Bond Sizing and Recovery Assumptions: Where Underwriting Breaks Down

Most underwriting models do not fail at the start. They fail at the end. On paper, everything looks structured. The bond is sized. The asset value is defined. Recovery assumptions are plugged in. The numbers align, and the deal moves forward. But underwriting is not tested when the deal is signed. It is tested when something goes wrong. And that is where bond sizing and recovery assumptions begin to unravel.

‍

The hidden dependency underwriting relies on

At the core of every bond sizing decision is a quiet dependency: the assumption that recovery will behave as expected. This assumption is rarely challenged deeply enough.

Recovery is treated as a derived outcome. A number that naturally follows from residual value and market conditions. But in reality, recovery is the most fragile part of the entire underwriting structure. Because it is not just about what the asset is worth. It is about what remains after everything else is accounted for. And that difference is where most models break.

The data makes this structural vulnerability visible. S&P Global's infrastructure default and recovery study found that senior secured infrastructure debt averages a 77.1% recovery rate, while subordinated infrastructure debt averages just 23.1%. That's not a small gap. It's a 54-percentage-point difference driven almost entirely by where in the capital structure a lender sits and whether the residual assumptions underpinning that position actually hold.

Why bond sizing looks more reliable than it actually is

Bond sizing gives a sense of control. It creates the impression that downside risk has been contained within a defined buffer. But the sizing itself is only as strong as the assumptions behind it.

Most models rely on stable residual values, predictable decommissioning costs, and the presence of a functioning secondary market. These inputs make the bond appear sufficient. The issue is that none of these variables are actually stable. Residual values shift with market demand. Decommissioning costs increase with regulation and inflation. Secondary markets tighten precisely when assets need to be liquidated. Therefore, the bond is not sized according to reality. It is sized against an optimistic version of it.

Where recovery assumptions quietly distort risk

Recovery is often simplified into a percentage. A clean estimate of how much value can be recovered at the end of the asset's lifecycle or in a default scenario. But recovery is not a percentage. It is a sequence of events. The asset must be evaluated, prepared, potentially dismantled, transported, and sold. Each of these steps introduces cost, time, and uncertainty.

And most models compress all of that into a single number. This stage is where distortion begins.

If decommissioning costs are underestimated, recovery is overstated. If timelines are ignored, value erosion is missed. If market demand is assumed rather than validated, liquidity risk disappears from the model entirely. The result is not a conservative estimate. It is a fragile one.

The real point where underwriting breaks

Underwriting does not break because teams lack sophistication. It breaks because the structure separates what should be connected.

Bond sizing is often done independently of detailed recovery modelling. Residual value is estimated without integrating full life cycle costs. Decommissioning is treated as a side calculation rather than a core input. This creates a disconnect. The bond reflects a level of protection that recovery cannot actually support. And that gap only becomes visible when the asset is stressed, delayed, or forced into liquidation. At that point, it is too late to correct the model.

Why this problem is becoming harder to ignore

This gap has always existed, but it is becoming more visible now. Assets are evolving faster, especially in energy and infrastructure. Regulatory environments are tightening, increasing end-of-life obligations. At the same time, capital providers are under greater pressure to justify their assumptions.

The broader infrastructure debt market has historically masked this problem by performing well on average. Only 3.9% of infrastructure debt is in default by year 10, compared to 14.3% of non-financial corporate debt. That resilience creates confidence. But it also creates complacency because the aggregate numbers obscure how badly individual recovery assumptions can break when an asset hits distress. Even within infrastructure, recovery rates vary significantly depending on lifecycle stage, leverage, and sector. The average holds. The tail risk doesn't.

What was once acceptable as a directional estimate now needs to be defensible. Lenders and insurers are being asked not just what their assumptions are, but where those assumptions come from and how they hold under stress. And most models cannot answer that convincingly.

What a more defensible approach looks like

The shift is not about adding more complexity. It is about connecting the right pieces.

Recovery needs to be modelled as a dynamic outcome, not as a fixed input. Residual value needs to reflect real transaction data, not static depreciation curves. Decommissioning costs need to be integrated into valuation, not appended later.

Most importantly, every assumption needs to be traceable. If a recovery number changes, it should be clear why. If a bond is sized at a certain level, it should be possible to explain the exact conditions under which that sizing holds. This is what turns underwriting from a model into a defensible position.

Where Buckstop fits in this shift

This is the exact layer Buckstop is built to address. Instead of treating recovery and residual value as assumptions, Buckstop structures them as outputs backed by data and scenarios.

It connects residual values to actual market behaviour, integrates decommissioning costs into the calculation, and models multiple recovery outcomes instead of relying on a single estimate. So when a lender or insurer sizes a bond, they are not relying on a static number. They are working with a model that reflects how the asset behaves under real conditions. And more importantly, one that can be explained, reviewed, and defended.

The Bottom line

Bond sizing does not fail because it is calculated incorrectly. It fails because it is based on recovery assumptions that do not hold under pressure. And recovery fails when it is treated as a simplified number instead of a complex outcome. For insurers and lenders, the shift is clear. Stop treating recovery as a percentage. Start treating it as a model of reality. This is where underwriting either holds or breaks. Want to strengthen your underwriting approach with a more realistic recovery model? Get in touch with us here.

FAQs

• What is bond sizing in infrastructure financing?
  Bond sizing is the process of determining the financial security required to cover risks associated with an asset, including decommissioning and recovery scenarios.
  
  
• What are recovery assumptions in underwriting?
  Recovery assumptions estimate the value that can be recovered from an asset after costs such as decommissioning, legal expenses, and market factors are considered.
  
  
• Why do recovery assumptions fail in underwriting?
  They often rely on static residual value estimates, ignore full lifecycle costs, and fail to account for changing market and regulatory conditions.
  
  
• How does bond sizing depend on recovery assumptions?
  Bond sizing is directly influenced by expected recovery. If recovery is overestimated, the bond may be undersized, increasing financial exposure.
  
  
• What is the biggest risk in bond sizing?
  The biggest risk is disconnecting bond sizing from realistic recovery modeling, leading to insufficient protection in downside scenarios.
  
  
• How can lenders improve bond sizing accuracy?
  By using scenario-based recovery modelling, integrating decommissioning costs, relying on real market data, and ensuring full traceability of assumptions.