How to Calculate Maintenance Capital Expenditure

Want to take the hassle out of finding, detailed number crunching and grading stocks? Streamline the process with an easy to use systematic method so that you can focus on the important tasks. Click here to use the live demo of old school value stock grading and valuation tool today.

Written by

Jae Jun

follow me on



To get this kind of information and other exclusive articles before regular readers, get on the VIP Mailing List today.


What You Will Learn

  • How to calculate maintenance capex, when is it useful and when is it not
  • How to calculate maintenance capex using Wal-Mart as an example

Bruce Greenwald’s Method to Calculate Maintenance Capex or Maintenance Capital Expenditure

How to Calculate Maintenance Capital Expenditure

In a previous post on how to calculate maintenance capital expenditures, I outlined the general aka quick and dirty method of calculating maintenance capex which is

Maintenance Capital Expenditure = Depreciation and Amortization

Although if you smooth out all the depreciation values across several years, you’ll be okay with your numbers, but what if you want to go one step further?

Since depreciation can be misleading, you may be missing out on opportunities by unknowingly using a unrealistically high number for maintenance capex calculations. This is where Bruce Greenwald’s method of calculating maintenance capex makes a lot of sense.

Calculate the ratio of PPE to sales for each of the five prior years and find the average. We use this to indicate the dollars of PPE it takes to support each dollar of sales. We then multiply this ratio by the growth (or decrease) in sales dollars the company has achieved in the current year. The result of that calculation is growth capex. We then subtract it from total capex to arrive at maintenance capex. – pg96 of Bruce Greenwald’s EPV book.

The above text in an understandable manner is as follows: (from my Value Idea)

  • Calculate the Average Gross Property Plant and Equipment (PPE)/ sales ratio over 7 years
  • Calculate current year’s increase in sales
  • Multiply PPE/Sales ratio by increase in sales to arrive to growth capex
  • Maintenance capital expenditure is the capex figure from the cash flow statement less growth capex calculated above, which is the true depreciation for the company

I also have been able to implement this into the stock investment spreadsheet to perform the calculations automatically with just a single ticker input. I’ll be using screenshots of the investment spreadsheet to take you through the examples. (The stock value calculation spreadsheet that supports EPV and the images below will be released in October)

Maintenance Capital Expenditure Calculations

Wal-Mart is an easy example to start off with because of its consistency.

Maintenance Capital Expenditure

Walmart Maintenance Capital Expenditure

  • From 2005 to 2009, WMT PPE as a percentage of sales ranged from 21.2% to 25.4%.
  • The 2nd line shows the sales increase. It is simply calculated by this years sales minus last year sales
  • Capex line is the actual stated amount of capital expenditures in the financial statements
  • To calculate the maintenance capital expenditures for 2009 you do 25.4% x $26,808 = $6,809
    • The $6,809 value is the growth capex so then subtract the result from Capex to get $11,499-$6,809=$4,690
    • $4690 is the maintenance capital expenditure amount WMT used in 2009.

Compare the value of $4690 to the stated depreciation and amortization amount of $6,739. Look at the difference in values between 2005-2009. Since depreciation and amortization is a straight line approach in accounting, the amount is increasing while maintenance capital fluctuates depending on the business cycle.

So it seems like that the method Bruce Greenwald teaches is a much more accurate representation of maintenance capex.

Heavy Capital Expenditure Companies

The only problem I see at the moment in applying this method is that the maintenance capex remains incorrect for capex heavy companies. Take Conoco Philips (COP) for example.

maintenance capex

The maintenance capex is negtive because the PPE % of sales applied to the sale growth is very high. Does this mean that capex has been understated where the maintenance capex was actually $19,099+$5435=$24,535 in 2009??

Please let me know if you have the answer to this.

Some Final Thoughts

So the Greenwald method is a good one and makes sense for consistent companies but won’t work if the sales growth number is greater than the capex figure.

The method isn’t perfect and still doesn’t seem to be able to determine the maintenance capital expenditure of commodity based businesses such as COP, ATW and NUE.

That is how you calculate Maintenance Capital Expenditure. Am I doing something incorrectly? If so, please leave a comment. I would love to know.

To get this kind of information and other exclusive articles before regular readers, get on the VIP Mailing List today.


Want to take the hassle out of finding, detailed number crunching and grading stocks? Streamline the process with an easy to use systematic method so that you can focus on the important tasks. Click here to use the live demo of old school value stock grading and valuation tool today.

37 responses to “How to Calculate Maintenance Capital Expenditure”

  1. G4nz0 says:

    I am trying to replicate your calculations for WMT and COP – specifically the PPE as a % of Sales. Are you using the net PPE (from the balance sheet) and dividing it by the Revenues (from the income statement)?

    My percentages are way off from yours. In 2004, COP had sales $136B and PPE of $50B (give or take). With the exact numbers, it generates PPE to Sales ratio of 37.2%. Not the 75.2% you have above.

    How did you get that 75.2%? Are you averaging the previous 7 years of data?

  2. Ryan B says:

    Jae Jun,

    Are you going to be emailing out the new excel templates that include the Maintenance Capex calculation.

    If you take the averages for COP the Maintenance Capex calculation works out. I calculated the average for the three year and five periods as per your example and it worked out to a reasonable number.

  3. Jae Jun says:

    Actually it seems I was selecting the wrong cells in excel. I was using 2002 numbers for 2005 and 2003 for 2006 and so on. But the numbers still look wrong. I’ll have to recheck it.

    Yes Im using net PPE/revenues

    @Ryan B
    I’ll be sending out the new intrinsic value model spreadsheet to anyone that bought after Oct 2008. It includes more than just a capex calculation. A full EPV valuation method as well as a summary page and many changes.

  4. Kyle says:

    COP wrote off a ton of stuff in 08 and 09, that might be affecting the number pulled into the spreadsheet.

  5. Jae Jun says:

    Unfortunately, since the number used is sales, impairments are ignored in the equation.

    I’m thinking that rather than multiplying the ppe/sales ratio to the sales growth, why not just multiply that ratio to capex directly.

  6. I don’t understand why you would multiply that ratio to capex. That ratio is used explicitly to find what percentage of sales is needed in PPE to support those sales, and as a result is used to find growth capEx as you’ve mentioned. Multiplying the ratio by capEx would be a misuse of the ratio I would think. Doesn’t make intuitive sense to me to use it that way. Please elaborate.

    Great article btw.
    .-= Jonathan Goldberg´s last blog ..Lecture Summary – Francis Chou =-.

  7. Jae Jun says:

    My thought is that the ratio is really only indicating growth in sales due to the level of PPE which is why the numbers are incorrect if PPE makes up a big amount of sales.
    Right now if I go through any company that has a log of tangible assets, the maintenance capex numbers will always be incorrect. So if maintenance capex is incorrect for companies like COP or NUE or other commodity based businesses, EPV will also not work.

    From what I see, the calculation essentially takes away a sales figure from capex. So to me it seems like I am taking an apple and trying to subtract it with an orange.

  8. I’m a little bit confused but I will try to make light of the way I understand what Greenwald is trying to show.

    PPE/Sales shows the level of PPE that is required to support a given level of sales. By multiplying this by the increase in sales we are finding “Growth CapEx” which is the CapEx that is being used to support the extra sales.

    Actual CapEx is made up of “Growth CapEx” and “Replacement CapEx” – so by subtracting growth capex from current year’s capex we are left with replacement capex.

    Replacement CapEx is what we want to use in our calculation of EPV as we are using “no growth” cashflows to find EPV.
    .-= Jonathan Goldberg´s last blog ..Lecture Summary – Francis Chou =-.

  9. Jae Jun says:

    I understand what Greenwald is saying but after reading it so many times and thinking about it, I’ve come to the conclusion that it isn’t correct.

    When you multiply PPE/sales to the changes in sales from the previous year, the resulting number does not equal growth capex. Capex is what goes out, sales is what comes in, and by multiplying to sales, Bruce seems to be using an income figure to say it is an expenditure. Just doesn’t make sense how it can be used as growth capex.

    So as you mentioned, if PPE/sales shows the level of PPE required to support a given level of sales, multiply PPE/sales to capex in order to get maintenance capex rather than growth capex.

    After going through several companies, it seems to work out better because since PPE/sales is always less than 1, it which means I’m not getting negative maintenance capex as I do when I use Greenwald’s method.

    What do you think?

  10. Jae Jun says:

    A further point I thought about was that with Greenwald’s method, my example above shows that if the change in sales is either very large compared to capex or if sales is negative, then the maintenance capex becomes negative which isn’t possible.

    Even if a company loses money in a given year or does worse than the previous year, the company obviously has spent money related to growth capex.

  11. I definitely see what you’re saying but to me multiplying the ratio by the change in sales definitely makes sense as growth capex. Capex had to be spent above and beyond the amount needed to maintain prior year’s sales in order to reach this new sales number.

    Regarding the negative aspect… I’m not too sure. I usually just let it be a negative in my model so that I am evaluating on a consistent basis.

    I would be interested in reading more information about this to really lay out the logic behind it as well as solidify any possible modifications. I don’t know anywhere else this is discussed besides Greenwald’s book however.

    If you do any more on this please let me know! [email protected]
    .-= Jonathan Goldberg´s last blog ..Lecture Summary – Francis Chou =-.

  12. Jae Jun says:

    Will do Jonathon. Maybe I’ll try emailing Bruce Greenwald and ask him directly.. He’s a professor so I’m sure his email is on his school website.

    It’s also a shame that there is only a handful of websites or blogs that discuss EPV. So far I only know of you, my Value Idea and me..

  13. Yeah would be great if more people went through all the trouble we do before buying a stock!

    Great idea to email Bruce. Let me know if you do and hear back.

    Regarding this CapEx issue it’s definitely a tough one to get my head around. The way I see it just do what makes most logical sense for the situation (until you hear from Bruce of course!). If you get a negative number then that obviously doesn’t make sense as companies can never go with zero nevermind earning money on their maintenance capex. If the number is positive and within reasonable range of depreciation then I’d go with the zero growth capex number as it does make sense intuitively to use change in sales to ppe/sales as a proxy for the extra capex required to support the additional sales.

    Now one other issue I realized that we haven’t discussed yet is the issue of a negative GROWTH capex number. This would happen when the company’s revenue decreases. This does not make intutive sense as by subtracting this negative number from current capex we are imlying that if the company maintained its level of sales it would have a greater level of capex than it does in a year of decreasing sales. I would think that a year of decreasing sales and a year of flat sales should have the same capex and it would make sense that this is by definition mainentance capex. So in either of these situations I would just use that years capex as the zero growth capex number.

    Would love to hear your thoughts.
    .-= Jonathan Goldberg´s last blog ..Position Update – TSE:BUI =-.

  14. Jae Jun says:

    That’s pretty much what I did with the spreadsheet. If the change in sales was negative, I just took that years entire capex as the maintenance capex. So no increase in sales means no growth capex.

    Now I got to find Mr Greenwald’s email and ask him these questions.

    You would also think that there would be someone, even among the pros and famous managers that would talk about this.. From what I’m experiencing, even the best value investor managers may not even calculate maintenance and growth capex.

  15. Travis says:

    Hey Something occurred to me in relation to your heavy capital expenditures question. For COP it could be more an issue of Price volatility. It take a certain amount of asset to produce a barrel of oil that value doesn’t change if the price of oil doubles. You might try using a barrels of oil/ppe instead of sales per ppe.

  16. Jae Jun says:

    But what about non oil companies? Because I would like to know how to apply a fair calculation to all companies.

  17. Ola says:

    After lurking for a long time, thought I’d leave a quick post.

    Since capex directly affects output rather than sales (a company may not be able to shift the additional outpu) would it not be better to use sales and increases in inventory (as a proxy for total increase in production)?

    Also if information on the unit amount of production is available (as travis suggested for COP) this should provide a more appropriate figure.

    Great blog by the way.

  18. Jae Jun says:

    Hi Ola,

    Thanks for taking the time to comment.

    I’m not sure I agree with your view that capex affects output. Growth capex, yes, which eventually leads to increased sales. But maintenance capex does not affect the output at all.

    I questioned Greenwald’s method of calculating maintenance capex a very long time, but I do see the logic and from everything I’ve considered, it is the best method so far, but still flawed in some areas.

  19. Jim says:

    Concerning the maintenance capex calculation. My understanding is that the PPE/Sales Ratio is supposed to be “averaged”?

    Bruce Greenwald: “We calculate the ratio of PPE to sales for each of the five prior years and find the average. We then multiply this ratio (the average) by the growth (or decrease) in sales dollars the company has achieved in the current year.”

    Warren Buffett states:

    “These represent (a) reported earnings plus (b) depreciation, depletion, amortization, and certain other non-cash charges…less (c) the average annual amount of capitalized expenditures for plant and equipment, etc. that the business requires to fully maintain its long-term competitive position and its unit volume….Our owner-earnings equation does not yield the deceptively precise figures provided by GAAP, since (c) must be a guess – and one sometimes very difficult to make.”

    Therefore, it would seem to me that Warren is describing a similar process as Bruce Greenwald describes. It also further reveals that Warren is interested in knowing the value of a business today assuming no growth.

    My question to this board is, and currently I’m still exploring maintenance capex so I haven’t formed an opinion yet, would it make sense in averaging the conclusion you come to regarding maintenance capex over a period of, say, 5 years? For example:

    Bruce says to find the previous 5 year average of PPE/Sales and apply that average as your factor against the current years increase or (decrease) in net sales. After doing so, does anyone think it would make sense that the resulting information (growth capex) be removed from a cyclically adjusted capex? I notice that most of the posts and examples here have proposed to remove growth capex from that year’s stated capex but to me, and so far its just an untested thought, would be to remove the averaged growth capex from a cyclically adjusted (averaged) capex that is stated for book purposes.

    Any thoughts?

  20. Jae Jun says:

    My opinion is that, yes it should be averaged.

    Reason being, in a cyclical environment companies reduce capex and it goes up in later years. So the norm would most like be that 5yr averaged number.

    Besides, the point of averaging is to try and eliminate the effect of any cycles so if I can get a better picture of the company at normalized capex levels, it should be a more accurate valuation.

  21. Jim says:


    That’s my feeling to. Just wanted to see what others thought. Thanks for your input.
    .-= Jim´s last blog ..Asta Funding – Postion Exited & Analysis =-.

  22. Ola says:

    Hi Jae,

    What I was getting at with regards to capex affecting output is that if machines are not maintained then they are likely to have increased downtime, which will likely reduce output.
    As sales revenue = units sold x price, and units sold is a function of output, then capex should affect production.

    Possibly a better way of putting it is if capex was reduced to 0, there would definitely be an effect on output (particularly on max capacity)

    So where possible, increases in output (or a suitable proxy) should be used as opposed to increases in sales. Especially when companies are price takers…

  23. Simon says:

    Hi Jae,

    Jacob say use GROSS PPE – “You can calculate maintenance CAPEX by:
    Calculate the Average Gross Property Plant and Equipment (PPE)/ sales ratio over 7 years”

    Yet, in your calculation, you are using PPE right???

    Has this problem been resolved on another thread?

    What is your latest thinking on this issue?


  24. Jae Jun says:

    What Greenwald says and what Jacob says is exactly the same thing except Greenwald says it’s ok to use 5 yr average.

  25. Prashant says:

    Hii Jae,
    I do not understand negative maintenance capex. I was valuating Conoco Phillips (COP) using EPV and found out this thing. I have a question; Bruce Greenwald did the same thing for Intel (INTC) in his book wherein he took just a conservative 25% of amort and depreciation and added back to normalized earnings and let the rest 75% of Amort and Dep take care of maintenance capex.

    I tried that for current Intel financials and found that maintenence capex is only 40% of stated amort and dep. Bruce really takes very conservative numbers hence conservative valuations.

    But I saw that if I do the same for Conoco Philips and take a 20% of Amort and dep and let the rest take care of maintenance capex, then my valuations fall in line with DCF(not exactly when I use WAAC; but when I take the bruce and buffett way of calculating discount rates) and graham number valuations.

    Would you please put some further light on this?

    A curious investor
    Prashant Sharma

  26. Steve says:

    Won’t this lead to an incorrect CAP X when sales are declining. I think that using revenues to determine Growth Capital expenditures is not an accurate method at all. Sales are a trailing indicator to new growth capital expenditures, and not all result in a proportionate increase in sales. Wouldn’t it be more accurate , at least for small to medium sized businesses, to find an average capital expenditures and assume that the average is what is required for the industry and any positive outliers are considered growth capital expenditures? You cited the oil industry as an example, and I would argue that even their exploration is necessary for them to maintain their business. Obviously you need to determine how much they are investing in exploration to determine future prospects and profitability, however, Greenwald’s method would do nothing to truly make the determination between maintenance capx and growth capx.You really need to simply dig deeper and determine what they are spending on exploration vs maintaining their current facilities.

  27. Hua says:

    Hi Jae,

    Any updates on EPV with the heavy capex companies where PPE/sales>1? Thanks.

  28. Jae Jun says:

    @ Hua,
    Not at this time. I still see it as a flaw, so you’ll have to use good judgement on what to use. Or use the full capex amount to be on the safe side unless the annual report mentions something about it.

  29. red. says:

    Did my comment get lost in the interwebs?

    A shorter version:

    1) maintenance capex is properly calculated using the ratio of fixed assets to sales. Therefore:

    (gross PP&E + capitalized operating leases) / sales

    That Walmart, in your example, came out okay is happenstance: Walmart happens to keep its ratio of owned PP&E to operating leases constant over time. Relatively few businesses do this. Therefore, one should include the capitalized value of leases in order to get reliable results.

    2) The Greenwald formula works for asset-heavy businesses. Why wouldn’t it? In the COP example, eliminate DEPLETION, from the DD&A figure in the data source, and you’ll get an accurate reading of maintenance capex.

    Cheers & happy investing.

  30. Mihir Shah says:

    If the over the five years (or a given period) say the company’s accumulated depreciation is less than the accumulated maintenance capex for the same period, does it mean the auditors should review the accounting policy as it might be too liberal?

  31. Ted Rasa says:

    Wow, great discussion here on Maintenance Capex.

    For technology companies, wouldn’t it make sense to include Goodwill & Intangibles?

    So the formula would look like this:
    (PPE + Goodwill + Intangible Assets) / Sales

  32. Hey Ted.
    Im not following the reasoning for this.
    The goodwill and intangible assets don’t need anything to maintain so there is no need to add it which will only increase the total maintenance capex.

  33. Ted Rasa says:


    My reasoning is… goodwill and intangibles assets are accumulated with acquisitions. Not all acquisitions provide growth.

    i.e. Google’s purchase of Motorola Mobilility and Waze. (this could be argued)

    Rather Google is maintaining the competitive position of the company. You would then consider this maintenance capex, not growth capex.

    Your thoughts?

  34. Since acquisitions are purely for growth reasons, I would still not include it. Maintenance capex is the cost of doing business. Having additional brand names doesn’t add to the cost of maintaining a business. Bad acquisitions are just bad purchases and should be marked down. Instead, maintenance capex would be the cost to keep those acquisitions running. New building, painting, updating interior, new signs, etc.

  35. Ted Rasa says:

    My interpretation of maintenance capex differs a bit. Maintenance capex is the capex spent to maintain revenues. The way this is done will vary from business to business.

    I definitly would not pigeon hole all acquisitions as purely growth related. There are value added acquisitions just to keep your business relevant, this in turn will “maintain” sales.

    This is the reason why depreciation expense can be a misleading metric to compute maintenance capex.

    PS- in a post manufacturing world, intangible assets extend far beyond brand names. Think pharma, SAAS, video games, google, Facebook. If these companies were acquired, the majority of there assets would be in intangibles and goodwill.

  36. Rob Urban says:

    I think figuring out a “base” level of owner earnings is the key to determining a company’s true worth. I also tend to believe that Buffett wants to earn his 15% and this is what he is focused on. By focusing on a base maintenance level of owner earnings you are required to take a companies long-term competitive position into account as well as to determine the size of the moat or sustainable competitive advantages. And since you are a passive investor and cannot control how the cash is allocated, spent or invested, you need to figure out if margins are sustainable and if the company can continue to generate sufficient levels of return on invested capital, and if management is wise with investing that cash and protecting it. There’s no need to do discounted cash flows or relative P/Es or anything like that, you just need to see if what you can earn in owner earnings is better than other investment opportunities out there. Sounds simple, but it’s not.

    Bruce Berkowitz said: “the value of a stock is the amount of cash that the underlying company generates. cash is important, because it’s the only thing my family can spend. you run out of cash, you’re dead, whether a business or an individual. cash counts, so i count the cash. it’s not just any kind of cash, it’s the cash that an owner can keep after all the bills are paid and after the capital used to maintain the franchise of the business. once i have the cash count, i try to calculate it. when i have a range, i try to figure out how much on average a company can generate in cash. the bad years, the good years, what’s normal? what should you expect over a cycle? and then, i try and kill it. what can stop the cash from coming in? what can kill the company? i then take the calculation and compare it with the price of the stock. because investing, after all, is nothing more than comparing what you give to what you get in the future. i try to figure out if there is a margin of safety. can i get hurt? am i buying this stuff cheap enough

  37. David Addison says:

    I have my own method for estimating CapEx. Basically I average CapEx over the past several year to arrive at a baseline estimate, CapEx_0. I then look at how much Long Term Assets must grow to keep pace with earnings growth. Call Long Term Assets, LTAssets_0 and Long Term Earnings Growth, g. Therefore the CapEx at some time t, CapEx_t should be approximately equal to CapEx_0 + LTAssets_0*(1+g)^t – LTAssets_0*(1+g)^(t-1).

    Simply this to get: CapEx_t = CapEx_0 + LTAssets*g*(g+1)^t

    Would love to hear feedback!

Leave a Reply

Ready to try Old School Value?