Jan 3, 2017

Used Guitar Resale Value Regression Analysis


The used guitar market isn't particularly difficult to understand, but is rife with opaqueness and rumors, and a little light on actual data and studies.  While this isn't a comprehensive or conclusive analysis, it's a step towards a database that can help us draw real conclusions in the future.

Not excited yet?  I'll try my best :p






Research techniques and initial disclaimers:

          For this study, I used the very helpful tools of Reverb.com to gather data on 250 past transactions.  I only used data from guitars that sold, and only the price they actually sold at, disregarding the original listed prices.

          I also had to gather the MSRPs of every guitar, and correct for inflation in some cases.  For that reason (it's difficult to find original prices on guitars more than a decade or two old) the vast majority of the guitars studied were from the mid 1990's - current production models.  Guitars much older than two or three decades likely need more specialized research; old Teiscos may go for a third of their original price, while a 50's Les Paul can fetch 30 times (!) its originally hefty MSRP (after adjusting for inflation!).

          To recap; look at this as a study on the differences primarily between manufacturers, features, and price ranges of modern guitars, and not as a tool for identifying priceless unusual antiquities, and we'll hopefully come to some interesting conclusions.  Disclaimers over - let's get to the (relatively) interesting stuff!

Used Guitar Market Analysis
Initial regression results

Used Guitar Market Analysis
Results made easier to read and understand :)


Interesting stuff:

The baseline intercept is 0.388.  (This would be a guitar from no particular brand, in "fair" condition).  The units for everything are converted to "resale value as a percentage of MSRP".

In layman's terms, this means that a typical used guitar in not particularly great condition will go for around 39% of the original price.  The upper and lower confidence boundaries are 31% and 47% with 95% confidence; so only about one in twenty guitars will either go for under a third or over half of its original price (in "fair" condition).

Condition:

The condition of the guitar is quite important and I was able to collect solid data on this using Reverb's standard system of "fair", "good", "very good", "excellent", and "mint".  Moving up each bracket increases the resale value by 6.59% on average, standard deviation of 1%.

Mint = +26.36%
Excellent = +19.77%
Very Good = +13.2%
Good = +6.59%
Fair = +- 0.0%



Now, these brackets are qualitative and up for interpretation, but it's as good a system as one can hope for in a market so varied.  So, combining this information with the baseline, we can understand that a typical guitar in "excellent" condition will go for between 50% and 67% of its original MSRP, without looking at any other factors.


Price Range:

The original MSRP showed little to no cohesive trend.  I thought that either very cheap or very expensive guitars might hold their value better, but the coefficient was so small and the confidence so low it indicates virtually no correlation at all.  So neither expensive nor cheap guitars have an advantage regarding resale value (but hold your horses, because brand still matters!)


Age:

The age of the guitar has a very slight negative correlation (with about 75% confidence).  It's not very predictive and most of the guitars studied were typical guitars from the 1980s-2010s, so don't throw out your '59 LP just yet.



However, looking at specific cases such as 1990's Fender Stratocasters vs 2010's Fender Stratocasters found that typically, at least regarding guitars produced since the 1980's, every decade of age will devalue the guitar by about 2%, FWIW.


Construction:




Acoustic:
While the majority of guitars studied were electric, acoustic guitars appeared to hold their value slightly worse, only about 3.5% difference, and with only 65% confidence.  Because less samples were taken and the margin of error so high, we can more or less disregard this result.




Hollow Body:
Hollow body guitars held their value noticeably better than average, at +8.3%.  Thus, holding everything else constant (as much as possible!) most hollow body guitars will be worth between 2% less and 19% more used than an equivalent solid body.  It's a wide range but we can say with 89% confidence that the relationship is a positive one.



Semi Hollow Body:
Semi hollow guitars (i.e. ES-335s) held their value best of all, at +13.0%, standard error of 3.5%.  That's a noticeable trend!  It should be noted that many of the semi-hollows were also Epiphones (more later) which may have inflated this measure.  Nonetheless, it appears that semi-hollows have a measurable advantage over the typical solid body in resale value.


Rareness:

Few guitars in the study were considered "rare", so take the small sample size into consideration. Indicating a guitar as "rare", be it a special edition or a less common vintage guitar, averaged an increase of 16.6% to the value, standard error of 4.6%.  That's certainly a noticeable increase; defining the threshold for "rare" proves to be much more difficult.  It's hardly a surprise that limited edition or hard to find guitars go for more, but now we have a rough estimate of how much more that might be.



Used Guitar Market Analysis
Results made easier to read and understand :)

Brands:

There was only one clear winner here; Epiphone, with an average increase of +14.3%.  I'm not sure exactly why Epiphone has such an advantage here; part may stem from the fact that more Epiphones were also semi-hollow guitars, part is likely the high demand for Gibsons by people without money for actual Gibsons that makes Epiphone so popular.  Anywho, I'm coining the term "the Epiphone-Effect" right now just in case it turns out to be part of a larger trend :p

Fender held up alright, at +4.1%, but with a fairly large margin of error.  Whether it's a Strat or Tele, it'll probably do a little better than an Ibanez or other "average" brand.

Gibson did more poorly at an average of -6.0%, in contrast.  This may be due to their extortionary MSRPs or due to recent complaints of unpredictable quality (not a huge Gibson fan here, surprise, and I do feel a touch of schadenfreude at this result).

Squier did pretty well at +5.5%, actually holding value slightly better than their more expensive Fender relatives.  This is a little surprising given how common Squiers are (you'd think the market would be flooded), but they have a reputation for being workable guitars at great prices.

Ibanez was almost indistinguishable from the mean, neither contributing nor detracting from the value as compared to an "average" guitar.  While researching MSRPs on a few older Ibbys I read a few forums chatting about how everyone knows that Ibanez guitars don't hold their value as well.  Huh.

Schecter did slightly worse than average, but with a high margin of error due to relatively low sample size.  As a Schecter fan I'm a little bit disappointed yet also little bit excited by the prospect of continuing to get some smashing guitars at slightly undervalued prices.

PRS also didn't do so hot at -6.7%.  Most likely they suffer from the same issue with high prices as Gibson; the market simply has to be smaller when typical guitars cost several grand.  (In the theme of separating Epiphone and Gibson, Fender and Squier, PRS SE was not included in the analysis of PRS resale).

Rickenbacker did poorly at -21.1%, but the sample size was small and original MSRPs fluctuate quite a bit.  Again, these are pricey guitars and tailored to a more specialized audience, but I'd still take the result with a grain of salt.

The one Teisco sale that I measured was the clear loser at -30%; although it was a fairly typical representation of a Teisco resale, it was only one instance.  While selling today for $100 what could be bought in 1965 for $69 may seem like a deal, inflation means it's most definitely not.  Manufacturing has come a long way in the past 50 years, and mass-produced guitars have improved by leaps and bounds while cutting costs dramatically along the way.  While older guitars are always more variable and some may increase in value quite a bit, most older, mass-produced guitars simply will not.



Used Guitar Market Analysis

Application:

Neato, we have a workable data pool and have extracted some moderately useful info from it.  How can we apply this?

Simple; take the intercept coefficient and add or subtract based on each of the other attributes.

Let's take a 2010 Squier Classic Vibe 60's Strat in "very good" condition - a $400 Squier solidbody (age is negligible).  Thus we start with:

Intercept = 38.8% + (no change for construction) + (no change for rarity) + 13.2% for condition +5.5% for brand = 57.5%.

Multiply that by the MSRP of $400 and we get an estimated resale value of $230.  Pretty nifty huh?  Of course, every guitar is a special snowflake, the market is never quite the same, and there are significant margins of error involved.  Still, it's nice to be able to get a workable rough average for nearly any guitar.

Let's try another example for completeness sake before wrapping this up - a Gibson ES-335, two years old but in mint condition.  38.8% + 13% for semi hollow, + 26% for mint condition, - 6% for brand.  71.8% times MSRP of $3500 gives us an estimated value of $2500.  That's a $1000 bath for it no longer being factory new, but that's just how the market works.

One last thing I analyzed out of curiosity - does color affect the resale value?  Obviously if it's a flame top that jacks the initial value up, that's different, but I wanted to see if there was any difference between colors as well.  The answer is "no, not really", so don't stress about resale value when trying to decide between racecar red and Pelham blue for your next Strat :)


Final words:

Again, this is far from comprehensive or conclusive with certainty, but it's a decent place to start and shows some interesting results.  I doubt too many other people will be excited to trawl webpages for transaction history data, but a database of at least 1000 instances would definitely yield more conclusive results, if anyone's feeling up to the task.  And for any webpage programmers out there, a simple in-browser calculator where people can input guitar attributes and get out a set of results would be a fantastic tool for the less mathematically-inclined.

Thanks for reading, hope you learned a little something, and stay tuned for next time!

3 comments :

  1. Schecter is a great guitar! I have four and would love to have more. Like them over Gib and my PRS SE

    ReplyDelete
    Replies
    1. Couldn't agree more! They're sometimes written off as "metal guitars" but their quality and versatility easily match many far more expensive brands.

      Delete
  2. Ok, this is super cool. I'm going to have give something similar a go--off to scrape Reverb's sale data, I guess!

    ReplyDelete

Questions, comments, concerns, complaints? Leave your thoughts below!