17.3 Rolling Forecasts and Variance Analysis

Rolling Forecast Versus Annual Budget Lock

Apex locked a calendar-2026 annual budget in November 2025: 2,000,000 units × $40.00 = $80,000,000 of revenue and the operating/financial package from section 17.1. That lock does not move when March actuals come in. If you rewrite the budget every time you have a new actual, you no longer have a control standard. You have a forecast wearing a budget label.

A rolling forecast always looks a fixed horizon ahead — on the FMVA and in most FP&A shops, 12 months. On 31 March 2026 the rolling forecast is April 2026 through March 2027. January–March 2026 drop off as completed actuals. January–March 2027 are added as new forecast months. The calendar-2026 budget is still January–December 2026. Both columns can sit on the same dashboard: actual, budget, forecast.

Exam trap: calling a reforecast of the remaining nine months of 2026 a rolling 12-month forecast. That is a stub-year outlook. It is useful for "will we hit the annual budget?" It is not rolling, because the horizon shrinks every month until December is a one-month forecast. Rolling means the horizon does not shrink.

Worked calendar: after April closes, the rolling window becomes May 2026–April 2027. The annual budget is still 2026. Q1 actuals stay in the variance pack; they do not get restated in the lock.

FP&A uses the rolling view to answer "what do we now think the next year of operations looks like?" and the lock to answer "how did we do versus the plan we promised?" Investment-banking models rarely keep an annual lock at all; they keep a live forecast. Do not grade an IB case on budget variances the prompt never asked for.

Variance = Actual − Budget

The identity is:

Variance = Actual − Budget

Then you label it favorable or unfavorable based on whether the gap helps profit.

• Revenue: actual > budget is favorable (positive variance, more sales).
• Cost: actual > budget is unfavorable (positive variance, more spend).

Some FP&A packs flip the subtraction for costs so that a positive number is always "good." That is a presentation choice. On the exam, compute actual − budget first, then apply the label. Do not memorize a sign until you know whether the line is revenue or cost.

Apex FY2026 full-year actuals versus the lock (section 17.2's driver outcome, now realized):

Actual COGS = 2,100,000 × $24.50 = $51,450,000. Actual gross profit = $86,100,000 − $51,450,000 = $34,650,000. The $2,650,000 gross-profit beat is not "everything is fine in the plant." It is a net of a large sales beat and a real cost miss. That is why you split price and volume.

Price and Volume Variance Split

For revenue:

Volume variance = (Actual units − Budget units) × Budget price

Price variance = (Actual price − Budget price) × Actual units

Apex:

• Volume = (2,100,000 − 2,000,000) × $40.00 = $4,000,000 favorable
• Price = ($41.00 − $40.00) × 2,100,000 = $2,100,000 favorable
• Sum = $6,100,000 favorable, which equals actual revenue minus budget revenue.

The cross-term is inside the price variance because price is evaluated at actual units (the extra 100,000 units are also valued at the extra $1). If you evaluate price at budget units you get only $2,000,000 and then need a $100,000 cross-term line to tile to $6,100,000. Either three-line bridge (volume at old price, price at old units, cross-term) or two-line (volume at budget price, price at actual units) is acceptable. Mixing them — volume at actual price and price at actual units — double-counts the $100,000.

For COGS, the same split is a quantity variance at standard rate and a rate variance at actual quantity:

Quantity (volume) variance = (Actual units − Budget units) × Standard unit cost

Rate (price) variance = (Actual unit cost − Standard unit cost) × Actual units

Standard unit cost is the budget $24.00.

• Quantity = (2,100,000 − 2,000,000) × $24.00 = $2,400,000 unfavorable (more units, more COGS dollars)
• Rate = ($24.50 − $24.00) × 2,100,000 = $1,050,000 unfavorable
• Sum = $3,450,000 unfavorable, matching actual COGS − budget COGS.

The flexible budget for COGS at actual volume and standard rate is 2,100,000 × $24.00 = $50,400,000. Actual versus flexible = $51,450,000 − $50,400,000 = $1,050,000 unfavorable — the same rate variance. The $2,400,000 quantity piece is the volume you expected once sales beat 100,000 units. Grading the plant on the full $3,450,000 static miss would punish it for the sales team's volume win.

Gross-profit bridge (should tile to +$2,650,000):

Contribution thinking: extra 100,000 units at budget prices and standard cost add 100,000 × ($40 − $24) = $1,600,000 of standard contribution. Then price adds $2,100,000 and the rate miss takes $1,050,000: 1,600,000 + 2,100,000 − 1,050,000 = $2,650,000. Same number, two presentations.

Exam trap: reporting the $6.1 million revenue beat and stopping. Half the value of the beat was supposed to be spent as extra COGS at $24; some of it was spent at $24.50. The dashboard has to show both bars.

Management Action

A variance pack that does not change a decision is a filing exercise. Tie each material bar to an owner and a next step.

If the rate miss is a one-time expedite fee, add it back when you build the rolling forecast (same add-back logic as section 17.2) but leave it in the budget variance pack — the lock does not get a free pass. If the rate miss is the new $24.50 run-rate, the rolling forecast should use $24.50, not $24.00, or the next 12 months will reprint the miss as a surprise.

Do not "help" the variance pack by restating the budget to actual volume and calling the gap zero. That restatement is the flexible budget, and it is a second column, not a replacement lock.

Connect to Dashboards

Chapter 18 (Presentation and Visuals) is where these numbers get a chart. The FP&A dashboard that belongs with this chapter is not a rainbow pie of SG&A. It is:

1. A three-column actual / budget / rolling-forecast strip for revenue, gross profit, and cash.
2. A waterfall of the $2.65 million gross-profit variance (the four bars above).
3. A volume versus price split so a price cut that bought volume is visible.
4. A landing chart: year-to-date actual plus remaining-year forecast versus the annual lock.

Excel Data Visualization and Dashboards is core. The variance math in this section is the data those dashboards plot. A dashboard that shows only actual versus last year, with no budget column and no rolling column, is a trend chart, not an FP&A control chart.

Cash-Flow Forecasting, Conceptually

Profit is not cash. Apex's $2,650,000 extra gross profit does not drop $2,650,000 into the bank. Extra volume raises AR and inventory. Extra Q4 seasonality (35% of the year) raises the year-end working-capital stock even if days are unchanged. Capex on the financial budget still spends cash that never hits SG&A.

Two honest cash views exist:

1. Indirect (three-statement): net income + D&A − ΔNWC − capex ± financing. This is already core 3-statement modeling. Annual cash on the statement of cash flows is this view.
2. Direct (cash budget): collections from customers minus payroll, vendor payments, tax, capex, and debt service. This is the treasurer's receipts-and-disbursements calendar.

A company can budget operating profit of $12 million and still need a revolver in October because Q4 inventory builds before Q4 collections arrive. That concept — cash budget ≠ operating budget — is fair game as Budgeting & Forecasting knowledge.

What is not on the final is the Monthly Cash Flow Modeling elective. After the 2026 refresh that course is elective, and elective content is not tested. Do not spend exam hours building a 13-week cash worksheet, a monthly revolver corkscrew from that elective, or a weekly receipts calendar as if it were core. If a case gives annual statements, use the annual indirect cash flow. If a case mentions that Q4 is 35% of sales, you may reason that year-end AR and the intra-year cash trough are worse than a flat-year model — without opening an elective monthly file.

Exam traps on rolling forecasts and variances

• Rewriting the annual lock when March actuals arrive, then claiming there is no variance.
• Calling a remaining-year stub a rolling 12-month forecast.
• Taking the $6.1 million revenue variance as the whole story and skipping the $1.05 million rate miss.
• Computing price variance on budget units and volume variance on actual price, which double-counts the $100,000 cross-term.
• Treating Monthly Cash Flow Modeling as a tested core course because cash still appears in FP&A discussions. Electives are not tested; the conceptual difference between profit and cash is enough.
• Building a variance waterfall and then never putting it on the Chapter 18 dashboard, where the examiner actually sees it.