Summary
- Forward valuations of PSI’s major stocks look interesting.
- Nonetheless, PSI suffers from a lot of structural flaws and there are better alternatives in this space.
- PSI's volatility quotient is also relatively elevated, and its ability to mitigate risk and deliver excess returns is questionable.
- The fundamentals in the semiconductor space look worrying, whilst the technical picture too suggests further weakness ahead for PSI.
Introduction
The Invesco Dynamic Semiconductors ETF ( PSI ) which covers 30 semiconductor manufacturing stocks has been a solid wealth generator over the years; since its inception over 17 years back, it has delivered almost 2x the returns of the Russell 3000.
This year, however, it has been an entirely different story, with PSI giving up more than a third of its market cap, and also underperforming the Russell 3000 in the process.
Nonetheless, given its long wealth-generating track record, I can see how the significant correction this year could attract a few buyers back to the table; more so, as the forward valuations of the heavyweights (the top 7) of this portfolio are all largely trading at attractive discounts to their long-term forward P/E average. As you can see from the image below, the current weighted average P/E of the top 10 names of PSI can be availed at a ~14% discount to the 5-year average.
YCharts, Seeking Alpha
Whilst I appreciate the valuations on offer, I still don't believe that investors should pursue PSI. Here's why-
Structural flaws
As a semiconductor option, this ETF has plenty of flaws, and there are other well-rounded options you could pursue. To reiterate my case, I've compared PSI to two of the largest semiconductor ETFs in the market today- The VanEck Semiconductor ETF ( SMH ), and the iShares Semiconductor ETF ( SOXX ), both of whom have over $7bn in AUM (individually) today, against PSI's AUM of only $0.6bn.
Perhaps one reason why PSI hasn't been unable to attract similar levels of AUM is because of its inefficient cost structure and relatively unstable portfolio. PSI's expense ratio of 0.55% is above the typical ETF median of 0.45%, whereas an SMH can be accessed by shedding out a figure of only 0.35%.
It also doesn't help that PSI resorts to ample level of portfolio churn every year; typically, around 84% of the assets are turned over every year; that is a lot when you consider that your average ETF only turns over 28% of its assets in a single year. In that regard, SMH comes out quite favorably, as typically only 1 in 5 stocks from its portfolio get turned over.
Granted, whilst the income theme will always be a secondary factor whilst pursuing a growth theme such as semiconductors, it can still make a difference over time. Amongst these three options, PSI's dividends have grown at the slowest pace, both over the short-term (in fact a decline of 2%), as well as the medium term (+11%), and its current yield of 0.35% is 2.7x lower than what you could get from a SOXX. SMH too provides yields that are over 2x better than PSI.
Below-par risk-adjusted return track record
As I've reiterated in some of my previous ETF-based articles, it isn't enough to just look at the absolute return track record of an investment product. It's also important to get a sense of the volatility quotient of the product (standard deviation), and whether it generates a return threshold than justifies the volatility associated with said product (Sharpe ratio). One also needs to gauge the potential of these ETFs to provide excess returns in the face of harmful volatility (Sortino ratio).
In that regard, I've looked at how PSI has fared, both over a recent 5-year span, as well as a longer time period of 10 years. What those stats tell us is that PSI is clearly the riskiest semi-option of the three (highest monthly standard deviations), regardless of the time frame. This puts it in a difficult spot to generate a certain excess return threshold (over the risk-free rate), and we can see how it falls short of its peers on the Sharpe ratio front. In fact, over a 10-year basis, it is the only one of the three ETFs that has generated a ratio below 1.0, clearly highlighting that it is taking more risk than warranted.
Finally, you have the Sortino ratio which gives you a sense of how these ETFs manage harmful volatility whilst generating returns. Once again, PSI falls short of the standards maintained by the other two products although in fairness, that relative differential over a shorter time frame of five years.
Nonetheless, PSI's weak risk-adjusted return profile is a function of the type of stocks it pursues. Unlike the other two portfolios that are heavily oriented towards the well-rounded and safer business models of large-caps (99% of SMH's stocks have a market cap of over 12.9bn, whilst it is 93% for SOXX), PSI devotes a sizeable chunk of its funds to mid-and small caps (large-caps account for 55% of the portfolio) that are just inherently more volatile and unidimensional.
YCharts, ETF.com
Underlying conditions don't look good for semiconductor stocks
Around a month ago, Gartner significantly cut its 2022 global semi-revenue growth estimates from 14% to 7%; worryingly next year's forecast points to an annual decline of -2.5%; I suspect there are further downside risks to that number as conditions in some of the key global semiconductor hubs look in bad shape.
Take for instance, Korea which is the largest producer of memory chips; chip exports here have been sliding for four straight months , and a couple of days back reports came out highlighting that factory shipments had declined for the first time in three years. The drop was quite brutal as well, coming in 22% lower (June had seen a 5% annual gain). Meanwhile inventories also remain at rather elevated levels. Taiwan, another notable semiconductor hub has seen manufacturing contract for two straight months, with recent export orders registering their biggest fall.
Bloomberg
Besides the fundamentals, you'd also have to be very intrepid to get on board with semiconductor stocks at this juncture, given their heightened sensitivity to market movements.
Look at PSI for instance, its beta around a year ago was 1.19x, these days it is at 1.6x, basically implying 34% growth. Since the 18th of August, the VIX has spiked by 34%, even as the S&P500 has fallen by 8% or so; meanwhile, PSI has fallen by an even greater margin.
And if you look at PSI's standalone chart, we can see that it broke down from its 22-month ascending channel in mid-Jan; since then, it has been trending lower in the shape of a descending channel and currently still looks quite some way off from its lower boundary.
Finally, also consider that despite the correction this year, PSI's portfolio of semiconductor stocks looks relatively overbought vs the broader markets as represented by the Russell 3000 (the ratio is still ~23% higher than the mid-point).
Stockcharts
For further details see:
Why You Should Shun The Invesco Dynamic Semiconductor ETF