Measuring Oil Sensitivity in Bond Portfolios While Accounting for Maturity
Summary
The document discusses ways to assess the relationship between oil prices and bond portfolio values. A simple starting point is to compare historical Brent prices with bond prices or fund net asset value, then calculate correlation or run a regression. The responses recommend examining regression fit, coefficient estimates, and statistical significance, while treating correlation as an exploratory step rather than a causal estimate.
Bond prices change over time for reasons beyond oil, including pull to par and movement along the yield curve as maturity shortens. One suggestion is to analyze yields rather than prices; another is to define a consistent maturity exposure, such as repeatedly using a benchmark bond at a fixed tenor or constructing a curve. A response argues that a more rigorous causal analysis is difficult because oil, inflation, production costs, borrowing, and bond prices may affect one another, and suggests an instrumental variable. It cites prior research but expresses uncertainty about its method. The document offers no completed empirical result or specific instrument.
Key ideas
- Historical correlation or regression can serve as an initial assessment of oil and bond portfolio co-movement.
- Regression coefficients, fit, and p-values help describe the estimated relationship but do not establish causation.
- Bond price comparisons can be confounded by pull to par and changing maturity exposure.
- Yield series or a consistently defined benchmark maturity can reduce some maturity-related complications.
- Causal analysis may require an instrumental variable because oil, inflation, and bond markets can influence one another.
Tags
Full text
# Correlation Oil Price with Bond Fund - question on how to tackle this # Correlation Oil Price with Bond Fund - question on how to tackle this I am trying to assess the sensitivity of a bond portfolio to the price of oil. The first intention was to 1. Get brent prices (CO1) for say a period of 5 years and 2.Get prices of each bond for the same period (or take the top level valuation –NAV- of the fund) and 3. do a CORREL in excel. However I wonder if doing so will get me the true sensitivity to oil price because of the maturity effect. Say that the portfolio is made of only one bond, looking at the change in price may not be right because as we go back in time the price of the bond is also affected by the fact that we are placed somewhere else on the yield curve. Or maybe I am overthinking it and comparing historical prices would be enough? How would you tackle this? I have access to Bloomberg and APT. Many Thanks ## Answer by Dave Harris (score 2) https://quant.stackexchange.com/a/37932 You have chosen a very difficult problem if you want to do it correctly. Oil prices are an important, but unstable, determinant of inflation. However, other inflation impacts oil production costs, which in turn impact oil. Both anticipated and unanticipated inflation impact bond prices. Bond prices are a determinant of oil output as money is borrowed for exploration and so is a factor cost. You need an instrumental variable. There is an article that attempts to tackle this at: > Cette, G., & De Jong, M. (2013). Market-implied inflation and growth rates adversely affected by the brent. Journal of Asset Management, 14(3), 133-139. I have not decided if I agree with the methodology, but there isn't a lot out there on this topic. ## Answer by Michael Hartmann (score 1) https://quant.stackexchange.com/a/37491 A good place to start would be correlation, then if there are interesting results you can put in the time and effort to break down where that correlation is coming from. I would start with a regression on your chosen bond portfolio. Significant numbers to look at will be your R-Square (of course) but also p-value and your coefficients. You'll not only be able to tell if there is a correlation but how important those correlations are. Like I said, that would be a good place to start. ## Answer by Lipton (score 1) https://quant.stackexchange.com/a/37930 I think you're absolutely right that a bond "changes" as time passes. I think there are at least two things you can address in this question: - Pull to par Pull to Par is the effect that price of a bond converges to par value as time passes. Obviously, if you can hold to maturity, at maturity the price of the bond should equal to its face value. Another way to see it is that assuming yield unchanged, the price of bond will change due to is reduction of maturity. As a consequence, I'd run the correlation in yield instead of price. - Roll down As you mentioned, as a bond gets closer to maturity, it's also rolling down the yield curve. Whether to include it into your analysis depends on how you construct the bond portfolio. For example, if you want to study say the correlation between TIPS and oil price, you may want to always pick a 10y bond, and roll the benchmark 10y bond. Or you can build a bond curve to do it, etc.
Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.