Maintenance capital and working capital

Revised practical article; source details and figures are dated to their cited filing or event, with current evidence reviewed through 24 September 2026.

A business can look cash-rich because it has stopped replacing equipment. It can also look cash-poor because it is building the next stage of capacity. If we call all capital spending “maintenance,” we can understate what a growing business might earn. If we call all spending “growth,” we can overstate what an owner can safely take out.

The accounts usually tell us how much the business spent, but not how much it had to spend to keep its existing competitive position. The 1986 Berkshire letter defines owner earnings by subtracting the average capitalized expenditure needed to maintain long-term competitive position and unit volume, plus extra working capital required to sustain that position and volume [1]. The definition contains two separate judgments: fixed investment and working capital. Neither should be replaced automatically by depreciation or a percentage of sales.

Why depreciation does not settle the question

Depreciation spreads an accounting cost over an estimated useful life. It is not necessarily the money the business must spend this year to remain productive. A delivery van may be depreciated over five years, but a sudden shortage of parts may make replacement cost rise. Software can become commercially obsolete before its accounting life ends. Conversely, an old building may keep serving the same customers with modest maintenance even as accounting depreciation continues.

The gap is not proof that management is manipulating results. It is a reason to inspect the assets and the operating task they perform. A business owner can ask:

Which equipment, systems, licenses or premises are essential to deliver today’s saleable output?
What repairs, upgrades and replacements have been made over several years?
Has capacity, location, service quality or product range expanded?
Are there overdue replacements or deferred maintenance?
Does the business plan rely on a capacity level that the current assets cannot sustain?

A useful operating history includes several years of capital spending, depreciation, disposals, capacity additions, service failures and, where possible, physical asset condition. Use cash paid for property and equipment from the cash flow statement, then reconcile it to additions in the fixed-asset note. The two can differ due to unpaid purchases, leases, acquisitions, disposals, foreign exchange, or construction timing.

Economic depreciation research makes a related point: the loss in an asset’s productive value may not follow a simple straight-line path [2]. A standard depreciation schedule can be an accounting convention rather than a map of replacement economics. It is evidence; it is not the owner-earnings answer.

A practical range, not a magic ratio

Suppose a small milling business reports KSh 1.5 million in depreciation and spent KSh 2.4 million on equipment during the year. It replaced one motor, installed a larger mill, and added a delivery vehicle. The filing or accounts may not allocate the KSh 2.4 million between upkeep and expansion.

A defensible draft might state:

Observed fact: total equipment cash spending was KSh 2.4 million for the year, based on bank records and invoices.
Observed fact: the new mill raised rated capacity; the motor replaced a failed unit; the vehicle added a delivery route.
Analyst estimate: maintenance and replacement are provisionally KSh 0.8–1.3 million, pending multi-year repair and replacement records.
Unresolved: no reliable evidence yet separates routine maintenance from the costs of keeping output at its former level.

Then show the owner-earnings range under both ends of the estimate. If earnings move materially, the uncertainty is material and belongs in the conclusion. Do not choose the number that makes the company look cheapest.

The same rule applies to listed companies. A public filing may show total additions to property, plant and equipment, but the report rarely gives a clean “maintenance capex” subtotal. Management may discuss growth projects or capacity plans, but that guidance is still a forecast and may not cover all replacement needs. The article should say “maintenance capex is not separately disclosed” and give a range only when the evidence supports one.

Working capital is investment too

Working capital is the cash committed to receivables, inventory and other operating balances, net of supplier credit and operating liabilities. It is part of the operating system. A shop that adds inventory to reduce stock-outs may tie up cash while improving customer service. A distributor that gives a retailer another month to pay is financing that customer. A growing manufacturer may need to buy raw materials before collecting from its buyer.

The basic cash-conversion-cycle estimate is:

Days sales outstanding + days inventory outstanding − days payable outstanding

For a period with suitable average balances:

DSO = average trade receivables ÷ credit sales × days in period
DIO = average inventory ÷ cost of goods sold × days in period
DPO = average trade payables ÷ purchases (or a disclosed cost proxy) × days in period

The denominator choice matters. Cost of sales is not always the same as purchases. An ending balance is not the same as an average balance. Seasonal businesses can make a year-end snapshot misleading. Write down the numerator, denominator, dates and source before interpreting the ratio.

Peer-reviewed studies find links between working-capital management and firm performance, including work on large Belgian firms and a large sample of Spanish SMEs [3], [4]. Later research on UK SMEs examines how cash flow changes the relation between net working capital and performance [5]. The message is not “push CCC as low as possible.” It is “measure how much cash the operating model absorbs and what the business receives in return.” Extending payable days may protect cash this month while damaging supply reliability next quarter. Cutting stock may improve reported working capital but cost sales if lead times are long.

Growth and maintenance can coexist

Classifying an asset purchase as “growth” does not make it optional forever. A new delivery vehicle can add a route today and become part of ordinary fleet replacement later. Conversely, an asset labeled “replacement” may also add capacity or lower cost. This is why an analyst should describe what changed in the operating system rather than relying on management’s label alone.

For an owner deciding what to distribute, combine the question with the next year’s operating plan. Required maintenance spend may arrive in lumps. A motor replacement every four years does not mean the business can ignore it for three years and then be surprised by the invoice. A reserve can smooth known replacement cycles, but it should be tied to a list of assets, estimated timing, and cost range.

What the workbook should make visible

A future model should show, for each asset group, opening carrying value, cash additions, disposals, depreciation, and closing value. A separate analyst schedule should identify the basis for maintenance low/base/high cases. Working-capital inputs should show actual balances, period dates, and the formula used. The owner-earnings output should respond to both maintenance and working-capital assumptions. It should not quietly insert a universal capex-to-depreciation ratio.

The next article turns the same cash-cycle logic into a weekly operating routine: what to forecast, how to see a shortfall early, and how to avoid rules that sound precise but do not fit the business.

Build the estimate from the asset’s job

A practical asset review begins with the job each item performs. The motor in a mill may be essential to the current output. A second mill may add capacity. A backup generator may protect the business from outages even if its use is low in a normal month. A point-of-sale system may reduce theft and processing time. The model needs to know what would happen to current sales and service if the asset failed or was not upgraded.

Imagine a cold-room business with one compressor installed six years ago. Its accounting depreciation is KSh 180,000 this year. The maintenance log records routine service and refrigerant work, while a major replacement is estimated to cost between KSh 700,000 and KSh 950,000 within the next two years. A second cold room, costing KSh 1.8 million, was installed to serve a new customer. The company spent KSh 2.5 million total this year, but the numbers do not map one-for-one to maintenance and expansion. A sensible worksheet would list the compressor and new cold room separately, link invoices and service records, state replacement timing, and show the effect of the plausible replacement range on owner earnings.

If the compressor has a documented eight-year service life and the current replacement estimate is KSh 800,000, a simple reserve planning estimate might spread that cost over the remaining period. That is a planning convention, not proof that annual cash spending will be KSh 100,000. The unit could fail sooner, the replacement price could change, or maintenance could extend its life. The owner should preserve both the expected timing and the risk range. Economic replacement investment and the accounting charge can differ because asset efficiency and age do not necessarily move in a straight line [2].

The working-capital bridge depends on the scale

Suppose a wholesaler normally sells 1,000 cases a month and needs KSh 2 million of stock to support that volume. It begins a new contract requiring another 300 cases every month. If supplier credit and customer terms remain unchanged, the extra sales may require a one-time build in stock and receivables. That incremental amount funds growth. By contrast, a higher receivables balance caused by slower customer payment can increase required working capital even when monthly sales are flat.

For owner earnings at the existing volume, ask what working capital is needed to preserve service, not what happened to the entire balance this year. For an actual historical cash bridge, show the full change in operating balances. Then distinguish the portion associated with growth, payment timing, acquisition, currency or an unusual event. This prevents a growing business from being penalized as if all growth capital were a permanent maintenance charge, and it prevents a stagnating business from claiming that overdue invoices are harmless.

Questions for the next owner review

What is the date, condition, and role of each material asset?
Which repairs or replacements happened over the last three to five years?
What open purchase order, lease, or supplier commitment is not yet in cash capex?
What additional inventory or credit is required to maintain current sales volume?
Which working-capital changes are growth-related, seasonal, overdue, or recoverable?
If this assumption is wrong, how does it change cash, debt needs, and the owner-earnings range?

These questions should be answered by invoices, asset logs, monthly balances, contracts and operations records. If the answer depends only on a generic capex ratio, the analysis is not yet grounded in the business.

Estimate maintenance spending from the work the assets do

A useful maintenance estimate starts with an asset register that connects each asset to a required operating task. Include the asset ID, type, purchase date, installed cost, current condition, useful life, service capacity, repair history, replacement lead time and next planned work. Record whether the business owns it, leases it, or relies on a supplier’s equipment. A refrigerated van may be essential to preserve current customer service even if it is leased. A repair contract may already cover part of the cash cost. The register makes those facts visible before the owner chooses an estimate.

Classify spending into four work categories:

Routine upkeep: repairs, service and consumables that keep equipment in safe working order.
Replacement: spending that renews exhausted or obsolete capacity already required for current sales.
Compliance: upgrades required by safety, environmental, tax, industry or customer rules to keep trading.
Expansion or improvement: spending that adds capacity, opens a new location or materially changes the service.

A single invoice can contain more than one category. For a mixed project, record the component and its share, with the basis. A KSh 900,000 refrigeration upgrade might replace compressors worth KSh 500,000 and add a new cold room worth KSh 400,000. If invoices or contractor schedules establish that split, the model can allocate it. If not, carry a low/high allocation and state what document would narrow the range. Compliance spending may not increase output, but the business cannot sustain legal operations without it; include it in the cost of maintaining the existing ability to trade.

Use three independent lenses to estimate the range:

Asset-by-asset replacement plan. Estimate next-year repair and replacement dates and current supplier quotes. This is the most practical short-run cash budget. It can be lumpy, so a single year might not represent long-run earning power.

Multi-year normalized spending. Compare actual capital additions, repair bills, disposal proceeds and capacity changes across a full replacement cycle. Exclude clearly evidenced new capacity from the no-growth case. Include replacement spending even when it is irregular. This method helps smooth lumpy equipment years, but acquisitions, currency movements and inflation need separate treatment.

Capacity and service test. Ask what assets are necessary to produce current units at the promised quality, delivery time and compliance level. A company may have very old assets and low current capex because maintenance has been deferred. A low-spend history is not proof that current operations can continue without investment. Record breakdowns, lost orders, service delays and customer specifications as well as accounting amounts.

Compare the three results instead of averaging them automatically. A current supplier quote may exceed historical spend because replacement cost has risen. A capacity study may identify a major system that accounting depreciation understates. Show the low, base and high estimates with the item, evidence, owner and confidence. The base case is an assumption, not a fact. Change it when service records or new quotes arrive.

Reconcile the working-capital investment at the same activity level

For a repeatable analysis, split working capital into balance, cash change, days measure and reason. A useful table contains opening and closing receivables, inventory, operating payables, accrued operating costs and customer advances. The operating working-capital balance can be written as:

Trade receivables + inventory + other operating assets − trade payables − operating accruals − customer advances

The cash investment for a period is usually the increase in that balance, after separately checking noncash movements and acquisition effects. An increase uses cash. A release provides cash. A one-time release from collecting old invoices should not be counted as a recurring annual owner-earnings boost.

Separate current activity from growth. If unit sales grow 20%, the company may need more stock and receivables even with unchanged collection days. Model the cash required to support that growth in the growth case. If receivable days rise while revenue stays flat, that is a collection or customer-credit issue. If payable days rise because a supplier renegotiated and remains current, that is an agreed source of operating funding. If invoices pass their due dates with no agreement, report arrears and supplier concentration separately. Do not give a struggling company an artificial owner-earnings benefit because it stopped paying its suppliers on time.

Seasonality changes the right comparison. A fruit packer may buy, harvest and pay labour before receiving export cash. The year-end inventory balance can be close to zero just after shipment even though the peak seasonal funding need was large. A school supplier may collect deposits months before delivery and then carry an obligation to provide goods later. Use monthly or weekly balances around the busy period and align them to customer and supplier terms. Year-end snapshots answer what was on the balance sheet that day; they do not reveal the peak cash required to complete a normal trading cycle.

A working-capital assessment should finish with a funding decision. If an order increases contribution but adds a receivable and stock requirement, calculate the peak amount and days financed. Then compare customer deposits, supplier terms, a revolving facility and the owner’s own cash. The facility needs to remain available until collection occurs. Do not use a six-month asset loan as though it were automatically the right instrument for a 45-day invoice gap.

A practical range that keeps the decision usable

Suppose a mill uses a motor, grinder and delivery vehicle for its current output. Its three-year normalized replacement schedule points to KSh 800,000 of annual upkeep. New supplier quotes show KSh 1.1 million at current prices. A larger mill costs KSh 1.8 million, of which KSh 700,000 replaces the current grinder and KSh 1.1 million adds a second line. The analyst might present maintenance of KSh 0.8–1.1 million, cite the three-year invoices and quotes, and classify the second line as growth. The total capex plan remains KSh 2.9 million. This lets an owner see that the base business needs replacement capital even though only a portion of the expansion improves volume.

If no asset condition report or purchase history is available, the result is a temporary range that can guide records collection, not a confident historical valuation. The next step is concrete: inspect the equipment, ask the supplier for written replacement terms, pull three years of invoices and tie each purchase to an asset ID and operating purpose. Then update the model. This produces a stronger business decision than substituting depreciation, total capex or a generic percentage for evidence.

Reconcile the asset register before estimating replacement cost

A practical fixed-asset review begins with an inventory of what the business actually uses. Assign an asset ID, location, custodian, purchase date, original cost, current condition, downtime history, remaining useful life, repair history, replacement lead time, and whether the asset is owned, financed or leased. For a vehicle, add mileage, fuel consumption, tyre history and route capacity. For a cold room, add temperature failures, energy use, compressor service and spoilage events. The register should reconcile to the general ledger and a physical inspection; a book value alone does not confirm that an asset exists or still works.

Then link each asset to capacity. How many production hours, customer orders or deliveries depend on it? What is the fallback if it fails? An inexpensive point-of-sale system can be more critical than a costly spare machine if it controls every checkout. Rank replacement needs by business interruption and cash cost, not only replacement price. List a realistic replacement date and lead time, then spread the expected cost over the period the asset supports. This produces an annual planning reserve, while the payment schedule remains lumpy in the cash forecast.

A useful three-estimate method is: (1) recent maintenance and repair spend, excluding one-off accident or expansion costs; (2) a unit-level replacement plan based on supplier quotations and expected timing; and (3) a capacity-and-condition review by the person operating the asset. Compare the estimates. If the business has unusually low repairs but old equipment, deferred work may explain why historical capex understates the future need. If annual capex is high because the firm is adding capacity, do not label all of it maintenance. Preserve a low, base and high case with a reason for each.

Close the estimate with a cash bridge: opening asset cost plus additions minus disposals equals closing gross cost; opening accumulated depreciation plus depreciation less disposals equals closing accumulated depreciation; and opening net book value plus additions less depreciation and disposals reconciles to closing net book value. This does not calculate economic maintenance capex by itself, but it reveals where the filing or ledger fails to reconcile. For owner decisions, the next question is whether cash available after that replacement plan still covers working capital and scheduled debt.

Use a maintenance calendar, not just an annual budget

Turn the asset estimate into a rolling 24-month calendar. List the expected replacement or overhaul month, quotation date, lead time, down-payment terms, shutdown days, and whether the work can be staged. Add a confidence flag: contracted, quoted, engineering estimate or placeholder. A KSh 600,000 replacement due in three months is not equivalent to KSh 600,000 evenly spread over a year if suppliers require a deposit now. The timing belongs in the 13-week and annual cash forecasts; the long-run estimate belongs in the owner-earnings bridge.

Revisit this calendar after downtime, a safety incident, a change in production mix, or a new supplier quote. The maintenance manager can propose the work, the finance lead can test funding, and the owner can approve the timing. A written deferral should state the operational risk and the next review date so a temporary cash squeeze does not silently become a permanent underinvestment policy.

Review the estimate after the operating plan changes

A maintenance estimate is attached to a business plan, not frozen at last year's spending. Revisit it after a large customer is added or lost, equipment utilization changes, a product standard changes, a site is expanded, or a supplier discontinues parts. If volume rises materially, some working capital and maintenance needs may rise even when the business has not opened a new location. “Same assets” does not always mean “same cash requirement.”

A practical quarterly review asks the operations lead to explain every material difference between budget and actual: was a repair delayed, did an asset fail, did a replacement price change, was a project partly growth and partly renewal, did a new customer change stock or credit terms? Keep the original estimate beside the revised one. That creates a record of whether management systematically underestimates or overestimates the cost of preserving operations.

A cash estimate should include timing. If a replacement motor costs KSh 1.2 million and the supplier needs 50% on order with a 12-week lead time, the business needs a cash date as well as an annual amount. A reserve can be built over time, but it is still an earmark against future use. Show whether the company has cash, an approved facility or merely an intention to save later. Likewise, a working-capital release cannot be counted twice: once as cash already collected and again as a future source of financing.

A compact decision note for each proposed release

Before reducing stock, delaying a repair, renegotiating supplier timing or extending customer credit, write a short note with six fields: the current problem, the cash amount and date, the operational mechanism, the affected customer or supplier, the downside if the assumption is wrong, and the person who will check the outcome. This makes the cash effect comparable with the service effect.

For example, an owner may plan to cut safety stock by KSh 300,000. The decision note should name the SKUs, lead-time variability, recent stock-outs, substitute availability and expected order fill rate. If three fast-moving parts account for most downtime, reduce the slow tail first and leave the critical items protected. For a customer credit extension, estimate the incremental receivable from sales volume multiplied by the additional collection days, then compare it with contribution margin and the funding cost. Do not treat gross sales as cash benefit.

The evidence on working capital and performance helps frame the question but cannot decide it for a specific firm. Deloof and García-Teruel and Martínez-Solano study firm-level relationships in particular samples, while Afrifa examines cash flow and UK SME performance [3–5]. Differences in industry, financing access, customer power and operating cycle matter. The correct practical conclusion is to measure the company's own cash conversion and service outcomes over time, then test whether a change improves total economics.

Workbook map

The Project 1 workbook has a Maintenance register that annualizes five illustrative replacement costs over their planning lives: KSh600,000 for a compressor over 24 months, KSh2.1m for a van over 36 months, KSh850,000 for a generator over 30 months, KSh480,000 for POS/network equipment over 24 months, and KSh300,000 for fittings over 48 months. The resulting annual reserve is KSh1.655m. Its separate three-month Working Capital example moves from KSh2.0m net working capital in October 2026 to KSh2.2m in November and KSh2.9m in December; the modeled cash cycle lengthens from about 31.0 to 41.0 days. Receivables, Inventory and Payables make due dates, stock age and supplier criticality visible. The Evidence Vectors sheet labels all these rows as fictional, records their example periods/as-of dates and points to the source cells; source and retrieval dates are not applicable because no real ledgers or asset records support them. These are not a low/base/high estimate for a real company; replace them with inspection, invoice, sales and contract evidence before using the cash result.

Technical appendices

The worked calculations and source boundaries below extend this article's examples. They keep assumptions and unresolved inputs visible so readers can reproduce the steps without mistaking an illustration for an issuer or sector estimate.

References

[1] Berkshire Hathaway, “Chairman’s Letter — 1986,” https://www.berkshirehathaway.com/letters/1986.html.

[2] D. Livdan and A. Nezlobin, “Investment, Capital Stock, and Replacement Cost of Assets When Economic Depreciation Is Non-Geometric,” Journal of Financial Economics, vol. 142, no. 3, pp. 1444–1469, 2021, https://doi.org/10.1016/j.jfineco.2021.05.021.

[3] M. Deloof, “Does Working Capital Management Affect Profitability of Belgian Firms?” Journal of Business Finance & Accounting, vol. 30, no. 3–4, pp. 573–588, 2003, https://doi.org/10.1111/1468-5957.00008.

[4] P. J. García-Teruel and P. Martínez-Solano, “Effects of Working Capital Management on SME Profitability,” International Journal of Managerial Finance, vol. 3, no. 2, pp. 164–177, 2007, https://doi.org/10.1108/17439130710738718.

[5] G. A. Afrifa, “Net Working Capital, Cash Flow and Performance of UK SMEs,” Review of Accounting and Finance, vol. 15, no. 1, pp. 21–44, 2016, https://doi.org/10.1108/RAF-02-2015-0031.

Companion resources

The workbook and technical appendices are available with research membership. The video production guide maps this article to a planned episode; a published video is not yet available.