« March 2005 | Main | May 2005 »

April 21, 2005

Honeywell Event Study

Yesterday, Honeywell released its quarterly earnings report. The stock price also tanked. Are those two items related? What is the influence of the overall market movement? Are there certain typical stock price trends around the release of earnings, and can those trends be exploited?

The quantitative finance community has a methodology (called Cumulative Abnormal Return, or CAR) that may be useful in answering those questions. The return of a stock (from day to day) can be decomposed into market and idiosyncratic components: r_t = alpha + beta * r_market_t + eta_t . r_t is the return of the stock at time t, alpha is the long-term idiosyncratic return, r_market_t is the return of the market at time t, beta is the correlation of the stock with the market, and eta_t is the abnormal return at time t.

Under normal circumstances, eta_t is zero. During some event, this abnormal return may vary. In particular, we are looking for the accumulation of eta_t (CAR) during a particular time frame. It makes sense to consider time before the event as well as after the event. The eta_t can be calculated by: eta_t = r_t - (alpha + beta*r_market_t). In other words, it is the actual returns minus the returns that the model predicts. Alpha and beta can be determined by choosing a period before the event time frame (called the "estimation period"), and regressing the actual returns against the market returns.

So, I grabbed the daily stock price from Yahoo for the time period from the beginning of 2000 to yesterday (I did this analysis last night). I used the excellent IBES/WRDS database to look up the quarterly reporting dates for that time period, as well as the reported EPS results. Then, because there were quite a number of "earnings events", I wrote a little macro to do much of the analysis.
Some details: I used the S&P500 as the market, and looked at daily returns (appropriate for this kind of event). I picked an event horizon of -20 days to +25 days, and an estimation period of 35 days prior to the beginning of the event. (It would have been nice to use a longer estimation period, but the estimation period would have included the previous quarter's event.) I excluded a couple of events, including the 3Q of 2001 (due to 9/11) and 2Q of 2000 (because it looked weird, maybe due to the GE merger thing). The graphs of each of the past 18 quarterly reports can be seen below.

CAR for each quarterly earnings report

There is a lot of noise in those, though, and it is difficult to see trends. The worksheet "analysis" averages the CARs over all of the events. Also, I tried grouping the series into the set that had positive returns on the event day and a set that had negative returns on the event day. There are a couple of interesting things in the graph: First, it seems like there is a drift in the returns before the event, generally upward in cases with positive events and generally downward in cases with negative events. This is probably not tradable, for obvious reasons. Second, there seems to be strong movement after the announcement. Good events yield maybe 5% returns over 15-20 days (post-event), while bad events lose maybe 3% over 15-20 days (post-event). (Unfortunately, I don't know of a good way of quantifying the errors in those numbers.) Both of these effects are somewhat typical for earnings announcement events.

CAR averages for Honeywell quarterly earnings reports

Historical performance does not guarantee future results. This is intended for amusement reasons only. Heck, the stock is up 2.5% today (although that has a lot to do with the overall market, as it translates into only a quarter of a point of abnormal return).

April 10, 2005

Parker Mesa Overlook

One of my favorite hikes near Los Angeles is the "back" way to Parker Mesa Overlook, part of Topanga State Park. (The venerable guide Afoot and Afield in Los Angeles refers to it as "Topanga Overlook", but I'll go with the trail markings.) Today was perfect for a hike, with cool ocean breezes, a sunny clear sky, and green hillsides (from the winter rains). So, I strapped on my backback (dusty, and unfortunately not necessarily mountain dust) and headed on a little walk.

It is just a short drive to the trailhead, west along Sunset Blvd, to Paseo Miramar (almost to PCH). This is in the Pacific Palisades neighborhood, which is quite affluent. One of the enjoyable parts of the trip is the drive up Paseo Miramar, as road is very steep, and winds past some incredible houses. Parking is available on the street. In the past I was able to park fairly close to the trailhead (at the very end of the street), but today I had to park a little ways down the hill.

The trail itself is really fire-road... wide, well-kept, and bare. Normally, I dislike fire-roads, as they lack character (and challenge), but the surrounding views are sufficient to override those concerns. Sneakers are acceptible footwear, although there are a couple of inclines that are dusty enough that one should be careful. I took me about an hour and a half to get from my car to the Overlook, but I spent some time taking pictures and enjoying the view. The trail has some great views of the ocean, and of the west side of Los Angeles.

The Overlook is located at N 34°03.552' W 118°34.426', with an error of 36 ft by my GPS. (Google Maps is awesome, as you can see.) It is 1558 ft above sea level, according to my read. The Park just installed some benchs for sitting and looking at the ocean, but as they were occupied by chatting Angelenos, I was happy with a sun-warmed rock.

I have hiked this trail several times. Once, I tried to run up it, but wasn't able to sustain the pace very long. Another time, I took my parents and sister with me, but the day was very foggy and they didn't get to see much. My favorite time was a hike that ended up at night, with Elizabeth and I making our way back under the light of the moon.

Near the beginning of the trail. The green of Los Angeles spring is not the verdant green of wetter areas. It hedged its bets against the long dry brown summer.
The Santa Monica coastline, looking south. The small line in the middle of the curve of the coastline is the Santa Monica pier. Elizabeth used to live just south of that. If you look closely, you might see fellow hikers on the trail.
A number of flowering trees and bushes could be seen.
The final approach to the Overlook.
A view of the ocean, while I ate my lunch.
The view north seemed slightly silvered by the sea mist.
Several boulders along the trail seemed to be aggregates of many smaller rocks. I guess that these boulders were formed in the ocean, and then thrust up in some cataclysm. I wonder if the incessant workings of erosion will return it to the sea, or future cataclysms will suddenly toss it further.