Data Warehousing and Data Science

4 July 2016

Historical Portfolio Positioning

Filed under: Analysis Services — Vincent Rainardi @ 7:08 am

One of the most difficult chart to create in asset management sector is Historical Duration Positioning like this:

Sample portfolio

First we have to get the numbers for each category for each month (at least each quarter), e.g. in January the portfolio was as follows:

  1. Government 25%, 1.1 years
  2. IG Corporate 20%, 0.7 years
  3. EM Sovereign 10%, 0.4 years
  4. EM Corporate 5%, 0.3 years
  5. High Yield US 10%, 0.2 years
  6. High Yield non US 5%, 0.4 years
  7. Securitised 10%, 0.7 years
  8. Convertibles 10%, 0.9 years
  9. Cash 5%, 0 years

The 2 numbers above are market value weighting (%) and duration (years). In the above the portfolio duration was 4.7 years (sum of all the durations from 9 categories).

Notice that the categorisation above is cutting across 6 different dimensions:

  1. Government/sovereign bond vs corporate bond
  2. Investment Grade bond vs High Yield bond
  3. US vs non US vs Emerging Market
  4. Securitised vs non-securitised
  5. Convertible bond vs non-convertible bond
  6. Cash

To create this report, first we need to identify cash and cash equivalent positions. Then convertibles and securitised. Then HY vs IG, sovereign/government vs corporate, then Emerging vs Developed Markets.

Because on point 3 we have EM Sov, it means that on point 1 the government is DM (developed market). Because on point 5 & 6 we have HY it means that point 1 to 4 are IG. The business analyst are expected to know that “sovereign” is equivalent as government, and “government” usually means investment grade debt in developed market.

So the actually meaning of the categories are:

  1. DM IG Government 25%, 1.1 years
  2. DM IG Corporate 20%, 0.7 years
  3. EM IG Sovereign 10%, 0.4 years
  4. EM IG Corporate 5%, 0.3 years
  5. HY US 10%, 0.2 years
  6. HY non-US 5%, 0.4 years
  7. Securitised 10%, 0.7 years
  8. Convertibles 10%, 0.9 years
  9. Cash 5%, 0 years

Total is not 100%

Point 1 to 8 above are debt, which means that they exclude cash, but they include debt derivatives (not just bonds), but not other derivatives. So if the portfolio has an equity index option (which is an equity derivative), it would be excluded from the calculation and that makes the total not equal to 100%. It could be less than 100%, or more than 100%. For example, if the portfolio has 2% equity, the total would be 98%. But if the portfolio has 2% equity index option, the total would be 102%.

In this case we would need to decide whether we would prefer to create a 10th category called Other and put the non debt, non cash asset into it, or scale it up to 100%. I would recommend the former than the latter because the latter is misrepresenting the weighting of a particular asset class, e.g. if we say that Securitised is 11%, it is actually 10%.

Usual definitions for cash, securitised and EM/DM

The usual asset class considered as cash are net proceeds, cash balances, FX forwards (but not FX swap), certificate of deposit, money market security, cash funds, commercial paper, etc. The grey area are treasury bills, short term government bonds, and marketable security i.e. for some portfolios such as global credit they could be considered cash equivalent, for other portfolios such as short maturity gilt funds they could not be considered as cash equivalent.

Securitised and convertible bonds are identified regardless of region. Securitised includes CMBS, RMBS (agency and non agency), mortgage passthrough, and other ABS such as car loan.

Definition for “Developed Market” differs from company to company. For example, is Jersey developed or emerging market? How about Cayman Islands? Do we use country of risk, or country of incorporation, or country of domicile? The country of risk might be US, but the country of domicile could be Cayman Islands.

Re-classify the past

Once it is all sorted out, we can calculate the % and duration for each category, for each month. At this point we can plot the timeline. But the security classification in the company changes from time to time. For example before 2011, they might not differentiate between EM Sov and EM Corporate, only EM. Or between HY US and HY non-US, i.e. they only labelled it as HY. This means that we need to reclasify the 2010 positions according to 2016 definitions.

This is one of the usage of a data mart (as oppose to a data warehouse), which is to reclassify the securities used in the position fact tables. We can create a new attribute in the security dimension called Bond Category, and label each security with one of the 10 categories above. Equity and equity derivatives, and currency derivatives would have this attribute set to null (or to their correct category). We can then use this attribute for the above reporting.

Measure: %, duration, spread and DTS

The above example uses % and duration as the measure to present in the chart. Duration is suitable to use for a global credit portfolio (whether it is buy and hold or active) particularly if the portfolio is leaning toward IG or has significant proportion of DM sovereign debt.

But if it is an emerging market debt portfolio, a more appropriate measure would be Duration Times Spread (DTS, which is the modified duration x the yield spread to government debt). And if it is a corporate debt it is more appropriate to use Spread Duration rather than Duration (particularly if it has a high proportion of high yield).

Different measures can also be used to monitor historical portfolio positioning, for example risk measures such as PV01 (interest rate risk), IE01 (inflation risk), DV01 (another interest rate risk). VAR (Value at Risk), portfolio volatility (3Y standard deviation) and TE (Tracking Error) are also common. Tracking these risk measures over the last 3 years (or 10) and compare them between portfolios (and to benchmark) is a good demonstration that the desk has incorporated a good risk management when managing the portfolio. And it is in client’s/investor’s best interest to know that hence such as report provide good value to both the asset management house, and to the investors.

Blog at