Financial Modeling - 2nd Edition

1. 1Build the full pro forma first: income statement, balance sheet, and the plug. In Farmers Bagels, no dividends and no new equity make debt the plug. FCF is read off the finished statements, not guessed separately. Check that your FCF matches the model's cash/debt movements.
2. 2Define FCF as the cash the firm would produce with no financial assets at all: no debt, no cash, no marketable securities. Every line you add or strip must serve that definition.
3. 3Run the seven-line calculation for each projected year: (1) profit after taxes; (2) add back depreciation; (3) add back interest paid, net of taxes; (4) subtract interest earned, net of taxes; (5) subtract the increase in current assets other than cash and cash equivalents; (6) add the increase in current liabilities; (7) subtract new fixed assets purchased, meaning the change in fixed assets at cost.
4. 4Take the interest tax effect from the same tax rate the model uses, so lines 3 and 4 strip financing out of PAT cleanly. Depreciation is a non-cash add-back, but the capex line (7) is measured at cost, not net of depreciation.
5. 5If profits are negative, rebuild the P&L as if the firm had no interest payments, including tax-loss carryforwards. Add the loss carryforward back as a non-cash charge. If you work from the actual P&L instead, net out the interest element embedded in the carryforward.
6. 6Compute WACC separately (the passage's example uses 20 percent) and discount FCFs at it. Use half-year discounting, treating each year's flow as arriving mid-year, because the NPV convention of year-end flows is less realistic.
7. 7Set the terminal value from year-5 FCF as a stand-in for all later cash flows. Gordon version: flows grow after year 5 at the sales growth rate (10 percent in the example), so TV = FCF5 × (1+g) / (WACC − g). Alternatives: year-5 book debt plus equity, a market/book multiple, P/E × year-5 profit plus book debt, or an EBITDA multiple × year-5 EBITDA.
8. 8Add year-0 cash and marketable securities to the PV of FCFs, or equivalently treat them as negative debt when you derive equity value. Then run sensitivity analysis on the base case (growth, WACC, terminal method).

Farmers Bagels, projected five years. Its first two years produced healthy profits but negative cash, because growing sales needed more net working capital and new fixed assets. The FCF build starts from profit after taxes, adds back depreciation (10 percent of sales) and after-tax interest paid, and subtracts after-tax interest earned. It then subtracts the increase in non-cash current assets, adds the increase in current liabilities (payables rising 1 point a year toward 20 percent, other current liabilities at 1 percent of sales), and subtracts new fixed assets. Only in 1999 is Farmers projected to turn FCF positive. The passage's table figures are not reproduced here, so the valuation mechanics are shown per $1 of year-5 FCF. With WACC = 20 percent and post-year-5 growth g = 10 percent, TV = FCF5 × 1.10 / (0.20 − 0.10) = 11 × FCF5. Discounted five years at 20 percent (1/1.2^5 ≈ 0.402), that terminal value is worth about 4.4 × FCF5 today. Terminal value therefore carries a large share of enterprise value when early FCFs are negative or small. Enterprise value is the sum of discounted FCFs plus discounted TV, plus year-0 cash and marketable securities, and half-year discounting raises it relative to year-end discounting. This dependence is why the sensitivity analysis matters.

• If profit is positive but FCF is negative, check growth: rapid sales growth consumes cash in working capital and fixed assets. Treat it as expected, as with Farmers' first two years, not as an error.
• If the model uses a debt plug and debt turns negative, don't read it as FCF error. It signals accumulated cash. Apply a minimum cash-to-sales ratio and set debt to 0 when the balance sheet would otherwise need negative debt, since debt at 12 percent costs more than the 6 percent earned on cash.
• If you use Gordon terminal value, then WACC must exceed g. Also confirm that year-5 FCF reflects a steady state; growth-depressed FCF understates TV.
• If year-0 cash is surplus not needed to run operations, add it to PV of FCFs (or treat it as negative debt). If cash is operating cash, do not add it.
• If the Gordon TV looks aggressive, cross-check it against another method: book debt plus equity, a market/book multiple, a P/E-based value, or an EBITDA multiple.

• Leaving interest in the cash flow: starting from PAT without adding back after-tax interest paid and subtracting after-tax interest earned. The tell: FCF changes when you change the debt/equity mix.
• Including cash in the working-capital change: the increase in current assets must exclude cash and cash equivalents. The tell: FCF moves one-for-one with the cash plug.
• Subtracting depreciation-netted capex: using the change in net fixed assets instead of new fixed assets at cost. The tell: capex that looks too low and FCF that looks too high.
• Mishandling losses: ignoring the carryforward, or leaving interest embedded in it. The tell: FCF for loss years doesn't reconcile with the no-interest P&L.
• Letting the terminal value dominate unchecked: a TV with g close to WACC, or taken from a year-5 FCF still distorted by growth. The tell: enterprise value swings hugely in sensitivity runs.

• ▸Profit after taxes (year t): take from the pro forma P&L.
• ▸Depreciation add-back: non-cash charge, e.g., 10 percent of sales in the Farmers case.
• ▸After-tax interest paid (add back): interest expense × (1 − tax rate).
• ▸After-tax interest earned (subtract): interest income × (1 − tax rate).
• ▸Increase in current assets excluding cash (subtract): receivables, inventory, prepaids.
• ▸Increase in current liabilities (add): payables, accrued expenses, taxes payable, other.
• ▸New fixed assets purchased (subtract): change in fixed assets at cost.
• ▸FCF = sum of the above; then discount at WACC with half-year timing (t − 0.5).
• ▸Terminal value: FCF5 × (1+g)/(WACC − g), or a cross-check method; add discounted TV and year-0 cash to get enterprise value.

1. 1Build the free cash flow (FCF) line for years 1–5 from the pro forma income statement and balance sheet. Derive it from sales growth, cost of goods sold, depreciation, taxes, and the changes in current assets, current liabilities and fixed assets. The model's inputs are sales growth, current assets at 15% of sales, current liabilities at 8%, net fixed assets at 77%, and depreciation at 10% of average fixed assets at cost.
2. 2Fix the discount rate as the weighted average cost of capital (WACC), computed as in Chapter 2. The source example uses 20%. WACC is the opportunity cost of the firm's risky cash flows, so use the same rate for every year's FCF.
3. 3Discount each projected FCF at the WACC. Decide the timing convention up front. The plain NPV formula treats flows as arriving at year-end. Half-year discounting treats them as arriving mid-year, which the source calls more logical because cash flows arrive smoothly through the year.
4. 4Compute the year-5 terminal value as a proxy for the present value of all FCFs from year 6 onward. Use the Gordon-model form: cash flows grow after year 5 at a stated rate (the example uses the 10% sales growth rate), so TV = FCF6 / (WACC − g), or equivalently FCF5 × (1 + g) / (WACC − g). Discount the TV back to today at the WACC.
5. 5Sum the present value of the FCFs and the present value of the terminal value. Then add the year-0 cash and marketable securities, but only if they are surplus balances not needed to produce future FCFs and could be paid out without hurting future performance.
6. 6Cross-check the terminal value against at least one alternative from the source: year-5 book debt plus equity, an enterprise market/book multiple times that book value, P/E times year-5 profit plus year-5 book debt, or an EBITDA multiple times year-5 EBITDA. If the Gordon TV and the multiple-based TV diverge widely, resolve the gap before reporting a value.
7. 7Move from enterprise value to equity value consistently. Either subtract debt from the enterprise value that includes the cash add-back, or treat the cash as negative debt, which the source calls wholly equivalent. Do not do both.
8. 8Run sensitivity tables on sales growth and on WACC versus growth. Guard the terminal value formula so it only evaluates when WACC exceeds the growth rate. The source builds a protected data-table cell for this.

The source gives one fully numeric discounting case, and it is the building block of enterprise value. An investment pays $100 at the end of each of five years, and the opportunity cost is 10%. Discounting each payment gives a present value of $379.08. A $400 price therefore overpays by $20.92, so the NPV is negative. Enterprise value applies the same logic to FCFs, with a WACC of 20% and a terminal value added. The source terminal value assumes cash flows grow at 10% after year 5, so TV = FCF5 × 1.10 / (0.20 − 0.10) = 11 × FCF5. The terminal value is eleven times year-5 FCF, so it is likely to dominate the total. Sensitivity matters here: if g rose toward 20%, the denominator would approach zero and the TV would blow up, which is why the source warns that WACC must exceed g. Moving to mid-year timing multiplies each discounted flow by (1.20)^0.5, about 1.095. Check whether your model also shifts the TV. The source text does not say. Then add year-0 cash if it is surplus. The source's FCF figures are not reproduced in the supplied passages, so apply this structure to your own FCF line.

• If WACC is less than or equal to the post-horizon growth rate, then do not use the Gordon terminal value. The formula breaks down, so guard the cell or use a multiple-based TV.
• If the year-0 cash is not needed to run the business and could be paid out without hurting future performance, then add it to the PV of FCFs. If you instead net it against debt, use that treatment once only.
• If cash flows arrive evenly through the year, then use half-year discounting rather than end-of-year NPV. Say which convention your headline number uses.
• If book value is expected to track market value in year 5, then book debt plus equity is a defensible terminal value. Otherwise use an enterprise market/book multiple, a P/E-based measure, or an EBITDA multiple.
• If a single input such as sales growth or WACC moves the value sharply, then show the data-table range, not a point estimate.

• Applying a terminal growth rate at or above WACC. The tell is a negative or enormous terminal value, or sensitivity tables that look erratic at high growth.
• Counting cash twice, once added to the PV of FCFs and again as negative debt when moving to equity. The tell is equity value above enterprise value by more than the debt difference explains.
• Mixing discounting conventions, such as year-end FCFs with a mid-year terminal value, or the reverse. The tell is that your result differs from a colleague's using the same FCFs and WACC.
• Letting the terminal value dominate without a cross-check. The tell is that the PV of the terminal value is most of the enterprise value and no multiple-based TV was run alongside it.
• Building FCFs from a model that does not balance or recalculate, for example with iteration turned off on the circular cash plug. The tell is an Excel circular-reference warning or cash that differs from the cash flow statement.

• ▸WACC: enter the cost of capital from your Chapter 2-style calculation; the source example uses 20%.
• ▸FCF years 1–5: list each projected free cash flow with its source line.
• ▸Timing convention: state end-of-year or mid-year discounting, and whether the TV follows the same convention.
• ▸Post-horizon growth rate g: enter the rate applied after year 5 (the example uses 10%) and confirm WACC > g.
• ▸Terminal value (Gordon): compute FCF5 × (1 + g) / (WACC − g) and discount it to today.
• ▸Terminal value cross-check: compute one alternative (book debt plus equity, market/book multiple, P/E, or EBITDA multiple) and note the gap.
• ▸Year-0 cash and marketable securities: enter the balance and tick whether it is surplus (not needed to produce FCFs).
• ▸Enterprise value: sum of PV of FCFs, PV of TV, and any surplus cash.
• ▸Equity value bridge: subtract debt, or treat cash as negative debt; mark which method you used.

1. 1Fix the valuation date and write down which cash flows you are discounting. Free cash flows to the firm are discounted at WACC and give enterprise value. Equity cash flows (dividends) are discounted at the cost of equity and give equity value directly. Never cross the two rates and flows.
2. 2Build the cost of capital first. Estimate the cost of equity with the Gordon model (r_E = D1/P0 + g) or the CAPM. Estimate the cost of debt with one of the chapter's three debt models. Combine them into WACC. The discount rate is a risk-adjusted rate, so every input carries a judgment.
3. 3Choose g as the anticipated growth rate of dividends, not the historical one. In the Abbott case the 10-year compound rate of 14.87% gives a 16.28% cost of equity, and the five-year rate of 12.03% gives 13.40%. Write down why you picked one. Use a pro forma (sales growth, capital needs, payout ratio) to test it.
4. 4Discount the projected free cash flows at WACC. This gives enterprise value. If growth is high early and slower later, split the stream (explicit years, then a Gordon-style terminal value discounted back over those years). Check that the long-run g is below the discount rate.
5. 5Pull net debt from the balance sheet at the valuation date: debt minus cash and marketable securities. In the pro forma model, debt is an input (constant over the horizon) and cash is the balance-sheet plug, so read both at the same date.
6. 6Compute equity value = enterprise value − net debt. Divide by shares to get a per-share figure.
7. 7Cross-check from the equity side. Compare the per-share result with a Gordon dividend value and with the observed price (Abbott: $49 at end-1998). A large gap means revisiting g, r_E or the cash flow forecast, not forcing the numbers to agree.
8. 8Re-run with the alternative growth rate or discount rate and report equity value as a range tied to the stated assumptions.

The passages give no complete enterprise-value-to-equity bridge, so the real numbers below come from the equity-side Gordon example, which you use as the cross-check for your bridge. A firm pays a current dividend of $8 per share. Dividends grow 35% a year for five years, then 8% a year indefinitely. The cost of equity is 18%. The five high-growth dividends are worth about $40 today. The years 6–∞ stream is a Gordon value (using the 8% long-term growth rate) discounted back five years. Together the two pieces give a share value of 230.33. By subtraction, roughly $190 of that, over 80%, comes from the terminal stream. That tells you where the assumptions matter: the 8% long-term g and the 18% rate. The Gordon formula needs g below r_E. If g exceeded r_E, the formula would give an infinite value, which signals either that long-term growth is overstated or that the discount rate is too low. For the bridge itself, take your DCF enterprise value, subtract debt, add cash and marketable securities, and compare the per-share result with this kind of dividend-based value. The Abbott case shows the same sensitivity: a growth choice of 14.87% versus 12.03% moves the cost of equity from 16.28% to 13.40%.

• If you discount firm free cash flows, use WACC and expect enterprise value. Subtract net debt to reach equity value. If you discount dividends or equity cash flows, use the cost of equity and treat the result as equity value already. Do not subtract debt again.
• If the long-term growth rate g is at or above the discount rate, then stop. The formula is invalid (infinite value). Lower the terminal g or revisit the discount rate before reading any output.
• If a historical growth rate and your expected growth differ, then use the expected rate and document why. The historical rate is only evidence. Abbott's 14.87% versus 12.03% moves the cost of equity by almost 3 points.
• If terminal value is most of the total (over 80% in the $230.33 example), then treat equity value as a range and test the long-term g and discount rate first.
• If the DCF equity value and the market price or Gordon value diverge sharply, then audit the cash flow forecast, WACC, and the net debt inputs before concluding the stock is mispriced.

• Discounting free cash flows to the firm at the cost of equity (or dividends at WACC). The tell: a per-share value that is far off the observed price with no explanation, and a bridge that double-counts or omits debt.
• Using historical dividend growth as g without judgment. The tell: your cost of equity swings by several points depending on which window you choose (16.28% versus 13.40% for Abbott).
• Letting long-run g meet or exceed the discount rate. The tell: terminal value that dominates everything, or a formula result that explodes or turns negative.
• Taking net debt from the wrong date. In the pro forma, cash is a plug that changes each year while debt may be held constant. The tell: net debt that does not match the balance sheet at the valuation date.
• Truncating the dividend stream and calling it exact. The 67-year NPV is only an approximation of an infinite series. The tell: a spreadsheet value that differs slightly from the closed-form Gordon result.

• ▸Valuation date: the single date for all inputs.
• ▸Cost of equity (r_E) and method: Gordon or CAPM, with the growth rate or inputs used.
• ▸Anticipated dividend growth g and rationale: one line on why this rate, not the historical one.
• ▸Cost of debt and WACC: the debt model used and the resulting weighted rate.
• ▸Enterprise value: PV of free cash flows to the firm at WACC, split into explicit years and terminal value.
• ▸Terminal value share of enterprise value: terminal PV divided by total, to flag assumption dependence.
• ▸Debt at valuation date: from the balance sheet.
• ▸Cash and marketable securities at valuation date: from the balance sheet.
• ▸Equity value = enterprise value − (debt − cash), plus per-share value.
• ▸Cross-check: Gordon or market price, the gap, and the reason for any gap.

1. 1Start with what the rate is for: in a sales-driven model, one input (Sales_n = Initial sales × (1 + g)^n) sets current assets, current liabilities, net fixed assets and, through them, the cash plug and the dividends. Treat g as the model's master assumption, not a line item.
2. 2Write g as a forward-looking judgment, not a history lookup. Benninga's rule for the Gordon growth rate applies here too: the correct rate is 'anticipated', not 'historical'. Compute the historical rates (for example, the 10-year and 5-year compound rates) and use them as bounds. Then state in one sentence why the future should look like one of them, or like neither.
3. 3Lay out the year-0 base and the sales-linked ratios before projecting: sales of 1,000; current assets 15% of end-of-year sales; current liabilities 8%; net fixed assets 77%; depreciation 10% of average assets. Put g in one parameter cell and anchor every reference with dollar signs, so the formula copies cleanly through years 2–5.
4. 4Keep policy items separate from sales-driven items. Assets and current liabilities follow sales. Debt and equity are policy decisions. Define the plug (cash and marketable securities = total liabilities − current assets − net fixed assets) so the balance sheet balances by construction, and read the plug's sign as the financing consequence of g.
5. 5Check the plug after setting g. A growing negative cash balance means the growth you assumed needs outside financing that the model forbids (no new debt, no new stock). Either revise g or change the financing policy deliberately. Do not let the plug hide the problem.
6. 6Choose a terminal growth rate separately from the explicit-period rate. After year 5, the terminal value is a Gordon-type value of cash flows growing at a perpetual rate. Require that rate to be below the discount rate (WACC). If you reuse the explicit-period rate, confirm it can last forever.
7. 7If the explicit-period rate is above the discount rate, split the horizon. Use a high rate for a defined number of years, then a lower long-run rate, and discount the second stage back to today (the 35%-for-five-years-then-8% structure).
8. 8Run a sensitivity pass. Re-value the firm at the alternative growth rates (the 10-year vs. 5-year Abbott rates show how far one choice moves the output) and report the range, not a single point.

Benninga's base case: year-0 sales are 1,000 and sales are expected to grow 10% a year. Sales_n = 1,000 × 1.10^n, so years 1–5 are 1,100; 1,210; 1,331; 1,464.10; 1,610.51. Applying the stated ratios to year 1: current assets = 15% × 1,100 = 165; current liabilities = 8% × 1,100 = 88; net fixed assets = 77% × 1,100 = 847. The cash plug then absorbs whatever imbalance remains, since debt and stock are held constant. The same 10% is then used as the post-year-5 growth of free cash flow in a Gordon-style terminal value, discounted at a 20% WACC; that gap of 10 points between WACC and g drives the size of the terminal value. A second illustration of why the rate is a judgment: Abbott Labs' dividend growth was 14.87% over ten years and 12.03% over five, which gives a cost of equity of 16.28% vs. 13.40% on a $49 share price. The choice of growth rate alone moves the answer by 2.88 points. Benninga says the right rate depends on what you expect the company to pay in the future, and that pro forma sales growth and capital needs can help you predict it.

• If the historical growth rates differ materially by window (14.87% over ten years vs. 12.03% over five), then choose the window you can defend as representative of the future and state why. If you cannot defend either, treat history as irrelevant and build the rate from sales and capital-need projections.
• If the long-run growth rate is at or above the discount rate (g ≥ r), then the Gordon formula breaks (negative or infinite value). Either the long-run rate is too high or the discount rate is too low, so fix one. Never read the negative output as a real value.
• If the firm is growing very fast now, then use two stages (high growth for a set period, then a lower perpetual rate, as in 35% for five years then 8%) rather than one blended rate.
• If the cash plug turns negative or balloons as g rises, then the growth assumption is outrunning your financing policy. Revisit g or the financing assumption, and say which one you changed.
• If the terminal growth rate is set equal to the explicit-period sales growth, then confirm that the rate is sustainable indefinitely and well below WACC before accepting the terminal value.

• Extrapolating the historical rate mechanically. The tell: the growth input equals a trailing average and no sentence explains why the future resembles the past.
• Using the same high growth rate forever. The tell: g is close to or above the discount rate and the terminal value is huge, negative, or unstable when you change WACC by a point.
• Letting the plug hide a financing gap. The tell: the balance sheet balances, but cash is falling or negative while debt and equity are frozen.
• Breaking the model's anchoring. The tell: year-2+ values drift or error because the growth-rate and ratio cell references lack dollar signs and shifted when copied.
• Reporting a single-point valuation. The tell: no sensitivity to g, even though a choice between two defensible rates (16.28% vs. 13.40%) moves the result materially.

• ▸Year-0 sales: the current-year sales base the projection grows from.
• ▸Historical growth rates: list each window (e.g., 10-year and 5-year compound rates) with the data behind it.
• ▸Chosen explicit-period g: the single rate (or stage rates) you will use for years 1–5.
• ▸Rationale in one sentence: why the future should look like this rate and not the alternatives.
• ▸Sales-linked ratios: current assets %, current liabilities %, net fixed assets % of sales, depreciation rate.
• ▸Financing policy: debt and equity treatment, and what the plug (cash and marketable securities) absorbs.
• ▸Plug check: sign and trend of the plug over years 1–5, and whether growth needs outside financing.
• ▸Terminal growth rate and discount rate (WACC): confirm terminal g is below WACC and state the gap.
• ▸Sensitivity range: valuation at the alternative growth rates, with the spread reported.

1. 1Pull three years of statements and write down the ratio each line has to sales. The passages show the pattern in Abbott Labs' 1996–1998 cash flow statement. Use the same three-year window for every line so the history is comparable.
2. 2Build the ratios from the items that actually move cash with sales. On the income side, use cost of goods sold and SG&A. On the balance sheet, use current assets (receivables, inventories, prepaid expenses), sales-related current liabilities (payables, taxes payable) and fixed assets. Keep each ratio as a separate, labeled input cell, not a number buried in a formula.
3. 3Separate sales-driven items from financing items. An increase in current liabilities counts as free cash flow only when it is directly related to sales, such as accounts payable or taxes payable. Do not drive debt or cash from the sales ratios.
4. 4Convert each ratio into its cash effect. Growth in sales raises current assets (a cash drain, though not a tax expense) and raises sales-related current liabilities (a cash source). Net fixed assets at cost also rise (a use of cash). Check that your ratios produce these signs in the free cash flow calculation.
5. 5Cross-check each assumed ratio against the cash flow statement. Multiply the ratio by projected sales and compare the implied dollar change in receivables, inventories and payables with the historical changes. For Abbott, receivables, inventories, prepaids, payables and taxes payable netted to a cash drain of $461,195K in 1998, $338,453K in 1997 and $239,861K in 1996.
6. 6Test fixed-asset intensity separately from working capital. Compare the implied fixed-asset investment with historical spending on property and equipment. Abbott spent $990,619K, $1,007,296K and $949,028K over 1998, 1997 and 1996, against depreciation and amortization of $784,243K in 1998.
7. 7Run the pro forma (income statement plus balance sheet) with the ratios as the only drivers. Then run a sensitivity analysis around each ratio. The source's point about cost of capital applies here too: the estimate is an art, not a science, so test the range, not a single point.

Take Abbott Labs' 1998 cash flow statement (dollars in thousands). The working-capital lines are trade receivables (143,470), inventories (111,649), prepaid expenses and other assets (239,533), trade accounts payable and other liabilities +178,979, and income taxes payable (145,522). Together they are a net cash drain of $461,195K, against $338,453K in 1997 and $239,861K in 1996. The passages do not give Abbott's sales, so the ratios themselves cannot be computed here. The procedure is to take sales from the income statement and divide each year's working-capital change by that year's change in sales. If the drain ratio is unstable across the three years (1997's prepaid build of $491,769K was far above 1996's $303,766K), you pick the assumption deliberately instead of copying the latest year. Fixed assets work the same way: gross property and equipment purchases of $990,619K in 1998 are set against sales to give the capital-intensity assumption, and depreciation of $784,243K shows how much of that spending only replaces existing assets. Each ratio then drives the pro forma, and the free cash flow it produces feeds the valuation.

• If a current-asset line (receivables, inventories, prepaids) has swung widely across years, as Abbott's prepaid and other assets did, then use a multi-year average or a stated trend, not the latest year.
• If an increase in a liability is tied to sales (payables, taxes payable), then model it as a ratio to sales and count it as a cash source. If it is not tied to sales, keep it out of the sales-driven ratios.
• If fixed assets at cost are growing faster than sales in the history, then do not assume a falling Fixed Assets to Sales ratio without a stated reason, because each increase in fixed assets reduces free cash flow.
• If the valuation is sensitive to a ratio, then run a sensitivity range around it and report the range, because the source treats these estimates as an art that should be stress-tested.
• If the ratio set gives a free cash flow that contradicts the historical cash flow statement in sign or order of magnitude, then revise the ratios before touching the discount rate.

• Treating the working-capital build as an expense. It is a cash drain but is not deducted for tax purposes, so profit after taxes ignores it. The tell: projected free cash flow equals profit after taxes plus depreciation, with no working-capital line.
• Dropping the liability offset. Modeling only the asset side overstates the cash drain. The tell: payables and taxes payable stay flat while sales grow.
• Using a single year's ratio as the forecast. One unusual year (such as Abbott's 1997 prepaid build) gets locked in. The tell: the forecast ratio equals the last historical column to the decimal.
• Confusing depreciation with capital spending. Net fixed assets can look stable while gross investment is large. The tell: capex in the forecast roughly equals depreciation despite a sales growth assumption.
• Presenting one point estimate with no sensitivity. The tell: a single valuation figure and no range around the ratio inputs.

• ▸Line item: name the income statement or balance sheet line (COGS, SG&A, receivables, inventories, payables, net fixed assets).
• ▸Historical ratio to sales, year 1: line item divided by sales for the earliest year.
• ▸Historical ratio to sales, year 2: same calculation for the middle year.
• ▸Historical ratio to sales, year 3: same calculation for the latest year.
• ▸Cash effect: mark whether growth in this line is a cash use (assets, fixed assets) or a cash source (sales-related liabilities).
• ▸Forecast ratio: the assumption you will use, expressed as a percent of sales.
• ▸Basis for forecast: average, trend or deliberate change, with one sentence of justification.
• ▸Low and high sensitivity case: the ratio range you will test in the pro forma.
• ▸Cross-check: implied dollar change versus the historical cash flow statement line, and whether it is reasonable.

1. 1Separate functional items from policy items before you build. Drive sales-dependent lines (receivables, and fixed assets if you model them as a step function of sales) from the sales forecast. The asset side and usually current liabilities are functional. Reserve the mix between long-term debt and equity for policy. Write the policy items in a visible assumptions block, not inside formulas.
2. 2Set the sales driver first. In Benninga's base case, year-0 sales are 1,000 and grow 10% a year. Every functional balance-sheet line inherits that path, so the financing need is derived from growth, not typed in.
3. 3Choose the plug deliberately and name it on the sheet. Options are cash and marketable securities, debt, or new equity. Define the plug mechanically as the amount that makes assets equal liabilities plus equity, so the balance sheet always balances.
4. 4Read the plug's financial meaning, not just its mechanical one. Choosing cash and marketable securities means the model absorbs any surplus or shortfall as a change in cash. Choosing debt or equity means the model absorbs it by borrowing or issuing. Write one sentence stating what the plug implies about how growth is financed.
5. 5Set the capital-structure policy explicitly. Decide the long-term debt versus equity mix as an input (for example a target debt-to-equity ratio) and keep it separate from the plug. If debt is a policy line and cash is the plug, you fix debt and let cash absorb the residual.
6. 6Set the dividend policy as an input and tie it to growth. The Gordon model's g is the anticipated dividend growth rate, not the historical one. Benninga notes the pro forma statements can help predict future dividends from anticipated sales growth and capital needs, so check the payout against the financing the forecast requires.
7. 7Run the model for every forecast year and inspect the plug line. A plug that swings to implausible values (very large cash, or a plug that goes negative) shows the policy set is inconsistent with the growth assumption. Adjust the policy inputs, not the functional relationships.
8. 8Test the policy by flexing one input at a time. Change the dividend assumption or the debt-equity mix and watch the plug and any dividend-based cost-of-equity estimate move. Keep the changes in the assumptions block so you can trace each effect.

Benninga's pro forma model starts from year-0 sales of 1,000, growing 10% per year, so year-1 sales are 1,100 and year-2 sales are 1,210. Receivables and other sales-driven items scale with those figures. Current liabilities are functional, so the only open financing choice is the long-term debt and equity mix, plus the plug. The base model sets cash and marketable securities as the plug. Mechanically, cash is whatever makes assets equal liabilities plus equity. Financially, it means that if growth consumes more funds than the policy-set financing supplies, the cash balance falls, and if the policy supplies more, cash accumulates. Dividend policy then links back to valuation. For Abbott Labs, the Gordon model gives a cost of equity of 16.28% with the 10-year dividend growth rate and 13.40% with the 5-year rate. Benninga calls the choice of g a judgment call about what the firm can and will pay out. A pro forma model built on the sales growth and capital needs above is the tool for making that judgment defensible.

• If an item moves with sales, model it as a functional relationship. If management chooses it, make it a policy input. The debt-equity mix is the standard policy item.
• If you want the balance sheet to show how growth is financed, choose the plug on purpose and state what it implies. Do not default to cash without checking that its implied financing story is acceptable.
• If the plug (for example cash) goes implausibly large or negative, change the debt, equity, or dividend policy. Do not distort the functional ratios to force a fit.
• If you use a dividend growth rate g in the Gordon model, use the anticipated rate. If historical rates disagree (5-year versus 10-year), pick the one that fits expected payout and say why.
• If you present a cost-of-equity figure, show how much it moves under the alternative dividend-growth assumption (16.28% versus 13.40% for Abbott). That range is part of the answer.

• Treating the plug as a purely mechanical device. The tell: nobody on the team can say what the plug choice implies about financing, yet the balance sheet balances.
• Hard-coding debt, dividends, or the plug inside formulas. The tell: changing a financing assumption requires editing formulas across several cells, and a policy change is hard to trace.
• Mixing policy and functional items, such as making long-term debt a fixed percentage of sales without deciding it is policy. The tell: the capital structure shifts with sales and no one chose that.
• Extrapolating historical dividend growth as the forward rate without checking it against what the forecast can fund. The tell: g is copied from a past period (such as the 10-year rate) and the cost of equity swings from 13.40% to 16.28% unnoticed.
• Never inspecting the plug line over the forecast horizon. The tell: the model balances every year, but cash or the plug balloons or turns negative in later years.

• ▸Sales driver: year-0 sales and annual growth rate (source example: 1,000 and 10%).
• ▸Functional items: list each balance-sheet line driven by sales and its ratio or step rule.
• ▸Debt-equity policy: target long-term debt versus equity mix or ratio, with the reason.
• ▸Dividend policy: payout assumption and the anticipated dividend growth rate g, with its basis.
• ▸Plug selection: cash and marketable securities, debt, or new equity.
• ▸Plug definition: formula making assets equal liabilities plus equity.
• ▸Financing implication: one sentence on what the plug choice means for how growth is financed.
• ▸Plug check by year: value of the plug each forecast year, flagged if negative or implausibly large.
• ▸Sensitivity note: result of changing the dividend growth rate or the debt-equity mix, and the effect on the plug or cost of equity.

1. 1Calculate the cost of equity two ways and reconcile. Use the Gordon model: cost of equity = D1/P0 + g, where D1 is next year's anticipated dividend, P0 is current price, and g is the anticipated dividend growth rate. Confirm g < the discount rate, or the infinite sum has no finite value.
2. 2Run the CAPM as the second cost-of-equity estimate. Regress the firm's periodic returns against the S&P 500 returns (Excel Trendline) to extract beta. For Abbott, expect a low-ish R^2 (~33%) — that is respectable in CAPM work, since the remaining variability diversifies away in a portfolio.
3. 3Choose SML parameters: set r_f to the risk-free rate (Treasury bill yield) and set the market risk premium E(r_M) − r_f to the historic average of a broad-based portfolio's return minus the risk-free rate.
4. 4Decide whether to tax-adjust the SML. Classic SML: cost of equity = r_f + beta*[E(r_M) − r_f], ignoring taxes. Benninga-Sarig tax-adjusted SML replaces the slope term with [E(r_M) − r_f] + T_C*r_f to account for the marginal corporate tax rate T_C.
5. 5Calculate the cost of debt using one of the three modeled approaches (including a CAPM-based estimate of debt's required return), representing the anticipated future cost of the firm's borrowing rather than its historic coupon.
6. 6Weight each component by its share of total capital and blend into WACC. Apply the corporate tax shield to the debt component consistently with your SML tax treatment.
7. 7Use the resulting WACC as the single discount rate on the firm's free cash flows (the cash the firm would produce with no debt and no cash/marketable securities) to arrive at enterprise value.
8. 8Sanity-check WACC against the opportunity-cost logic: it must represent the return required to make the firm a viable alternative to similar-risk investments.

Benninga computes the cost of capital for Abbott Laboratories. He first estimates the cost of equity via the Gordon model (D1/P0 + g) and separately via CAPM by regressing Abbott's returns on the S&P 500. The regression yields an R^2 of about 33 percent, meaning roughly a third of the variation in Abbott's returns is explained by market movements and the rest is diversifiable. He then feeds the estimated beta into the SML — first the classic form r_f + beta*[E(r_M) − r_f], then the Benninga-Sarig tax-adjusted form that replaces the premium with [E(r_M) − r_f] + T_C*r_f. He estimates the cost of debt with the modeled approaches, weights equity and debt by their capital shares, and produces the WACC. That WACC becomes the discount rate applied to Abbott's free cash flows to value the firm.

• If dividend growth g is greater than or equal to your discount rate, then do not use the Gordon closed form — the infinite dividend sum diverges and has no finite value.
• If the economy's marginal corporate tax rate materially affects the cost of capital, then use the Benninga-Sarig tax-adjusted SML rather than the classic tax-ignoring SML.
• If you are discounting firm (unlevered free) cash flows, then use WACC; if you are discounting equity cash flows only, then use the cost of equity.
• If the CAPM R^2 looks low (around 33%), then do not reject the estimate — that is normal, because the unexplained variance is diversifiable in a portfolio.
• If two cost-of-equity estimates (Gordon vs. CAPM) diverge sharply, then investigate the inputs (g, P0, beta, premium) before averaging — the discount rate is not trivial to pin down.

• Applying the Gordon formula when g >= discount rate — the tell: the model returns a negative or nonsensical price, or an unbounded value.
• Treating cost of capital as a single obvious number rather than a risk-adjusted estimate requiring theoretical models plus numerical computation — the tell: no reconciliation between Gordon and CAPM outputs.
• Ignoring taxes in the SML when the corporate tax rate is significant — the tell: WACC understates the required return because the debt shield and the equity premium are mistreated.
• Discounting the wrong cash flow with WACC (e.g., using it on equity cash flows) — the tell: enterprise and equity values don't reconcile with the capital structure.
• Rejecting a CAPM beta because R^2 is 'only' ~33% — the tell: discarding a respectable estimate and chasing a spuriously high fit.

• ▸Risk-free rate r_f: enter the Treasury bill yield.
• ▸Market risk premium E(r_M) − r_f: enter the historic average market-portfolio return minus r_f.
• ▸Beta: enter the slope from regressing firm returns on S&P 500 returns (note the R^2).
• ▸Corporate tax rate T_C: enter the marginal rate; leave blank if using classic SML.
• ▸Cost of equity (CAPM): compute r_f + beta*premium (classic) or with the tax-adjusted premium.
• ▸Cost of equity (Gordon): compute D1/P0 + g; confirm g < this rate.
• ▸Cost of debt: enter the anticipated future borrowing cost from the debt model used.
• ▸Capital weights: enter equity share and debt share of total capital.
• ▸WACC output: blend weighted after-tax debt cost and equity cost; use as the FCF discount rate.

1. 1Fix the forecast horizon first. In Benninga's model it is year 5. The terminal value at that date stands in for the present value of all FCFs from year 6 onward, so the growth rate you pick applies only after the last explicit year.
2. 2Pick g from expectations, not from history. The source says the correct growth rate is the anticipated rate and is a judgment call. Abbott's 10-year dividend growth (14.87%) and five-year growth (12.03%) imply costs of equity of 16.28% and 13.40%. Use history as evidence only after you have argued it is representative of the future.
3. 3Anchor g to the model's own drivers. Benninga sets post-year-5 cash flow growth equal to the model's sales growth of 10%. The source also notes that pro forma statements, which project sales growth and capital needs, can help you predict future growth.
4. 4Test the constraint before you compute: g must be below the discount rate (WACC for FCF, cost of equity for dividends). Write the check into the sheet. If g is at or above the rate, the Gordon formula returns an infinite or negative number, and that signals a violated condition, not a negative value.
5. 5Compute TV at year 5 with the Gordon form: FCF5 × (1+g) / (WACC − g). At g = 10% and WACC = 20%, the multiplier is 1.10 / 0.10 = 11× year-5 FCF. Then discount TV to today at the year-5 factor, together with the explicit FCFs.
6. 6If your growth in the forecast window is above the discount rate, use two stages. Discount the high-growth years explicitly, then apply the slower long-term rate g2 from the next year onward, and discount that Gordon value back over the explicit years.
7. 7Cross-check TV with an alternative method from the source: year-5 book value of debt plus equity, market/book multiple × book value, P/E × year-5 profits + book debt, or EBITDA ratio × year-5 EBITDA. Vary the EBITDA ratio (the source suggests 6 to 14) and compare the equity value it implies against your Gordon result.
8. 8Run sensitivity on g and WACC jointly, as the source does with a data table. Build the table's calculation cell so it handles g ≥ WACC (the source redefines cell B107 for this) instead of showing nonsense values.

The source's supernormal-growth case shows how much value sits behind the terminal rate. A firm pays a current dividend of $8, grows it at 35% for five years, then slows to g2 = 8%, with an 18% discount rate. The five explicit dividends are worth $40 today. The share is valued at 230.33. By subtraction, about 190 of that value, roughly 83%, comes from years 6 to infinity, and that portion is the Gordon formula discounted back over five years. Now see how sensitive the formula is to g in the source's plain Gordon case: D1 = $3, g = 10%, r = 15% gives 3 / (0.15 − 0.10) = $60. If g rises to 14%, the same formula gives 3 / 0.01 = $300. A four-point change in g multiplies value fivefold. At g = 15% the formula fails. In Benninga's FCF model, WACC is 20% and g is 10%, so TV = 11 × FCF5. The spread of 10 points is wide enough to leave a margin, but you still document why 10% holds in perpetuity.

• If g ≥ the discount rate, stop. Either the long-term growth is lower than you assumed, or the discount rate is too low (the source's two diagnoses). Do not plug a number into the formula.
• If the explicit window still has growth above the discount rate (the 35% case), then shorten the claim: use a multi-stage model and apply a lower g2 only after growth fades.
• If a historical growth rate and a recent one disagree (Abbott 14.87% vs 12.03%), then choose the one you can defend as representative of future payouts. Say why the other is not.
• If the Gordon TV and a multiple-based TV (EBITDA, book value, P/E) diverge a lot, then reconcile by checking which assumption the gap implies. Do not average them blindly.
• If you present the valuation, then show the sensitivity of equity value to g and WACC together, because small changes in the spread (WACC − g) drive large changes in TV.

• Using trailing historical growth as g without checking it is representative. The tell: g equals a past CAGR and there is no argument about future payouts or sales.
• Letting g approach or exceed WACC. The tell: a negative or enormous TV, or a sensitivity table with absurd cells.
• Treating TV as a detail. The tell: the explicit forecast is built carefully while terminal value silently makes up most of enterprise value (about 83% in the dividend example).
• Applying a high near-term growth rate in perpetuity. The tell: the same 35%-style rate drives both the explicit years and the terminal calculation, with no fade stage.
• Not cross-checking TV against another method. The tell: only one Gordon figure exists and no implied multiple or book-value comparison appears anywhere in the model.

• ▸Discount rate (WACC for FCF; cost of equity for dividends): enter the rate you discount at.
• ▸Proposed terminal growth rate g: enter the perpetual rate and one line on why it is expected.
• ▸Spread test (rate − g): compute it; if zero or negative, stop and revisit g or the rate.
• ▸Year-5 FCF: enter the final explicit-year cash flow.
• ▸Terminal value = FCF5 × (1+g) / (rate − g): compute and record the implied multiple of FCF5.
• ▸Share of enterprise value from terminal value: PV of TV divided by total value; flag if dominant.
• ▸Alternative TV (book value, EBITDA ratio, or P/E method): enter one and its resulting value.
• ▸Gap between Gordon TV and alternative TV: record it and name which assumption explains it.
• ▸Sensitivity range: list the g and WACC values tested and the resulting equity values.