Tuesday, 23 February 2016


Liquidity refers in several different ways to the degree to which a person or organisation can transform securities of one type into securities of a more generally or widely accepted type.  This sounds vague, but I wanted to start my examination of liquidity by stating it thus.  

Next, I'd like to point out that any security in the world can be liquid at one moment and then later can be illiquid.  Third, I'd like to state that liquidity can be considered a ranking measure applied to all assets, from most to least liquid and that this ranking is context dependent.

The context can be time-specific (what now is liquid may not be liquid tomorrow; what now is illiquid may later be perceived as liquid), market-specific (the market associated with a particular security can have its own liquidity measures), holder-specific (what may be liquid to you because you only own one unit may be dramatically less liquid for me, if I own a large fraction of the amount outstanding of the security) and scenario-specific (holdings may be considered liquid under certain 'normal' market conditions by simultaneously less liquid under others - e.g. in a forced sale).  In this last case there is sometimes a real need or an imagined need to perform the transformation to perform some obligation.  Two examples of this are when a bank run occurs and the institution struggles to transform its assets back into cash to meet lender cash requirements; and when investors in hedge funds demand the return of their capital or regulators of those hedge funds demand to see how the manager could plan to liquidate its assets under management in a way which satisfies investor demands.

Finally, in estimating the liquidity of an entity's set of holdings, there may or may not be any additional liquidity constraints in place, which would alter the liquidity profile of those assets - I'm thinking here of clauses in hedge fund offering memoranda to investors which aim to remain fair of the average investor at all points during a forced unwind.

At its most general, liquidity is a relative measure between two arbitrary securities.  But, rather like the extension of the capital asset pricing model by Sharpe to the concept of a correlation between a security and the market index, it is convenient to consider all of these relative liquidity reads to be between the security in question and the single most liquid instrument, often considered to be cash in the local currency of interest.

One perfectly acceptable result of a liquidity analysis is a simple ranking of securities where the ranking is in effect for a known period of time.  Another is an arbitrarily scaled measure, with 0 occupying the value associated with the currently most liquid security.   This is like the ranking approach, but with the distances between securities having a common method of interpretation.   These two can be applied at the level of the market.

A third, appropriate only at the holding level, per security, would be a quadruple of fraction sold, time period, percentage cost (what fraction of the asset value will be lost in performing this transform at this time and in this size), constraint set (which a common theoretical target liquidity asset being cash in the local currency).  Let me explain all  four parts.

For a given constraint set (e.g. the company needs to raise 1,000,000 GBP in the next 4 weeks to meet a bond coupon payment; a hedge fund needs to satisfy an investor stampede to the exits, which means that a gating schedule sets up a demand schedule for cash over the coming 6 months), a company can examine its holding set and decide to fix two of the three remaining dimensions, and examine the effect on the third.  It could, for example fix the fraction sold to be 100% and the percentage cost tolerance to be <1% then see how long it would take to sell all the holdings of the asset.  It could, alternatively, hold the time to be <1d and the fraction to be again 100% and see just how much the percentage cost tolerance increases.  Thirdly, it could keep the cost tolerance at <5% and the time to be <1d to see how much they could sell of their holding set.