Three years ago the information technology sector was priced at a 50% discount to energy (see graph below). Since then, the discount has collapsed and the ratio of IT over energy (represented by XLK/XLE) is approximately 1.
Given the expectation of a “leaner” Fed and rates going up, energy might outperform technology and the ratio could pull back going forward. Over the last 30 years the energy sector has outperformed the overall market by almost 2.5% when rates are between 2.75% and 4%. That outperformance is even more pronounced when rates are higher. The main reason for this outperformance is that energy costs are included in the consumer price index and the energy sector is positively correlated to an expectation of higher rates and inflation. Thus, going forward, the above ratio may experience some head winds.
It is critical for investors to measure the potential impact to their portfolios if the ratio were to start a downtrend from here.
One simple way to check how exposed your portfolio is to these sectors is to aggregate the net exposure from stocks that are classified as ‘IT’ and ‘energy’. However, there are 3 main shortcomings with this approach:
- These classifications are mere labelings and no correlation effects are taken into account.
- Stocks that are not classified as information technology and energy are excluded from the tally. For example: Tesla is not an IT or energy company but we know it will react to an IT correction.
- Other non-equity assets cannot be included in the aggregation.
Instead of simply looking at exposures, we prefer to look at measures that help diagnose the explanatory power of technology in “normal” and “extreme” times. This way we can gauge how much of IT factor might creep into the portfolio when/if technology begins to underperform energy. Let’s make it more concrete:
Step 1: MCTR
In normal times we decompose the risk of a portfolio (or a single security) into an on-the-fly factor model comprised of: technology sector represented by the XLK ETF, energy sector represented by the XLE ETF and residual. Residual is the amount of risk that cannot be explained by XLK and XLE.
We use a measure for risk called Marginal Contribution of Total Risk (MCTR), which takes into account the position sizes, the volatility of each position on a standalone basis, and the correlation of that position with other securities in the portfolio.
The table below shows various securities and their amount of technology risk in normal times.
|XLK Factor||XLE Factor||Residual||MCTR||XLK/MCTR|
|CORP:TSLA 20231002 3 98||0.19%||0.29%||4.50%||4.98%|
* XLK included as a “control” security
It can be seen from the above table that the methodology allows us to rank securities that are not classified as “information technology” or “energy”. For example, the technology component in Tesla’s risk is similar in magnitude to that of Netflix. We can also classify non-equity securities, such as Tesla’s corporate bond.
Given that correlations tend to increase in extreme market conditions, we also need to calculate the influence of the factor model: [XLK, XLE, residual]in scenarios whereby technology is under pressure. This is not a trivial calculation as we are not simply interested in measuring the expected profit and loss (P&L) when XLK is down 5% (for example), but we are interested in decomposing that expected P&L to determine what portion can be explained by XLK, XLE what portion would remain unexplained. Decomposing the P&L under stressed scenarios is akin to a “stressed” factor model, or more precisely, a factor decomposition that is conditional on a set of posterior probabilities representing XLK‘s expected drop to be 5%. The following table summarizes the results:
Step 2: Expected P&L when XLK down 5%
|XLK Factor||XLE Factor||Residual||P&L||**XLK Factor/P&L|
|CORP:TSLA 20231002 3 98||-0.15%||-0.09%||0.00%||-0.24%||0.62|
The table shows how the component of expected P&L explained by technology is more pronounced during scenarios of a technology decline. Investors should closely monitor this “contagion” effect.
The tables above can be easily replicated with a few lines of our API code (available upon request).
Please contact us if you would like to calculate the above measures for your investment program(s) or individual securities.