What a currency peg does

The undertaking behind the number 

An earlier lesson in this module set out what a fixed exchange rate costs an institution in domestic policy independence. This one is about the other half: what the institution actually does, day after day, to keep the rate where it says the rate is. A peg is announced as a policy and maintained as an operation, and the two are easy to conflate, because the announcement is public and the operation is not.

The operation is a standing offer on both sides. A central bank running a peg deals with the commercial banks in its jurisdiction, not with the public, and it will buy domestic currency from them for anchor currency, or sell domestic currency to them for anchor currency, at the published rate or within a narrow published band around it. Because that offer exists, no bank has any reason to deal with another bank away from it, and the market rate arrives at the official rate without the central bank having to trade at all on most days. The obligation does the work; the transactions are what happens when somebody tests it.

The two sides of that offer are not symmetrical, and the asymmetry is the single most important structural fact about pegs. Meeting demand for domestic currency is unconstrained, because a central bank issues the domestic currency and can create as much of it as the offer requires. Meeting demand for anchor currency is constrained, because the central bank cannot create the anchor currency and can only pay out what it holds. A peg is therefore open ended in one direction and finite in the other, which is why every discussion of a fixed rate arrangement eventually becomes a discussion about reserves.

How the rate is held 

Three instruments hold a peg, and they are used in roughly this order. The first is the domestic policy rate, kept close enough to the anchor's that holding money in one currency rather than the other does not pay for itself. That is the mechanism of the earlier lesson, and it is a preventive instrument: it removes the reason for a one way flow before the flow starts.

The second is intervention, which is the buying and selling described above once it is actually happening. When holders present domestic currency for conversion, the central bank takes it in and pays out anchor currency from reserves, and reserves fall. When holders present anchor currency, the central bank takes that in, issues domestic currency against it, and reserves rise. The arithmetic is a single multiplication by the parity in each direction.

That last exclusion is the third instrument. Intervention changes the quantity of domestic currency in the banking system as a by product of holding the exchange rate, and a central bank that wanted the exchange rate held but not the domestic liquidity changed will offset the effect, conventionally by issuing or redeeming its own short dated bills, by adjusting what banks must hold on deposit with it, or by moving the terms of its liquidity operations. Practitioners call the offset sterilisation. It is why the balance sheet of an institution running a peg can be doing something quite active while the price it publishes does nothing at all.

Two refinements are worth holding. Most pegs are a central parity with a narrow band rather than a single number, so the institution is inactive inside the band and deals at the edges; a band is a stated tolerance, not a different regime. And a peg can be written against a basket of currencies rather than one, in which case the arrangement is still fixed, but fixed to a weighted average that itself moves against each of its members.

The dirham and the riyal 

The dirham and the riyal are each held at a fixed published rate against the United States dollar and each is freely convertible into it, and both arrangements have stood for decades. They are maintained by the mechanism above, with the domestic policy rate carrying most of the load, which is why the announcements out of the two institutions follow the anchor's so closely. The Kuwaiti dinar is the regional exception and is managed against an undisclosed basket instead, so it moves in small amounts against the dollar where its neighbours do not.

One consequence of a fixed spot rate is easy to miss and is the part most worth knowing. A view about a pegged currency cannot be expressed in its spot price, because the spot price is administered. It is expressed instead in the forward market, where the price of exchanging the two currencies at a future date is quoted freely, and in the domestic money market, where the difference between local and anchor interbank rates is not fixed by anybody. Those two places are where the pressure on a fixed rate arrangement becomes visible as a price, and the spot chart is where it is invisible by construction.

What it does to a pair 

An administered price has no discovery process behind it, so the observed volatility of a pegged pair is close to nothing, and the distance it travels in a session is measured in the last decimal places of the quote. The cost of transacting it does not fall to nothing alongside the movement. It is set by the firms willing to quote the pair, and there are fewer of them precisely because there is so little turnover in it, so the quoted spread on a pegged pair is commonly wide relative to a pair that moves a great deal more.

The ratio is the point rather than either figure. When the distance a price travels shrinks toward zero and the cost of transacting it does not, cost stops being a deduction from the arithmetic and becomes the dominant term in it. The same reasoning extends to time. A position held past the daily cut off carries a financing adjustment derived from the difference between the two currencies' interest rates, and under a peg that difference is small and stable rather than absent, so on a long hold the financing term can be larger than anything the price itself contributes.

There is a second, quieter effect. Because the pair barely moves, the ordinary chart reading described earlier in the curriculum has very little to read: ranges, averages and breakouts are all computed from price variation, and a price held still by policy supplies almost none. Technical description of a pegged pair is therefore description of an administered number, which is a different object from the one those tools were built for.

Concentration, not removal 

A peg is often described as removing currency risk for holders in the pegged economy. What it does is more specific, and it runs in two directions at once. Against the anchor, and only against the anchor, the exchange rate is held, so a position denominated in the anchor currency converts back into the domestic currency through a constant. Against everything else the domestic currency inherits every move the anchor makes, for reasons decided elsewhere, as the earlier lesson set out.

The consequence for a portfolio is a correlation result. Positions across different instruments, different asset classes and different regions look diversified when they are listed, and if every one of them is denominated in the anchor currency then the currency component of each is the same component repeated. Diversification across instruments does not diversify a currency exposure; it multiplies the number of places one exposure is held. A peg makes that concentration comfortable to hold rather than making it smaller, because the conversion step that would ordinarily reveal it is a constant.

The second direction concerns the shape of the risk rather than its size. Risk measured as observed price variance is a statement about how a price has behaved, and the price of a pegged pair has behaved the way it has because it is managed. The pressures that would otherwise move it do not stop existing when the rate is fixed; they show up in the quantity of reserves, in the forward curve and in the domestic money market instead. Any measure built from spot variance therefore reads a managed price as an unusually quiet one by construction, which is true as a description of the past and is not evidence about anything else.

Peg, managed float and free float 

Exchange rate arrangements are usually taught as three boxes and are better read as a spectrum, with two variables running along it: how much the authorities commit to, and how much of that commitment they publish.

• At the most committed end sits a hard peg. A currency board issues domestic currency only against anchor currency reserves held one for one and is constrained by law rather than by policy, and a country can go further still by adopting another currency outright and issuing none of its own.
• A conventional peg states a parity against one currency, or against a basket, holds it within a narrow band, and maintains it with the policy rate and the intervention described above. Both regional arrangements in this lesson are of this kind.
• A crawling peg or crawling band keeps the mechanism but moves the parity itself, either on a published schedule or in response to a stated variable such as an inflation differential, so the commitment is to a path rather than to a level.
• A managed float publishes no target at all. The authorities intervene, sometimes heavily, but because no level is announced the market cannot compute what is being defended. Discretion is retained and the credibility that comes from a published commitment is given up.
• A free float leaves the rate to transactions between willing parties. Intervention is rare and is conventionally described as addressing disorderly conditions rather than a level, and the currencies most heavily traded in the sessions covered earlier in this module sit here.

Two complications sit on top of that list. The first is that what an institution says and what its data shows can differ in both directions, which is why international bodies publish a classification based on observed behaviour alongside the announced one: a currency described as floating can move so little against one anchor that its behaviour is indistinguishable from a peg, and a currency described as pegged can be adjusted often enough that the label flatters it. The second is that a currency can travel along this spectrum without any announcement at all, because the observed behaviour changes first and the classification follows.

Where practitioners disagree 

The first argument is whether a pegged pair is a market instrument at all. One position holds that an administered price with no discovery process is not a market, that the spot quote carries no information, and that whatever is left in the pair is financing. Another holds that the currency is traded actively, just not in the spot quote: the forward curve and the domestic interbank market are where positions are expressed and where pressure becomes visible. Neither side disputes the facts. They disagree about what counts as trading a currency, and the disagreement is unresolved because the two definitions never meet.

The second argument is whether a peg reduces risk or relocates it. Measured by the variation of the spot price, it plainly reduces it, and that measurement is not in dispute. Measured by the shape of the distribution, the objection is that variance and discontinuity are different things and that a low reading on the first says nothing about the second. Both measurements are correct about what they measure. There is no test that settles which is the better description of an arrangement while the arrangement is in place, which is exactly why the argument recurs.

The third argument is about the taxonomy itself. One tradition treats the categories as nearly empty, on the grounds that almost every float is managed to some degree and almost every peg allows a band, so only observed behaviour carries information. Another argues that the announcement is itself a mechanism, because a published commitment changes how banks, importers and lenders behave, and two currencies with identical observed behaviour are not in the same position if one of them has made a promise and the other has not. The second view has the harder thing to measure and the first has the harder thing to explain.