewise, the description of the conservative-corporatist model as creator of social segmentation and less egalitarian than the universal Beveridgian model of the Nordic countries and the idea that the French welfare state model is a product of hybrid logic in which the State and the social partners are in an ill defined relationship of variable geometry that impairs the quality of social dialogue, are in conformity with existing research on the subject, but the role played by generalized trust, at once upstream and downstream of this process, is altogether unclear.

Stated otherwise, between the beginning and the end of the study, the authors’ theoretical model seems to have changed: while the lack of civism of the French was made responsible for their distrust at the beginning of the work, toward the end, it is the French social model put in place in the post-war period that explains this distrust (“the deficit of trust is intimately linked to the functioning of the State and to the social model”). Unless it is considered that the functioning of the State and of the social model engenders the lack of civism of the French (which cannot be determined from the data presented in the study), it would seem that we are dealing with a new internal contradiction – all the more so that, once more, these two explanations borrow from two models that do not coincide in the typology of existing literature on generalized trust: the explanation by civism and the institutional explanation. When is added the fact that the passage between the 1930s and the Liberation to a “Society of Distrust” in France seems also to mobilise a cultural explanation, it is the entire framework of the study that becomes dubious.

The French and the Market: Sophism and “Economic Moralism”

The part of the work bearing on the difficulty of French social relations is probably the most convincing, but it is tempting to link it to existing research, such as Thomas Philippon’s studies. But it is a subject, theoretically and empirically, distinct from the issue of generalized trust in the French society. However, at least one point in the argumentation is worthy of discussion: the one of the necessary evolution of the French social model toward the “Nordic model”, which is affirmed here as in numerous studies. Let us only point out that this evolution is not all that simple and supposes, notably, that the size of the country and corresponding growth strategy be taken into consideration. In addition, Rothstein shows that at the origin of the “puzzle” of the Nordic societies’ adoption of a universal model of social protection can perhaps be found a “neo-corporatist” model, combination of statism and corporatism, the very mixture that Algan and Cahuc consider responsible for all the French social and economic ills. [17]

The passage dedicated to the “fear of the market” and “competition” falls into the weakness already identified of strong affirmations backed up by weak empirical elements and great theoretical indetermination. As an example, the idea that “the efficacy of the market is to a great extent founded on mutual trust” proceeds from a limited or particularized form of trust, but the authors attempt to support it with help from data on generalized trust. The authors’ theoretical shortcuts occasionally border on sophistry, notably when they write: “Mutual trust and its quid pro quo, the capacity to respect ones’ commitments, seem to play a decisive role in market efficacy. It is therefore logical that distrust toward the market be stronger in the countries in which mutual trust is little developed. From this point of view, the distrust of the French toward the market and competition may be connected to the weak trust of the French toward others”. In reality, the logic of the reasoning is specious, and none of the sequential steps in this sentence is certain because limited trust and generalized trust are merged, and because distrust is interpreted as the contrary of trust. So it is that the authors seem to say that because the measured level of generalized trust is low in France, the distrust with respect to the market must be strong, which would explain that interpersonal economic transactions, which precisely rely on limited trust, would supposedly be made more difficult. The interpretation of the responses of the French on the beneficial or harmful character of competition as denoting a “fear of the market” is likewise questionable.

Taking into account the reservations expressed up to this point, the affirmations put forth to close the study, they too founded on the “epidemiological approach” to generalized trust to which is added for the circumstance the uncertainty of a counterfactual approach to economic development (“economics-fiction” of some sort) seem hyperbolical: “Thus, our deficit in trust explains 66% of our gap in per capita income in comparison with Sweden. The French GDP would be increased by 5%, which is to say an increase of nearly 1,500 euros per person if the French had the same trust with respect to their fellow citizens as the Swedish”. To the knowledge of the author of this review, not a single empirical study exists founded on a model of development that attributes two thirds of a difference in growth to only one identified (non-residual) variable, a fortiori not to a difference in the level of generalized trust.

The theme of Algan and Cahu’s study is without doubt interesting, but their application of partial empirical methods to uncertain data in an indeterminate theoretical framework to draw radical conclusions considerably diminishes its scientific import. The lingering feeling that the authors are in fact speaking about something else (moreover ill-assorted) than trust in France, never leaves the reader. In any case, at the term of this assessment, the idea of a French society gnawed by generalized distrust, of the French mistrustful of everything and everybody, does not seem to have either a precise theoretical meaning, or a solid empirical basis. On the other hand, it conceals a paradoxical moralistic backdrop: the ultima ratio of the moral whose orders are to be followed here is economic efficacy. [18] Trust in France [19] is most likely a mystery, but it remains mostly intact once Algan and Cahuc’s work closed.

Translated from the French by Ken Ritter