The Liquidity Coverage Ratio, from HQLA to net cash outflows, explained for banking professionals

The Liquidity Coverage Ratio (LCR) is the metric a treasury day gets built around, and this post takes it apart the way a treasury team actually reads it: buffer over stressed net outflows. By the end you should be able to trace every line on your PRA110 return rather than just quote the headline percentage in a committee pack.

Watch the video first for the walkthrough, then read on for the detail you will want when you are staring at your own return and trying to trace where each number comes from.

What the LCR is and why it matters

The LCR answers one question: if funding dried up hard for 30 days, could the bank cover the cash going out from its own stock of liquid assets, without relying on emergency central bank facilities to get over the line?

The ratio is simple to state:

LCR = High Quality Liquid Assets / Total net cash outflows over 30 days

Expressed as a percentage, with a 100 percent minimum. Worth remembering that the full 100 percent was phased in. In the EU and UK regime the minimum stepped up over several years and reached the full 100 percent in 2019, so older analysis may reference a lower threshold. At 100 percent the buffer exactly covers the modelled stressed outflow. Most firms run well above that, both for comfort and because supervisors expect headroom.

It matters because it is a hard constraint on the balance sheet. Every deposit you take, every wholesale line you roll, every asset you pledge as collateral moves this number. If you sit in treasury or liquidity reporting, this is the metric that shapes your week.

Why regulators built it after the crisis

In 2008 the problem was not that banks lacked capital on paper. It was that funding vanished faster than anyone had modelled. Overnight and short wholesale markets froze, depositors moved, and firms that looked solvent could not find cash. Northern Rock is the obvious example, but the pattern repeated across the system.

Basel III introduced the LCR as a direct answer. The logic is deliberately conservative:

• Assume a severe but plausible stress lasting 30 days.
• Do not assume you can lean on emergency central bank facilities to pass. Note this does not exclude all central bank interaction: central bank reserves that can be drawn in stress do count as Level 1 HQLA. The point is that you cannot rely on discretionary emergency support to meet the ratio.
• Require a stock of assets you can actually sell or repo when everyone else is trying to do the same thing.

The 30 day window is a judgement call, not a law of nature. The idea is that it buys management and supervisors enough time to organise a fuller response. That is why the whole design fixates on assets you can turn into cash quickly.

The ratio, piece by piece

Two moving parts sit on either side of the line.

The numerator is your HQLA: the stock of liquid assets after haircuts and caps.

The denominator is net cash outflows: gross stressed outflows minus capped inflows, both over the same 30 days.

The subtlety is almost entirely in how you build each side. Get the classification and the run off assumptions right and the arithmetic is trivial. Get them wrong and a clean looking spreadsheet reports the wrong number with total confidence.

High quality liquid assets: Level 1, 2A and 2B

HQLA is not a single pool. It is a stack of tiers, and each tier is treated differently.

Level 1

The most liquid, highest quality assets. Cash, central bank reserves that can be drawn in stress, and high quality government bonds. Level 1 assets take no haircut and there is no cap on how much can count. This is the core of most buffers.

Level 2A

A step down. Certain lower rated sovereign and public sector exposures, and some very high quality covered bonds and corporate bonds. Level 2A takes a 15 percent haircut.

Level 2B

Lower quality still but with observed depth. The haircut depends on the asset type. As a guide, qualifying residential mortgage backed securities sit around a 25 percent haircut, while qualifying equities and lower rated corporate bonds sit nearer 50 percent. So the range you hear quoted, 25 to 50 percent, is really two different asset families with different treatment, not a vague spread.

The tiering is doing a specific job. It tells you both how much of an asset counts and how much confidence the rules place in your ability to sell it under pressure.

Haircuts, caps and the usability test

Two mechanisms stop the buffer being flattered.

First, haircuts. A holding valued at 100 does not contribute 100 if it sits in Level 2A. You count 85. Level 2B counts even less. The haircut is a proxy for the price you would take in a fire sale.

Second, caps on Level 2. The total Level 2 contribution is limited to 40 percent of total HQLA, and within that Level 2B is capped at 15 percent. So you cannot fill your buffer with high yielding lower quality paper and still pass. Level 1 has to do the heavy lifting.

Then comes the part supervisors care most about: usability and monetisation. It is not enough for an asset to sit in an eligible category on a spreadsheet. Under Basel and PRA rules, HQLA must be both unencumbered and under the control of the liquidity management function. That second requirement is easy to miss. An eligible bond sitting in a subsidiary or a business line that treasury cannot direct at short notice fails the test even if it is technically unencumbered.

That is why encumbered assets are generally excluded. If a bond is already pledged as collateral, you cannot sell it to raise cash. To be precise, the exclusion bites where an asset is encumbered for a period, typically beyond the 30 day horizon, so a very short encumbrance that unwinds inside the window is treated differently. But the practical point stands: if it is spoken for over the stress period, it is not in your usable buffer. This trips people up constantly, because the asset is high quality and clearly HQLA eligible in the abstract, but it is not available.

Net cash outflows over 30 days

The denominator is where most of the modelling work lives.

You take every category of funding and apply a run off rate: the share assumed to leave over the 30 days under stress.

Illustrative run off assumptions look like this:

• Stable retail deposits: a low run off, often around 5 percent.
• Less stable retail deposits: higher, perhaps 10 percent or more.
• Operational corporate deposits: a lower rate on the operational portion, with the non operational portion treated much more harshly, illustratively around 25 percent or more.
• Non operational wholesale funding: high, sometimes 100 percent, because it is assumed to walk straight out of the door.

On the other side you have inflows: contractual money coming in over the 30 days, such as maturing loans repaying and reverse repos unwinding. Reverse repo maturities do count as inflows, but be careful: secured funding inflows have their own rates by collateral type, and inflow rates in general vary by counterparty and collateral. Do not apply a single blanket rate.

The rule that shapes everything: inflows are capped at 75 percent of gross outflows. You cannot net your outflows down to nothing by assuming a wall of cash arrives. At least 25 percent of gross outflows must be covered by the buffer itself. This is deliberate. Regulators do not want banks assuming counterparties keep paying them on time in the middle of a market wide crisis.

A worked example with illustrative numbers

All figures illustrative, to show the arithmetic end to end.

HQLA

Before caps, that is 800 + 170 + 50 = 1020. Now check both caps.

Level 2 total (170 + 50 = 220) must not exceed 40 percent of total HQLA. Here Level 2 is about 22 percent of the total, so it passes and nothing is cut back.

Then the Level 2B cap. Level 2B (50) must not exceed 15 percent of total HQLA. Here 50 out of 1020 is about 5 percent, comfortably under 15 percent, so it passes too. Both caps checked, nothing cut.

Total HQLA = 1020.

Outflows and inflows

• Gross stressed outflows: 1400
• Gross contractual inflows: 600
• Inflow cap: 75 percent of 1400 = 1050. Our 600 is below the cap, so we use the full 600.

Net cash outflows = 1400 minus 600 = 800.

The ratio

LCR = 1020 / 800 = 127.5 percent.

Comfortably above 100 percent. Now change one thing. Suppose 200 of that Level 1 stock is pledged as collateral and therefore encumbered over the stress period. HQLA falls to 820. The ratio drops to 820 / 800 = 102.5 percent. Same balance sheet on paper, very different position, purely because of usability.

Where teams get it wrong

The recurring errors are not exotic.

• Counting encumbered assets in HQLA. If it is pledged over the horizon, it is out.
• Ignoring the control test. An unencumbered asset that treasury cannot actually monetise at short notice still fails.
• Misclassifying deposits. Treating a less stable or non operational balance as stable understates the outflow and flatters the ratio.
• Forgetting the inflow cap. Modelling large inflows and letting them net outflows below the 75 percent limit.
• Missing the Level 2 caps, so a Level 2 heavy buffer looks larger than it can count for.
• Stale collateral data, where an asset was freed up or newly pledged and the report has not caught the change.

Managing the number in practice

Day to day, the LCR is a moving target, not a month end snapshot. Intraday collateral movements, a large corporate withdrawal, or rolling a wholesale line all shift it in real time, and all of it eventually has to reconcile back to your PRA110 return.

Good teams do three things. They forecast the ratio forward, not just report it backward, so they can see a dip coming before it lands. They watch encumbrance closely, because that is where usable buffer quietly leaks away. And they reconcile classifications regularly, because a deposit reclassified in the source system can move the number more than any market event.

If you are building this from data, working directly with the raw cash flows in Python is the clearest way to see the mechanics. The LCR is the 30 day survival view, so it pairs naturally with the structural funding view of the net stable funding ratio (NSFR), and both sit inside the wider ILAAP, where a firm sets out its own liquidity adequacy.

The practical takeaway: never quote the headline percentage without being able to point at the buffer, the caps, and the run off assumptions behind it. The number is only as good as the classifications feeding it, so spend your time there.