Capex is money you put into a building to create or extend useful life, record on the balance sheet, and recover over time through depreciation or amortization. Tenant improvements (TI) are leasing-driven Capex: cash you spend to deliver space to a tenant, usually priced per square foot and paid on a construction draw schedule.
Those two lines decide whether a value-add deal becomes a steady compounding machine or an expensive lesson in timing. They move real cash, on real schedules, under real legal definitions. The model either respects that, or it becomes a sales document.
Capex vs. TI: what counts (and why it matters)
Capex and TI feel like simple inputs, but the definitions drive funding mechanics, lender reserves, and investor expectations. When you classify costs correctly and tie them to timing, you reduce execution risk and make returns more believable.
What counts as Capex and TI and what does not
In underwriting, Capex typically includes base building improvements, major system replacements, landlord work tied to repositioning, and unit turns above routine make-ready. TI is a subset: build-out dollars tied to a specific lease event, paid as allowances or reimbursements.
Leasing commissions (LC) often travel with TI as “tenant inducements,” but they behave differently. TI buys physical work or reimburses it. LC pays brokers for transactions. Both reduce distributable cash and usually increase the size of the financing stack, which shows up as interest carry and covenant pressure.
Capex and TI are not operating expenses like payroll, utilities, repairs, or recurring maintenance. And they are not “one-time.” A value-add plan often turns leasing friction into recurring capital needs through rollover, re-tenanting, and periodic repositioning. If your pro forma treats them as isolated events, the plan will look prettier than the bank account.
One boundary matters more than people admit: your classifications need to match the lender and appraisal regime you’re actually using. Credit agreements often define “Capital Expenditures,” “Tenant Improvements,” and “Leasing Costs,” then tie reserves, future funding, and covenant add-backs to those terms. If your model uses sponsor-friendly labels that can’t be reconciled to the loan, you raise execution risk for no benefit.
How Capex and TI swing value-add outcomes
Capex and TI drive three variables that dominate value-add results: the time it takes to stabilize NOI, the peak cash burn (and therefore how much capital you must control), and the exit narrative that survives buyer diligence. Two deals can reach the same stabilized NOI and still produce very different IRRs because the path matters: when cash leaves, when cash returns, and how certain the schedule is.
These lines are also the easiest to optimize on paper and the hardest to compress in the field. Code, delivery standards, and lender requirements impose minimum scopes. When teams cut budgets to hit an IRR target, they often buy schedule risk and rework risk. The model improves, but the project does not.
Institutional buyers and lenders press on Capex and TI because these are common sources of drift between underwriting and actuals. The job isn’t to sandbag every assumption. The job is to separate what you control from what you merely trigger, then show how contracts and governance keep the plan inside guardrails.
A practical taxonomy that improves forecasting
A decision-useful model groups Capex by causal driver, not by construction trade. This structure makes downside behavior easier to predict and helps you explain variance when actuals come in.
- Base building Capex: Roof, facade, elevators, chillers, boilers, parking structures, life safety, lobbies, and amenities that you can plan and sequence.
- Deferred maintenance catch-up: “Optional” items that surface after close through tenant complaints, municipal inspections, and insurer engineering reports.
- Turnover and ready-for-occupancy: Make-ready, unit turns, and refresh work tied to leasing velocity, either frequent and granular (multifamily) or lumpy (office/industrial).
- TI and LC package: TI, free rent, and LC as market-negotiated concessions driven by tenant credit, term, vacancy, and competing supply.
- Compliance and insurer Capex: Fire/life safety, ADA, energy rules, flood mitigation, seismic retrofits, and insurer mandates that can be binary and schedule-critical.
Keep these buckets in the model even if reporting collapses them into “Capex.” They have different levers, different downside shapes, and different financing implications, especially when you size the capital stack.
TI economics: treat it like a concession with construction risk
Underwriting often reduces TI to a single “$/SF” line. Economically, TI is landlord-funded tenant value exchanged for a rent stream over a term, with downtime, build-out, and tenant performance layered on top.
When TI is material, model three components separately:
- Signing commitment: The maximum exposure per the lease before change orders and delays, and a number you can track and govern.
- Cash draw profile: Milestone-based draws with lien releases and retainage, where timing drives interest carry and reserve sizing.
- Residual value: The portion of improvements that is reusable at rollover, which affects exit diligence and future re-tenanting cost.
TI also interacts with free rent. Both are concessions, but they hit cash differently: TI is cash out, while free rent is cash not received. If you bundle them, you can miss the cash burn shape that drives financing risk. Separate them, and negotiations become stress-testable.
Don’t force a shortcut from Capex to NOI
Value-add stories often imply “spend Capex, get NOI.” Real underwriting needs the middle steps: leasing friction, operating cost impacts, and tenant demand. Capex is often necessary, but it isn’t automatically productive.
A clean structure is to show three channels:
- Revenue uplift: Higher face rent, higher effective rent via lower concessions, improved occupancy, ancillary income, and premiums for renovated units.
- Cost reduction: Energy savings, fewer repairs, lower turnover, lower insurance risk profile, and more controllable systems.
- Risk reduction: Lower probability of system failure, fewer compliance events, and better lender and insurer acceptance.
Some Capex keeps the building functioning and won’t lift NOI. It can still protect value, reduce future cash drains, and support refinance terms. Treat it honestly as value creation, value preservation, or both.
Model the flow of funds so timing can’t hide
Many Capex models miss not because totals are wrong, but because timing is wrong. Capital partners care about peak funding need, reserve mechanics, and covenant compliance during the messy middle. If you need a clean scaffold for a property model, build it off a real real estate cash flow model rather than a static budget table.
A flow-of-funds frame should show:
- Sources: Equity contributions, debt at closing, future funding, tenant reimbursements, seller credits, and legally usable deposits.
- Uses: Acquisition costs, Capex, TI, LC, interest carry, operating deficits, reserves, fees, and taxes.
- Cash controls: Who approves draws, what documents are required, and how retainage and lien waivers work.
Model Capex as draw schedules tied to milestones, not as straight-line annual numbers. For TI, link draws to executed leases and construction timelines, with a realistic lag from lease signing to cash out.
Timing mechanics that separate good deals from tight deals
Three lags drive most of the gap between underwritten returns and realized returns. When you model these lags explicitly, you see the true liquidity requirement and avoid surprise capital calls.
- Lease-to-cash lag: Lease execution is not rent commencement, so model downtime, build-out, and free rent so “leased occupancy” does not masquerade as “paying occupancy.”
- Retainage timing: Retainage reduces near-term cash out but can create a completion spike that breaks liquidity if ignored.
- Change-order behavior: Contingency is a decision buffer, so show base scope, contingency, and a governance rule for releasing it.
When you run sensitivities, don’t only stress total Capex. Stress the schedule using a simple scenario framework, like the approach used in stress test cases for deals. A six-month delay can hurt more than a moderate overrun because it compounds interest carry and delays NOI.
Model lease events, not TI averages
Averages are convenient and often wrong at the tails, and the tails decide the outcome. Value-add performance often hinges on the marginal leases needed to clear the market, not the median tenant you pictured in underwriting.
A better approach is simple:
- Lease cohorts: Build cohorts by space type or unit type and target tenant segment.
- Concessions by cohort: Underwrite TI and LC per cohort, tied to lease term and tenant credit.
- Renewal logic: Assign renewal versus new-lease probabilities because renewals often need less TI and LC, but still move with the market.
If you can’t build cohorts, at least split first-generation lease-up from steady-state rollover. First-gen usually carries higher concessions and longer downtime, especially after repositioning.
A quick TI sanity check: annualize the concession
When comparing TI packages across different lease terms, convert upfront TI into an annualized burden for intuition. Think of it as an economic amortization plus the landlord’s cost of capital, not an accounting treatment.
Take TI per square foot, spread it over the lease term, and compare the annual equivalent to rent per square foot net of recoveries. If the annualized concession looks meaningfully worse than comps, the “headline rent” may be doing most of its work in marketing, not in value.
Align the model to lender underwriting and the credit agreement
Transitional lenders increasingly require hard reserves for TI, LC, and Capex. Many facilities use holdbacks or future funding, meaning the lender funds eligible costs only after draw conditions, lien releases, and leasing tests are met. If you want a deeper comparison of risk and control across capital providers, see real estate private equity vs. direct lending.
Your model needs to reflect that reality:
- Eligibility mapping: Separate eligible from ineligible uses because exclusions pull equity forward in time.
- Controlled accounts: Model lender-controlled reserves (TI/LC, Capex, debt service) that reduce sponsor discretion.
- Leasing tests: Reflect lease-based advance conditions, such as minimum term, rent, and tenant credit criteria.
If the loan includes an interest reserve, model it as a real cash account with replenishment rules. Treating it as a plug invites a liquidity surprise.
Documentation and governance: where control actually lives
Capex and TI are controlled in documents, not in committee decks. Equity agreements set approval rights and budget thresholds. Debt documents set draw mechanics and reserve rules. Contracts and leases decide who pays for what, and when.
- JV agreement: Approval thresholds, capital calls, defaults, and reallocation rights between Capex buckets.
- Loan documents: Definitions, permitted uses, draw conditions, cash management, and overrun triggers.
- Construction contract: Scope, allowances, contingency ownership, schedule, and change-order governance.
- Lease forms: Allowance structure, delivery condition, reimbursement timing, and delay remedies.
- Management agreements: Concession authority limits, broker terms, and sponsor approval rights.
Execution order matters. Align loan terms and JV governance before signing leases that commit you to TI packages the facility won’t fund.
Fresh angle: use a “peak cash day” metric to spot fragile deals
Most models report “total Capex” and “total TI,” but the metric that often predicts pain is the single worst liquidity moment. A simple addition to your underwriting is to compute the day (or month) when cumulative cash out is highest before stabilization, then name it and manage to it.
In practice, the “peak cash day” usually occurs when TI draws hit, interest carry is still high, and rent commencement has not caught up. Once you can point to that moment, you can negotiate structure instead of arguing about budget decimals:
- Right-size reserves: Fund reserves to the peak, not to annual averages.
- Sequence work: Move sponsor-controlled Capex earlier or later to flatten the cash curve.
- Match funding: Align holdbacks and future funding to the same timeline your leases create.
This is also a clean way to communicate risk to an investment committee because it turns a spreadsheet into an operational story: “Here is when we are most exposed, and here is what protects us.”
Closeout: keep records clean and defensible
When the deal winds down, closeout discipline matters. Archive the index, version history, Q&A threads, user list, and full audit logs, then hash the archive so you can prove it wasn’t altered.
Next, set retention terms that match the JV agreement, loan docs, tax requirements, and audit timelines. After retention is satisfied, require vendor deletion and a destruction certificate. If there’s a legal hold, it overrides deletion every time.
Key Takeaway
Capex and TI are not just budget lines; they are timing, legal definitions, and cash controls that determine whether a value-add plan survives the messy middle. When you model lease events, draw schedules, and lender mechanics, you stop “optimizing” returns and start underwriting execution.