The furnace in your rental is twenty two years old. It just failed. The contractor says he can get it running for 900 dollars, or put in a new one for 4,800.
Most investors reach for return on investment here, and it is the wrong tool. Return on investment compares a use of capital against not using it. That framing only works when doing nothing is an option, and on a broken furnace in a rented house in January, doing nothing is not an option. The alternative to spending money is not keeping the money. It is losing the rent.
So throw out the ROI question. There is a better one.
The test: cost per remaining year of service
Every mechanical decision on a rental reduces to a single comparison. What does each option cost, divided by how many years of service it buys?
| Repair | Replace | |
|---|---|---|
| Cost today | 900 | 4,800 |
| Years of service bought | 3 | 20 |
| Cost per year | 300 | 240 |
Replacement wins, and it is not close, even though it costs more than five times as much cash on the day you write the check. The 900 dollar repair is the more expensive option per year of heat delivered.
That is the base case and it is the one most people get backward, because the 4,800 is vivid and the 300 dollars a year is not.
Then hold period flips it
Now introduce the variable almost nobody puts into this equation. How long are you keeping the property?
If you are selling in two years, you do not consume twenty years of furnace. You consume two. And you will not recover the rest at sale, because appraisers credit mechanical systems thinly. A new furnace does not add 4,800 dollars to an appraised value. On a residential rental, call the recovery about 1,000 dollars, and that is generous.
| Two year horizon | Repair | Replace |
|---|---|---|
| Cost today | 900 | 4,800 |
| Recovered at sale | 0 | 1,000 |
| Net cost over the hold | 900 | 3,800 |
| Cost per year of ownership | 450 | 1,900 |
Same furnace. Same contractor. Same two quotes. On a ten year hold, replacement is the obvious answer. On a two year hold, repair wins by a factor of four.
The right answer is not a property of the furnace. It is a property of your plan. Which means the first question is not “how bad is it,” it is “when am I selling this,” and if you cannot answer that, you cannot answer the furnace question either.
One caution on the short hold. The repair has to actually reach your sale date. A repair that buys three years on a twenty two year old unit is a reasonable bet. A repair that buys eight months is not a repair, it is a deferral with a surcharge, and it will fail during a showing.
Why a repair dollar is not a renovation dollar
We made an argument recently in which renovations actually return their cost that a discretionary renovation dollar on a rental has to clear a real hurdle: its return on cost has to beat your loan constant, which on a typical investor loan today sits somewhere in the high sevens to mid eights. An 18,000 dollar kitchen that adds 85 dollars a month of rent returns 5.7 percent and makes you poorer, even though rent went up. The full mechanics of that hurdle are in the loan constant.
A repair dollar has no hurdle at all. That is the distinction and it matters.
The kitchen competes against leaving the money in your pocket. The furnace competes against an unheated house, which in most jurisdictions is an uninhabitable house, which means no rent, a tenant with a legal remedy, and potentially frozen pipes.
Put a number on the downside instead of a return on the upside. Skip the furnace in February on a 1,600 dollar unit and the realistic cost is a month of lost rent plus a make ready plus leasing, which is the same turnover arithmetic we ran in what an eviction actually costs a rental owner. Roughly 2,900 dollars, and that assumes nothing freezes.
So the furnace does not need to return anything. It needs to be cheaper than 2,900 dollars plus whatever the burst pipe costs. It is.
The two terms everybody leaves out
The cost per year table is the skeleton. Two more terms change the answer often enough that they belong in every one of these decisions.
The insurance term
This is the one that quietly costs the most, and it applies hardest to roofs.
A twenty two year old roof is not just a leak risk. It is a non renewal risk. Carriers have tightened roof age rules substantially, and an inspection that turns up a roof past its stated life can end the policy. What replaces it is either a surplus lines policy at a much higher premium or, if you are slow about it, force placed coverage arranged by your servicer, which is typically the most expensive coverage available and covers the lender rather than you.
Say replacing the roof is 9,000 dollars and the premium difference between a standard policy and what you will otherwise be paying is 1,400 dollars a year.
The roof pays for itself in 6.4 years on the insurance line alone, before a single drop of water comes through. Against a twenty five year roof, that is not a close call. And the premium delta flows straight through your operating expenses into your coverage ratio, which is the part that shows up on your next refinance.
The full sequence, including what happens when this lands on a file already in underwriting, is in roof age and your rental loan, and the broader cost trend is in how rising insurance and tax costs are reshaping deals. What your policy does and does not cover when something goes wrong is in landlord liability and umbrella insurance.
The retention term
A mid lease system failure is one of the most reliable predictors of a tenant not renewing. Not because of the failure itself, but because of how it is handled. Three days without heat in February converts a satisfied tenant into a tenant who starts looking, and a repeat failure six months later finishes the job.
Price it. If turnover on the unit costs about 2,900 dollars, and a second failure during the lease raises the probability of non renewal by twenty five points, that is 725 dollars of expected cost attached to the repair option that the 900 dollar quote does not mention.
Add it to the repair column and the three year repair goes from 300 dollars a year to something closer to 540. The replacement was already cheaper. Now it is not close.
A working decision rule
For any system that fails, in order:
- Get the age and the expected remaining life, not just the quote. A contractor will tell you if you ask directly.
- Compute cost per year of service for both options. Two divisions.
- Cap the horizon at your hold period. If you are selling in three years, nothing beyond year three counts, minus a thin recovery on sale.
- Add the insurance term if the component is one carriers underwrite: roof, electrical service, plumbing supply lines, heating source.
- Add the retention term if the failure is one the tenant experiences: heat, hot water, cooling in a hot market, the only bathroom.
- Replace on the second failure of the same component, regardless of what the arithmetic says. Two failures means the unit is at the end, and the third one arrives at the worst possible moment.
Rule six is the one to write down. Everything else is a calculation. That one is a policy, and it exists because the third failure always seems to land in a downturn or during a vacancy, which is exactly the argument in what actually happens to a rental in a downturn.
Where this hits the loan
Three places.
Your coverage ratio cannot see any of it. Capital expenditure does not appear in a rental coverage ratio, which is built from gross rent against principal, interest, taxes, insurance and association dues. A property can be quietly eating 4,000 dollars a year of deferred systems and still underwrite at 1.15. You can confirm that on the DSCR deal calculator or in how to calculate DSCR. The money the ratio leaves out is the entire subject of capex reserves on a rental, and the operating income it does use is built in net operating income on a rental.
At purchase, the appraiser can force your hand. If an appraisal comes back subject to repairs, a lender will typically require the work before funding or hold money back in escrow. That converts a decision you were planning to make on your own schedule into a condition on the closing, and it is why the age of the mechanicals belongs in your purchase diligence rather than your first year budget.
Reserves are what make the right answer available. The investor who replaces the furnace is not smarter than the one who repairs it. Usually they just have 4,800 dollars. Cash on hand is what converts this from a forced choice into an actual decision, which is most of the argument for funding a reserve at all.
The short version
Stop asking what the return is. Nothing about a failed furnace is optional, so there is no return to compute. Divide each quote by the years of service it buys, cap the horizon at your hold period, and add the insurance and retention terms if they apply. Replace on the second failure no matter what the numbers say.
The furnace decision is not really about the furnace. It is about whether you know how long you are keeping the property, and most of the time the reason this feels hard is that you do not.
If you are weighing a capital project against a refinance, or want a property underwritten with a realistic capex line rather than a placeholder, FMC Capital is a Pittsburgh-based brokerage financing investment property for investors across Pennsylvania and additional states, NMLS #1840537. Get in touch and we will model it with the real reserve in it.
FMC Capital LLC, NMLS #1840537. This article is general information for real estate investors and is not legal, tax or investment advice. All figures are illustrative worked examples, not quotes. Component costs, useful lives and insurance pricing vary widely by market, property and carrier. Loan terms, program parameters and rates vary by lender, property and borrower profile and are subject to change. Habitability standards and landlord repair obligations are governed by state and local law; consult a licensed attorney in the property’s jurisdiction.