The surface water trap: risk for banks and insurers - Part 3

Part 3: What does water damage to the electrical system cost?

In part 1 we saw how fragmented insurance cover for surface water is. Part 2 looked at what the duty under capital requirements regulation to document adequate insurance of collateral can cost the bank in capital. Part 2 calculated the capital cost for a standardised bank on a residential mortgage. Larger banks with internal models (IRB) see the effect through loss given default (LGD), and that is what we calculate here, using an example from commercial real estate.

The LGDs in the table only show the part of the loan the collateral does not cover. They are not regulatory LGDs. For an IRB bank’s own LGD estimates, the capital requirements rules set a floor of 10 per cent for the part of the exposure secured by commercial real estate and 25 per cent for the remainder. The table shows the mechanism, and what matters is the change between the scenarios.

Assumptions

Let us assume a combined warehouse and office building with a market value of NOK 20 million. The basement houses the building’s main switchboard, a UPS (battery-backed emergency power system) and the control system for ventilation and lifts. Electrical equipment that has been under water often cannot simply be dried out and switched back on. We assume a replacement cost of NOK 3.3 million for water damage to the system.

We also assume the water arrives as surface runoff and sewer backflow during a cloudburst, not from a river. It is then unlikely to count as natural perils damage. The damage is covered only by the water damage section of the building insurance, with an excess and a risk of a reduced payout.

Commercial property insurance often has a fixed sum insured rather than full value cover, and the sum does not automatically follow construction cost inflation (see part 1). We assume the settlement for the system ends at NOK 3 million. That is NOK 300 000 below the replacement cost, which the owner must bear. The gap may be due to:

• a low sum insured

• a separate limit for water damage

• a deduction for the age of older equipment, which some policy terms apply to electrical damage

• the excess

Two owners each have an almost identical building in the same industrial area. Both buildings lie outside NVE’s flood zones and therefore pass the bank’s usual map check without comment. Vestre Kai Eiendom AS has paid down part of its loan and owes NOK 13.4 million, a loan-to-value ratio of 67 per cent. Nordre Lager AS bought recently and owes NOK 17 million, a loan-to-value ratio of 85 per cent.

When commercial property is sold off, the buyer market is thinner than for housing. We therefore assume a 20 per cent forced sale discount and 3 per cent selling costs.

Four scenarios

The realisable amount is the market value less a 20% forced sale discount and then 3% selling costs. Interest and costs during the default period are not included. For comparison, 7Analytics and Housing Lab found prices around 5.6 per cent lower when a home moves from 0 to 1 on the flood risk scale (see part 1).

Scenario A does not change the LGD. The damage hits technical equipment, not the building structure, so the property value is unaffected. Still, the NOK 300 000 shortfall comes out of the owner’s own cash. That buffer may not be there the next time something happens.

In scenario B, the insurer requires a documented improvement as a condition for full cover at the next claim, for example moving the switchboard out of the basement. If this has not been done, the payout can be reduced. The insurer may also refuse the claim entirely, citing an exclusion for inadequate or failing drainage. The system is out of action during repairs, and tenants start considering alternatives. We assume the property value falls to NOK 19 million because of increased letting risk, not physical deterioration. Cheaper measures exist, but we assume they have either been carried out without helping enough, or are not feasible on this site.

In scenario C, the insurer has given notice that continued cover of the system requires it to be moved to the first floor. We estimate that this costs NOK 4 million, which the owner must fund. That is more than the replacement cost of the equipment itself (NOK 3.3 million), and we assume the amount also covers new cabling and building work. The investment is not included in the LGDs in the table. It does drain the owner’s cash, which is part of the reason we set PD higher in scenario C (see below).

If the system is not moved, the building is effectively uninsured against the next water damage to the system. The property value falls to NOK 16.5 million, because buyers can now see the risk of lower value and lower cash flows. If the owner instead asks the bank to finance the measure, Nordre Lager’s exposure rises from NOK 17 million to NOK 21 million. Against a realisable amount of NOK 12.8 million, the LGD then becomes 39 per cent. On the other hand, if the building becomes insurable again, some of the lost value may come back, so financing the measure may still pay off for the bank.

The hidden cost: lost rent

Say the building brings in NOK 150 000 a month in rent and is out of use for three months after serious damage to the switchboard. That means NOK 450 000 in lost rental income.

Loss of rent is often covered by the building insurance itself, but only when the damage is covered. Even then, it is covered only for the period and up to the amount stated in the policy schedule. If the claim is rejected, the lost rent is not covered either, and delays caused by lack of financing are excluded. The tenant’s own business interruption is covered by a separate policy.

The bank cannot see whether the owner has loss of rent cover, or for how long, without asking for the policy schedule. Without such cover, the cash flow that services the loan may weaken.

Three months is an assumption, and it may be optimistic if the repair needs approval from several parties. The Coop Extra shop in Nesbyen is one example. It was destroyed by floodwater during Storm Hans, and the owner applied for a building permit to renovate it. He only received a final refusal in late autumn 2024, more than 15 months after the storm, and only then could he pursue his claim with the insurer. It is not known whether the insurance claim itself has been settled since then, and that is the decision the owner and the bank actually care about.

Expected loss

Expected loss is the probability of default (PD) times the loss given default (LGD) times the exposure. The PD figures below are assumptions, not estimates from any model, so feel free to replace them with the bank’s own.

We set PD at 0.4 per cent in the normal situation. For a building that has been through scenario C, with weakened insurance cover and a real risk of the tenant leaving, we set it at 2 per cent. In practice, PD is driven by cash flow and interest coverage, not by insurance status alone. Without loss of rent cover, three months of lost rent (NOK 450 000) is a quarter of the annual rent of NOK 1.8 million. For an owner with an 85 per cent loan-to-value ratio, 2 per cent is therefore probably on the low side.

For Nordre Lager, this gives:

• Normal situation: 0.4% × 8.7% × NOK 17 million = roughly NOK 5 900 a year in expected loss

• Scenario C: 2% × 24.7% × NOK 17 million = almost NOK 84 000 a year

The expected loss is more than 14 times higher, on the same building.

And then the portfolio

Assume the bank holds collateral in 25 commercial buildings in an industrial area, all in the same catchment. Each is similar to Nordre Lager, with a loan of NOK 17 million, which adds up to NOK 425 million in exposure. Suppose around a quarter of them, six buildings, end up in something like scenario C. Expected loss then rises by only about NOK 0.5 million a year.

Expected loss hides the tail, though. The six buildings share a catchment, so their tenants may leave at the same time and the owners may all default together. If that happens, the loss is about NOK 25 million (six times NOK 4.2 million).

The example simplifies in two ways, both of which probably understate the loss. In part 1, severity varied from building to building in the same area. And six forced sales in the same area at the same time are unlikely to go through at only a 20 per cent discount.

For a large bank, NOK 25 million is a small number. For a regional bank with many commercial loans in the same catchment, it is no trifle (see part 1 on geographically concentrated banks).

Green, yellow, red: a loan classification the bank could already use

After Storm Hans, Nesbyen municipality divided properties into three groups. Some were cleared, some received a permanent ban on building and subdivision, and the rest were placed in a “yellow zone”, neither cleared nor resolved (see part 2).

The bank can borrow this classification for its own portfolio management of loans exposed to surface water:

• Green: insured

• Red: not insured

• Yellow: insurance status neither confirmed nor ruled out

The yellow category must be defined by what the bank actually has. For example, a loan is yellow if its insurance status has not been documented since approval, or if the documentation is older than, say, twelve months.

The concentration risk is unknown

The whole range in the example above, from 0 to almost 25 per cent LGD, depends on factors that, as far as we know, Norwegian banks do not systematically monitor today:

• whether the sum insured actually covers the replacement of technical equipment

• whether business interruption cover exists at all

• how many other commercial buildings share the same catchment and the same vulnerable placement of electrical infrastructure

What the bank can do, and the upside

There are several steps a bank can take now:

• Layer building-level surface water data on top of the flood zone map, so the two risks are shown separately.

• Ask for insurance certificates and claims history for buildings in the same catchment. An insurance certificate from the customer does not reveal reduced cover (see part 2), so the claims history should ideally come from the insurer, with consent and an agreed data flow.

• Check whether the bank is registered as mortgagee on the policy. Some policy terms then give the mortgagee special protection and notice.

• Mark loans with missing or outdated documentation as yellow.

• Decide how to respond when a customer asks for financing for risk-reducing measures, as in scenario C.

There is also an upside. A measure that substantially reduces the building’s physical climate risk can qualify under the climate change adaptation objective of the EU Taxonomy. The financing of the whole building can then qualify, not just the adaptation itself, as long as the building also meets one of these energy requirements:

• Built in 2020 or earlier: energy performance certificate C, or among the top 30 per cent for energy efficiency

• Built in 2021 or later: energy performance certificate A, or a net zero energy building

Table 2 in Finance Norway’s guide to taxonomy-aligned property shows how to calculate this.

The figures above are stylised, but the concentration risk is not. It is there long before the next cloudburst, just as it was in Krokstadelva before the flames.

This is part 3 of three. Part 1 covers what surface water does to collateral, and why the risk falls into several different coverage categories. Part 2 covers the documentation requirement in capital requirements regulation.