Vulnerable Tech Sector

Feb 19, 2004

Exuberant Gains

The DJIA is up 47% from its lows of October 2002.  That's pretty good, but not nearly so strong as the Nasdaq Composite, (up 87%).  The small-cap Russell 2000 (up 82%) is nearly at all-time highs.  Sentiment polls are bullish, and cash has been gushing back into mutual funds.

It's no surprise, talk of a "new bubble" is in the air.  Smith Barney strategist Tobias Levkovich has said "...references (are) being made to investor speculation and the possibility that a new stock market bubble is forming,"  In an article headed Stock Market Bubble, Take Two, Fortune.com calls the recent rally "almost indiscriminate."  The implication is that the run-up is excessive and must end in a eventual collapse.

Bubble factors often cited include positive sentiment, easy money, high valuations, and unrelenting market advance... most especially in small caps and technology stocks. The DJ Technology index is up 99% from its lows, and some individual issues are up hundreds of percent.

A Reprise of 2000?

With all the bubble talk, let's look at some stats.  Based on analysis of mutual fund portfolio behavior, our statistics suggest there's substantial risk of substantial Tech sell-off.  Aggregate fund exposure to the Tech Sector is above-average and heading lower, and funds with historically successful Tech tactics are withdrawing faster than the rest.   First consider aggregate exposure.

Click image to see full size graph

The above chart shows a composite index of aggregate Tech exposure across a sample of more than 100 general equity funds.  In this index, positive readings mean these funds show above-average Tech exposure (above their own average), and negative readings indicate below-average Tech exposure.

Currently at +.1850, aggregate exposure is clearly above-average, and down from about +.50 some six months ago.  Back in 1999 the index reached up to about +.65 some 6 months before getting down to today's level (and at that point the bubble still had some 4-to-5 months to go).

The index

We measure Tech exposure in a statistical model called TXP.   Ebbs and flows of a fund's Tech exposure are reflected in its partial covariance with Tech sector stocks.  Based on micro analysis of daily returns, TXP performs rolling multivariate exponential regressions, generating for each fund a continuous path of net  effective Tech exposure. Aggregate series are compiled from individual fund series.

This model has been around for about a decade, but just recently emerged from an exclusive client contract.  It also happens just now to be telling an important story.  It's a story of professional portfolios lightening up in Tech, suggesting a downturn in the making.

Tech Timing

It's not just the aggregate exposure that's heading lower.  In addition to aggregate, TXP also measures each fund's tactical management of Tech exposure.  High exposure when Tech is strong and low exposure when Tech is weak implies tactical gain.  The opposite pattern (high exposure in weakness and low exposure in strength) implies tactical loss.

Compare the TXP exposure histories of Janus Twenty and Legg Mason Value Trust.

The Janus funds are by now notorious for their Tech bubble excess in 1999-2000.  But their outsized exposure and miscued timing were visible in TXP as they were developing.  Here is a five-year picture of the TXP path of Janus Twenty Fund, together with the DJ Technology index.

The graph plainly shows the  enormous Tech exposure right into the height of the 1999-2000 Technology bubble.  At one point in 1999 our index of the fund's Tech exposure almost reached 2.5x  the Tech index.  Exposure was still almost at 2.0x as late as the end of 2000.  The fund finally began reducing exposure in 2001.  The retreat continued through the entire bear market...and through most of the 2003 year-long rally.

Click image to see full size graph

Now compare that Janus performance with the TXP exposure path of Bill Miller's Legg Mason Value Trust.  Miller's TXP never reached more than about 1.5x, and his retreat was more orderly.  Better yet, the Tech reduction in LMVTX anticipated the market top by months.  By the time the bubble was about to burst in late 2000, the Value Trust's exposure was almost down to zero.

Click image to see full size graph

These charts illustrate just two Tech sector histories.  Other strategies and patterns are apparent in comparing TXP exposures across a broader range of funds.   Tracking fund-by-fund Tech exposures is a great advantage.  But TXP also goes further.  

Unique Indicator

Analyzing fund-by-fund Tech exposure together with fund-by-fund Tech timing shows which way the "proven" smart money is running.  We say "proven" because we can track the record.  The issue is whether the positive timers or negative timers are committing assets to the sector.  The relationship of current exposure to timing capability gives a statistic called TXP Slope.

The Slope is positive when good Tech timers have higher exposure than bad Tech timers.  It's negative when the good timers have lower exposure than the bad timers.  This TXP Slope relationship is compiled weekly, and the result is a remarkable indicator of strength or weakness in the Tech sector.  Here is a five-year chart of TXP Slope.

Click image to see full size graph

The TXP Slope fell notably as the Tech market was topping in 2000.  Funds with the better Tech timing capability were bailing out while the lesser tech timers were staying in (or even still coming in).  In that case, the change from positive to negative was dramatic.

Here in late February, the TXP Slope has been declining for months, and is almost at zero. As in 2000, overall Tech exposure is receding, and the funds with the better Tech histories are retreating faster than others.  This is a clear sign of weakness, and a vulnerable time for the Tech sector.

Nevertheless, 2004 does seem different from 2000. This time the decline occurs in context of less aggregate exposure at the peak...as shown back at the top of page.  And the Slope has been eroding for months rather than overnight.  In fact, the Slope has not yet crossed notably negative. So there are differences, and the differences point to less negative Tech sector inference. But "less negative" than 2000 leaves a lot of room for loss.

For further informaton on TXP, use the Contact link below.

Nothing at this site is offered as advice or recommendation.
All site content is subject to full Disclaimer statement and Terms of Use.
Copyright© CarpenterAnalytix.com 2004. All rights reserved.