Measuring a state, reading a trajectory.

The assessment of sovereign risk poses two distinct questions that financial institutions seldom confuse through carelessness, but often through a lack of proper instrumentation. The first concerns long-term solvency: will a state, over a horizon of several years, remain capable of honouring the debt it issues today? The second concerns trajectory: in what direction, and at what speed, are the structural factors of sovereign risk moving? These two questions appear close in kind. They belong, in fact, to different analytical disciplines, one to the stabilised measurement of a condition, the other to the directional reading of a movement. The first has found, over the past century, its answer in sovereign credit rating as practised by the three major international agencies and their peers. The second remains, within the European institutional ecosystem, insufficiently equipped.

Sovereign credit rating is a remarkable analytical production when considered for what it is. Built upon a century of observation of sovereign defaults, supported by tested probabilistic models, subject to standards of methodological transparency in continual improvement, it provides bond markets with a common grammar without which the international allocation of capital would, in practice, be unworkable. The institutions of Basel have integrated it into their prudential frameworks; European regulators have framed it without calling it into question; investment committees mobilise it as a standardised input within a broader decision. Its usefulness is not in dispute. It constitutes an acquired asset whose value far exceeds the episodic criticisms it attracts.

A further observation, no less well established, nevertheless deserves serious attention. Carmen Reinhart demonstrated as early as 2002, in a study that the specialised literature treats as a reference, that sovereign ratings rarely precede banking and currency crises. They tend, rather, to register them after the fact, sometimes with a delay of several quarters or even years. The Securities and Exchange Commission restated this finding in July 2020, at the height of the pandemic, by describing rating actions as lagging indicators of the cost of capital. The Bank for International Settlements and the Agence Française de Développement have, on several occasions, documented the same property across other data sets. None of these works reproaches the agencies. They describe, with increasing precision, what sovereign rating is and what it is not.

The nature of the instrument explains this property. A rating is, by design, a measurement built for stability over time. It seeks to protect the investor from excessive informational volatility, which justifies its measured pace of response to changes in fundamentals. This characteristic, methodologically embraced by the agencies under the term through-the-cycle rating, is claimed as an advantage and is indeed, for the purpose to which it is dedicated, a real advantage. It carries a direct and inescapable consequence. An instrument that produces a stable measurement of an average state cannot, simultaneously, produce a sensitive reading of a movement in progress. These two requirements are, in measurement theory, mutually exclusive.

It is upon this observation that the present analysis rests. European financial institutions engaged in sovereign allocation decisions hold today a first-rank instrument for answering the first question, that of long-term solvency. They hold a more fragmentary instrumentation, more dependent on internal or outsourced analyses, for answering the second question, that of trajectory. Private strategic analysis finds in this space its proper object, not in order to contest an instrument whose usefulness has been demonstrated, but in order to propose a complementary discipline whose function is of a different kind. Three movements articulate what follows. The first specifies what sovereign rating, by its very construction, cannot deliver. The second outlines the principles of a reading of movement, requiring different sources, different weightings, different standards of traceability. The third examines what this articulation, between the measurement of a state and the reading of a trajectory, signifies for the governance of an institutional decision exposed to sovereign risk.

To understand what sovereign rating does not deliver, one must first understand what it does deliver, and the manner in which it does so. Three architectural properties, claimed by the agencies themselves in their published methodological documents, structure the instrument. They define its strength for the purpose to which it is dedicated and, by direct consequence, the questions it cannot address.

The first of these properties is through-the-cycle rating. The agencies, in their official methodological frameworks, state explicitly that they seek to assess the solvency of a sovereign issuer not at a given moment in time, but across the span of a full economic cycle. This stance, formalised under the term through-the-cycle rating, aims to neutralise the cyclical fluctuations that might otherwise produce rating changes which the real evolution of solvency would not warrant. For prudential use and for the stability of the Basel regulatory frameworks, it is a substantial advantage. The International Monetary Fund, in its Global Financial Stability Report of October 2010, documented with precision the consequences of this property. The temporal smoothing of assessments contains, mechanically, at the moment when reality eventually catches up with convention, an adjustment amplitude that the literature designates by the term cliff effect. The longer the rating has deferred the recognition of a latent deterioration, the more brutal the final realignment will prove. This property is consubstantial with the method. It is not a weakness to be corrected; it is the necessary obverse of an assumed virtue.

The second property concerns the articulation, within the rating exercise, of a quantitative dimension and a qualitative one. No serious model of sovereign analysis reduces itself to an algorithmic function of accounting indicators alone. Institutional variables, the quality of governance, political stability, legal predictability enter into the assessment on the same footing as the ratio of debt to gross domestic product or the current account balance. This element of judgement is not only legitimate, it is necessary: no state can be read solely from its accounting fundamentals. But judgement, within the context of rating, exerts itself under a constraint of defensibility. The analyst must be able to justify his rating before the internal committees of the agency, before regulators, before the issuers who contest unfavourable assessments. This constraint produces a particular cognitive economy, one that favours prudence, reproducibility, and convergence with the assessments of peers. Andreas Fuchs and Kai Gehring demonstrated in 2017, in the Journal of the European Economic Association, that this cognitive economy comes with a geography of its own: ratings reflect a cultural proximity between the analyst and the rated object, observable empirically across a wide corpus. This property, here again, does not constitute a fault. It belongs to the very nature of an institutionally framed rating apparatus. It does, however, limit the instrument's capacity to register the inflection points that occur precisely at the margins of the dominant convention.

The third property bears on the horizon of measurement. A sovereign rating addresses, by construction, solvency over several years. It interrogates a state's capacity to honour its debt within a long temporal window, which is the horizon of institutional bond investors, the principal audience of the instrument. This long horizon grounds the legitimacy of rating as a reference point for strategic allocation. It also grounds a relative indifference to movements of short and medium duration. A profound transformation of a state's political landscape, a recomposition of its strategic alliances, a shift in economic doctrine, an accelerating deterioration of confidence in its institutions may all occur without modifying, within a timeframe useful to the investment decision, its accounting fundamentals. The rating looks far, and that is its strength. It dispenses, in the same gesture, with looking near.

These three properties are not three methodological choices that could be reversed without consequence. They form the architectural core of a discipline that has been constituted as such, over a century, in order to produce a stable, defensible, and publicly usable measurement of sovereign solvency. To read a state's trajectory, by contrast, demands an instrumentation built upon different principles: heightened sensitivity to movement, hierarchical weighting of non-accounting factors, a shorter horizon, and different standards of validation. These principles do not contest the rating, they occupy its blind spot.

To outline a discipline is not to expose its procedure. It is to articulate its founding principles, those that orient concrete choices without exhausting them. The reading of sovereign trajectory rests, in the practice of those firms seriously engaged on this object, upon three articulated principles, each of which responds, by displacement rather than by opposition, to one of the properties set out above.

First principle: the reading of movement rather than the measurement of the state. An anticipative discipline interrogates less the level of sovereign risk at a given moment than the second derivative of that level, that is, the speed at which it evolves and the acceleration of that evolution. The question is not only where is this state today, but at what speed, and in what direction, are its foundations now displacing themselves. Carmen Reinhart and Kenneth Rogoff devoted the substance of their book This Time Is Different, published in 2009, to demonstrating that sovereign crises follow trajectories that are observable over the long run, with leading signals discernible quarters and sometimes years before rupture. These signals are not mysterious. They reside in the trajectory of foreign-currency debt, in the evolution of the current account balance set against the composition of its financing, in the progressive deformations of the maturity structure of public debt. A reading that attaches itself to these derivatives rather than to instantaneous levels does not supplant the rating, it complements its angle of view. It answers a question that the stable measurement does not pose.

Second principle: the hierarchical weighting of structural factors beyond the accounting indicators. Recent sovereign ruptures documented by the literature, from abrupt political transitions to collapses of institutional confidence, have not been preceded chiefly by accounting deteriorations. They have been preceded by transformations in the structural balance of forces that underpin the solvency of a state. Daron Acemoglu and James Robinson, in Why Nations Fail, published in 2012, systematised the proposition that the long-term solidity of a state is read in the inclusive or extractive nature of its institutions, well before it is read in its accounts. Robert Blackwill and Jennifer Harris, in War by Other Means, published in 2016 by Harvard, demonstrated how the economic use of state power reconfigures the conditions of sovereign solvency according to logics that the classical accounting grid does not capture. Jean-François Bayart, in his now-classical work on the political modes of state functioning, has long established that the routes of failure in fragile states pass through clientelist and patrimonial channels whose accounting trace arrives only in due course. An anticipative discipline weighs these dimensions on equal footing with the economic indicators. It treats politics and political economy as explanatory variables in their own right, not as qualitative context appended to the figures.

Third principle: the epistemic traceability of sources. A discipline that claims to anticipate without being able to document its sources reduces itself to an opinion in dress. Western military intelligence has developed, over more than half a century, standards for source rating that the professional literature regards as the most rigorous available today. The STANAG 2511 standard of the Atlantic Alliance formalises a two-dimensional rating: the reliability of the information producer, graded from A to F, and the plausibility of the content reported, graded from 1 to 6. This double rating, inherited from the older practices of intelligence and adapted to contemporary requirements, makes it possible to distinguish a well-sourced but implausible piece of information from a plausible but poorly-sourced one, a distinction that current commercial methods frequently conflate. Richards Heuer, in Psychology of Intelligence Analysis, published by the Central Intelligence Agency in 1999 and now a standard of the profession, has shown that the systematic application of such sourcing disciplines substantially reduces the cognitive biases that affect unframed analytical judgements. To transpose these standards to private strategic analysis is not an academic refinement; it is the condition for a production in which every assertion can be traced back to its source, rated by its reliability, and defended independently of the subjective conviction of the analyst who produced it.

These three articulated principles define a discipline distinct from sovereign rating, complementary to it within the institutional decision, and operative upon a century of professional intelligence practice transposed to private analysis. Arden Cole has formalised this transposition under the name STRATUM, a method that organises its seven operational phases in the production of notes addressed to European institutional decision-makers. The method is in no way unique in its inspiration, which draws upon the Western tradition of structured intelligence. It is singular in its application to the field of European sovereign risk, where few comparable instruments exist to this day.

The usefulness of an analysis, in sovereign matters as elsewhere, is not measured by the sophistication of its method but by its capacity to illuminate a decision. The three principles set out call for, when applied to the actual functioning of European investment committees, three implications of governance that serious institutions gain from making explicit in their internal policy. These implications are not tactical prescriptions. They are questions that any decision-making apparatus exposed to sovereign risk has an interest in posing to itself, and to which the quality of its answers fixes a substantial part of its robustness.

The first implication concerns the place that the institution accords to official rating within its decision-making process. The rating is useful, as has been said. It does not suffice. An investment committee that treats the rating as a conclusion of analysis, rather than as a standardised input within a broader analysis, voluntarily deprives itself of the information that the rating does not supply. The crisis of structured products of 2007 and 2008 demonstrated, in another field, what such a confusion produces: the delegation, to a third party, of the analytical responsibility that properly belongs to the institution. The practical question that follows is simple to formulate and harder to address. How does our decision-making process articulate the official rating, which measures the state, with other readings, of which at least one bears on the trajectory? The most rigorous institutions construct a formalised answer to that question, one that does not wait for the next crisis to write itself. The others discover their exposure in the tally of losses.

The second implication concerns the value of the temporal differential. Between the moment when a sovereign deterioration becomes observable through an attentive anticipative reading and the moment when it is recorded by an official rating, several quarters typically elapse, and sometimes several years. This interval is not a technical detail; it is the zone within which the performance of significant sovereign allocations is played out. For a fund engaged on a position of several hundred million euros, a few quarters of lead or lag in arbitrage translate into spreads measured in tens of basis points, sometimes more. The practical question reformulates itself accordingly. By what mechanisms does our institution identify, within that critical interval, the inflection points that the official rating has not yet consecrated? The answer calls for sources of analysis built for that kind of reading, whether internalised within a dedicated team or contracted from a specialised firm. It calls, above all, for explicit recognition of the temporal asymmetry between the production of an analysis and its official ratification by the agencies.

The third implication touches upon the governance of disagreement. It happens, and the frequency of the occurrence is not negligible, that an anticipative reading diverges from the official position of the agencies on a given issuer. This situation places the institutional decision-maker before a dilemma that is not analytical but institutional. To act before the agencies exposes one to criticism in the event of error, because the institution has departed from the common reference point. To wait for the agencies exposes one to late arrival in the event of being right, because the institution will have acted only with the market. This dilemma is not resolved in the finesse of the analysis; it is resolved in the prior formalisation of a source-use policy. Such a policy specifies the weight that the investment committee accords to the various readings available, the conditions under which an anticipative reading may justify an action contrary to the dominant convention, and the criteria by which the performance of such decisions will be assessed after the fact. A policy of this kind is not an administrative refinement. It is the condition under which an institution can, in full awareness, use a reading of movement alongside the measurement of the state. Without it, the most rigorous anticipative analysis remains unusable, because no one is authorised to use it.

These three implications converge upon a simple observation. The quality of an institutional decision exposed to sovereign risk does not rest upon the quality of a single source, however excellent. It rests upon the architecture of the apparatus that articulates several sources, assigns them differentiated roles, and formalises the manner in which they concur to produce a composed judgement. The rating, the anticipative analysis, the internal readings of investment teams, the contextual information drawn from institutional dialogue with issuers all constitute inputs of which none suffices, and whose methodical combination produces the only level of quality that serious decision-makers can legitimately aim for. This architecture is a matter for the executive committee, not for the analyst.

The distinction between the measurement of a state and the reading of a trajectory, set out at the threshold of this analysis, extends beyond the single field of sovereign risk. It runs through the entire set of situations in which an institution must decide under uncertainty: political risk, regulatory risk, reputational risk, geoeconomic risk. Within each of these fields cohabit instruments of stabilised measurement, indispensable to the ordinary conduct of affairs, and anticipative disciplines whose function is to read the inflection points before they become matters of record. These two orders of instrument do not substitute for one another. They complement one another within an analytical architecture whose design belongs to institutional governance, not to technique.

This observation leads to a restatement of what a strategic analysis worthy of the name actually is. It is neither a substantiated opinion nor a compilation of public information. It is a production in which the share of subjective judgement is fully traced, in which every assertion is linked to a rated source, in which every competing hypothesis has been examined before the retained conclusion is settled, and in which the reader is able to reconstruct the reasoning rather than to accept its verdict. Only under that condition does the analysis become usable within an institutional decision, because it can be defended, contested, and revised according to the standards that govern the other components of a serious decision.

A rating measures one quantity. A strategic analysis measures another. The distinction is not of degree, but of kind. To recognise this difference is to restore to each discipline its proper function, and to enable institutional decision-making to rest upon each of them, in its rightful place.