Multivariate Regression Analysis of Used Com-Pac Prices

[NateD]

My background is in economics, and particularly in statistics and modeling, so any time I am thinking about a significant purchase of an item, I like to collect a little data on similar units and see what a "fair" price would be (then I try to find one for much lower than the fair price). It works particularly well when looking at used cars and figuring out whether the price should be higher or lower based on mileage and model year. I've started to day dream about a CP-23, so I started poking around Sailing Texas (http://www.sailingtexas.com/cboats99.html) to see what they have gone for in the past. I put together a little model to see what kind of price I should expect based on how old the boat is and what kind of condition it is in. For age, I used the number of years old it was at the time of the sale (Sailing Texas records when the boat was posted for sale). Condition is a qualitative thing, good condition to one person is poor condition to someone who expects high quality. So I went through and looked at the pictures of the boats and rated the condition on this scale:

      0 = Not in sailing condition, missing sails/rig, or significant damage structural damage
      1 = Below average, old sails, very few upgrades or additions
      2 = Average for a boat of it's age, a few upgrades
      3 = Above average, newer sails, newer cushions, many upgrades

It's not perfect, but it separates the really superb well cared for boats (3's) from the ones that are missing all their cushions and have 30 year old sails (1's). I recorded what state the boat was in, whether it had a trailer, what kind of motor it had (gas outboard/electric trolling/diesel inboard), and whether it was a I/II/III model. Not everything came up as statistically significant when I actually ran the models. I was initially interested in looking at the 23's, but I figured I would do the 19's and 16's for the heck of it too.

The results. For the 23, the base price was $13,765. For every year old it is, deduct $508 (so a 1990 being sold today would subtract (2009-1990=19*$508=$9,652). For every point of "condition", add $2,392 (so add 2*$2,392=$4,784 for an average boat). If it has a trailer, add $1,234. Bringing it all together, a 1990 Com-Pac 23 in average condition with a trailer should sell for about $10,135.

For the 19, the base price was $6,760. For every year old it is, deduct $137. For every point of "condition", add $1,204. If it is Mark I (they were made pre-1985), subtract another $782. So a 1990 Com-Pac 19 in average condition should sell for about $6,571.

When doing multiple regression analysis, the more observations you have, the more detailed your model can be, and a whole lot of 16's have been sold over Sailing Texas' website (91), so I was able to do a more detailed model. The base price is $4,203. Subtract $110 for every year old it is. Add $659 for every point of condition. If it is one of the rare models with a center board, add $2,243. If it has a gas powered outboard, add $299 (don't add anything if it has a trolling motor). If it is located in Florida, subtract $284. So a 1990 Com-Pac 16 in average condition, no center board, with a gas outboard, and not located in Florida, should sell for about $3,726.

I tried grouping states into regions and using those in the regressions to see if prices varied across the country, but the only thing that came up as statistically significant in any of the models was the deduction for 16's in Florida. There is some evidence that Com-Pac's on the West coast are worth a little more, but it's wasn't statistically significant. So if your looking to sell your boat, don't expect to get more or less based on your geographic location (unless it is a 16 in FL).

Over on the SailNet forum, there has been a lot of talk about the huge drop off in boat prices during the recession. Since Sailing Texas records when the ads were posted, I was able to isolate boats sold during 2008 and 2009 compared to boats prior to that. Prices for similar 23's and 19's sold prior to the recession were not statistically different from prices of boats that were sold during the recession. But, prices for 16's sold during 2008 and 2009 look like they were down $200 compared to what they would have sold for in prior years.

One flaw in all of this is the prices posted on the website are only "asking" prices, we don't know what the actual boats sold for. I did remove a few boats from the analysis where it looked like the owner was asking way more than they would ever get. The other issue is the condition ranking. To get a 3 ranking, it had to have newer sails AND upgrades (like VHF, bimini, anchor, compass, cushions, roller furler, stereo). There were many boats that I ranked as a 2 (average) that had either new sails or upgrades, but not both. If I saw them in person, I might have been willing to bump them up into a 3 after closer inspection.

I don't think I uncovered anything that a casual observer of Com-Pac boat prices didn't know already, but I found the results interesting anyway. I put all of this into an Excel sheet that I can email to anyone who wants to play with it. It lets you type in different years/conditions to see what it should sell for. I also put it up on Google Documents, but it doesn't work as well in Google as it does in Excel.

http://spreadsheets.google.com/ccc?key=0ArG8egtFCmyRdHp4MEYweUZiUk95Rm9xSGMwTFRRWmc&hl=en

[Bob23]

Nate:
  Thanks. I already own a 1985 23/2. I was so fortunate to find a 20 year old 23 in very good condition for a very reasonable price...owner just wanted to unload it. (It was a 2.5)
  Your information is timely. I just spotted an older 23 this afternoon that I'd like to buy, fix up and resell. I don't know the year yet...it's dirty sitting on it's trailer. I'll stop to speak with the owner this weekend.
Bob23

[Steve Ullrich]

Very impressive Nate.  Good to see you have something to keep you busy through the winter!

[brackish]

Great stuff NateD.  My background is also steeped in statistics having been in manufacturing my entire career, and thanks to a boss and company that believed in functional cross training, having done five years as a Quality Manager using statistical tools to define, control and improve processes.  I find my statistical knowledge comforting in this time of constant media bombardment with sensational, often horrific,  but random,  mutually exclusive events.  My first thought is always "what are the odds.......?" or "in the total population this represents....?"

Is 1990 considered base in your spreadsheet or just being used as example?  I've noticed that at some point the aging factor does become somewhat linear, however, early on it seems to be a curve, reflecting more rapid depreciation from new to a point where the curve goes straight.  Irrelevant if your targeted purchase is older than say twelve to fifteen years old.

Frank

[NateD]

I was playing with it last night to see if there is a better or worse time of the year to buy or sell. I isolated sales that took place between April and July, and July to December. It didn't make any difference for the 23's and 19's, but for the 16's, buying between April and July lowered the price by $240, with no significant impact on prices later in the year. I thought prices would dip toward the winter, and be higher in the spring/early summer. My only explanation is that there is probably more supply of boats up for sale in the spring and early summer which pushes down prices a little. Or maybe people excitedly buy their next boat in the spring before they've sold the old one, and are willing to take a lower price to get rid of the old one while very few people would buy a second boat in the fall without have adequate storage space for it.

brackish,

   Yeah, the relationship is nonlinear between age and value when the boat is relatively new, but I wanted to keep things simple and have a rule-of-thumb number to discount the price by for age so I stuck with a linear model. The other issue is not a whole lot of newer ones have been sold yet, so it's hard to get a good estimate of that nonlinear relationship. When looking at used cars that are only a few years old I needed to include the price of buying it new and of extremely low mileage (<10,000) ones to get an accurate estimate of the drop off for 4-6 year old cars. I suspect the same is true for boats, but since anything that new is out of my price range, I didn't invest much time in exploring it. I was using 1990 just as an example. It's late enough that all of the models had bow sprits and other improvements.

[Nicolina]

Pretty impressive. Unfortunately, data on boat sales is not easy to get by, and most boat types sell too rarely for meaningful quantitative analysis (plus, as you said, factors that are difficult to quantify, such as condition, influence price too). Brokers have access to a database that records sales (as opposed to listings), but I think it covers only boats sold through formal brokers. But it would be interesting if you could get your hand on that dataset and see what you'd find for CP models 23, 25, 27 etc.

Good luck, and keep us informed on your research!

[tmolik]

research means diddely squat....you search for a boat that is in your price range, well kept, and you like it based on your experience.

Been there, done that.
tom

[ka8uet]

I find this information helpful in setting a price for my 23/3.  It came out just about where I was thinking it would, about $7300.  I don't know if the fact that she's on her trailer, which goes with her, makes a difference or not.  I hate to let her  go, but my increasing disability makes it unlikely I will sail her again, and she deserves to be sailed. 

[Bob23]

Ka8:
   Sorry to hear of the impending sale of your 23. I can tell from many of your previous posts, that you truly love her. I know the day is far off but coming when I'll have to part with "Koinonia" but hopefully it's a long way off.
   I think that $7300.00 is too cheap. I haven't gone over the analysis presented in this thread but I wouldn't let my 23/2 go for that price. And it's older than yours. Look around the web and try to see what others are going for. Bad enough that you are forced to sell; don't give her away.
  Statistics are one thing. Emotional attachment is quite another. In my opinion, the Compac 23 is a unique boat and if someone is looking for one, nothing else will do. I think the base price for a new 23 is 35k or so.
   Sorry to hear of your increasing disability. I don't know the nature of it and I don't think that you live too close to my home port of NJ. But a welcome always stands to come sail on my 23 if your'e in the area.
Bob23

[curtisv]

Nate,

You just might be overthinking this.

But while you're at it how about doing the same for the CP27, adding error bars and assigning a confidence estimate.

:-)

Curtis