FIG Modeling Tests: What to Expect and How They Differ

Introduction

FIG modeling tests evaluate two competencies simultaneously: your technical modeling ability (can you build a functioning model under time pressure?) and your sector-specific judgment (do you understand why FIG models are built differently?). Most candidates who fail FIG modeling tests do not fail because they cannot use Excel. They fail because they approach a bank model the same way they would approach a corporate DCF, and the outputs immediately reveal that they do not understand the fundamental structure of financial institution economics.

The root difference is this: financial institution models are balance sheet-driven. You start by projecting the balance sheet (loan growth, deposit growth, securities holdings), then derive the income statement from the balance sheet (net interest income = earning assets x NIM, provision = loans x provision rate), then build to net income, then calculate regulatory capital adequacy, then determine how much can be distributed. This is the reverse of how standard corporate models work, where the income statement drives the balance sheet. Interviewers see within the first five minutes of your model whether you understand this architecture. If you start with a revenue line and project down, you have already shown them you do not understand the business.

The conceptual foundation that every FIG modeling test requires you to demonstrate is why traditional DCF breaks down for financial institutions. Understanding this is not just interview knowledge: it determines how you structure the entire model.

In a standard DCF, you project unlevered free cash flow (EBIT x (1 - tax rate) + D&A - capex - change in working capital) and discount at WACC, which blends the cost of debt and cost of equity according to the target capital structure. The output is enterprise value, from which you subtract net debt to get equity value. This framework requires that debt be separable from the business.

For a bank, debt is not separable. A commercial bank's deposits are not a financing choice. They are the raw material that funds the loan portfolio that generates net interest income. Removing deposits removes the business. There is no "unlevered" bank to value. Free cash flow to the firm is undefined because interest expense is an operating cost, not a financing cost. WACC cannot be computed meaningfully because the debt-to-equity ratio is a regulatory variable, not a management preference.

Free Cash Flow to Equity (FCFE) vs. Free Cash Flow to the Firm (FCFF)

FCFF is cash flow available to all capital providers (debt and equity), discounted at WACC to get enterprise value. FCFE is cash flow available only to equity holders after debt service, discounted at cost of equity to get equity value directly. For banks, only FCFE is meaningful, and it takes the form of dividends plus share buybacks (total shareholder payout), constrained by regulatory capital requirements. This is exactly what the DDM models. The absence of a meaningful FCFF definition is why EV, EBITDA, and WACC cannot be used for financial institutions.

The implication for modeling tests: if you use WACC anywhere in a bank model, or try to calculate EBIT (since there is no EBIT for a bank), or try to separate operating from financing activities, you have made a fundamental error. The correct framework is cost of equity as the discount rate, dividends plus buybacks as the distributable cash flow, and equity value (not enterprise value) as the output.

The most common FIG modeling test archetype for analyst and associate roles at bulge bracket and specialist FIG firms is a bank operating model that feeds into a DDM valuation. Time limit is typically 2-4 hours. You receive a set of historical financials (3-5 years), a balance sheet, and assumptions (growth rates, rate environment outlook, management efficiency ratio targets).

What You Build

Balance sheet projection: Start with earning asset growth. Total loans grow at an assumed rate (typically 4-8% annually for a regional bank, 2-5% for a money-center bank in the current environment). Securities holdings are a residual plug or explicitly targeted as a percentage of assets. Total interest-earning assets is the starting point for income.

Net interest income: NII = average earning assets x NIM. NIM projection is the most judgment-intensive assumption in the model. You need to take a view on the rate environment (are deposit costs rising faster than loan yields?), the loan mix (CRE vs. C&I vs. consumer have different yields and repricing dynamics), and funding mix (core deposits vs. wholesale funding have different beta to rate moves). A regional bank that funded itself heavily with brokered CDs during the rate hiking cycle will see faster NIM compression than one with a strong core deposit franchise.

Non-interest income: Fee revenue from treasury services, wealth management, mortgage banking, and trading. Model as a percentage of assets or grow at an assumed rate. Do not over-complicate this section: interviewers are not testing whether you can model 12 fee line items. They want to see a clean, defensible approach.

Provision for credit losses: Provision = beginning allowance for credit losses x provision rate (net charge-offs as a percentage of loans), adjusted for loan growth to maintain coverage ratios. In a normalized environment, use 30-50 basis points of average loans for most regional banks. In a stress scenario, 100-200+ basis points. CECL adds complexity here because you are reserving for expected losses over the life of the loan rather than incurred losses, but in a modeling test context, keep the provision logic simple: provision rate times average loans.

Efficiency ratio: Non-interest expense divided by total revenue (NII plus non-interest income). Well-run regional banks target 50-60%. Model operating leverage by growing expenses at a slower rate than revenue if management has announced efficiency initiatives. The efficiency ratio is how interviewers assess whether your income statement is internally consistent.

Net income: NII + non-interest income - provision - non-interest expense - taxes. This flows directly to the capital calculation.

DDM valuation: Once you have net income and excess capital by year, apply the payout ratio (or calculate it from the capital math) to get dividends per share. Discount at cost of equity (typically 9-12% for US banks, calculated with CAPM). Stage 1 is the explicit forecast (3-5 years). Stage 2 is a transition where dividend growth decelerates. Stage 3 is the terminal value using the Gordon Growth Model: D(T+1) / (cost of equity - terminal growth rate). Terminal growth is typically 2-4%.

Key Judgment Calls

The assumptions that most affect output: (1) NIM trajectory over the projection period, especially if the rate environment is unclear; (2) provision rate, particularly whether you use a through-the-cycle normalization or current-quarter actuals; (3) efficiency ratio improvement, which can swing earnings by 100-200 basis points over a 5-year period; (4) CET1 target, since a more conservative target reduces distributable capital; (5) terminal ROE relative to cost of equity, since this drives the DDM terminal value, which represents 60-80% of intrinsic value.

Senior bankers look first at NIM and then at the capital bridge. If your NIM is flatline across all 5 years in a falling-rate environment, you have not thought about the model. If your CET1 ratio somehow stays constant while you are paying out 70% of earnings and growing loans at 6%, your model has a math error.

Bank merger models are the most FIG-specific exercise in the testing universe. They combine standard accretion/dilution analysis with components that exist nowhere else in investment banking: core deposit intangible (CDI) valuation, regulatory capital impact analysis, and TBV earn-back modeling. Time limit is typically 2-4 hours for a full model or 60-90 minutes for a simplified version.

Structuring the Deal

You receive transaction parameters: acquirer and target financial profiles, purchase price (or price per share), consideration mix (stock vs. cash vs. combination), and any deal-specific information (synergy targets, management commentary on cost saves).

Step 1: Purchase price allocation. The total consideration paid, minus the target's tangible book value (equity minus goodwill minus other intangibles), gives you the tangible book value premium. This premium must be allocated to identifiable intangible assets (principally the core deposit intangible) and goodwill.

Core Deposit Intangible (CDI)

The CDI represents the economic value of the target bank's below-market-rate deposit base. Core deposits (checking, savings, money market accounts) are a cheaper funding source than wholesale alternatives. The CDI quantifies how much more valuable those deposits are to the acquirer than their face value. CDI is calculated using a cost-savings approach: the present value of the savings from using core deposits (at their all-in cost including interest and overhead) versus an alternative wholesale funding source over the estimated life of the deposit base. CDI is typically 1-3% of core deposits. It is amortized over approximately 7-10 years using an accelerated method (sum-of-years-digits or double-declining balance), because the deposit runoff is front-loaded. CDI amortization reduces pre-tax income but does not reduce CET1 capital (intangibles are already excluded from regulatory capital). This creates a difference between GAAP EPS accretion/dilution and cash EPS accretion/dilution that sophisticated FIG models present separately.

Step 2: Goodwill calculation. Goodwill = total consideration paid - fair value of net assets acquired (including any mark-to-market adjustments to the loan portfolio and securities book) - CDI. Goodwill is not amortized under GAAP but is excluded from regulatory capital, creating an immediate CET1 dilutive impact.

Step 3: Funding the deal. For a stock deal, you issue new shares at the current acquirer price (or the deal-implied price). For a cash deal, the acquirer uses balance sheet cash or issues debt. The consideration mix matters enormously for CET1: cash consideration drains capital directly, while stock consideration replaces capital at risk (you are issuing equity, not deploying it). Most bank deals in the current environment are primarily stock-funded to preserve capital ratios.

Pro Forma Income Statement

Combine acquirer and target income statements, then add merger adjustments:

Cost synergies: Bank mergers derive 60-80% of their value from cost takeout. Branch consolidation, technology platform rationalization, and back-office elimination are the primary sources. Model synergies phasing in over 2-3 years (year 1 at 25-35% of full run-rate, year 2 at 70-80%, year 3 at 100%). Restructuring charges typically equal 50-100% of annualized cost synergies, recognized in year 1.

Revenue synergies: Cross-sell opportunities, fee income expansion, and deposit repricing benefits. Conservative models assume modest revenue synergies (15-25% of cost synergies). Aggressive models include geographic expansion benefits and product cross-sell. FIG bankers are skeptical of high revenue synergy assumptions: they are the first number that deal opponents attack.

CDI amortization: Add back as a pre-tax expense item in the combined income statement. CDI of $50 million amortized over 10 years on an accelerated basis might produce $10 million of amortization in year 1, declining to $4 million by year 10. This reduces GAAP earnings but not cash earnings.

Loan fair value marks: If the target's loan portfolio has a below-market yield (common when interest rates have risen since origination), fair value the loans to market, creating a discount that accretes back into income over the remaining loan life. This is a benefit that partially offsets CDI amortization.

EPS Accretion/Dilution and TBV Earn-Back

TBV dilution at close: Total consideration paid minus target's tangible book value gives you the TBV per share dilution at close for the acquirer. This is the most scrutinized metric in bank M&A. If the acquirer has TBV per share of $60.00 before the deal, and the deal reduces TBV per share to $53.00, the TBV dilution is ($7.00), or (11.7%). Boards and investors want this dilution to be earned back within 3-5 years.

[TBV earn-back period](/guides/fig-investment-banking/tbv-dilution-earn-back-period): Model the acquirer's standalone TBV per share and the pro forma combined TBV per share on a year-by-year basis. The earn-back period is the year when the two lines cross: i.e., when the pro forma TBV per share exceeds what the standalone TBV per share would have been. The crossover method is the preferred approach at most FIG shops. An earn-back period above 4 years is increasingly difficult to get board approval for. Deals structured at 1.5-2.0x P/TBV typically earn back within 2-4 years with realistic cost synergies. Deals above 2.0x P/TBV stretch the earn-back to 5+ years.

Key Judgment Calls

Where candidates make costly mistakes: (1) Calculating CDI on total deposits rather than core deposits (brokered CDs and time deposits above $250,000 are excluded from the CDI base); (2) using straight-line amortization for CDI rather than accelerated (produces materially different year 1 and year 2 EPS numbers); (3) forgetting to model restructuring charges (creates a year 0 / year 1 capital draw that is often overlooked); (4) ignoring the capital ratio impact of goodwill (goodwill is excluded from CET1, so a goodwill-heavy deal dilutes CET1 even if equity value is preserved); (5) using the EPS accretion method for earn-back instead of the crossover method (they produce different results, and the crossover method is what most FIG bankers and boards use as the primary metric).

Insurance modeling tests are rarer than bank operating models and merger models, but they appear in interviews at firms with dedicated insurance coverage (Goldman, Morgan Stanley, Lazard, Barclays, Piper Sandler). They test whether you can simultaneously apply two fundamentally different valuation frameworks within the same parent company.

The Structure

A diversified insurance holding company with a life insurance segment and a P&C segment. You are asked to value the consolidated entity using a sum-of-the-parts approach.

Life insurance segment: Valued using embedded value (EV). EV = Adjusted Net Asset Value (ANAV) + Value of In-Force Business (VIF). VIF is the present value of expected after-tax profits from the existing book of policies, discounted at a risk-adjusted rate. Key inputs: discount rate for VIF (typically 8-10%), assumed persistency rates (what percentage of policyholders keep their policies in force, period by period), and mortality assumptions (for life policies) or lapse rates. VIF is sensitive to the discount rate: a 100-basis-point increase in the discount rate can reduce VIF by 15-25%. Multiples of EV (price/EV or price/ANAV) are the primary relative valuation metric for life insurers.

P&C insurance segment: Valued using a P/E multiple applied to normalized earnings, which requires normalizing the combined ratio to mid-cycle. P&C insurer earnings are highly cyclical: in hard markets (combined ratio of 93-96%), earnings are strong. In soft markets or post-catastrophe years (combined ratio of 103-108%), earnings are weak or negative. Normalizing requires picking a through-the-cycle combined ratio assumption and calculating what earnings would be at that level. Apply a P/E multiple of 11-14x (current median insurance industry P/E is approximately 13.6x). The P&C segment may also be valued on price-to-book or price-to-tangible-book if the business has significant investment portfolio exposure that drives returns independent of underwriting.

Capital allocation: One complexity specific to insurance SOTP is allocated capital. Each segment operates under risk-based capital (RBC) requirements. You need to allocate the holding company's total capital to each segment based on RBC minimums plus a buffer. Excess capital at the holding company level (capital above segment-level minimums) is valued separately, typically at face value or with a modest discount for deployment uncertainty.

Time limits: Most FIG modeling tests run 2-4 hours. Senior banker tests (associate, VP laterals) tend toward 3-4 hours with more complete model requirements. Analyst-level tests tend toward 2-3 hours with a pre-built skeleton provided.

Data provided: Historical financials (3-5 years), balance sheet, selected assumptions or guidance from management, and sometimes a partially built model with certain formulas already in place. For bank merger tests, you receive the target and acquirer standalone models and deal parameters. You are not expected to source data: everything you need is provided.

What they expect you to build: A functional model with correct logic (balance sheet that ties, capital ratios that are calculated correctly, DDM that discounts at cost of equity rather than WACC) and output pages that an MD could use in a meeting. Presentation-quality formatting is not the priority in a 2-hour test, but clearly labeled sheets, visible assumptions, and outputs that answer the questions posed (EPS accretion/dilution, TBV earn-back, DDM intrinsic value) are non-negotiable.

What to skip: Do not spend time on elaborate waterfall formatting or color-coded headers if you are running short on time. A clean, functional model with correct outputs beats a beautifully formatted model with a balance sheet that does not balance. If you have to choose between completing the sensitivity table and checking that your CET1 ratios are correctly calculated, finish the capital math first.

Bulge bracket FIG tests (Goldman, Morgan Stanley, JPMorgan, Barclays, Citi): Higher expectations on modeling completeness. More likely to receive a partially built skeleton with specific outputs requested. Will test whether you understand the interaction between the income statement, balance sheet, and regulatory capital. May include a bank merger model in addition to an operating model. Expect to present your work verbally and defend your assumptions.

FIG specialist firms (Piper Sandler, KBW / Stifel, Hovde, Stephens): More focus on the merger model since bank M&A advisory is a larger portion of their business. KBW and Piper Sandler are the highest-volume bank M&A advisors by deal count: in 2024, Piper Sandler worked on approximately 53 disclosed bank transactions, KBW on approximately 62 (through Stifel). The modeling tests at these firms reflect that deal focus: expect a bank merger model with CDI, TBV earn-back, and synergy sensitivity as the primary exercise. Community bank specifics (thrift structures, mutual-to-stock conversions) may also appear.

Boutique advisory firms (Lazard FIG, Evercore financial services, Perella Weinberg): More likely to focus on a specific deal scenario (comparable to a real transaction they recently worked on) rather than a generic bank operating model. Expect to be tested on your reasoning and judgment as much as on your Excel mechanics. The expectation is that you can defend every assumption with a fundamental rationale, not just populate formula cells.

Mistake 1: Using WACC instead of cost of equity. If WACC appears anywhere in your bank DDM, you have already demonstrated a fundamental misunderstanding of why standard DCF does not apply to financial institutions. Cost of equity is the only appropriate discount rate because dividends are equity-level cash flows. WACC reflects the blended cost of debt and equity for unlevered enterprise value, which is not a meaningful concept for a bank.

Mistake 2: Not linking the income statement to the capital ratios. Candidates who project dividends as a flat percentage of earnings without checking whether those dividends are feasible given CET1 requirements will produce incorrect DDM outputs. The bank DDM is only as good as the capital constraint embedded in the payout calculation. If your bank is growing loans at 7% annually and paying out 60% of earnings, you need to verify that CET1 does not fall below target. If it does, the payout must come down.

Mistake 3: Ignoring CDI amortization in the merger model. CDI amortization is a real GAAP expense that reduces reported earnings but does not reduce cash earnings or regulatory capital. Forgetting it understates the GAAP EPS dilution in early years and overstates accretion. Models that ignore CDI also misstate the TBV dilution calculation at close because CDI is an intangible that reduces tangible book value.

Mistake 4: Using a terminal growth rate inconsistent with the sustainable growth formula. Terminal growth rate = ROE x retention ratio. If your terminal ROE is 11% and your terminal payout ratio is 60%, sustainable growth is 4.4%. A terminal growth rate assumption of 3.5% is consistent. A terminal growth rate of 6% with those parameters means the bank distributes more than it earns, which is internally inconsistent and will confuse any interviewer who checks the math.

Mistake 5: Not stress-testing NIM. NIM is the most sensitive single assumption in the bank operating model. A 10-basis-point change in NIM on a $50 billion earning asset base is $50 million of pre-tax income. Interviewers expect to see a NIM sensitivity table showing how intrinsic value changes if NIM is +/- 20 basis points from the base case. If you submit a model without any NIM sensitivity, you have signaled that you do not understand which assumption matters most.

These mistakes are not obscure technical failures. They are the first things a senior FIG banker checks when they review your model. Avoiding them is the clearest signal that you have prepared correctly and understand why bank models are different from everything else in investment banking.