Hedging. When I started out as an ETRM consultant it took me a while to get the difference between trading and hedging – or at least get a feeling that I understand. For the sake of an ETRM implementation the exact differentiation does not matter that much (as long as everyone agrees hedging is about removing risk from portfolios). What matters is to allow traders and risk managers to report exposure against the right products (and what is “right” is a whole other different topic).
You are a trader and have an open position against a less liquid product (eg 10 MW power delivery in Czech Republic several months from now – let’s assume Sep 2015). There is no liquid market for Czech power derivatives and there is no way you can hedge that position with a perfect hedge.
What do you do?
You hedge with a liquid product that is highly correlated with the underlying you need to hedge – and you would eventually re-hedge if things in Czech Republic become liquid closer to delivery or leave things as they are if you are happy.
What could such a product be?
- It could be the same product in a neighboring market – for example, you have found out a high correlation between Czech and German prices – so you can hedge with a German futures if they are liquid enough
- It could be a different product in the same market – for example if Q3-15 is liquid you can buy roughly 3.4 MW of Q3-15 and then re-hedge when the individual months become liquid. And if you are wondering why a 10MW Sep contract will be hedged with 3.4 MW Q3-15 contract, you should read this
- It could be a combination of the two (ie a Q3-15 German contract, which will then have to be re-hedged, but if it was easy it wouldn’t be fun) or any other thing that you have found to be correlated enough and liquid enough (if Czech gas turns out to be very liquid and very correlated, there you go)
Implications for reporting
The implications of this scenario (and it is a very viable one across all energy markets – examples further down) are the following:
- The system needs to be able to “translate” one exposure into another – ie report the German power equivalent of a Czech contract – so it can tell you exactly how much of German power you should buy
- The system should be able to work well with different contract lengths for flow commodities – ie figure out that 10 MW Sep-15 and 3.4 MW of Q3-15 are the same thing – as you need to know how much of Q3-15 you should buy to hedge your position
- The system should be able to handle either / or reporting – ie be able to report original position and hedges as German power (or as Q3-15) OR both as Czech power equivalent – and be able to keep the respective positions against the respective markets
Now imagine this in a scenario where you have a gas TPP in Germany, selling its output in Czech Republic, working on low calorific gas in the NCG market area, and your gas is supplied via a still unchanged old contract and is priced based on the German HEL publications (leichtes heizoel = gasoil in German). If such a scenario actually exists it is very helpful to my point for the following reasons:
- You might need to hedge your Czech power sales to eliminate the power market risk – and you are likely to do it with German products
- Low calorific gas is not a liquid product – there are some pipelines in Germany in both gas market areas that transport it, but the liquid products are the high calorific gas derivatives in NCG and Gaspool. Hence, if you want to hedge the gas price exposure, you might figure out that the low calorific pipelines are linked to the Dutch gas pipeline network, and in general NCG L gas is very highly correlated to the TTF price in the Netherlands – and TTF trading is liquid
- Finally, the HEL prices are published long after the actual delivery – and it is also not a liquid product. Fortunately, HEL actually comes either via transport or via pipelines from the Netherlands, and is traditionally hedged using Rotterdam FOB barges delivery Gasoil 0.1% sulfur. The GO FOB Rotterdam 0.1% swaps are traded liquidly and might be able to provide a good hedge. However, if your risk managers do not like it you might want to consider hedging using the ICE ULS (ultralow sulfur) Gasoil futures
All of the above means:
- You have to translate Czech power into German one (and vice versa)
- You have to translate German L Gas into TTF (and again vice versa)
- You will need to calculate the GO 0.1% FOB Barges Rotterdam equivalent of your HEL exposure and then turn this into ULS Gasoil futures (by roughly splitting a month into 40% of this months’ futures contract and 60% of next months’ futures contract – but this is a separate topic, that goes in between through the Gasoil 1st line swap on ICE, so I will keep things simple for now)
Implications for the ETRM solution
For the ETRM solution the above might not mean anything – it will not be an exception if the main solution traders capture deals in offers reporting positions only against the traded positions or the direct underlying asset prices. This would translate the example above in the best case in a position in Czech power, a position in NCG L Gas (or NCG only if it cannot or is not configured to differentiate between L Gas and H Gas grid within a market area) and HEL.
However, if you are a bit more aspiring about functionality, then the example about would mean several interesting things:
- Capturing correlations between asset prices (eg German and Czech power, NCG L Gas and TTF, HEL and GO FOB Barges Rotterdam 0.1%)
- If we take capturing correlations one step further – driving curves from each other within the system (a standard case is a curve being modelled as a spread off a highly correlated liquid market)
- Capturing differences in products on the same underlying: specifically in the oil case above – if you want to be flexible in hedging the HEL exposure with different gasoil products, the system should be model relationship between several curves – the Gasoil futures, the Gasoil 1st Line Swap (which is a monthly swap settled against futures settlement prices for each working day of the month and therefore has exposure against 2 futures contracts – as the futures contract settles in the middle of the month), and the Gasoil FOB Barges Rotterdam swap(which is typically modelled as a spread off the Gasoil 1st Line Swap)
- The last point makes for 3 products on the same underlying (FOB barges delivery of gasoil in the Rotterdam area), BUT with different price settlement (expiration date for the futures, monthly average of first nearby futures daily closing prices for the 1st line swap, and monthly average of PLATTS published prices for the OTC market). In the oil market this relationship expands across multiple products via crack spreads and diffs and you end up with a curve tree possibly starting with Brent or Gasoil futures (although you might want to model the Gasoil futures using a crack spread from Brent) and then adding up cracks and diffs to arrive at a large number of fuel oil, gasoil, and diesel product variations. If such a tree is to be modelled in the ETRM system, the latter needs to allow for multiple steps in the curve modelling including unit conversions and spreads at each step. The good news is that if mostly financial oil is traded, remembering 3 to 5 conversion factors and what unit the price of a crack spread is in and what unit the notional is in takes you a long way in designing a well working solution
The bottom line
Traders will most of the time not have a perfect hedge available in the market. In such cases they would go to the next best thing that covers their exposure and re-hedge later if market conditions change. This means a solution covering risk reporting (whether in the direction of Middle Office or Front Office) should be very flexible in reporting exposure against the right curve in the right units. The “right” curve might mean either the curve representing the traded underlying or the liquid “best hedge” one. Finally, the relationship between a traded underlying and a best hedge for it might be fairly direct and straightforward or go through several steps of adding spreads, unit conversions, or multiplying with specific correlations that a regression model shows.