The Price of Illiquidity

01 — Origins: Why Liquidity Has a Price

The formal study of liquidity as a priced characteristic of assets is relatively recent. Classical finance, from Markowitz through the CAPM, treated markets as frictionless. Assets were assumed to convert to cash at their fair value without cost or delay. The spread between bid and ask, the depth of the order book, the time required to find a counterparty — none of these entered the model. Return was compensation for systematic risk. Everything else was noise.

The breakdown of that assumption became evident not through theory but through practice. Institutional investors operating in thin markets — small-cap equities, municipal bonds, emerging market debt — observed systematically higher returns than CAPM predicted. The excess could not be explained by beta. Something else was being priced. Amihud and Mendelson were the first to formalise it rigorously, demonstrating in 1986 that expected return increases with the bid-ask spread, that clientele effects emerge (long-horizon investors hold less liquid assets), and that the relationship is concave — the marginal premium per unit of illiquidity diminishes as illiquidity rises (Amihud & Mendelson, 1986).

The intuition behind the premium is not complicated. An investor who holds an asset she cannot sell without incurring material cost or delay bears a risk that does not appear in a single-period return: the risk of needing liquidity at the wrong moment. This is not credit risk or duration risk. It is exit risk — the possibility that the price at which she can actually transact diverges materially from the price at which the asset is marked. In a crisis, that divergence is not marginal. It is the entire trade.

Kyle's 1985 model of informed trading introduced a complementary mechanism. In Kyle's framework, liquidity — defined as the ability to trade without moving the price — depends on the information structure of the market. Where informed traders are present, market makers widen spreads to protect themselves against adverse selection. Illiquidity is not just a transaction cost; it is a signal of information asymmetry (Kyle, 1985). Assets with thin trading are thin precisely because few people are confident in their valuation. That uncertainty is itself a cost to the uninformed buyer.

These two strands — transaction cost and information asymmetry — together explain why illiquidity commands a premium. The investor is compensated for (i) higher round-trip transaction costs, (ii) a reduced ability to rebalance in response to changing information, and (iii) exposure to the possibility that, at the moment she needs to sell, the market for her asset has effectively ceased to function.

02 — Deriving the Illiquidity Premium: Models and Mechanics

The most tractable derivation of the illiquidity premium begins with the bid-ask spread as a proxy for transaction costs. If an investor pays the ask and must sell at the bid, the round-trip cost reduces the net return of the trade. For a buy-and-hold investor with horizon T, the cost is amortised over the holding period. A shorter horizon means a higher annualised drag; a longer horizon means a lower one. This is why illiquid assets migrate to long-horizon holders — they are the rational owners.

The Amihud Illiquidity Ratio

The most widely used empirical proxy for illiquidity is the ratio of a day's absolute return to its dollar trading volume: ILLIQ = |r| / Volume. A high ratio indicates that a small amount of trading moves the price substantially — the asset's price is sensitive to order flow. This can be aggregated across days and stocks to produce a market-level illiquidity measure, which exhibits strong counter-cyclical properties: it spikes during crises and compresses in bull markets.

Acharya and Pedersen extended the CAPM to incorporate liquidity risk directly, producing the liquidity-adjusted CAPM (LCAPM) (Acharya & Pedersen, 2005). Their model yields an expected return that has four components: the standard market beta premium; a premium for the covariance of the asset's illiquidity with market illiquidity (assets that become illiquid precisely when the market does are most dangerous); a premium for the covariance of the asset's return with market illiquidity (assets that fall in value when liquidity dries up); and a discount for assets whose illiquidity covaries negatively with market returns (rare; these are natural hedges). The model's central insight is that not all illiquidity is equivalent. An asset that is merely hard to trade in normal times is different from one that becomes untradeable in a crisis. The latter commands a much higher premium — or should.

For private assets — unlisted equity, real estate, private credit, infrastructure — the derivation of an illiquidity discount takes a different form. The premium and the discount are the same phenomenon viewed from opposite sides of the transaction: the discount is the lower price paid at acquisition; the premium is the higher return earned over the holding period as a consequence. One is the cause, the other the effect — and in a well-functioning market, they should be equivalent in present value terms. The standard approach to measuring this is the restricted stock discount: studies of transactions in which identical equity was sold both freely tradeable and subject to a lock-up consistently show a discount of 20–35% on the restricted shares (Silber, 1991). The discount reflects the expected cost of illiquidity over the lock-up period, capitalised into the initial price.

More sophisticated models derive the illiquidity discount from option theory. Longstaff's 1995 model treats the liquidity constraint as forgoing the option to sell at the optimal time (Longstaff, 1995). The value of that option is bounded by the value of a lookback option on the asset — since the optimal exit point is, in retrospect, the peak. For a volatile asset with a long lock-up, this bound is not trivial. Even for moderate asset volatility of 25% and a one-year lock-up, the upper bound on the illiquidity discount exceeds 15%.

The determinants of the illiquidity premium, drawn from both empirical and theoretical work, can be organised as follows:

• Asset-level: bid-ask spread — the baseline transaction cost; the primary driver in Amihud-Mendelson
• Asset-level: trading volume and depth — thinner markets imply greater price impact per unit traded
• Asset-level: price volatility — higher volatility increases the option value of timing, thus the cost of losing it
• Asset-level: lock-up duration — longer constraints compound the discount non-linearly
• Market-level: systemic liquidity — Amihud's market ILLIQ; conditions under which all assets become hard to sell simultaneously
• Investor-level: horizon — shorter-horizon investors require higher compensation; the premium is partly an endogenous clientele effect
• Investor-level: funding liquidity — investors who may face redemptions or margin calls cannot tolerate illiquidity; their required premium is higher

The interaction of the last two factors deserves emphasis. The 2008 crisis was, among other things, a demonstration that funding liquidity and market liquidity are endogenous to each other. When investors face forced redemptions, they sell what they can — typically their liquid holdings — which concentrates the illiquid residual, raises its effective weight in the portfolio, and increases the urgency of exit precisely as the market for those assets disappears. Brunnermeier and Pedersen formalised this spiral in 2009, showing that margin requirements tighten as prices fall, forcing further deleveraging, which further depresses prices (Brunnermeier & Pedersen, 2009).

03 — Accounting Treatment: The Architecture of Optimism

The accounting framework governing illiquid assets is structurally disposed toward overvaluation. This is not a design flaw. It is an unavoidable consequence of requiring entities to report fair values for assets that have no observable market price. The result is a set of rules that delegate the most consequential measurement decisions to the entity whose interests are served by a higher number.

Both IFRS 13 and US GAAP ASC 820 require fair value measurement and organise inputs into a three-level hierarchy (IASB, IFRS 13, 2011) (FASB, ASC 820, 2011):

• Level 1 — quoted prices in active markets for identical assets; no adjustment permitted
• Level 2 — observable inputs other than Level 1; includes quoted prices for similar assets, or identical assets in inactive markets
• Level 3 — unobservable inputs; valuation based on the entity's own assumptions about what market participants would use

Level 3 is the problem. It captures precisely the assets whose fair value is most uncertain and most subject to managerial discretion. Private equity NAVs, illiquid structured credit, direct real estate, infrastructure assets, and private credit positions all commonly sit at Level 3. The reporting entity determines the discount rate, the comparable transaction set, the exit multiple, and whether — and by how much — to apply an illiquidity discount. External auditors review the model and the inputs; they do not independently derive the value.

The smoothed mark is not a measurement of value. It is a record of what management believed the value to be, at a point in time, using assumptions that no transaction has tested.

Three specific accounting dangers follow from this architecture.

Stale pricing and volatility suppression

Private asset valuations are typically updated quarterly, or upon a triggering event. In the interim, the carrying value does not move. This creates the well-documented phenomenon of return smoothing: private equity and real estate funds report returns that appear to have lower volatility and lower correlation with public markets than is economically real. Getmansky, Lo, and Makarov demonstrated in 2004 that observed serial correlation in hedge fund returns is largely an artefact of stale pricing rather than genuine persistence (Getmansky, Lo & Makarov, 2004). The same mechanism operates in private equity. A fund's reported Sharpe ratio is flattering not because the underlying assets are genuinely less risky, but because the measurement frequency artificially compresses the denominator.

The institutional consequence is that allocation models that include private assets will systematically understate portfolio risk and overstate diversification benefit. Mean-variance optimisers will over-allocate to the smoothed series. This is not a theoretical concern — it is a documented feature of the modern endowment model and of retail multi-asset funds with illiquid sleeves.

NAV inflation and deferred loss recognition

Because the illiquidity discount is management's to apply, the incentive structure runs toward minimisation. A lower discount produces a higher NAV, which supports fundraising, reduces the appearance of impairment, and delays the recognition of losses. This is not fraud in the normal case — it is the exercise of discretion within a permitted range. But it means that carrying values for illiquid portfolios systematically overstate realisable value in a forced-sale scenario.

The gap between carrying value and realisable value is not a rounding error in stress conditions. During the 2008 crisis, AAA-rated tranches of residential mortgage-backed securities carried at par — or close to it — by major banks sold at 20–40 cents on the dollar when forced transactions occurred. The accounting hierarchy had permitted a fiction that the market ended abruptly.

Duration mismatch in open-ended structures

Perhaps the most structurally dangerous configuration is the open-ended fund holding Level 3 assets. The fund offers daily or weekly liquidity to investors — a Level 1 or Level 2 obligation — while holding assets that require months or years to liquidate without material price concession. The fund's NAV is stated at the mark; redemptions are met at that mark. As long as the fund is in net inflow, the mismatch is invisible. Once net redemptions begin, the fund must sell liquid assets first, concentrating the illiquid residual, which raises its weight in the portfolio and accelerates the divergence between NAV and realisable value. The gating decisions imposed by Woodford Investment Management in 2019 and by several real estate funds in 2022–23 were the mechanical consequence of exactly this structure (FCA, 2020).

IFRS 13 requires disclosure of the sensitivity of Level 3 valuations to changes in unobservable inputs. In practice, these disclosures are of limited utility — they describe what happens if a particular input changes by a specified amount, but do not address the scenario in which the market for the asset ceases to function entirely, which is the risk that actually matters.

04 — Historical Analysis: When the Premium Is Collected

The illiquidity premium has a distinctive temporal profile: it is promised in normal conditions, appears to be earned in bull markets, and is collected — involuntarily and violently — in crises. The historical record across asset classes is consistent on this point.

The long-run evidence

Amihud's 2002 empirical study of US equities demonstrated a robust positive relationship between stock-level illiquidity (measured by his ILLIQ ratio) and subsequent excess returns, controlling for size, beta, and prior returns (Amihud, 2002). The premium is economically significant: moving from the most liquid to the most illiquid quintile of stocks yielded an annualised return differential of roughly 4–8% over the 1964–1997 sample. Subsequent replications in international markets confirmed the pattern, with the premium tending to be larger in markets with weaker investor protections and less developed secondary trading infrastructure.

In fixed income, the liquidity premium shows up most clearly in the spread between on-the-run and off-the-run US Treasury securities — bonds with identical credit risk and near-identical duration that trade at a persistent yield differential solely because the on-the-run issue is more actively traded. This spread, typically 5–15 basis points in normal conditions, widened to over 50 basis points during the 2008 crisis — a stark illustration of how fast the premium can reprice (Krishnamurthy, 2002).

2008: the structured credit collapse

The 2007–2009 crisis was, in its financial mechanics, a liquidity crisis as much as a credit crisis. Structured products — CDOs, CLOs, RMBS tranches — had been priced on the assumption of continuous market function. When interbank funding markets froze in August 2007 and then catastrophically in September 2008, the secondary market for these instruments did not just widen; it ceased to exist for extended periods. Dealer balance sheets, already leveraged, could not absorb the inventory. The absence of price discovery meant that Level 3 marks sustained values that any informed participant knew were fictional.

The crisis revealed a second dimension: liquidity risk is correlated across asset classes precisely at the moment diversification is most needed. Equities, credit spreads, commodity prices, and currency pairs all moved together in the September–October 2008 period. The liquidity-adjusted CAPM's prediction — that assets which become illiquid when the market does deserve a higher premium — was validated in the worst possible way.

2020: the COVID dislocation

March 2020 produced a shorter but in some respects sharper liquidity dislocation. Even the US Treasury market — the deepest, most liquid market in the world — experienced severe dysfunction, with bid-ask spreads widening to multiples of their normal levels and the Federal Reserve ultimately intervening to purchase Treasuries directly. Corporate bond ETFs traded at discounts to their stated NAV of 5–7%, revealing the gap between the price at which the underlying bonds could theoretically be valued and the price at which they could actually be sold (O'Hara & Zhou, 2021).

The episode was instructive because it demonstrated that illiquidity is not confined to genuinely illiquid assets. Given sufficient selling pressure and balance sheet constraints, liquidity can evaporate from instruments that institutional investors treat as effectively cash. The illiquidity premium, in that sense, is latent in every asset and becomes manifest when the aggregate demand for liquidity exceeds the aggregate supply of it.

2022: the UK gilt crisis

The UK gilt market disruption of September–October 2022 illustrated the specific danger of duration mismatch in institutional structures. Defined benefit pension schemes, operating under liability-driven investment strategies, held leveraged gilt positions as duration hedges. When gilt yields spiked sharply following the Truss government's fiscal announcement, the collateral calls on those positions exceeded the liquid assets available to meet them. The funds were forced to sell gilts into a falling market, which depressed prices further, which triggered further collateral calls. The Bank of England's intervention in the long-end gilt market on 28 September was not a policy choice — it was a financial stability necessity (Bank of England, 2022).

The LDI crisis is a pure expression of the liquidity spiral described by Brunnermeier and Pedersen. The pension schemes held assets they believed they could sell; under stress, the selling itself destroyed the market. The illiquidity premium they had not explicitly priced was collected in full over two weeks.

05 — Outlook: Structural Elevation in the Late 2020s

The conditions entering the late 2020s are structurally more illiquid than any prior period in the institutional investment era. Three forces account for this.

First, the scale of private market allocation. Assets under management in private equity, private credit, and real assets exceeded $13 trillion globally by 2024, having roughly tripled over the preceding decade (McKinsey Global Private Markets Review, 2024). Institutional portfolios that held 5–10% in alternatives in 2010 routinely carry 20–30% today. This is not inherently dangerous — long-horizon capital should hold illiquid assets. But the speed of the expansion has outrun the development of secondary markets, which remain thin, information-asymmetric, and accessible primarily to the largest participants.

Second, the democratisation of private assets. Retail-accessible structures — European Long-Term Investment Funds (ELTIFs), interval funds, perpetual NAV vehicles — have extended private market access to investors who have neither the liquidity reserves nor the horizon patience that the underlying assets require. The structural mismatch is identical to Woodford and the 2022 real estate funds, simply at greater scale. When redemption pressure arrives — and it will — the absence of a deep secondary market will translate directly into forced discounts or imposed gates.

Third, the interest rate reset. The 2022–2024 rate cycle significantly altered the economics of private credit and real estate. Assets underwritten at low yields now carry both a mark-to-market impairment and reduced debt serviceability. The slow repricing of private portfolios — which Level 3 accounting permits to occur gradually rather than immediately — has masked the extent of the adjustment. The gap between carrying values and transaction values in secondary private equity markets has been material and persistent, with secondary buyers typically pricing portfolios at 85–92 cents on the dollar of stated NAV.

The secondary market gap as a real-time illiquidity signal

Secondary private equity transaction prices provide the most direct observable measure of the illiquidity discount embedded in private portfolios. When well-managed institutional funds with diversified portfolios transact at 88 cents on the dollar of NAV, that eight-to-twelve point discount is the market's estimate of the combined cost of illiquidity, information asymmetry, and uncertainty about the accuracy of the mark itself. It is not a distressed signal. It is the normal price of private market participation.

The macro environment adds a further dimension. Central bank balance sheet normalisation has withdrawn a significant source of liquidity from fixed income markets. Dealer balance sheets — constrained by post-2010 regulation — have not expanded to fill the gap. The ratio of corporate bond outstanding to dealer inventory has risen sharply, meaning that the market can absorb less selling pressure before prices dislocate. The next credit cycle downturn will encounter this thinner market structure.

The mitigant — partial and imperfect — is the growing institutionalisation of secondary markets and continuation vehicles in private equity. GP-led secondaries, NAV lending, and tender offer structures have created more exit optionality than existed in 2008. But volume in these markets remains a fraction of the stock of illiquid assets that would need to transact in a genuine stress scenario. They provide exit for the prepared seller; they do not provide liquidity for the system.

— The Premium That Prices Itself

The illiquidity premium is unusual among risk premia in that it compounds its own danger. An investor who earns the premium does so by accepting reduced optionality — fewer rights to exit, fewer mechanisms to respond to new information. This is tolerable when the investment performs as expected. When it does not, the illiquidity that was supposed to be compensated becomes the binding constraint.

The accounting framework, rather than disciplining this dynamic, accommodates it. Level 3 valuation permits the carrying value of an impaired illiquid asset to remain close to its acquisition price until a transaction forces otherwise. The reported premium appears to be earned — consistently, smoothly, with low volatility — right up until the exit. At that point, the premium is either confirmed or consumed.

What the historical record suggests, and what the current structural configuration implies, is that the premium has been systematically underpriced across the private markets expansion of the 2010s and 2020s. The gap between NAV and secondary transaction prices is not noise. It is the market's estimate of what the illiquidity cost actually is, stated plainly in the only currency that matters: money changing hands.

Investors who hold illiquid assets without an explicit, independently derived illiquidity premium — priced at the portfolio level, not borrowed from public market analogues — are not earning a premium. They are deferring a reckoning.