The Desk Between: Syndicate Judgement and the Making of Sovereign Borrowing Costs
Executive Summary
Recent commentary on the return of expensive sovereign Eurobond borrowing has focused on the factors associated with high borrowing costs, including macroeconomic fundamentals, global liquidity conditions, investor sentiment and climate vulnerability. This literature has generated important insights into the determinants of sovereign bond yields, yet it typically begins with the coupon as an observed outcome. The institutional process through which that coupon is produced has received far less attention.
This essay examines sovereign Eurobond issuance from the perspective of institutional decision-making. It argues that investment bank syndicate desks occupy a pivotal position within the issuance process by aggregating investor demand, interpreting market conditions and recommending the final pricing of a sovereign bond. The coupon therefore emerges through a process of institutional judgement that translates assessments of sovereign creditworthiness into a market price. Although sovereign borrowing costs have been studied extensively, the organisational setting in which they are formed remains comparatively underexamined.
Drawing together sovereign debt research, market practice, World Bank guidance and organisational theory, the essay continues to develop the concept of evaluative capability that is being built through my recent essays. This refers to the capability of sovereign debt managers to understand, interrogate and negotiate with the institutional processes through which external judgements are formed and transformed into financial outcomes. The World Bank’s practical recommendations for debt managers consistently point towards such a capability being required, although the concept itself has not been explicitly articulated within the sovereign debt literature.
The essay concludes that greater attention should be paid to the institutional production of sovereign borrowing costs. Understanding sovereign debt requires more than explaining the determinants of yields after issuance. It also requires examining the organisational processes through which assessments of sovereign creditworthiness are translated into the coupons that ultimately shape market access and debt service obligations. This broader perspective positions syndicate desks within the wider architecture of evaluative institutions that mediate sovereign participation in international capital markets.
The Desk Between: Syndicate Judgement and the Making of Sovereign Borrowing Costs
Gabon returned to the Eurobond market in February 2025, with its debt pricing at a punishing 12.7 per cent yield - the highest debut yield ever recorded for an African Eurobond issue. Kenya followed a month later, raising $1.5 billion at a 9.95 per cent yield. This frontier-market debt wave continued through the second half of 2025 as international issuers, including Angola, the Republic of Congo, and Laos, pushed a further two billion dollars of high-cost paper into the market. High borrowing costs persisted into the next year, with the Republic of Congo returning in February 2026 to issue another $700 million bond carrying a 9.5 per cent coupon. While Dryden and Volz aggregate these specific transactions under a dataset tracking 'coupons,' their text confirms that figures like Gabon's 12.7 per cent represent the market-clearing yield at issuance. This essay explicitly separates coupons from yields to maintain technical precision throughout the upcoming institutional analysis (Dryden and Volz, 2026). The researchers assemble this run of issuance into an argument that deserves to be taken on its own terms: that the reopening of capital markets to developing sovereigns since 2025 has been a reopening on punitive conditions, that climate-vulnerable states pay a further premium largely independent of their macroeconomic fundamentals, and that a correction in the architecture of development finance is overdue. The data behind the claim, an updated series stretching back to 2009, is not in question here.
What is in question is a habit of framing that the article shares with most of the literature it draws on. The question Dryden and Volz ask, and that Presbitero et al. (2015) ask in a more econometric register, and that Hardy (2022) asks with reference to secondary-market liquidity, is why sovereign borrowing costs are high. Fiscal fundamentals, government effectiveness, global liquidity conditions, the jurisdiction of issuance, the presence of collective action clauses: each enters as an explanatory variable in a model whose dependent variable is a yield or a spread already realised in a completed transaction (Presbitero et al., 2015; Hardy, 2022). The coupon exists, in these accounts, before the analysis begins. It is a number to be explained rather than a number to be traced back to the room in which it took shape.
This essay asks a different and more institutional question: not why the yields are high, but where, in the sequence of decisions that ends in a signed termsheet, the number is actually produced. The World Bank’s own guidance note for sovereign debt managers offers an answer that the sovereign-spreads literature has been slow to take up. The price is set, in the first instance, on the syndicate desk of a lead-managing investment bank, in a process that debt management practice treats as a logistical stage to be worked through and that academic economics has treated, when it has treated it at all, as a black box that translates fundamentals into a coupon without residue (van der Wansem, Jessen and Rivetti, 2019). The claim explored here is not that syndicate desks manipulate sovereign issuers or extract rents through misconduct. It is an institutional claim: that the desk aggregates dispersed information, interprets investor demand, negotiates internally and with the issuer, and arrives at a recommended price through a process closer to judgement than to calculation in any narrow sense, and that this process has received strikingly little sustained attention from the disciplines that study sovereign debt.
Outcomes without an origin
The asymmetry in the literature is not accidental. Sovereign bond yields are public, comparable across countries and time, and available in the data formats that econometric research requires. Presbitero et al. (2015) construct a panel of emerging and developing economies issuing between 1995 and 2013 and regress the probability of issuance and the spread at issue on GDP per capita, growth, the current account balance, reserves in months of imports, the fiscal balance, external debt, an index of government effectiveness and participation in IMF-supported programmes. Hardy (2022) works at a finer grain, modelling the bid-ask spread and yield of individual Eurobonds against issue volume, remaining maturity, the credit rating at issue and subsequently, the governing law of the bond, and the presence and strength of collective action clauses, finding that New York law issuance raises yields for low-rated bonds and that an enhanced collective action clause can lower them by tens of basis points. Both papers are careful and their findings are not in dispute. What both share, along with Dryden and Volz, is a starting point in which the yield is a fact about the bond rather than a fact about a negotiation.
Callon and Muniesa (2005) offer a vocabulary for naming what this framing leaves out. A market, on their account, is not simply the abstract meeting point of aggregate supply and demand that neoclassical theory imagines, but a concrete arrangement of devices, procedures and calculative agencies that together perform the work of aggregation. The relevant distinction, in their formulation, runs between the abstract market that economists describe statistically and the microstructure, the actual architecture of exchange, through which particular transactions are concluded (Callon and Muniesa, 2005). A sovereign Eurobond yield belongs to both registers at once. It can be treated as a data point in a cross-country panel, and it can be treated as the output of a specific microstructure: an order book, a syndicate desk, a roadshow, a set of investor relationships accumulated over years of prior issuance. The regression literature has occupied the first register almost exclusively. The guidance literature produced for debt managers occupies the second, in far greater and more practical detail, yet rarely frames what it describes as a subject of theoretical interest in its own right.
A comparable inversion has already been diagnosed once in the study of financial evaluation, in work on the credit rating agencies that spent decades treating a published letter grade as a self-explanatory fact about a sovereign, a corporation or a security, while paying comparatively little attention to the deliberations of the rating committee that produced it. The parallel is worth naming rather than developing at length, because Eurobond pricing is not rating and a syndicate desk is not a rating committee. Both, however, are sites where a number that circulates publicly as though it simply described an underlying reality is in fact assembled through an internal process that combines dispersed information with a discretionary act of judgement, and both have been understudied for the same reason: the output is more tractable, as data, than the process that generates it.
The instruments the outcome-focused literature does examine reinforce the point rather than undermine it. Hardy’s finding that collective action clauses affect both liquidity and yield is a finding about a contractual technology negotiated, drafted and priced by the same set of institutional actors whose pricing behaviour concerns this essay: an enhanced clause, agreed between issuer, lead-managers and legal counsel well before a book ever opens, is subsequently treated by investors as information about how orderly a future restructuring would be, and that expectation is then folded back into the coupon the desk recommends (Hardy, 2022). Participation in an IMF-supported programme functions in a related way in Presbitero et al.’s model, entering as a variable capable of working in either direction, a reassurance to investors or a signal of distress depending on how the market reads it, precisely because its effect on price is not given by the fact of participation itself but by an act of interpretation performed somewhere between the fact and the yield (Presbitero et al., 2015). Both variables, in other words, are already the residue of prior institutional judgements. Treating them as inputs exogenous to a regression on the coupon is one further step in the same habit of beginning the analysis downstream of where the interesting work has already taken place.
Inside the book
The World Bank’s guidance note describes the structure of a Eurobond transaction in terms that map cleanly onto the distinction between abstract market and microstructure. Three categories of party are involved: issuers, intermediating banks and investors. Within the bank, an internal boundary separates the originator, who has direct and sometimes confidential contact with the issuer, from what the note calls the outside of the Chinese wall, comprising the syndicate desk, the salesforce and secondary market traders. The syndicate desk carries formal responsibility for new issue execution, including pricing, allocation and risk management, while the salesforce is the client-facing arm responsible for gathering investor orders (van der Wansem, Jessen and Rivetti, 2019). Before any of this begins, a credit rating from one of the three dominant agencies is treated as a precondition for international issuance, sought in a process lasting eight to twelve weeks that runs in parallel with, but prior to, the pricing work of the syndicate desk itself (van der Wansem, Jessen and Rivetti, 2019).
The book-building process through which a price is actually reached is worth describing in the sequence the ICMA Primary Market Handbook and the World Bank note lay out, because the sequence is where the theoretical claim of this essay is grounded. Ahead of any public announcement, bookrunners may conduct pre-sounding, approaching a small number of investors under conditions that formally cross them over a regulatory boundary into possession of material non-public information, in order to gauge demand and calibrate an initial price range (ICMA, 2012). Once a transaction is announced, price guidance is issued and progressively tightened through one or more rounds of what the handbook calls initial price talk, before the order book itself opens (ICMA, 2012). Investors then submit orders, frequently with an associated limit price, and the syndicate desk aggregates these into a demand curve from which a final price and an allocation are chosen. Cornelli and Goldreich (2003), studying sixty-three international equity issues from a single major bank, find that the price is not set according to a fixed formula but at the discretion of the investment banker in consultation with the issuer, and that the quantity-weighted average of limit prices alone explains over eighty per cent of the variation in where the final price lands within its initial range. Oversubscription matters too, but far less: on their estimates, demand would need to rise by roughly sixty-eight per cent to move the price by the same amount that a one-dollar rise in the average limit price achieves directly.
Price is only half of what the desk decides during this window. Allocation, the question of which investor receives how much of an oversubscribed book, is decided through a further and largely informal act of judgement, in which the World Bank note describes syndicate desks sorting orders by investor quality: solid real money names such as pension funds and insurance companies, expected to hold a bond through its life, are given priority over trading accounts and certain hedge funds, whose reputation for exiting quickly can undermine secondary-market performance and, by extension, the terms available to the sovereign the next time it returns to market (van der Wansem, Jessen and Rivetti, 2019). This categorisation is not disclosed to the issuer in advance of the deal and rests on the desk’s accumulated, largely tacit knowledge of individual investors’ past behaviour, which is itself a form of institutional memory unavailable to a first-time issuer and only partially available even to a frequent one. Set against the Gabon and Kenya transactions with which this essay opened, the mechanism takes on a sharper edge: a coupon of 12.7 per cent or 9.95 per cent is not simply a number the market handed down, but the outcome of a book that a desk assembled, weighted by investor type, and priced according to a demand curve constructed from bids the issuer itself never directly saw (Dryden and Volz, 2026; van der Wansem, Jessen and Rivetti, 2019).
This is book-building understood, in Callon and Muniesa’s terms, as a distributed calculative agency rather than the act of any single mind. The desk does not calculate a price from first principles. It assembles a device, comprising an electronic order-management system, a roster of institutional relationships built over prior transactions, real-time reference pricing from comparable sovereign curves, and internal risk limits, and reads a price off the interaction among these elements as the book develops (van der Wansem, Jessen and Rivetti, 2019; Callon and Muniesa, 2005). Beunza and Stark’s ethnography of a Wall Street arbitrage desk describes an analogous configuration in a different corner of the same institutions: calculation there is distributed across traders, screens and models, coordinated through what they call an emergent traffic control of cues exchanged among desks working in close physical and informational proximity (Beunza and Stark, 2004). The syndicate desk during a live book-building exercise fits this description closely. Pricing decisions are reached through conference calls among lead-managers and the issuer, through the reading of an evolving electronic order book, and through the informal exchange of colour about which investors are behind particular orders, none of which reduces to the mechanical execution of a formula, however much the formula, once a spread over a reference rate has been agreed, gives the final number its numerical form (van der Wansem, Jessen and Rivetti, 2019).
Translating creditworthiness
The essay has so far used the word judgement loosely, trading on an intuitive sense of the term that now needs tightening before it can carry the weight the argument places on it. A judgement, in the sense meant here, exists where multiple outcomes remain defensible even after all available evidence has been assembled, so that an institutional actor must still weigh competing readings of that evidence against one another, with no fixed rule available to settle the weighing in advance. This is what separates judgement from calculation, which applies an agreed procedure to settled inputs and returns a determinate answer, and from mere aggregation, which sums or averages existing signals without resolving any conflict among them. While Cornelli and Goldreich (2003) derive their learning from international equity book-building, the underlying mechanics of distributed calculative agency apply cleanly to sovereign debt syndicates. They find that the price is set at the discretion of the investment banker, not according to a prespecified rule, is evidence of judgement in exactly this sense. The order book supplies abundant information, and the average limit price explains most of the variation in where a price lands within its range, but neither fact determines the final number on its own, since the desk must still decide how much weight investor consensus deserves against oversubscription, against the issuer’s own tolerance for a slower deal, and against a reading of how a book still open tomorrow might behave (Cornelli and Goldreich, 2003). Pricing and recommendation are the acts through which a judgement, once reached, is delivered. The judgement itself is the weighing that precedes them, and it is this weighing, rather than the number it eventually produces, that the essay is concerned to locate.
Judgement, so defined, is not quite the most precise word for what the desk does with it. The desk neither produces creditworthiness out of nothing nor simply observes a value already latent in a sovereign’s fundamentals waiting to be read off. It translates a diagnosis, partly its own and partly inherited from the credit rating that preceded it in the pre-phase, into a market-facing number that investors are willing to act on. Translation, rather than judgement alone, is the more exact description of the operation: not creation, which would overstate the desk’s independence from the fiscal reality it is pricing, and not passive registration, which would understate the discretion Cornelli and Goldreich document, but the carrying of an evaluation from one register, fiscal and institutional, into another, a coupon a market will trade.
Even translation understates something the act of pricing does. Once a syndicate recommends a coupon and the market accepts it, that coupon does not simply report the sovereign’s credit standing at a moment in time. It becomes part of what that standing now is. The next investor to weigh the same sovereign’s debt will read the closing coupon, and the secondary-market price that follows it, as evidence of creditworthiness in its own right, quite apart from the fiscal fundamentals that fed into the original book, and the next rating review will not be immune to that evidence either. Callon and Muniesa’s claim that markets are collective devices which calculate compromises on value, rather than mechanisms that discover a value fixed elsewhere in advance, applies here in its strongest form: the syndicate desk’s price is not only a translation of an antecedent diagnosis but a constitutive act, one that helps set the terms on which the sovereign’s creditworthiness will next be assessed (Callon and Muniesa, 2005). The desk translates, but the translation does not stay outside the reality it describes.
Pichler and Wilhelm (2001) supply a structural reason why this discretion sits where it does. They model the underwriting syndicate as a solution to a problem of moral hazard in team production: individual bankers within a syndicate have private incentives to under-invest in cultivating investor relationships unless the syndicate’s own stability across repeated deals, and the reputational stake of a designated lead bank, discipline that incentive over time. The lead-manager’s authority to set price and allocation is, on this account, not an incidental feature of how underwriting happens to be organised but the mechanism that makes the whole arrangement function, because it concentrates the reputational consequences of a badly judged price on the party best placed to prevent one. The World Bank note describes the same arrangement from the issuer’s side of the table in more cautionary language, warning that lead-managers occupy an inherently dual role, advising the issuer while also executing on their own institution’s behalf, and that a generous late-stage adjustment to price guidance in the investors’ favour is a recurring feature of the final calls before an issue is launched (van der Wansem, Jessen and Rivetti, 2019).
Why the issuer, and the market beyond it, treats this translation as authoritative rather than as one bank’s opinion among several is a question the essay has so far left implicit, and it deserves a direct answer. Part of the answer is the reputational and repeat-transaction capital that Pichler and Wilhelm describe: a lead-manager’s recommendation carries weight because the bank’s standing in future syndicates depends on investors finding, deal after deal, that its books were priced honestly against demand (Pichler and Wilhelm, 2001). Part of the answer is more structural than reputational. The syndicate desk and its salesforce hold something close to exclusive, continuously maintained relationships with the pool of long-term institutional investors capable of absorbing a sovereign issue of any size, relationships built over years of non-deal roadshows and periodic contact that no debt management unit could replicate on its own even if it wished to (van der Wansem, Jessen and Rivetti, 2019). An issuer has no second, independent channel into that same investor base against which to check the desk’s reading of demand, and part of the answer is simple convention: book-building is the procedure the entire market recognises as how a Eurobond gets priced, so that departing from it, even where an issuer suspects the guidance offered has been softened, carries a risk to execution that few debt managers are willing to bear. Authority of this kind rests on a working monopoly over access to investors, combined with a convention nobody has an incentive to break unilaterally. It is a thinner and more provisional form of authority than a licensed profession’s, resting on no accreditation and no published methodology, but it is authority nonetheless, and it is exercised over a number that will determine a sovereign’s debt service for a decade or more. Taken together, these three features mean that the syndicate desk’s authority does not derive from statutory delegation or formal regulatory recognition of the kind that underpins a licensed profession. It derives from an unrivalled position within the information and relationship architecture of the sovereign bond market itself. Its authority, in other words, is infrastructural rather than legal, which is precisely what makes it easy to overlook and difficult to contest.
None of this amounts to an accusation of misconduct. It is closer to what Muniesa, working from Dewey’s pragmatist theory of valuation, calls a flank movement away from the assumption that a valuation either reports an independent fact about worth or merely projects a social convention onto an otherwise neutral object. Dewey’s alternative, in Muniesa’s reading, treats valuation as itself a form of action performed under conditions of uncertainty, continuous with the practical work of anticipating and shaping the situation being valued rather than standing apart from it as a detached description (Muniesa, 2012). A syndicate desk’s recommended price is a valuation of this pragmatist kind, translated rather than discovered, produced through a sequence of practical acts, aggregating limit prices, weighing investor quality, testing tolerance through successive rounds of price talk, that count as judgement precisely because no antecedent rule specifies their outcome in advance.
Abbott’s account of how professions establish jurisdiction over a domain of problems offers a further way of naming what is distinctive about this judgement, and where its authority falls short of a profession’s. A jurisdictional claim, for Abbott, rests on a body of abstract knowledge capable of diagnosing a case, drawing an inference from that diagnosis, and prescribing a treatment, with the abstraction doing the work of justifying why this occupational group, rather than some other, should hold authority over the domain (Abbott, 1988). Syndicate desks perform a compressed and transaction-bound version of the same sequence. They diagnose market conditions and investor appetite for a specific credit; they infer, from the shape of the order book, what price the diagnosis supports; and they prescribe that price as a recommendation the issuer is functionally, if not formally, bound to accept. What they lack, relative to a profession in Abbott’s sense, is persistence: the jurisdiction is asserted and dissolved within a single transaction, exercised by a syndicate that convenes for the deal and disperses once settlement is complete, its authority renewed each time only by reputation and convention rather than by any standing accreditation (Pichler and Wilhelm, 2001).
The weak hand
The World Bank note is candid about where this leaves an infrequent issuer. In the pre-phase of a transaction, before any pricing decision has been made, investment banks frequently secure direct access to a Minister of Finance, a channel that the note warns can sideline the debt management unit to administrative support and displace planning that ought to run through a medium-term debt strategy (van der Wansem, Jessen and Rivetti, 2019). Mutize adds to this, arguing that Finance Ministers should not be the point of access precisely because of their political standing, relatively short tenures, and potential desires to fund visible projects that may be in conflict with the long-term health of the sovereign (Mutize, 2025). By the time the transaction reaches its final conference calls, the note observes, the issuer is often reluctant to abort a deal that has already consumed months of preparation, a reluctance the syndicate managers are aware of and one that leaves the issuer, in the note’s own phrase, in a very weak negotiating position at the last moment (van der Wansem, Jessen and Rivetti, 2019). The formal governance of the transaction assigns the issuer final say over pricing. The practical governance of the transaction, conducted through calls the syndicate desk generally leads, gives that formal authority little to work with once the book has been built around a set of expectations the desk itself has shaped.
Beckert (2010) offers a way of understanding why a debt management unit without deep prior experience of the primary bond market would find it difficult to resist this framing even where its formal authority is intact. Under conditions of what Beckert calls radical uncertainty, where the range of possible future states and the likelihood of each cannot be calculated from existing information, actors do not simply optimise against a known distribution of outcomes; they draw on socially available scripts, models and cognitive frames supplied by the actors around them in order to act at all (Beckert, 2010). For a first-time or infrequent sovereign issuer, the syndicate desk is not merely a counterparty with adverse incentives. It is also, for practical purposes, the only available source of a cognitive frame for interpreting an order book, a reference yield curve, or the significance of a widening spread on a peer country’s outstanding bond. The World Bank note’s own account of a task force needing training and guidance from the very banks it is negotiating against captures this dependency exactly, and it captures why building internal capacity is described throughout the note not as a matter of technical proficiency alone but as a precondition for the issuer to hold any interpretive ground of its own during the calls that matter (van der Wansem, Jessen and Rivetti, 2019).
The guidance note draws its own distinction between two types of issuer that maps directly onto this problem of interpretive dependence. An infrequent issuer with an attractive credit reputation can, in the note’s words, afford an aggressive hit-and-run approach to pricing without concern for how investors fare afterwards, extracting the tightest possible spread because it does not need the market’s goodwill for a follow-on deal any time soon. An issuer that must return to market regularly, or whose reputation is less secure, cannot afford to behave this way, since a coupon squeezed too hard in one transaction will be remembered against it in the next (van der Wansem, Jessen and Rivetti, 2019). Nearly all of the transactions Dryden and Volz catalogue, Gabon, Kenya, Angola, the Republic of Congo, belong unambiguously to the second category: fiscally constrained, likely to need market access again within a small number of years, and therefore structurally unable to adopt the one negotiating posture, indifference to future access, that the guidance note identifies as giving an issuer real weight in its own book.
Dryden and Volz’s finding that climate-vulnerable sovereigns pay a systematic premium in dollar bond markets, largely independent of the macroeconomic fundamentals that dominate the Presbitero et al. and Hardy models, sits well against this account of asymmetric interpretive power. A premium of this kind could in principle be lodged entirely in investor preferences prior to any negotiation, a pure demand-side fact the syndicate desk simply transmits. It could equally be lodged, in part, in the bargaining position that a climate-vulnerable and typically low-income issuer brings to the book-building calls: precisely the states least likely to have returned to market often enough to accumulate the internal expertise, the investor relationships and the reference points that would let them contest a syndicate’s reading of demand rather than accept it. The data available in the public literature cannot yet distinguish these two channels, and this essay does not claim to resolve the question. It claims only that the second channel is institutionally plausible, consistent with everything the guidance literature says about how weak an infrequent issuer’s position becomes once a book is open, and almost entirely absent from the empirical models built to explain sovereign spreads.
Evaluative capability
The World Bank note recommends, at several points and without ever gathering the recommendations under a single heading, that a debt management unit develop internal technical understanding of pricing and yields, question the motivations behind advisers’ recommendations even on routine matters, collect and internally report market intelligence gathered from the banks it works with, and avoid excessive reliance on external advisers whose interests are not fully aligned with the issuer’s own (van der Wansem, Jessen and Rivetti, 2019). Read individually, these are pieces of practical advice addressed to a narrow professional audience. Read together, they describe a capability that the guidance literature never names and that the academic literature on sovereign debt has not, to this author’s knowledge, theorised as a distinct object of study.
The capability in question is not the same as compliance capacity, the ability to satisfy the disclosure, documentation and reporting requirements that a credit rating agency, a stock exchange or a securities regulator imposes on an issuer. Compliance capacity is necessary and the guidance note treats it at length, but it is oriented toward meeting a standard set elsewhere. What the recommendations above describe instead is an interrogative capacity: the ability to understand how a judgement, whether a credit rating or a syndicate desk’s recommended price, is actually produced, well enough to anticipate it, question it, and negotiate against it rather than simply receive it. Callon and Muniesa’s observation that calculative power is unevenly distributed, and that a weakly positioned agency can sometimes acquire the equipment needed to shift that balance, describes precisely what evaluative capability would amount to in this setting: not an exit from calculation but the acquisition of enough of the same equipment, market intelligence, technical fluency in pricing methodology, an institutional memory of prior transactions, to participate in the calculation as something closer to a peer than a subject (Callon and Muniesa, 2005).
Framed this way, evaluative capability sits beside, rather than inside, the concept of debt management capacity that already occupies a central place in the literature on low-income sovereign borrowing. Debt management capacity is typically measured by whether a country maintains a coherent medium-term strategy, monitors debt sustainability, and executes issuance according to sound fiscal planning. Evaluative capability is narrower and more specific: it concerns whether the state understands, well enough to contest, the internal workings of the institutions, syndicate desks among them, that translate its fiscal position into a market price. A state can hold a sophisticated debt management strategy and still enter a book-building call with no more capacity to read the desk’s diagnosis than the desk has incentive to explain it. The distinction is not academic. Rather, it is the difference between a strategy document and a seat at the table when the price is actually decided.
There is a larger claim implicit in the argument that deserves to be stated directly rather than left for the reader to assemble. A state that depends on external evaluative institutions for access to global finance is not simply a price-taker whose fundamentals are read off by a neutral market. It is subject to a chain of translation: fiscal reality rendered into a credit rating, a rating and a set of fundamentals rendered into investor demand, that demand rendered into a syndicate’s judgement, and that judgement rendered into a coupon that will determine debt service for a decade or more. Each point in the chain, the rating committee, the syndicate desk, and by extension the index providers and collateral frameworks that determine what a bond is subsequently worth to a regulated investor, is an evaluative institution in its own right, with its own conventions, its own thin or thick form of authority, and its own degree of openness to outside scrutiny. Understanding how a state’s fiscal condition moves through this chain, and where along it a state retains or loses the capacity to contest the translation being performed on its behalf, is no longer a matter internal to the study of sovereign debt narrowly conceived. It is part of what state capacity means under contemporary conditions of market-based development finance, in which a government’s fiscal choices are worth, quite literally, whatever the institutions along this chain translate them into.
The evidentiary limits of this argument should be stated rather than smoothed over. Cornelli and Goldreich (2003) reached their findings only because a single European bank granted access to sixty-three proprietary order books relating to IPOs, a degree of access that sovereign Eurobond research, dependent on public yield and spread data, does not currently enjoy at any comparable scale. The World Bank note is written from experience and observation rather than from a systematic sample of transactions, and its account of syndicate behaviour, however consistent with the ICMA handbook’s own description of the process, cannot be treated as a controlled empirical finding. The ICMA handbook itself is locked behind a near £6000 paywall aimed at institutions and not individuals. What can be said with more confidence is that the institutional process described across these sources, pre-sounding, price talk, book-building, allocation, pricing, is not incidental to the sovereign spread that eventually appears in a dataset. It is the process that produces it, and it remains almost entirely outside the frame of the literature that studies the spread as an outcome.
The argument of this essay is not that syndicate desks should be regarded with suspicion, still less that book-building be replaced with some more mechanical alternative. It is that a research agenda organised around sovereign bond yields as data points has, by its own design, looked past the room in which those yields are made, in much the same way that an earlier generation of scholarship looked past the rating committee while studying the rating (and to some extent, still does. My forthcoming monograph focusing solely on the Credit Rating Committee is one of the first works to do just that).
A different agenda would look more directly at that room. It would ask debt management offices to keep, and eventually share in anonymised form, the record of price guidance as it moved across successive rounds of a live book-building exercise, so that the gap between initial talk and final print could be measured systematically rather than inferred from published termsheets alone. It would treat the selection and mandate of lead-managers, currently visible mainly through the league tables banks use to market themselves, as a decision worth studying on its own institutional merits, since the choice of which bank sits inside the Chinese wall shapes the composition of the eventual order book before a single order is placed. It would sit alongside, rather than replace, the econometric tradition that Presbitero et al. and Hardy represent, supplying the institutional account of mechanism that a coefficient on government effectiveness or collective action clause strength cannot by itself provide. None of this is a small undertaking, and the confidentiality that surrounds live book-building will make much of it difficult to pursue through public data alone.
Dryden and Volz are right that the return of expensive Eurobond debt deserves attention and that its consequences for climate-vulnerable and fiscally constrained states are severe. What their article, and the wider literature it belongs to, has not yet asked is where, precisely, inside the institutions that intermediate between a sovereign and the market, the coupon comes into being, and what it would take for the states paying the highest prices to understand that process well enough to meet it on more equal terms. Perhaps it is not their role, at least in the forms those works have taken. A scholarship of evaluative infrastructures, attentive to syndicate desks as it has slowly learned to be attentive to rating committees, would not lower the cost of a single bond. It would make visible one link in the chain that set it, and give the states most exposed to that chain a clearer sense of where, within it, contestation remains possible.
Bibliography
Abbott, A. (1988) The System of Professions: An Essay on the Division of Expert Labor. Chicago: University of Chicago Press.
Beckert, J. (2010) ‘How do fields change? The interrelations of institutions, networks, and cognition in the dynamics of markets’, Organization Studies, 31(5), pp. 605-627.
Beunza, D. and Stark, D. (2004) ‘Tools of the trade: the socio-technology of arbitrage in a Wall Street trading room’, Industrial and Corporate Change, 13(2), pp. 369-400.
Callon, M. and Muniesa, F. (2005) ‘Economic markets as calculative collective devices’, Organization Studies, 26(8), pp. 1229-1250.
Cornelli, F. and Goldreich, D. (2003) ‘Bookbuilding: how informative is the order book?’, Journal of Finance, 58(4), pp. 1415-1443.
Dryden, A. and Volz, U. (2026) ‘Sailing into stormy waters: the return of expensive Eurobond debt’, Financing for Development Lab, 24 July.
Hardy, D.C. (2022) Sovereign Eurobond Liquidity and Yields. IMF Working Paper WP/22/98. Washington, DC: International Monetary Fund.
International Capital Market Association (ICMA) (2012) Primary Market Handbook, Section 6B, Appendix III: Pre-Sounding, Bookbuilding and Allocations (revised April 2012). Zurich: ICMA.
Muniesa, F. (2012) ‘A flank movement in the understanding of valuation’, The Sociological Review, 59(s2), pp. 24-38.
Mutize, M. (2025) ‘African finance ministers shouldn’t be making bond deals: how to hand over the job to experts’, The Conversation (Jun 23).
Pichler, P. and Wilhelm, W. (2001) ‘A theory of the syndicate: form follows function’, Journal of Finance, 56(6), pp. 2237-2264.
Presbitero, A.F., Ghura, D., Adedeji, O.S. and Njie, L. (2015) International Sovereign Bonds by Emerging Markets and Developing Economies: Drivers of Issuance and Spreads. IMF Working Paper WP/15/275. Washington, DC: International Monetary Fund.
van der Wansem, P.B.G., Jessen, L. and Rivetti, D. (2019) Issuing International Bonds: A Guidance Note. MTI Discussion Paper No. 13. Washington, DC: World Bank.