Estimating Boat Maintenance Costs
There are a couple of common rules of thumb regarding annual boat maintenance expenses: 10% of the used purchase price, or 2% of the original purchase price. These are often repeated, but as far as my research can tell, neither has ever been put to the test. So I decided to apply a little science – I put out a survey on social media, asking boat owners to tell me about their real-life ownership costs, and then I crunched the numbers. The results are surprising!
The 10% Rule 🔗
Often phrased as, “it costs 10% of the purchase price of a used boat to maintain it every year.”
The 2% Rule 🔗
Often phrased as, “it costs 2% of the price of a new boat to maintain it every year,” or “it costs 2% of the original purchase price of a boat to maintain it every year.” I’ve also seen variations with higher numbers – 3% or even 4% – and at least one that draws a distinction between regular maintenance and longer-term major overhauls (each of which are attributed 1.5%, amortized). But most commonly it’s the simple “2% rule,” so that’s what I’m analyzing.
What is Maintenance? 🔗
Sometimes you’ll see variations on these rules that claim to include all ownership expenses. I don’t think that’s what potential boat buyers are asking about when they want to know how much a boat costs to maintain. Most ownership expenses are easy to estimate with relatively high accuracy. Maintenance costs are the big unknown. I consider maintenance to be: repairing or replacing worn/broken equipment, regularly-scheduled tasks such as oil changes and bottom paint, varnishing brightwork, repair of normal wear-and-tear, winterization, spring commissioning, and things of that nature. I do not consider spending on upgrades (replacing functional equipment or adding new equipment), repairs from major accidents, insurance, dockage, fuel, or other consumables to be “maintenance”.
At first glance, the two rules of thumb seem to be completely incompatible. How can they both be correct? Clearly one of them must be wrong the day the original owner sails off the dealer’s dock! But we don’t often hear the 10% rule discussed in relation to new boats, so maybe it only comes into play when boats are somewhat older. It might be possible to reconcile the two rules due to depreciation.
Generally, boats decrease in value over time, just like cars. When the boat’s value decreases to 20% of its original value, the two rules of thumb will yield the same estimate. Since it’s widely assumed that boats depreciate on an exponential curve, it’s possible that a boat spends a significant portion of its lifespan near this 20% value, giving both rules a shot at being useful. Let’s see if the data bears this out.
The graph shows a rather nice fit of an exponential decay curve onto the data (the coefficient of determination, r2, is 0.83). So, boats do depreciate on an exponential curve after all. The regression formula suggests 18% of the value of the boat is lost almost immediately upon purchase, and the value continues to fall at roughly 6% per year. Now we know why they say boats aren’t a good investment from an accounting point of view (but who can put a price on fun?). This is about in line with my expectations, and slightly better than the depreciation of cars.
A boat’s value stays close to that 20% mark from roughly 15 years old until about 30 years old. That’s a pretty long time. My gut feeling is that a significant portion of used boats are bought and sold in that age range, but the data doesn’t bear that out. I wonder if this could be a fluke due to the small sample size in my survey.
Testing the “10% rule” 🔗
The heart of the 10% rule is a straightforward comparison of the maintenance costs vs the purchase price. If the rule is true, we should see a nice flat line with a slope of 10% (0.10).
The calculated regression line has a slope of 7% plus a small constant. The constant is a nice sanity check: it is telling us that even a free boat will require some maintenance, which is obviously true. If you compare the regression line with the 10% line, you’ll see that one or two boats are on opposite sides, so neither is clearly better. The regression line has an r2 of only 0.34.
Both lines fare better than I expected, but the data is so varied that neither can offer much prediction to a single boat. My own experience tells me the expenses for an individual boat vary substantially year to year as big-ticket maintenance items come due (standing rigging, sails, engine). It’s possible that the regression formula predicts a reasonable amortized maintenance spend, but unfortunately I don’t have enough data to prove that.
Older Boats 🔗
If I filter the list of boats to include only those which were at least 15 years old when purchased, the r2 value goes up significantly to 0.75!
The line is higher than the rule of thumb, at 14%, and has a slight negative offset. My best guess is that this was introduced because the linear regression algorithm doesn’t want a systemic bias one way or the other in the results. For our purposes, we actually do want that bias, because we’re looking for an upper bound on the maintenance costs.
Testing the “2% rule” 🔗
We can directly look at how well the survey results fit the “2% of new price” rule. If it holds true, we’ll see a flat line with a slope of 2% (0.02).
The calculated regression line is 1%, but with a pretty significant positive offset of about $3,000. Just as before, this indicates that even a cheap boat will have some maintenance. The constant puts this line within spitting distance of the 2% rule of thumb for much of the range the average buyer is going to be concerned with (say, up to $1 million), so this rule of thumb can’t be counted out yet. But with an r2 of only 0.29, it’s not looking good.
I applied the same trick as above to this data, filtering to only older boats. The resulting chart had a worse r2 and nothing interesting to show for itself, so I’m not even going to waste your time with it.
Predicting from boat length 🔗
It’s another common saying among boaters that bigger boats cost more than the incremental increase in length would suggest – much more. Is that really true? And if it is, can we use length as a predictor of maintenance costs?
The chart shows longer boats do cost exponentially more to maintain! This model has an r2 of 0.39, which is not great, but it’s holding its own against the others. You can also see the variation in the maintenance costs rises significantly with length as well. I conjecture this is due to a compounding effect of the year-to-year variation in maintenance costs.
- The “10% of used price” rule of thumb has an r2 of 0.34
- This rule rises to an r2 of 0.75 for boats over 15 years old
- The “2% of new price” rule of thumb has an r2 of 0.29
- Estimating costs based on the boat’s length has an r2 of 0.39
If you’re buying a used boat over 15 years old, you might actually get in the ballpark using the “10% rule”, or as I calculated it, “a bit less than 14%”. For everyone else, any of these formulas will give you food for thought, but your actual expenses are likely to be quite different. There is a lot of variation in maintenance costs that none of these rules can fully estimate.
Methodology & Future Research 🔗
I’ll be the first to admit this study has too few survey responses. There were a total of 33 (n = 33), 18 of which provided a “price when the boat was new” value (the only optional question on the survey). All of the dollar values provided were adjusted for inflation to 2019 dollars, except the maintenance cost, which is recent and thus shouldn’t need adjustment.
I asked respondents if they had a monohull or multihull, and all but one had monohulls, so I wasn’t able to do any useful analysis of relative costs of the different designs. Similarly, I asked about sail vs power, and only six respondents had power boats (counting one who answered “motorsailer” as a sailboat), so I didn’t have enough data to dig into that dimension either.
One of the most promising aspects of the survey that ended up going nowhere was asking the respondents to rate the condition of their boat. I was expecting better condition to correspond with higher maintenance costs. You can see for yourself in the graphs above that there is no pattern in the condition. I was very surprised by this, and if I run the survey again, I’m going to try to find a more objective measure of condition than self-rating to see if it falls in line.
The code I used to perform the analysis is open source. I encourage you to check my math! Unfortunately, the survey data is not open source because I promised the survey respondents anonymity. I asked people to tell me about major financial decisions, and I thought the number of respondents would be very low if I also asked to publish the information. Given the low number of responses, I think that was the right move.
Clearly, there are more factors that go into maintenance costs that this survey has been able to uncover. In the future, I plan on running another survey that requests more years of maintenance cost data and enough info to adjust them for inflation. I’m hoping that will take out the year-to-year variation. I’d also like to get enough information to separately analyze sail vs power, monohull vs catamaran, and variation by location. If you’d like to participate in a future survey, please sign up for the email list below.
Thanks to everyone who filled out the survey!