ADR and Cross-Listing Arbitrage

Thesis (edge)

Some companies are listed in more than one place. A firm might trade on its home exchange in its own currency, and also trade in New York as a depositary receipt, which is simply a certificate representing a fixed number of the home shares.

Economically these are the same claim on the same company. If one receipt represents two home shares, then the receipt should be worth two home shares converted at the current exchange rate. Nothing else makes sense.

In practice the two prices wander apart. The receipt trades in one time zone with one set of buyers, the home line trades in another with different buyers, and demand in one venue is not the same as demand in the other. When the gap gets big enough, the mechanism that closes it is conversion: a holder can surrender receipts and receive home shares, or deposit home shares and receive receipts. Someone who can do that will buy the cheap version and convert into the expensive one.

The trade is to spot a gap that exceeds the total cost of converting, and take both sides.

Where it works (regimes)

• Works best: in liquid, freely convertible listings in developed markets, where the mechanism functions smoothly and the barrier to converting is small. Here the gaps are tiny and are competed away in seconds, mostly by market makers.
• Gaps are larger: in emerging markets, in names with restricted convertibility, or where capital controls exist. These are the eye-catching deviations, and they are also usually the ones you cannot capture. The deviation persists because the mechanism is blocked, and a blocked mechanism means the price difference is a real cost, not an error.
• Fails: when the two markets do not trade at the same time. If the home market has been shut for hours, the receipt price is doing price discovery on the home line's behalf. The gap you are measuring is not a mispricing, it is new information the home market has not seen yet.
• Breaks down: in crises, when convertibility gets suspended, borrow disappears, or currency markets seize up.

Signals

• The converted price. Take the home line price, apply the conversion ratio, convert into the receipt's currency. Compare with the receipt's traded price. The difference is the raw deviation.
• The no-trade band. This is the honest part. The deviation must exceed everything it costs to close: the depositary bank's conversion fee, the spreads on both legs, the cost of hedging the currency, borrow on the short leg, taxes and settlement frictions. Most raw deviations sit inside this band and are not trades.
• The staleness flag. Only trust the signal when both markets are actually trading. Outside the overlap window, treat the deviation as unreliable.
• Persistence history. For each name, ask how deviations have behaved in the past. Did they close within days? Or has the name traded at a persistent premium for years, because conversion is restricted? A persistent premium is a structural feature, not an opportunity, and history is the way to tell the difference.

Portfolio construction

• Both legs, always. Long the cheap listing, short the expensive one, with sizes matched by the conversion ratio.
• Hedge the currency. This is not optional. The position spans two currencies, and unhedged, the exchange rate move will typically swamp the small deviation you are trying to capture. Hedging costs money and that cost belongs in the no-trade band.
• Diversify across names. Any single deviation may take a long time to close. A portfolio of many such positions is far more robust than a large bet on one.
• Size conservatively. The apparent risk looks tiny because the legs are the same company. That appearance encourages leverage, and leverage is what turns a slow convergence into a forced liquidation.

Risk model

• Convergence risk. The gap widens before it closes, or never closes. If the barrier that created the gap is structural, it may persist indefinitely.
• Currency risk. Unhedged, this dominates everything. Hedged, it costs money and reduces the edge, and the hedge itself can go wrong in a currency crisis.
• Convertibility risk. If the depositary suspends issuance or cancellation, or a government imposes capital controls, the mechanism that was supposed to close your trade simply stops existing while you are in the position.
• Borrow risk. Shorting the expensive leg requires borrow, and in exactly the situations where the premium is largest, the borrow tends to be scarce and expensive. The premium and the borrow cost are often two views of the same shortage.
• Settlement and operational risk. Two markets, two currencies, two settlement cycles, one depositary bank. There is a great deal to go wrong that has nothing to do with your model being right.

Costs & implementation

• Fees are explicit and material. Depositary banks charge for conversion and often levy periodic fees on holders. These are published and must be modelled precisely.
• Two spreads plus a currency hedge. The round trip is expensive relative to the size of the typical gap.
• The overlap window is short. For many cross-listings, the two markets are open together for only a few hours a day, and that is the only time you can execute both legs cleanly.
• Data is a nuisance. You need synchronised intraday prices from two exchanges, live exchange rates, accurate conversion ratios, and knowledge of corporate actions in both venues. Getting the ratio wrong produces fake signals.
• Competition is stiff. In the liquid names, dedicated market makers already do this and are faster than you.

Failure modes

• Trading stale gaps. Comparing a closing price in one time zone to a live price in another produces enormous apparent deviations that vanish the moment both markets open. This is the classic beginner error, and it makes backtests look wonderful.
• Ignoring the currency. A trade that appears to be a clean two percent convergence can be swamped by a five percent currency move.
• Chasing restricted names. The largest premiums exist because arbitrage is blocked. The premium is not the reward for effort, it is the price of a barrier you cannot cross.
• Assuming shortability. Many of the interesting expensive legs are hard or impossible to borrow.
• Overleveraging on apparent safety. Same company on both sides feels riskless. It is not, and the leverage is what converts a slow trade into a blown-up one.
• Forgetting depositary fees. They quietly consume the edge in the exact names where the gap is small enough to be capturable.

Our Notes & Suggestions

The most valuable output of studying this trade is learning to distinguish a mispricing from a barrier. When two prices for the same asset differ persistently, the default assumption should be that something prevents them from converging, and your job is to find out what it is. If you cannot name the barrier, you probably have not found the trade yet, you have found a data problem.

Start by restricting yourself to developed-market, freely convertible listings with a proper overlap window, and accept that the gaps there are small and heavily contested. That is the honest version. The tempting version, with big fat premiums in restricted markets, is where the money goes to die.

If you do trade this, respect the operational side. More losses in cross-listing arbitrage come from settlement problems, conversion delays, borrow recalls and currency hedging errors than from the pricing model being wrong.