Global Macro: Trading the Economy Rather Than the Company

Global macro is approach 1 of the eight described in this sequence. The position is taken on an economy or a whole market, on an interest rate, a currency, a commodity or a broad market measure, rather than on any single company's results. The position is expressed through instruments that require a fraction of the money up front, so a small sum of cash carries a much larger exposure.

The approach turns on one change, and the change is one of level rather than one of skill. Every other way of looking at money that a reader is likely to have met so far asks a question about a business. Is this company earning more than it did last year. Is it earning more than the company across the road. Is the price being asked for a share of it high or low against what the business actually produces. Each of those questions is a good one, each has an answer, and each is asked about an organisation with a name, a set of accounts and a person who can be held responsible for it. Global macro asks a question one level up, about a country or a market as a whole, and at that level almost none of those questions can even be formed.

Start with something ordinary. A vegetable seller who has kept the same pitch for twenty years knows a great deal about that pitch. She knows which days are busy, what the shop two lanes over charges, whether her supplier has been short lately, and roughly what her stall would fetch if she ever sold it. All of that knowledge is about one small business. Now ask her a different question. Not how her stall is doing, but whether borrowing money has become dearer for everybody in the city over the last year. The question about the city's borrowing costs is a real one with a real answer, and not one thing she knows about her own stall helps her answer it. The second question is not a harder version of the first. It is a question about a different object.

The change of object is the whole shift, and once a reader has it, the rest is consequence. Change the object a position is taken on and four things change with it: what can be read about that object, what a position in it is made of, how the position is paid for, and what can go wrong with it. Those four consequences come in order below, with one arithmetic example worked in full and the three risks named at the end. Whether the approach works, whom it suits, when it might be used and what it has ever returned are not things a mechanism can settle, here or anywhere else in this sequence.

One more thing before the mechanism. Every rupee figure below belongs to Nilgiri Absolute Return Fund, an invented open-ended fund that the rest of this sequence is built on, and that fund does not run this approach. The fund runs approach 6, long-short equity. Everything worked below is a constructed case showing what the same net assets would look like if approach 1 were run instead. A reader who forgets that will come away believing this invented fund does something it does not do.

What is this approach actually looking at, if not at a company?

At an economy, or at a market taken as a single object. The formal way to say it is that the unit of analysis changes. In every company-level approach the unit of analysis is one business: the thing a view is formed about, the thing an analyst can be right or wrong about, the thing whose fortunes decide whether the position makes money. Here the unit is a country's borrowing costs, or one currency measured against another, or the price of a commodity, or the level of a whole market rather than of any company inside it.

Treating the change as a difference of scale is tempting, as though a macro position were simply a very large company position. It is not. A company and an economy are different kinds of object and the difference shows up in what is available to read. A company files accounts. A company has directors who can be asked questions, and if the contract gives a fund a seat, somebody from that fund sits in the room where decisions are taken. The company has competitors doing something similar, so its numbers can be set beside theirs. Above all the company produces earnings. A price for a share of it can therefore be divided by something the business actually generates, and the answer is a figure a person can argue about.

An interest rate has none of those things, and the absence is total rather than partial. There are no accounts for a rate. There is no management to ask, and nobody at all is accountable to somebody holding a position in it. There is no competitor rate doing a comparable job that could be set beside it in a table. And there are no earnings, so there is no division to perform and no ratio to argue about. The only comparison available is the same measure at some earlier level, and an earlier level shows where the rate has been without saying anything at all about where it should be.

Read that table once more with a particular reader in mind: somebody who has spent two years learning to read a set of accounts. Almost every skill that reader has built is a skill for the left column. None of it is wasted, but none of it transfers. The natural instinct on meeting this approach is to ask what the macro equivalent of a good set of accounts is. There is no answer. The question assumes an object this level of analysis simply has not got.

What kinds of thing does a position actually get taken in?

Four, conventionally, and it is worth naming them separately because each one sits at a different level of the same economy and a reader who lumps them together will misread the third and the fourth.

The first is an interest rate: what it costs to borrow inside one economy, quoted for a stated period of time. A currency has no price except in terms of some other currency, so the second is a currency measured always against another. The third is a commodity, meaning a physical good with a single quoted price, whose level is not the result of any one producer's decisions. The fourth is a broad market indexA measure of a whole market's level rather than of any one company., one number standing in for a whole market rather than for any company inside it.

Notice what all four share. The shared property is the definition, and the four items are only examples. Not one of the four belongs to anybody, not one of them files anything, and not one of them can be asked a question. A share belongs to a shareholder and is a claim on a company. A commodity price is not a claim on anything and belongs to nobody. A rate, a currency, a commodity and a market level therefore sit in one category despite looking nothing alike: they are all levels rather than claims.

Which of the four is easiest, which is most crowded, which is most rewarding and which anybody should look at first are not questions the mechanism answers. Nor does the mechanism say what any of the four will do next. A level carries no statement about its own future, and reading one out of the structure is not possible.

Why is a position like this posted against rather than bought outright?

Because there is usually nothing to buy. The absence of anything to buy sounds like a technicality, and it is in fact the structural fact that makes the whole approach look the way it does. If a fund wants a position in a company, it can buy shares in that company and hold them: something exists, it changes hands, and the fund pays for all of it. If a fund wants a position in the level of a whole market, or in what borrowing costs inside an economy, there is no object to hand over. A level is not a thing that can be delivered in a box.

So the position is expressed as a contract instead: an agreement between two parties whose value moves with the level in question. How such contracts are priced and settled is covered in a different subject area. One consequence of using a contract rather than a purchase matters more than any other. A contract is not paid for in full at the start. Cash is posted against it as collateral, and the amount posted is a fraction of the size of the position it supports.

The cash posted is called marginCash posted with the broker as collateral, which is a fraction of the size of the position., and margin is the hinge of the whole mechanism. Margin is not a payment and not a part-payment. The cash is security held by the broker against the possibility that the position moves the wrong way, and it comes back if the position is closed at the level it was opened at. Nothing has been bought and nothing has been sold: the fund has posted security against a contract, and until that contract is closed the fund's books carry cash with the broker and an obligation, not a holding.

Three structural consequences follow, and each of them is mechanical rather than a matter of taste. The first is that the instruments involved are usually liquid and exchange-traded, so a position can be opened and closed at a quoted price on any day the exchange is open, without a counterparty having to be found privately for each side. The second is that a position can be large relative to a fund's net assets while the fund holds very little, precisely because the money posted is a fraction of the size carried. The third is that a view can be expressed in either direction with equal ease: taking a contract that gains when the level falls involves no borrowing of anything, so there is no separate step required to be short in the way there is when the position is in shares that must first be borrowed. A short position's loss is unbounded while its gain is not, and the mechanics and arithmetic of that asymmetry are covered separately in this sequence.

Look at the two panels together and notice what is easy to miss. In the upper panel every bar is a value of something the fund actually has or actually has to give back. In the lower panel the long outline is not a value of anything the fund has. The outline is the size of a contract, and the only real number on that panel is the small solid bar at the bottom. A person reading a summary of the second fund and looking for what it holds would find Rs 50,00,00,000 of cash with a broker and conclude, quite reasonably and quite wrongly, that a tenth of the fund was engaged.

How much is posted, and how much is actually carried?

Here is the worked case, and the label on it is load-bearing. Nilgiri Absolute Return Fund is an open-ended fund with net assets of Rs 5,00,00,00,000 at its record date. The fund runs approach 6, long-short equity. Its books at that date carry shares of Rs 6,50,00,00,000 and borrowed shares sold of Rs 2,50,00,00,000, a gross exposure of 180.0 per cent of net assets and a net exposure of 80.0 per cent. Those figures are the fund's own. Everything that follows is a counterfactualA constructed case showing what would follow if something were different, clearly labelled as not having happened.: a constructed case showing what the same Rs 5,00,00,00,000 of net assets would look like if approach 1 were run instead. The constructed case has not happened and this fund has not done it.

In the constructed case the fund holds a single position under its margin arrangement with Marudhar Securities Private Limited, an invented broker. The fund posts Rs 50,00,00,000 of cash and carries a position of Rs 5,00,00,00,000. The size carried is called the notionalThe size of the position being carried, as opposed to the cash posted against it., and the margin rate is the posted cash divided by that size: Rs 50,00,00,000 divided by Rs 5,00,00,00,000, being 10.0 per cent. The 10.0 per cent belongs to this constructed position and to nothing else. A margin rate is set broker by broker and position by position, so no rate read in one place carries across to another.

The same position can now be said two ways. Both sentences are true on the same day about the same contract, and a reader will meet both. The notional is Rs 5,00,00,00,000, or 100.0 per cent of the fund's net assets. The cash posted is Rs 50,00,00,000, or 10.0 per cent of them. Neither is a rounding of the other and neither is a simplification. The two figures measure two different objects: one is the size of the contract, the other the security lodged against it, and the ratio between them is fixed by the margin rate.

The figure is worth staring at because the geometry does something prose struggles to. The pale nine tenths of that bar look like part of the position, and they are part of the position, but no money of the fund's sits there. In a purchase the shaded part and the whole bar are the same object, so a reader who has only ever bought things outright has never seen a picture in which most of a position is unpaid for. Here they are two objects, and the gap between them is not a debt in the ordinary sense either: nobody has lent the fund Rs 4,50,00,00,000. The fund has simply agreed to a contract of that size and lodged security against it.

What does one move look like when it is read two ways?

Suppose the thing traded moves 10.0 per cent against the fund in the constructed case. The loss is 10.0 per cent of the notional, so 10.0 per cent of Rs 5,00,00,00,000, being Rs 50,00,00,000. The rupee amount is a single figure and there is nothing ambiguous about it. How large that figure sounds is the part that carries two honest readings.

Read against the fund, Rs 50,00,00,000 is 10.0 per cent of net assets of Rs 5,00,00,00,000. An investor holds a share of the fund and not a share of the position, so the 10.0 per cent is the loss an investor in the fund would feel, and a serious one. Read against the cash posted, the same Rs 50,00,00,000 is 100 per cent of the Rs 50,00,00,000 that was lodged with the broker. Every rupee posted has gone. Both sentences describe the same event on the same day, and which of the two gets quoted decides entirely how large the event sounds.

The ratio between those two readings is not a coincidence of these particular numbers and it is worth stating as a rule. Because the margin is 10.0 per cent of the notional, every move is exactly ten times as large measured against the margin as it is measured against the notional. A 2.0 per cent move is 20 per cent of the margin. A 5.0 per cent move is half of it. A 20.0 per cent move is Rs 1,00,00,00,000, or 20.0 per cent of the fund and twice everything the fund posted. The multiplier is one divided by the margin rate, and it does not change while the margin rate does not change.

One caution about those ratios before the risks. Arithmetic on a constructed position says nothing about whether a move of any particular size is likely, unlikely, expected or worth planning for. Every row is arithmetic performed on a constructed position, and arithmetic is all it is. A probability quietly attached to a loss figure would be a forecast wearing a calculation's clothes, so no probability attaches to any of it.

What is the first risk, and why does nothing bound it?

The position is directionalA position that gains when one thing goes one way and loses when it goes the other.. The word directional does a lot of work, so take it slowly. A directional position gains when the thing traded moves one way and loses when it moves the other, and there is no second leg inside the position doing the opposite. Some approaches described elsewhere in this sequence are built out of two legs for exactly that reason: a general move affects both and largely cancels. A macro position has no such second leg. The view is expressed at full weight in one direction.

Now add the arithmetic already worked above, and the first risk states itself. A move against the position costs the notional times the move, and nothing in the structure caps how far the level can go. A share can only fall to nothing, and that floor at least holds under a position bought outright. A level being traded through a contract has no such courtesy: it can go a long way and it can keep going, and where the position is the other way round from the level, the loss on it is not bounded at all. The mechanics and arithmetic of an unbounded loss are covered under short selling.

The first risk needs nothing else to go wrong, and that is what makes it distinct from the other two. Nobody has to have made a mistake, no broker has to change anything, no valuation has to be disputed. The level simply moves, and the arithmetic does the rest at ten times its apparent size against what was posted.

What is the second risk, and what exactly is missing?

An anchorSomething a price can be compared with, such as a company's earnings, which an interest rate does not have.. The missing anchor is the risk that sounds abstract and is in fact the most concrete of the three, so being exact about what is absent matters more than gesturing at it.

When somebody holds a share and the price falls a long way, an uncomfortable but genuinely useful question is available to them: is the price now low against what this business earns. The question can be answered, argued about, got wrong, and revisited. An answer does not tell anybody what the price will do. An answer gives the holder a second number, produced by the business rather than by the market, against which the first number can be set. The second number is the anchor.

An interest rate produces no such second number. Neither does a currency, a commodity level or a broad market index. There are no earnings, so there is no ratio, so there is nothing the level can be called high or low against except its own history. A holder of a macro position whose level has moved a long way has exactly one comparison available, and that is where the level used to be. Where something used to be is not a reason for anything.

The absence is worth stating exactly. A missing anchor is not a criticism of the approach and not a claim that the approach is harder or worse. It is a description of what is and is not available to read while a position is open, and it follows directly from the change of level the approach begins with. Remove the company and the accounts go; remove the accounts and the anchor goes. Every step of that follows from the first.

What is the third risk, and who actually controls it?

The broker. The broker holding a position sets what must be posted against it, and the broker can reset that requirement while the position is open. Marudhar Securities Private Limited is Nilgiri Absolute Return Fund's broker, and in the constructed case it is the party that decides the margin on this position. The fund does not set that number and cannot refuse a change to it.

Follow the consequence through. The obvious reading is not the one that bites. Suppose the level being traded has not moved at all. The position is exactly where it was, the fund's view is unchanged, and nothing about the thing traded is different from yesterday. The broker raises what must be posted. The fund now has two choices and only two: find more cash and post it, or reduce the size of the position until the smaller requirement is met. A position can therefore force a decision on a fund without the thing being traded having moved a single unit.

The third risk stacks on the first rather than standing beside it. A rising requirement is most uncomfortable at exactly the moment the position has already moved against the fund. At that moment the fund has least spare cash and least appetite to sell anything to raise more. The three risks are separate objects and any one of them can bite alone, but the arithmetic of the first makes the third harder to meet. Which party holds the position, lends against it and sets that requirement is covered under prime brokerage.

How would somebody actually use any of this?

Not by running one. Most people who ever meet this approach meet it on paper rather than on a screen. The realistic use is reading. An analyst is handed a factsheet. A person at a pension pool is handed an offering document. Somebody at a bank's treasury is asked whether two funds on a shortlist are carrying comparable amounts of risk. In every one of those cases the skill being used is the ability to tell which number is being quoted.

The sentence that does the most damage when it is misread is a statement of what a fund holds. A summary of the constructed position could truthfully say that the fund holds Rs 50,00,00,000 of cash with its broker and nothing else. The same summary could equally truthfully say the fund is carrying a position of Rs 5,00,00,00,000. A reader who takes the first sentence as the size of the risk has understood the fund's position to be a tenth of what it is. So the first question to ask of any exposure line is what it is struck on: the value of things actually held, or the size of contracts carried.

The second question follows from the first and is the one that lets two funds be compared at all. Where a fund runs holdings, an exposure figure is the value of what is on its books. The 180.0 per cent and 80.0 per cent figures for Nilgiri Absolute Return Fund at its record date are struck that way. Where a fund runs contracts, the same word points at notional instead. Two funds quoting the same exposure percentage on those two different bases are not describing comparable things, and no amount of care with the percentages fixes it. The exposure arithmetic itself, gross against net and what each of the two means, is worked separately in this sequence.

The third question is about the margin rate and is the one most often skipped: what fraction of the size is being posted. The fraction posted is the multiplier on every move. At 10.0 per cent the multiplier is ten. At a different rate it is a different number, and it is one divided by whatever the rate is. A reader who has that single division can work out how far a level has to move before the cash posted against a position has entirely gone.

A fourth question is worth asking even though it has no numerical answer: what would be looked at if the position moved a long way. For a company position there is at least a set of accounts to open. For this one there is not, and the honest answer is that the level would be compared with its own history. Somebody reading a document that implies otherwise, that suggests a macro position can be checked against some underlying worth, has met a claim the structure does not support.

Why does an account of this approach stop at the mechanism?

Because the alternative is advice, and advice needs a licence, a client and somebody accountable for the outcome. The distinction sounds like a formality until the ease with which the line is crossed is taken in. An account describing eight approaches could say which one has been quiet lately, which one suits somebody with a long horizon, which one behaves differently from a share portfolio. Every one of those sentences reads like description and every one of them is a recommendation with the verb removed.

The line itself is narrow and worth stating outright. A mechanism covers what the approach takes positions in, what a position is made of, how it is paid for, and what can go wrong with it. Whether the approach works, whether it is better or worse than any of the other seven, whom it suits, when it might be used, and what it has ever returned lie outside it, for this approach and for all eight. A mechanism and the risk it carries are one job; the rest is another, done elsewhere by somebody who is allowed to advise.

There is a second reason, and it is about honesty rather than about rules. Saying what an economy is going to do would require a basis, and a description of a mechanism supplies none. Not a weak one, not a cautious one: none. An account built on invented numbers that then hinted at a real direction of travel would be pretending to a foundation it does not have. Pretending is the worse fault of the two.