How to account for large, irregular expenses in FI planning

We have a pretty decent cash flow projection tool built in google sheets.

for large home expenses or remodeling projects, I put about 2.5% of our home value into our annual spending budget/plan. For us that is around $8000-$10000. Many years we are under that but some years we can be over. Keeping this constant at about 2.5% of home value is good especially for expenses that are infrequent, and that you will not have to buy again after you have bought the first time (roof, fence, pool, etc.)

For health expenses, I just average what we've spent in the last 5 years. (We've had 4 kids in 5 years, so that should be conservative enough). When we eventually switch from employer coverage to a bronze ACA plan, our deductible will go up, so I will re-evaluate then.

For vehicles, we do low interest loans, but there are ways to put one-time expenses into cash flow tools, like for guessing in the future when you would need to replace a vehicle. Even if we paid cash for vehicles, we wouldn't save specifically for it, it would just come out of our brokerage.

We don't budget car maintenance and repairs separately from other categories. Would be too much detail for us, and fortunately we haven't had significant repairs needed.

For the truly random large expenses that you can't predict or plan for, we would just view that as a brief delay in our FI journey, or something that we can't control. Another example is market returns, we can't control the market returns either.

We have no savings account or emergency fund. We just have a checking account with 30-60 days of expenses in it, and our brokerage account is our "emergency fund" where we transfer to/from as needed.

So basically, we do have an annual spending number, but it's on the pessimistic side. Plus our family is growing, so I have to put in assumptions that our expenses will rise faster than inflation.

The big thing is separating planning from execution.

• Planning: smooth those big, irregular costs into your annual number (cars, home repairs, etc.) so your FI target is realistic
• Execution: you don’t actually spend that money each year, just draw from your portfolio when those expenses hit

That’s how people keep their plan without overcomplicating cash flow.

One thing that caught me off guard is how much taxes can shift this too, especially in years where you’re taking larger withdrawals. It’s worth factoring in alongside the lumpy expenses. Good resource on that:

The One Big Beautiful Bill - Webinar — BGK Financial

My wife and I used the sinking fund methodology because the math of how much things cost (taking the expense and dividing over however many years) didn't also account for the complication of an expense coming up unexpectedly (where does the income come from and does that bust our ACA budget, etc.).

You definitely have to budget for it because when working it feels like a one-time expense but the reality is that when retired you take that money out of your 'everything else' budget and pay for it, you have to replenish it somehow, which means it impacts your total FI # and your SWR (even if it is lumpy). So it has to be part of your flow of money in and out.

Our first year of retirement was a crash course in this because our furnace, fridge, and washing machine all broke, and we had major medical bills that insurance didn't cover. We handled these by spending from a large cash buffer rather than taking more income from our retirement accounts, which kept our reported income low enough to protect our ACA health insurance subsidies. By including a set amount in our annual budget to top that buffer back up, we ensure these infrequent costs are factored into our safe withdrawal rate and don't bust our long-term financial plan. You don't have to have enough for all infrequent expenses in cash, but you have to have a large enough cash buffer if you do it this way that it can be topped off without having negative consequences.

Some good comments from Frank and Robert already. Personally I looked at my past expenses (which we’ve tracked for 20 years) and saw that major items (over 3k one offs - car, roof, hvac, lasick, dental implants, deck rehab) occurred about every other year and typically 5k and then the random 20-30k item every 5.

So from this I plan 10k a year for unexpected expenses, which to Roberts note is just under 10% of our expenses. The key thing is some of them are discretionary (in down years could vary the quality of car or push off house rehab, and good years can splurge or make the bigger expenses proactively).

so yes use your data and don’t blindly add which is the laziness that can turn conservative on top of conservative. So you seem on the right track.

I follow risk parity (thank you Uncle Frank) so plan to keep everything invested and sell what’s high when these items come up.

Typically they even out – so you budget for them as a discretionary expense that can be spent on something else if it doesn't materialize.

So figure out how they already run in your life. That is not going to change in retirement unless you radically change your lifestylr.

The biggest problems I see in retirement planning are Laziness and Learned Helplessness. People are Lazy in that they refuse to add up their actual expenses from their current records for the past two or three years, divide that into mandatory and discretionary and use that as their baselines, form which you should also calculate your PERSONAL inflation rate. People engage in Learned Helplessness (and forms of catastrophizing) in that they pretend like their future expenses are going to be some weird unknowable and unfathomable thing that is not directly related to their current expenses and will likely be very similar in many aspects.

I've recently been pondering something similar. I think that there are essentially 2 ways to handle this. One way would be to take a year of "regular" expenses (i.e., food, mortgage/housing, taxes, insurance, healthcare, transportation, leisure, etc) and add something like 10% or 20% for "lumpy" expenses. The other way would be to project forward using data on the standard service life of various "major" things. So, new vehicle after 15 years, new HVAC after 20 years, new roof after 25 years, etc.

I've been taking something more like the first approach. However, I'd also be interested to know how others are handling this sort of thing.