The worse the data points get, the more bullish we become, and we think that the key lesson of the last six months is that the value of data point investing in semiconductor stocks has now been eliminated.
Markets are amazingly efficient at arbitraging excess profit opportunities away, and we’ve seen it in semiconductor stocks before. Ten years ago all one had to do was correctly call the inflection point in industry revenue growth, and the stocks would follow. By 2001 investors had started to figure that out, and over the next several years the relationship broke down. Few semiconductor analysts bother with the sweeping sector calls that used to be so popular, as the market has figured out that adding value that way become almost impossible.
Data point investing in semiconductor stocks is now headed for a similar fate, in our view. Consider the fact that no amount of channel checking would have turned up weakness in January despite the fact that January was the selling opportunity. Meanwhile, publicly available SIA data showed the industry hitting dangerously high unit shipment levels, and valuations were too high as well.
There are too many investors taking too many airplanes to Taiwan and China, and the exercise is now undifferentiated and valueless. Oddly enough, some of the tools that used to work better – watching industry data closely, and paying attention to valuation – seem to be working better. No doubt that we’ll see another shift in semiconductor stock behavior that will render those measures useless as well. For now, though, the formula for success seems to favor watching earnings multiples and SIA data, and staying off those flights to Taipei.
We will set aside our confusion over why published industry sales data does not constitute a “data point” in Osha’s parlance. Or, for that matter, why the P/E multiple for a stock does not count as a “data point.” It seems amazing that an equity analyst would suddenly realize that valuations (earnings multiples) and macroeconomic data such as that delivered by the SIA are more important to an investment thesis than some random tidbit gleaned in a channel check. (For the uninitiated, a channel check is basically asking someone who buys semiconductors “how’s business?”) If there was ever an anomaly in the stock market, it was a time when valuations didn’t have an enormous impact on investing returns. On the other hand, perhaps we’re being too harsh on Osha. His pay - no, his very job - is driven by what tiny edge he can glean for hedge fund managers that none of the other sell side analysts were able to glean.
For the most part, hedge fund managers are not investors - they are traders. For traders, a moment’s swing in sentiment due to a slow sales day at one customer is enough to jump in for a quick profit. All the better that they get the data from the sell side - that leaves more of the 20 percent incentive pay for their own profits.
If you are an investor rather than a trader, you are less likely to be concerned by the weekly fluctuations in this or that. If you are a dedicated investor, you paid very close attention to the valuation when you bought, and were probably aware of the potential short-term downside to that valuation. You are less concerned about whether customer X ordered fewer semiconductors today than you are whether customer X is consistently ordering fewer semiconductors. You’ll suffer the first couple of points of downside waiting for the trend to become apparent, but you won’t waste buy and sell commissions trading on each wiggle in a volatile chart.
To some extent the whole concept of trading tech stocks is an ongoing bubble-era hangover. We used to be able to make so much money in tech stocks that we believe if we only try harder, we will be able to do so once again. So we jump on the plane to Taiwan and interview sweatshop factory workers to see whether they assembled more iPods today than they did yesterday. Thing is, it ain’t the bubble anymore. In fact, the tech bubble was only the tech bubble because the growth days in semiconductors were already gone and nobody believed it. Check out the chart:
This is nearly 30 years worth of data from the Semiconductor Industry Association [SIA] - yes, the same SIA that publishes the data that Osha suddenly has discovered may be of value. Plotted on a logarithmic basis, it shows a very clear deceleration in the growth rate that began about 10 years ago. When Osha wonders what changed when “Ten years ago all one had to do…” he need look no further than the same old SIA data. It would have given the right data then - although acting on that data would have (and did) make some investors look stupid. For a while. But Osha still hasn’t learned. He goes on to say:
Now, of course, the negative data points are myriad, although we’re mystified as to how much value-add there can be with the SOX down 22% from its January peak and valuations now reasonable. With a seasonal uptick in PC build activity beginning to show up, the predictable appearance of negative wireless data points justifying a trade into PC stocks has been especially amusing to watch.
Only someone who still thinks in terms of the bubble can believe that valuations are now reasonable. The chart shows how much growth trends have slowed. In fact, if you take the April 1996 number and the April 2006 number, you will find an average annual growth rate of just over five percent. Nominal. GDP has grown faster on a nominal basis, with less volatility. Just about any industrial segment - chemicals, steel, homebuilders - has had more growth in the last 10 years than semiconductors. So don’t tell us semiconductor valuations are reasonable until their P/E ratios are similar to those of Dow Chemicals (8.7), US Steel (12.7) or Toll Brothers (5.1). ‘Til then, you’re living in dreamland.
Note: That is not to say we think semiconductor valuations will head straight to zero. The vast majority of investors were at the same kegger Osha went to, and are still as hung over. The industry remains cyclical and thus will have ups and downs within that longer-term valuation compression. So we’ll keep on watching that SIA data, and we’ll try to call those turning points. Just don’t expect us to hang on for more than a few months. We’d rather miss a little upside within a wiggle than participate in more of the longer-term downtrend.