Lombard or Sell: The Real Arithmetic

An investor who needs EUR 1.5M for a purchase, and holds it inside a securities portfolio, has two ways to reach it: sell part of the portfolio, or borrow against it. Put that way, the choice sounds like a matter of temperament. It is not. It is a piece of arithmetic, and it is almost never done properly, because each side of the comparison hides costs that never appear on a confirmation slip.

The sale looks free because no interest is charged. The loan looks expensive because interest is charged and visible. Most decisions are made on that single contrast, which is roughly like choosing between two houses by comparing their door handles. The real ledger has more lines on both sides, and the purpose of this article is to write them all down, without pretending that either column is innocent.

What selling actually costs

The first line in the ledger is tax. A portfolio built over years carries unrealised gains, and a sale converts them into realised, taxable ones. How much tax depends on the investor’s residence, the holding structure and the applicable regime, which is precisely why the calculation must be run with real figures rather than instinct. What matters here is the shape of the cost: it is paid immediately, it is paid out of the proceeds, and it is paid on gains that would otherwise have continued to compound untaxed. The amount that arrives at the notary is not the amount that left the portfolio.

The second line is income. Securities produce dividends and coupons, and sold securities produce nothing further. An investor who liquidates part of a portfolio to buy a property has not simply moved value from one asset to another. He has switched off one income stream before the new one, if the property is let at all, has been switched on. That gap is rarely priced into the decision, because it never arrives as an invoice. It arrives as an absence, which is the easiest kind of cost to ignore.

The third line is timing. A sale driven by a completion date is a sale on the calendar’s terms, not the market’s. If the market happens to be strong in the week the funds are needed, that is luck, not planning. If it is weak, the investor sells low for reasons that have nothing to do with the portfolio, and the value given up is as real as any interest charge, though no statement will ever label it as a cost of the purchase.

The last line is the hardest to see: re-entry. In theory, a sold position can be repurchased once the property is bought and finances settle. In practice, the cash rarely goes back, and when it does, it goes back at whatever prices then prevail. A strategy interrupted is very often a strategy abandoned. Compounding works by being left alone, and a forced interruption at an arbitrary moment is the exact opposite of being left alone.

What borrowing actually costs

The loan’s first cost is visible and honest: interest. It accrues from drawdown, it is paid in cash, and over a long holding period it adds up to a serious figure. Any comparison that treats the Lombard route as costless is making the same mistake as the one-line comparison, in the opposite direction. The interest bill has to be set against what the retained portfolio actually produces and can reasonably be expected to produce, not against optimism about future performance.

The second cost is contingent: the margin call. The portfolio stays invested, which means it stays capable of falling, and if it falls far enough the bank will demand additional collateral or repayment. That risk is genuine and deserves its own analysis before anything is signed. It is managed principally through sizing, by borrowing well below the maximum the facility allows, so that ordinary market movement never reaches the coverage test. A borrower who cannot leave that headroom is answering the sell-or-borrow question with the wrong transaction.

The third cost is discipline. A sale ends the matter; a loan begins one. The facility needs a repayment source identified before drawdown: rental income from the acquired asset, planned distributions, a maturity, a future sale on chosen terms. A Lombard line with no designed exit tends to become permanent, and permanent debt secured on market-priced collateral is a position, not a bridge. The cost of that discipline is real even when everything goes well, because it constrains decisions for the life of the loan.

The final line is capacity. The bank lends against the portfolio at an advance rate, not at full value, so the facility must cover the purchase with room to spare. A purchase that only just fits inside the maximum available leaves no buffer for markets, and the buffer is what makes the contingent cost above tolerable. If the numbers only work at the ceiling, the honest reading is that they do not work.

How much room the bank offers is not a mystery to be discovered at the end. An advance rate is set against the specific portfolio in front of the institution: how liquid the positions are, how diversified across issuers and sectors, how volatile, and in which currencies they are held. A concentrated position in a single security supports far less borrowing than a diversified portfolio of liquid instruments of the same value, and some holdings support none at all. The practical consequence is that the borrowing column of the comparison can be written with real figures before anything is decided, because the custodian bank can state in writing what the portfolio will support. An investor comparing a sale against a guess is not doing arithmetic.

The comparison, done with real numbers

The method is simple to describe. On the sale side: the proceeds after tax, the income given up, and an honest view of what the deadline does to execution. On the loan side: the full interest cost over the intended holding period, the income the retained portfolio continues to pay, and the discipline the facility imposes. The two columns are compared over the same horizon, using the investor’s actual tax position and the lender’s actual terms, never generic assumptions. The exercise fits on a page, which makes it remarkable how rarely the page exists.

In one transaction we structured, an investor held a EUR 3M diversified portfolio with a Luxembourg institution and wanted to acquire a Spanish property for EUR 1.5M. Selling half the portfolio would have crystallised years of gains and dismantled a strategy that was performing. The bank applied a 60 per cent advance rate, which made EUR 1.8M available; the client drew EUR 1.5M, the portfolio remained invested and kept generating dividends and interest, and repayment was planned from rental income and future distributions.

The instructive part is not that borrowing won. In that case, on those numbers, it did. The instructive part is what made the arithmetic favourable: material unrealised gains, a portfolio worth keeping, income arriving on both sides of the balance sheet, and a purchase that fitted comfortably inside the available capacity. Change any of those inputs and the answer moves with them, which is the entire point of doing the arithmetic instead of applying a slogan.

Scenario testing belongs in the exercise too. Run both columns against a market fall. Sell, and markets fall afterwards: the seller looks clever, by accident. Borrow, and markets fall: coverage tightens and the repayment plan is tested. The comparison is ultimately between certain, immediate costs and contingent, future ones, weighted by how much buffer the structure carries. Neither column is risk-free. They are risky in different currencies.

Where the exercise fails in practice, it fails on the inputs. A tax figure taken from a general article instead of from the adviser who actually files the returns. An advance rate assumed instead of confirmed. An income line that quietly compares the portfolio’s best year against the property’s promised rent. Each shortcut favours whichever answer the investor privately preferred, which is why the two anchor figures, the real tax cost of the sale and the real terms of the facility, should be obtained in writing before the columns are compared. Both exist, and both are a request away. A decision of this size should not rest on estimates of numbers that somebody will happily put on paper.

When selling is the honest answer

The arithmetic does not always favour the loan, and pretending otherwise would repeat the door-handle error with better manners. When the unrealised gains are modest, the tax cost of selling shrinks, and with it the strongest single argument for borrowing. A portfolio bought recently, or one that has gone nowhere, gives up little by being partly sold. In that situation the loan’s interest and obligations buy the investor almost nothing.

Conviction matters as much as tax. If the portfolio no longer deserves to be held, the purchase is an occasion, not a reason. Borrowing to preserve a portfolio the investor would not buy today is momentum dressed up as strategy, and paying interest to sustain it compounds the original mistake. The sell-or-borrow question assumes the asset being protected is worth protecting. That assumption should be examined, not inherited.

Repayment capacity decides the rest. If no credible source exists to service and reduce the loan, borrowing converts a portfolio question into a solvency question, and the conservative answer is to sell. A facility serviced from hope is not a structure. Likewise, if the purchase only fits at the very top of the available capacity, the prudent versions of the loan do not exist, and the choice is really between selling and overstretching.

It is also worth remembering that the choice is not binary. Selling a portion with low embedded gains and borrowing a smaller amount against the rest often beats either pure route: less tax than a full sale, less debt and more headroom than a full loan. The right mix is an output of the arithmetic, not a compromise between temperaments.

Letting the numbers decide before the deadline does

The weakness of most sell-or-borrow decisions is not that the wrong column won. It is that no columns were ever written. The debt-averse sell, the tax-averse borrow, and both then assemble justifications for a decision that temperament already made. The deadline finishes the job: as completion approaches, whatever route can be executed fastest becomes the route, and speed is a poor proxy for arithmetic.

There is a simple procedural defence. Run the comparison at the moment the purchase becomes probable, before any binding commitment, and put a date on the page. If the completion terms later change, the columns are updated rather than rebuilt, and the funding route survives contact with the calendar. The investors who handle this well are not the ones with the strongest views on debt. They are the ones who did the sums while the sums could still change the decision.

Done early, the exercise takes little time and changes more than the funding route. A buyer who knows precisely how the purchase will be funded can commit to dates with confidence, negotiate without the seller sensing improvisation, and avoid the discounts that hesitation invites. The analysis is not only a way to choose between two costs. It is part of what makes the purchase itself go well.

Both routes are legitimate. Selling is simple, final and sometimes exactly right. Borrowing preserves the strategy at a price that must be measured, not assumed. What is not legitimate is the one-line version of the comparison, in either direction. The portfolio took years to build. The decision about how to draw on it deserves an afternoon.

Montclare does not act as a lender. It structures transactions, prepares the file and coordinates financing with authorised institutions.