The complexity trap and how volatility is saving banks in capital markets

Why a record quarter (Q126) does not settle the structural question, and what the decline of Rome can tell us about the trajectory of capital markets 

Executive Summary 

The first quarter of 2026 was, by any reasonable measure, an excellent quarter for the Wall Street franchises. JP Morgan posted record market revenue of $11.6 billion, up 20% year on year. Goldman Sachs reported its second-highest quarterly revenue ever, with record equities trading of $5.33 billion and investment banking fees up 48%. Citi crossed $7 billion in market revenue for the first time in a decade. Bank of America and Morgan Stanley both set records on the equities side. Volatility around the March oil shock and the AI-driven sector rotation, combined with a long-awaited rebound in advisory and ECM, produced something close to ideal trading conditions, and the franchises that were poisoned for it captured the upside fully.

It would therefore be premature, and probably wrong, to argue that the major banks are in structural decline in their capital markets businesses. They have just demonstrated, in public, that when conditions favour the franchise model, the franchise model still works very well. The question this paper sets out is a different and more careful one. It is whether a record quarter (particularly one driven by macro volatility rather than by underlying franchise growth) tells us very much about the structural trajectory of the business over the next decade. My view is that it does not, and that the more useful framework for thinking about that trajectory is the one Joseph Tainter set out in his 1988 study of how complex societies evolve. 

Tainter argued that complex systems invest in complexity to solve problems, that the returns on each new layer of complexity are diminishing, and that systems can look strongest in precisely the period when their underlying cost base has begun to outrun the value it produces. Rome had genuinely good years late in the imperial period; the empire's fiscal accounts were in surplus during several decades when historians now identify the structural fragility most clearly. Strong cyclical performance and weakening structural foundations are not mutually exclusive. They can, and o]en do, coexist for an extended period. 

Applied to the major capital markets businesses, the framework suggests three things worth taking seriously even in a quarter as strong as Q1 2026. 

• First, the cost base that supports a tier-one markets franchise has grown substantially over the past two decades, and a meaningful share of that cost base is fixed regardless of trading conditions. 

• Second, the parts of the business that benefit most from volatility ( flow equities, prime brokerage, commodities, FX in volatility-sensitive currencies ) are also the parts where non-bank competitors are most active and most cost-competitive. 

• Third, the parts of the business that have struggled in this quarter, notably rates and mortgages, are the structurally heavier products where balance sheet, regulatory capital and operational complexity press most directly on returns. The pattern visible inside the Q1 numbers is, on close inspection, the same pattern the structural argument would predict. 

I therefore offer the Tainter framework not as a prediction of decline but as a diagnostic question. When volatility recedes (as it eventually does) and the steady-state mix of business reasserts itself, will the cost base that produced this quarter's records will be earning its keep?

That is not a question whose answer is sealed by a single set of results, however strong. It is the question that prudent management of a franchise of this size has to keep asking, in good quarters as well as bad.

Tainter's Framework 

Joseph Tainter is an American anthropologist whose 1988 book The Collapse of Complex Societies has had a long second life among people who think about institutions, infrastructure, and economic systems. His subject was nominally the Roman empire, with case studies on the classic Maya and the Chacoan culture of the American southwest, but his argument was general. 

Tainter began from an observation that strikes most readers as obvious in retrospect. Human societies face problems like defending a frontier, feeding a city, selling disputes, allocating capital, and they solve those problems by adding complexity. More rules, more specialised roles, more institutions, more infrastructure. Complexity works; it is, as he put it, a problem-solving tool. But each new layer of complexity carries an ongoing cost: it has to be paid for, staffed, maintained, and in many cases defended against the new problems that complexity itself creates.

His central, less-obvious observation was that returns on additional complexity are not linear. They diminish. The first legion solves a real problem on a contested frontier. The tenth legion solves a smaller one. The fourth legion is consuming substantial resources to solve a problem that the previous thirtynine have largely already addressed. At some point the marginal cost of the next unit of complexity exceeds the marginal value it produces, and from that point onward the system is investing more and getting less. It is not yet failing in any visible sense. It can all post good years. But it has crossed a line that, on a long enough horizon, matters. 

Tainter's word for what happens next is collapse, which is misleading because it sounds catastrophic. What he actually describes is rational simplification; populations, capital, and effort migrating away from the high-complexity arrangement and toward simpler alternatives whenever those alternatives are available and adequate. Rome did not lose a single decisive war.

Over a long period, the people on the periphery of the empire concluded, individually and unspectacularly, that the imperial bargain (i.e. taxes, administration, conscription, in exchange for protection and infrastructure) was no longer worth what it cost them, and they began to opt out. The Goths offered a simpler arrangement, and a meaningful share of the population took it. 

The framework has been usefully applied to a range of modern systems: the late Soviet bureaucracy, the post-war American hospital, the European university system, large legacy industrial conglomerates.  

None of these collapsed in any sudden sense. Each began to look structurally fragile during a period when its surface metrics were still acceptable, and in retrospect the diminishing-returns curve was visible considerably earlier than the moment of obvious decline. I think the framework is useful for thinking about modern capital markets businesses for the same reason: not because they are about to collapse, but because the question of whether the marginal cost of complexity has begun to exceed the marginal value it produces is the right question to ask, regardless of what the most recent quarter looks like.

How the Markets Franchise built its Complexity

The modern tier-one capital markets business is, by any historical standard, an extraordinarily complex operation. Each layer of that complexity was added in response to a real problem, and I want to be clear that I am not arguing any particular layer was unjustified. The cumulative effect, however, is worth describing plainly because it is the necessary context for any structural assessment. 

• On the front office side, a contemporary franchise operates a split between voice, electronic, structuring, sales, syndicate and product specialists, each with its own management chain. 

• Risk runs across market, credit, counterparty, liquidity, model, operational and conduct disciplines, with separate enterprise risk teams coordinating between them. There are three lines of defence. There is compliance and surveillance. 

• There is a control room. There are new product approval committees, transaction approval committees, regional risk committees, global risk committees. Internal audit, external audit, regulatory examiners on site. Finance covers product control, valuation control, capital, liquidity and regulatory reporting as separate disciplines. 

• Technology runs thousands of applications, often with overlapping functions, connected by middleware that itself requires teams to maintain. 

• Above all of this sits a matrix in which most senior people report into a product head, a regional head, a coverage head and increasingly a client-segment head, each owning a piece of the performance review.

Each of these existed for a reason. The 2008 crisis demanded materially better capital management and risk infrastructure. The conduct events of the 2010s required surveillance and governance at a level that did not previously exist. Operational risk events drove resilience programmes that became permanent.

Model failures produced model risk governance. Settlement failures and trade-break events required the reconciliation infrastructure now embedded in every operations function. Regulatory fragmentation across the major jurisdictions has forced jurisdiction-specific reporting and capital regimes that now coexist with the underlying business. None of these additions were wrong. The point is only that they accumulate.

The financial signature of that accumulation is now well understood by anyone who has looked at the numbers. 

• Cost-income ratios across investment bank capital markets franchises have run broadly between 65% and 80% in recent years. 

• Average return on equity in the franchise has sat in the 6% to 8% range over the cycle, against a cost of equity that most analysts place around 10%. 

• Q1 2026 results pulled several franchises meaningfully above those long-run averages, which is what a strong quarter does. 

The relevant question is not whether good quarters exist (they manifestly do) but whether the long-run averages reflect the underlying economics of the business and what would happen to them if the volatility that produced this particular quarter receded. 

There is one number from Q1 2026 worth lingering on. JPMorgan, which posted the best market quarter in its history, simultaneously lowered its full-year 2026 net interest income guidance by $1.5 billion, with the entire revision attributed to its markets business. 

The non-markets portion of the guidance held steady. The single firm that had the most successful trading quarter in the industry was, in the same release, signalling caution about whether the conditions producing that result would persist. 

That is not a contradiction. It is precisely the diagnosis the structural framework would lead us to expect: cyclical strength running alongside cautious structural commentary, in the same set of numbers.

Where the Non-Bank Competitors sit

Any honest discussion of structural change in capital markets has to address the rise of the non-bank liquidity providers like XTX Markets, Citadel Securities, Jane Street, Jump Trading, and a small number of others. 

The tendency in industry commentary is to oscillate between two extremes on this topic. One narrative holds that the NBFIs are running away with the business and the banks are sleepwalking into irrelevance. The other holds that the NBFIs operate in a small set of commoditised products, lack the balance sheet and client coverage to compete in the parts of the market that actually matter, and will plateau as their preferred products mature. Both narratives contain some truth; neither, in my view, is the complete picture.

What is empirically clear is that the NBFIs have built market share in specific products at a meaningful pace over the past decade. Roughly 90% of spot FX is now electronic, and a significant share of the liquidity in major pairs is provided by non-bank firms. Citadel Securities executes close to a quarter of all US stock trades. Principal trading firms now account for around 60% of trading on electronic interdealer venues in US Treasuries, up from a small fraction a decade ago. Anonymous all-to-all trading on platforms like MarketAxess Open Trading has grown from negligible to over 17% of US corporate bond electronic flow in ten years. None of these data points is decisive on its own. Together they describe a steady, product-by-product expansion of non-bank participation in markets that were historically the preserve of dealer franchises.

What is also clear is that the NBFIs benefit from the same volatility that produced the strong Q1 2026 numbers for the banks. By multiple accounts, several major NBFIs are reporting record results for the same quarter, in many cases growing their share of the volatility-driven flow rather than ceding it. Both the banks and the non-bank firms can win in a high-volatility environment because the addressable revenue pool expands; the question is what happens to the marginal share when conditions normalise.

The structural argument about NBFIs is not that they are about to displace banks across every product. It is that they operate with materially lower fixed costs in the products where they compete, and that the products where they have most expanded share (i.e. spot FX, equity market-making, US Treasury electronic trading, increasingly corporate credit on all-to-all venues) share a common feature: they are products where balance sheet, credit intermediation, prime brokerage, settlement complexity and regulated capital are decreasingly the gating factors. 

Where those factors remain decisive ( FX swaps, NDFs, options, emerging market currencies, complex derivatives) bank franchises retain a structural advantage. The relevant question is what happens to that advantage as those gating factors evolve, which is the subject of the following section.

Reading the Q1 Split

The composition of the Q1 2026 results is itself useful diagnostic information, and I want to spend a paragraph on it because it sharpens the structural question rather than dulling it. 

The standout business across nearly every franchise was equities. Prime brokerage balances at records, derivatives strong, cash equities volumes elevated. Goldman Sachs's equities business hit a record $5.33 billion, JP Morgan's equities revenue was up 17%, Morgan Stanley's equity sales and trading rose roughly 25%, Citi’s equities was up nearly 40%. 

The macro setup of Q1 (i.e. extreme volatility, AI sector rotation, the March oil shock and the geopolitical tension around Iran) was tailor-made for flow desks. EquiKes is also the business where the franchise model and the non-bank model overlap most directly: prime brokerage is increasingly a balance-sheet-light, technology-intensive business; cash equities and derivatives market-making are precisely the products where Citadel Securities and similar firms have built their share. The banks won this quarter in equines. So did the non-banks. The volatility lifted everyone.

Fixed income, by contrast, was bifurcated in a way that maps directly onto the structural argument. Commodities, credit, currencies and emerging markets were strong almost everywhere; these are the volatility-sensitive, flow-driven parts of FICC where the macro backdrop was unusually favourable. 

Rates and mortgages, which are the steadier, more balance-sheet-intensive products, were weak across the franchises. Goldman's overall FICC revenue was actually down 10% on the quarter, weighed down by exactly these products. JPMorgan's FICC was up 21%, but the strength came from commodities, credit, and currencies; rates contributed less than the headline implied. The pattern is consistent across the franchises that have reported. 

The volatility-sensitive products won; the balance-sheet-heavy, capital-intensive products struggled.

This is precisely the pattern the structural argument predicts. The products where banks earn most easily are the products where their structural cost advantages (i.e. settlement, balance sheet, capital, prime brokerage, multi-product client coverage) matter least. The products where those cost advantages should matter most are the products where returns have been weakest.

A reasonable interpretation is that the volatility around Q1 was doing the work that, in less volatile quarters, the structural cost base used to do. When the volatility recedes (and it eventually does), the question of whether the structural cost base is still earning its keep becomes the dominant one again.

Which Products are Most Exposed

If the structural argument has merit, it suggests that some products are further along the diminishing - returns curve than others. I think this is broadly true, and the rough ordering is worth sketching, with the caveat that timing is genuinely uncertain and that strong cyclical conditions can extend the runway substantially in any given product.

• Spot FX is the product where the structural picture is most advanced. Electronification is essentially complete; non-bank participation in major pair liquidity is well established; and the marginal economics of providing flow have compressed to a level at which only the most efficient operators are comfortable. Banks retain meaningful franchises in spot, but the part of the business that earns a real margin has shrunk to specific client segments, internalisation flows, and value-added services around the basic execution. The structural transition here is not coming; it has largely happened. 

• Equity market-making in US cash equities is at a similar stage, with non-bank firms providing a meaningful share of liquidity and the banks retaining strong franchises in prime brokerage and derivatives. US Treasury electronic trading on interdealer venues has moved further than most observers outside the business realise, with principal trading firms now accounting for the majority of activity on those venues.

The products that have so far resisted the dynamic (I.e. FX swaps, NDFs, FX options, emerging-market currencies, complex credit derivatives) share a set of features that is worth naming clearly, because they are also the features that may evolve over the next several years. 

• Bilateral credit relationships, which currently keep most non-bank firms outside these markets. Settlement infrastructure, where banks have historically held a near-monopoly on the relevant services. 

• Capital treatment, where the regulatory framework currently advantages banks for inventory-heavy products through netting and prime broker arrangements that non-banks cannot easily replicate. 

• Operational complexity around multi-leg, multi-jurisdiction trades, where banks have built substantial infrastructure that is genuinely hard to copy.

Each of these is in slow evolution rather than imminent collapse, but each is moving in the same direction. Major banks have already pulled back from some prime brokerage activities post-Archegos. 

Non-bank firms increasingly hold member status at central counterparties, which removes the need for bank intermediation in cleared products. The migration toward central clearing of FX forwards and swaps, however contested its pace, is itself a barrier-removal process. 

The migration to T+1 settlement and the slow rise of stablecoins for cross-border payments chip away at the settlement infrastructure that banks have historically owned. Capital frameworks are evolving toward activity-based rather than entry-based treatment, which over time reduces the structural advantage banks enjoy in capital-intensive products. 

My base case, held with appropriate humility about timing, is that FX swaps and NDFs in 2030 will look more like spot FX does today than like FX swaps did in 2020 (i.e. high-volume, lower-margin, more electronic), with meaningful non-bank participation alongside continued bank franchises. Options and emerging market currencies follow on a longer arc, perhaps over the back half of the decade, because the credit, settlement and risk-pricing barriers there are genuinely harder to dismantle. 

None of this implies that banks exit these products. It implies that the part of the spread that compensates the bank for the underlying balance sheet, settlement and capital infrastructure compresses, and that the firms that earn well in this environment will be the ones whose cost structures are calibrated for a more compeKKve end state.

What this Means in Practice

If the diagnosis has merit, the strategic question for any major capital markets franchise is not whether it can earn record numbers in volatility quarters. The Q1 2026 results have settled that question. The strategic question is what its earnings look like through the cycle, and how its cost base will sit when the conditions producing this quarter recede. 

I see three broadly viable responses to that question, and I offer them as patterns I observe rather than prescriptions. None is comfortable, and each requires conviction about a specific end state for the business.

— Be the scaled survivor. A small number of franchises globally have the volume, the technology investment and the client breadth to sustain a high-fixed-cost, lower-margin model in the way that the largest non-bank firms do. For these institutions, the strategic priority is to invest aggressively in the integrated front-to-back platform, to internalise as much flow as possible, and to drive cost-income ratios materially below current industry averages. This is hard, capital-intensive, and meaningfully easier when the cyclical environment is strong, which is part of the case for using a quarter like Q1 2026 to fund the investment rather than to declare the question resolved. 

— Specialise where the franchise still earns. For most mid-tier banks, the more sustainable response is to retain a defensible franchise in the products and client segments where the structural advantages of being a bank still matter — regulated balance sheet products, complex multi-asset solutions, regional flow franchises with deep client coverage — and to actively reduce investment in the products where the structural advantage has eroded. This is psychologically difficult because it requires accepting that parts of the business that were strategic ten years ago no longer are. The alternative, however, is to defend everything and earn well in none of it.

— Become the platform provider. A small number of banks will reposition toward providing infrastructure to the rest of the market — clearing, custody, prime services, settlement regulatory reporting — running these as utility-economics businesses with the cost discipline that requires. This is the path of greatest structural change but it has the advantage that the resulting business model is well understood and is not dependent on volatility for its returns. It is also the path that builds on capabilities most banks already possess, even if those capabilities are currently buried inside larger franchises. 

What I think is harder to defend is the position of trying to remain a full-service franchise in everything, with the same cost base, on the assumption that strong cyclical conditions will continue. Q1 2026 was a strong quarter. So was Q3 2008, in retrospect, for several franchises that did not survive the year. I am not suggesting any analogy with 2008 as there is no comparable solvency dynamic in play, but the broader observation that a single strong quarter is poor evidence on which to settle a multi-year structural question is worth holding onto.

In Closing

I started this paper with the observation that Q1 2026 was, by any reasonable measure, an excellent quarter for the major Wall Street franchises. I want to end on the same note. Record market revenue, record equities trading, the best advisory quarter in years; these are real numbers, achieved by businesses that have invested heavily in capability and client relationships and have demonstrated that they can extract meaningful value when conditions are right. None of that is in dispute, and none of it should be. 

What I am trying to argue is more measured. A single strong quarter, however good, does not answer the structural question. The question is what the underlying cost base of these franchises looks like when measured against the long-run revenue mix rather than the favourable mix of any particular quarter. 

The Tainter framework is a useful diagnostic tool here precisely because it does not require any prediction of immediate decline; it asks only whether the marginal cost of complexity has begun to exceed the marginal value it produces, and whether the system has the capacity to recognise that and respond. 

In my view, the Q1 2026 results contain useful internal evidence on both sides of the question; strong volatility-driven performance in the products where structural cost advantages matter least, weaker performance in the products where they should matter most. 

The franchises whose leadership is using this quarter to reflect on those structural questions, while the cyclical wind is at their backs, are doing the more difficult and more useful work. The franchises that conclude from this quarter that the structural question is settled in their favour may be making the more comfortable choice in the short term. 

My experience, and Tainter's, suggests that the more comfortable choice is rarely the right one when the underlying environment is changing as quickly as this one is. Strong quarters are the easiest time to do hard structural work, because the resources are there and the pressure is not. Weak quarters force the same work under conditions in which it is much harder to do well. 

I offer these observations not as a forecast and not as criticism. I offer them as the questions we would want to be asking if we were running a Ker-one franchise in 2026, alongside the considerable satisfaction of aving just posted the numbers. Both things can be true at the same time, and the institutions that hold both in mind will, I suspect, be the ones best positioned for whatever the rest of the decade holds.

About the Author

Pierre Pourquery

is a seasoned financial sector professional with over 25 years of experience spanning capital markets, digital assets, and strategic transformation. He formerly led Capital Markets at EY and served as a Partner at the Boston Consulting Group for seven years. Earlier in his career, Pierre was the Global Head of Risk and Compliance at IBM, and held senior roles including Principal in Risk Management at a leading U.S. consultancy, Head of Market and Counterparty Risk at a major French bank, and Fixed-Income Trader. A published author and thought leader, Pierre regularly contributes to industry dialogue through articles, books, and speaking engagements at global conferences.

-------------------------------------------------------------