by Carrie
Fri Jan 17th, 2014 at 07:52:53 PM EST
Cross-posted from The Court Astrologer.
I saw something in the Financial Times which is screaming for an SFC treatment...
Beyond BRICS blog: Brazil: net debtor to the world (Jonathan Wheatley, January 15, 2014)
It has been a proud boast of Brasília for several years that it is a net creditor to the world because it holds more in foreign exchange reserves than it owes in overseas debt. However, it is far from clear that this is still the case. The issue is just one example of the vulnerabilities investors must include in their calculations of how Brazil and other emerging markets will fare as monetary policy in the developed world becomes less accommodating.
The gist of the argument is that the Brazilian Central Bank has been using currency swaps to try to drain domestic demand for dollars in an attempt to support its exchange rate. Analysts at BNP Paribas argue that the short sollar position accumulated by the BCB in the process must be subtracted from the foreign reserves, which now (after a few months of sliding Real exchange rate and increasing swap positions) means that Brazil is no longer a Dollar creditor.
Below I argue that this is wrong: Brazil remains a net creditor to the world since the swap positions with the domestic economy cannot affect the external balance (a point on which the BCB is correct). The conclusion I draw is that swaps is not a viable way to support the exchange rate. But we know a central bank cannot support its exchange rate indefinitely in any case. All this shows is that currency swaps with the domestic economy, however "clever", still cannot affect the exchange rate. You need to use reserves for that.
The argument from Beyond BRICS:
At about the same time as the country became a creditor, Brazil's central bank began using a nifty new method of intervention on foreign exchange markets. Instead of buying and selling dollars on the spot market - the standard method of central bank intervention - it used currency swaps. This is a clever alternative because it achieves the same result as buying or selling dollars with no impact on the stock of reserves.
...
The method works because it satisfies demand for foreign exchange contracts by financial market participants looking to hedge foreign exchange exposure or to speculate on movements in the exchange rate. By doing so, it removes demand from the market and has the same effect on the exchange rate as if that demand had been met by buying or selling dollars.
...
But when the US Fed began talking about tapering its QE programme last year, the real went on a slide. Since then, the central bank has upped its currency swap programme to a different order of magnitude. As Gabriel Gersztein and Thiago Alday at BNP Paribas in São Paulo pointed out in a recent note, between May 31 last year and January 10, the bank accumulated a short position on the US dollar through currency swaps of more than $77bn.
...
... Gersztein and Alday at BNP Paribas think a reasonable indication of the cost is to net out the central bank's short dollar position through currency swaps from its foreign reserves...
If we do that, we discover that, thanks to the use of its bazooka, Brazil ceased to be a net creditor to the world in October last year. The central bank's latest figures, for November 2013, show external debt at $312bn and foreign reserves at $362bn, giving a cushion of $50bn. Net out its short position through swaps of $68bn at the end of November and the cushion is gone.
I am not sure I am convinced. If I understand this correctly, the Banco Central do Brasil is offering swaps in the
domestic market with one leg indexed to the USD, but settled in BRL. I'm assuming swaps with foreign counterparties would settle in dollars, not in reals.
Assuming that's the case, then, the only effect I see from the swaps in case of a deterioration of the exchange rate is an increase in the domestic money supply in reals. This will just add to the demand for dollar assets, putting even more downward pressure on the real which is exactly the opposite of the intended effect. What this indicates is that the conclusion to draw is not that Brazil is less able to meet its foreign debt commitments, but that swaps are not an adequate way to support your exchange rate. But then we already knew central banks cannot support their exchange rate indefinitely anyway.
To sum up: if the central bank enters into currency swaps with foreign counterparties, the dollar leg of those swaps must indeed be added to the foreign reserves. However, if the central bank is conducting its operations in the domestic market with BRL-settled swaps, this has no effect on the reserve position, but it is also useless for supporting the exchange rate.
In order to make sure, I have conducted an elementary Stock-Flow-Consistent analysis. What this means is to look at macroeconomics with:
- precision regarding time
- tracking of stocks
- several assets and rates of return
- modelling of financial and monetary policy operations
- Walras' law and adding-up constraints
(from James Tobin's
Nobel Lecture)
One thing this has revealed is that I am not sure by what mechanism trade flows are supposed to affect the exchange rate within an SFC model. I should read chapter 12 of Godley&Lavoie for that, I suppose.
To isolate what's going on with the exchange rate, it is possible to trim down the SFC model substantially, to just three sectors: Brazil (BRA), Brazil Central Bank (BCB), and Rest of the World (RoW). We lose track of the GDP, but that's not much of a loss. This can be seen by looking at the larger model in which the domestic Brazilian economy is described in more detail. This would likely be unavoidable if one were to try to close the model to simulate its dynamics, but for an analysis at the level of accounting the following simplified model appears to suffice.
The only part of GDP that we're interested in is external trade. As money flows between sectors, in fact, domestic demand and income net out in the following table:
| BRA (BRL) | BCB (BRL) | RoW (USD) |
|---|
| Exports | +xR | | -xR/φ |
| Imports | -m$φ | | +m$ |
Here lower-case latin letters denote flows, superscripts denote denomination in USD or BRL, and
φ denotes the exchange rate USDBRL which, multiplied by a USD amount, yields a BRL amount. Flow variables are denominated in the currency of the seller, and each column is denominated in the appropriate currency. The table can be added row-wise by first multiplying the Rest-of-World column by the exchange rate
φ (or dividing the other two columns).
Now, imports are financed by issuing dollar-denominated liabilities and the central bank balances this by accumulating dollar reserve assets:
| BRA (BRL) | BCB (BRL) | RoW (USD) |
|---|
| New $ Liabilities | +λ(ΔL$)φ | | -λΔL$ |
| New $ Reserves | | -ρ(ΔR$)φ | +ρΔR$ |
Here upper-case latin letters are dollar-denominated stock notionals and lower-case greek letters are unit prices. The external stability goal of the Banco Central do Brasil is a positive asset balance with the Rest of the World, namely
λL$ ≤ ρR$
Next, the Brazilian private sector balances its income by issuing debt and holding cash. We assume the Rest of the World only holds dollar-denominated assets, so net real-denominated debt appearing on the table is the amount monetised by the central bank:
| BRA (BRL) | BCB (BRL) | RoW (USD) |
|---|
| New BRL Debt | +βΔBR | -βΔBR | |
| New Cash | -ΔCR-(ΔC$)φ | +ΔCR | +ΔC$ |
Update [2014-1-18 4:46:45 by Migeru]: Here's the resulting balance sheet:
| BRA (BRL) | BCB (BRL) | RoW (USD) |
|---|
| BRA $ Liabilities | -λL$φ | | +λL$ |
| BCB $ Reserves | | +ρR$φ | -ρR$ |
| BRA BRL Debt | -βBR | +βBR | |
| Cash | +CR+C$φ | -CR | -C$ |
The accumulated asset stocks give rise to interest flows. We assume interest flows are set at the start of each accounting period in the currency in which assets are denominated, but are paid at the end of the period. Start-of-period variables are denoted by a '0' subscript. Note the exchange rate does not carry a subscript. Also, held cash pays no interest.
| BRA (BRL) | BCB (BRL) | RoW (USD) |
|---|
| Interest | -λ0L$0φ | | +λ0L$0 |
| +ρ0R$0φ | -ρ0R$0 |
| -β0BR0 | +β0BR0 | |
Optionally, the central bank may enter into currency swaps with its domestic sector:
| BRA (BRL) | BCB (BRL) | RoW (USD) |
|---|
| Swaps New $ Leg | +σ(ΔS$)φ | -σ(ΔS$)φ | |
| Swaps New BRL Leg | -σ'ΔSR | +σ'ΔSR | |
| Swaps Interest | -σ0S$0φ0 | +σ0S$0φ0 | |
| +σ'0SR0 | -σ'0SR0 | |
Here the exchange rate acts at the beginning of the period, because although one of the swap legs is denominated in dollars it is all settled in Reals so the entire Real flow is determined at the start of the period. The condition on the swaps is that they are issued at par, that is,
σ(ΔS$)φ = σ'ΔSR
We can put all these tables together into one, adding one final line for the distribution of central bank profits back to the treasury:
| Transaction Flow | BRA (BRL) | BCB (BRL) | RoW (USD) |
|---|
| Exports | +xR | | -xR/φ |
| Imports | -m$φ | | +m$ |
| New $ Liabilities | +λ(ΔL$)φ | | -λΔL$ |
| New $ Reserves | | -ρ(ΔR$)φ | +ρΔR$ |
| New BRL Debt | +βΔBR | -βΔBR | |
| New Cash | -ΔCR-(ΔC$)φ | +ΔCR | +ΔC$ |
| Interest | -λ0L$0φ | | +λ0L$0 |
| +ρ0R$0φ | -ρ0R$0 |
| -β0BR0 | +β0BR0 | |
| Swaps New $ Leg | +σ(ΔS$)φ | -σ(ΔS$)φ | |
| Swaps New BRL Leg | -σ'ΔSR | +σ'ΔSR | |
| Swaps Interest | -σ0S$0φ0 | +σ0S$0φ0 | |
| +σ'0SR0 | -σ'0SR0 | |
| Central Bank profits | +p | -p | |
Update [2014-1-18 4:52:31 by Migeru]: This is supplemented by a balance sheet (although currency swaps, being derivatives, are "off-balance", here we are modelling them as two separate legs and will add each leg to the balance sheet):
| Balance Sheet | BRA (BRL) | BCB (BRL) | RoW (USD) |
|---|
| BRA $ Liabilities | -λL$φ | | +λL$ |
| BCB $ Reserves | | +ρR$φ | -ρR$ |
| BRA BRL Debt | -βBR | +βBR | |
| Cash | +CR+C$φ | -CR | -C$ |
| Swaps $ Leg | -σS$φ | +σS$φ | |
| Swaps BRL Leg | +σ'SR | -σ'SR | |
Update [2014-1-18 5:14:30 by Migeru]: And, to complete the accounting specification of the model, a reconciliation account (recording capital gains) is necessary in order for changes in total asset stock values to be equal to the negative of the 'net new assets' cash flows (from transactions, above), plus capital gains below).
| Capital Gains | BRA (BRL) | BCB (BRL) | RoW (USD) |
|---|
| From $ Liabilities | -L$0Δ(λφ) | | +L$0Δλ |
| From $ Reserves | | +R$0Δ(ρφ) | -R$0Δρ |
| From BRL Debt | -BR0Δβ | +BR0Δβ | |
| From $ Cash | +C$0Δφ | | |
| From Swaps $ Leg | -S$Δ(σφ) | +S$0Δ(σφ) | |
| From Swaps BRL Leg | +SR0Δσ' | -SR0Δσ' | |
The BCB's initial swap strategy is to lend σ(ΔS$)φ to the domestic sector as a substitute for the latter's need to fund its imports in the external market (net of domestic demand for $ cash): (λΔL$ - ΔC$)φ. This reduces the need to issue new dollar liabilities, and consequenty the need for the BCB to accumulate new dollar reserves.
However, when the BCB tries to support its exchange rate with swaps, it works in the opposite direction. In this case, the Central Bank is responding to excess demand for $ cash from its domestic sector. In that case, ΔS$ must be negative, which is why this works at cross-purposes with the ordinary use of swaps as a substitute for foreign reserves. If the stock S$ becomes negative as a result, the swap interest income -σ0S$0φ0+σ'0SR0 to the private sector is larger, the larger φ becomes. That is, as the Real depreciates, if the BCB tries to fight this with swaps it is forced to inject more and more BRL interest income into the domestic economy. This expansion of BRL income with a depreciating Real can only increase the domestic demand for $ cash from the domestic sector, which is what the BCB was trying to counter, so it is forced to issue even more swaps to support the exchange rate. This can't work, and thus it is not possible for the BCB to support its exchange rate by issuing swaps.
However, nothing that happens in the swaps sector can change the external balance, and therefore the BCB is right that Brazil remains a net creditor to the world, and BNP Paribas is wrong to add the BCB's $ swap leg to the $ reserves.
If we wanted to simulate this we'd need a simple way to extract FX demand/supply (and hence, price pressures) from this scheme. This has to come from the RoW balancing condition, as the CB profits act as a buffer variable for the BRL and BCB sectors.
