Detect Tax Rate Normalization Risk Through Multi-Year Effective Tax Analysis

Reported earnings can change for reasons that have little to do with the underlying operating performance of the business. One of the easiest places for that distortion to appear is the effective tax rate.

A company may report stronger or weaker net income because its tax expense moves sharply from one year to the next, even when pretax earnings follow a more stable path. For an equity research team, the important question is not simply whether the tax rate changed, but whether the change looks temporary enough that reported earnings may not represent a sustainable run rate.

In this workflow, we use Claude with Financial Modeling Prep's MCP server to compare pretax income, income-tax expense, and net income across three completed annual periods. Claude calculates the effective tax rate, establishes a company-specific normalization baseline, estimates the earnings impact of applying that normalized rate, and flags unusual periods for targeted filing review.

The objective is to identify tax-rate normalization risk before unusual tax effects are carried into earnings interpretation, screens, or financial models. It is not a tax-planning exercise, a forecast of future tax rates, or a broader assessment of earnings quality across the financial statements.

Defining the Tax-Rate Normalization Methodology

The methodology separates two questions: how volatile a company's effective tax rate has been and whether an unusual tax rate supported or reduced reported earnings.

For each company, Claude uses the latest three completed annual periods and calculates:

• Effective tax rate, or ETR: income-tax expense divided by pretax income
• Tax bridge: pretax income minus income-tax expense, compared with net income from continuing operations
• Three-year ETR range: the highest annual ETR minus the lowest annual ETR
• Normalized tax rate: the median of the three valid annual ETRs
• ETR deviation: the annual ETR minus the normalized tax rate
• Normalized tax expense: pretax income multiplied by the normalized tax rate
• Normalized continuing-operations net income: pretax income minus normalized tax expense
• Estimated normalized earnings impact: reported continuing-operations net income minus normalized continuing-operations net income

Net income from continuing operations is used for the normalization comparison rather than total net income. If continuing-operations income is unavailable or differs materially from total net income in a way that prevents reliable comparison, the period returns Review Required rather than substituting total net income.

The median ETR provides a company-specific baseline that is less sensitive than an arithmetic average to one unusually high or low year. It is an illustrative normalization assumption, not a forecast of the company's future statutory or effective tax rate.

The earnings impact is interpreted separately from tax-rate volatility:

• Reported Earnings Supported: The annual ETR is below the normalized rate, producing a positive normalized earnings impact.
• Reported Earnings Reduced: The annual ETR is above the normalized rate, producing a negative normalized earnings impact.
• Neutral: The annual ETR is approximately equal to the normalized rate.

Tax-rate volatility is classified symmetrically using the absolute deviation from the normalized ETR:

• High Tax-Rate Volatility: The three-year ETR range is at least 10 percentage points, or any annual absolute ETR deviation is at least 5 percentage points.
• Moderate Tax-Rate Volatility: The ETR range is 5 to less than 10 percentage points, or any annual absolute deviation is 2 to less than 5 percentage points.
• Low Tax-Rate Volatility: The ETR range is below 5 percentage points and no annual absolute deviation is at least 2 percentage points.
• Review Required: Required data is incomplete, the tax bridge is unreliable, continuing-operations net income is unavailable or materially inconsistent, or the ETR cannot be interpreted reliably.

A separate denominator safeguard applies to pretax income. If pretax income is positive but less than 1% of annual revenue, the period returns Review Required: Small Pretax Denominator because a small pretax base can generate an extreme and misleading effective tax rate. Zero or negative pretax income also triggers Review Required.

These thresholds are illustrative screening rules for research prioritization, not accounting-materiality or tax-advisory standards. For High- or Moderate-volatility cases, filing evidence is reviewed separately to determine whether the unusual tax effect reflects a disclosed one-time item, a potentially recurring factor, or another identified tax driver.

FMP Financial Data and Claude MCP Setup

The analysis was run on August 29, 2026. The numerical screen uses annual financial-statement data from FMP, while supporting tax explanations are retrieved only for companies whose tax-rate volatility warrants additional review.

Source	FMP MCP Tool and Endpoint	Returned Fields Used	Purpose
Income Statement	FMP → income-statement, annual	revenue, incomeBeforeTax, incomeTaxExpense, netIncomeFromContinuingOperations, netIncomeFromDiscontinuedOperations, netIncome, fiscalYear, period, date, filingDate	Calculate ETR, validate the tax bridge, apply the pretax-income safeguard, estimate normalized earnings impact, and align fiscal periods.
SEC Filings by Symbol	FMP → search-by-symbol	formType, filingDate, acceptedDate, link, finalLink	Identify the relevant annual filing and retrieve the official SEC document link.
Earnings Transcripts	FMP transcript data, fallback only	Not used in this run	Provide additional explanatory context only if filing evidence is insufficient.

The FMP Income Statement API supplies the annual financial fields used in the calculations. In this run, the exact continuing-operations field returned was netIncomeFromContinuingOperations, which is separate from netIncomeFromDiscontinuedOperations and total netIncome.

For each fiscal year, Claude records both the fiscal period-end date from date and the associated filingDate. These dates remain attached to the source observations so the annual periods can be aligned before the effective tax rate or normalization impact is calculated. This distinction is important when aligning fiscal periods with filing and reporting dates.

Supporting filing evidence is handled separately. The FMP SEC-filings tool returns filing metadata and document links rather than the complete filing narrative. Claude uses FMP to identify the relevant filing and its finalLink, then reads the corresponding official SEC document for the tax disclosure. The fiscal period end and filing date of that source are reported with the evidence.

An earnings transcript is only a fallback if the filing does not provide sufficient explanation. No earnings-transcript tool was required for Apple or Meta because their FY2025 10-K filings provided the necessary tax evidence.

To use these datasets through Claude, first connect the FMP MCP server. An active FMP API key is required. In Claude, open Settings, select Connectors, choose Add custom connector, and enter:

https://financialmodelingprep.com/mcp?apikey=YOUR_FMP_API_KEY

Once connected, start a new Claude conversation with the FMP connector enabled.

For this analysis, Claude retrieves FY2023-FY2025 annual income statements first. It verifies that pretax income minus tax expense reconciles with netIncomeFromContinuingOperations, checks that pretax income is large enough for a meaningful ETR calculation, and only then applies the volatility and normalization rules.

No missing financial value is estimated or silently replaced. If continuing-operations income is unavailable or materially inconsistent, the tax bridge fails, or pretax income is too small for reliable interpretation, the workflow returns Review Required.

Building the Tax-Normalization Prompt

The prompt uses Apple and Meta as a two-company sample to keep the run compact while testing the same normalization logic across different tax-rate patterns. The objective is not to compare the companies as peers, but to determine whether the same screening framework can identify unusual tax effects consistently.

The analysis uses the latest three completed annual periods because that provides enough history to identify a temporary outlier or a wider tax-rate swing without expanding the run unnecessarily.

The normalized tax-rate assumption is based on the median of the three valid annual effective tax rates. This reduces the influence of a single unusually high or low year and gives the screen a company-specific baseline rather than imposing one tax rate across different businesses.

The prompt also separates the numerical screen from the explanation. Filing or transcript evidence is requested only for Moderate- or High-volatility cases, and Claude is limited to one supporting source per company. This keeps the workflow efficient while preventing an unusual tax rate from being assigned a cause without evidence.

Using the connected Financial Modeling Prep (FMP) MCP server, evaluate tax-rate normalization risk for: Apple Inc. (AAPL) Meta Platforms, Inc. (META) This is an earnings-quality screening exercise, not tax advice, tax planning, or an investment recommendation. Use FY2023-FY2025 completed annual fiscal periods only. Keep the workflow compact. Use only the minimum FMP calls required and no more than one retry total. Financial data Retrieve annual income-statement data for: revenue incomeBeforeTax incomeTaxExpense netIncomeFromContinuingOperations netIncomeFromDiscontinuedOperations netIncome fiscalYear period date filingDate Use continuing-operations net income for all normalized earnings comparisons. Do not substitute total net income if netIncomeFromContinuingOperations is unavailable or materially inconsistent. Return Review Required instead. Tax-bridge validation For each fiscal year calculate: tax bridge = incomeBeforeTax − incomeTaxExpense Compare this directly with netIncomeFromContinuingOperations. Report any difference. If continuing-operations net income differs materially from total net income, flag the period for analyst review rather than assuming the two measures are equivalent. Do not estimate missing values. Pretax-income safeguard Calculate: pretax income / revenue × 100 If pretax income is: zero or negative: Review Required positive but less than 1% of revenue: Review Required — Small Pretax Denominator Do not interpret the effective tax rate normally when this safeguard is triggered. The 1% threshold is an illustrative screening safeguard, not an accounting-materiality standard. Effective tax rate For valid periods calculate: effective tax rate = incomeTaxExpense / incomeBeforeTax × 100 For each company calculate: normalized tax rate = median of the three valid annual ETRs The median is a company-specific screening baseline, not a forecast of the company's future effective or statutory tax rate. For every valid fiscal year calculate: normalized tax expense = pretax income × normalized tax rate normalized continuing-operations net income = pretax income − normalized tax expense estimated normalized earnings impact = reported continuing-operations net income − normalized continuing-operations net income Interpret the earnings effect separately: Positive impact: Reported Earnings Supported Negative impact: Reported Earnings Reduced Approximately zero: Neutral Symmetric tax-rate volatility Calculate: ETR range = highest annual ETR − lowest annual ETR ETR deviation = annual ETR − normalized ETR absolute ETR deviation = absolute value of ETR deviation Use these illustrative classifications: High Tax-Rate Volatility: ETR range ≥10 percentage points; or any annual absolute ETR deviation ≥5 percentage points Moderate Tax-Rate Volatility: ETR range 5 to <10 percentage points; or any annual absolute ETR deviation 2 to <5 percentage points Low Tax-Rate Volatility: ETR range <5 percentage points; and no annual absolute ETR deviation ≥2 percentage points Review Required: required data is incomplete; continuing-operations net income is unavailable or materially inconsistent; pretax income is zero, negative, or triggers the small-pretax safeguard; a tax benefit or unusual presentation makes the ETR unreliable; the tax bridge cannot be reconciled reliably. Apply the deviation thresholds symmetrically to unusually high and unusually low ETRs. Do not treat volatility and earnings direction as the same measure. Report separately whether the unusual tax rate supported or reduced earnings. Supporting evidence Only for companies classified Moderate or High, identify the year with the largest absolute ETR deviation from the normalized rate. Use at most one supporting filing per company. Use the relevant annual SEC filing identified through FMP SEC-filings data. Use the official SEC EDGAR document linked by that filing record for the narrative tax disclosure. Look only for disclosed explanations directly relevant to the unusual tax rate. For Apple, use the FY2025 Form 10-K to evaluate the FY2024 State Aid-related tax effect. For Meta, use the FY2025 Form 10-K and keep these concepts separate: the disclosed one-time non-cash FY2025 tax charge; potentially ongoing tax-rate effects such as CAMT and share-based-compensation-related tax effects. Do not imply that a disclosed one-time charge itself is expected to persist. If the annual filing provides sufficient evidence, do not retrieve an earnings transcript. Do not infer a tax cause that the filing does not support. Confidence High: all three annual periods are complete; continuing-operations income is available; tax bridges reconcile; no denominator safeguard is triggered; relevant abnormal periods have sufficient filing evidence. Medium: financial calculations are complete but the cause of an abnormal ETR cannot be verified clearly. Review Required: core financial data, continuing-operations income, tax-bridge comparability, or denominator reliability is insufficient. Output Table 1 — Multi-Year Effective Tax Analysis | Company | Fiscal Year | Period End Date | Pretax Income | Income Tax Expense | Continuing-Operations Net Income | Effective Tax Rate | ETR Deviation | Table 2 — Tax Normalization Summary | Company | 3-Year ETR Trend | Normalized Tax Rate | ETR Range | Most Deviant Year | Earnings Direction | Estimated Normalized Earnings Impact | Tax-Rate Volatility | Supporting Evidence | Confidence | Analyst Follow-Up | For Estimated Normalized Earnings Impact, report the amount and whether reported earnings were Supported or Reduced. For Supporting Evidence, identify the filing and filing date. Keep analyst follow-up to one specific action tied to the tax result. Do not provide broader investment commentary. Keep the entire response concise.

Tax-Rate Normalization Results and Interpretation

The analysis separates effective-tax-rate volatility from the direction of the resulting earnings effect. All six annual periods passed the tax-bridge and pretax-income safeguards, and continuing-operations net income equaled total net income in every period.

Table 1A: Multi-Year Tax Inputs

Company	Fiscal Year	Period End Date	Pretax Income	Income Tax Expense
AAPL	FY2023	2023-09-30	$113.736B	$16.741B
AAPL	FY2024	2024-09-28	$123.485B	$29.749B
AAPL	FY2025	2025-09-27	$132.729B	$20.719B
META	FY2023	2023-12-31	$47.428B	$8.330B
META	FY2024	2024-12-31	$70.663B	$8.303B
META	FY2025	2025-12-31	$85.932B	$25.474B

Table 1B: Continuing-Operations Income and Effective Tax Rates

Company	Fiscal Year	Continuing-Operations Net Income	Effective Tax Rate
AAPL	FY2023	$96.995B	14.72%
AAPL	FY2024	$93.736B	24.09%
AAPL	FY2025	$112.010B	15.61%
META	FY2023	$39.098B	17.56%
META	FY2024	$62.360B	11.75%
META	FY2025	$60.458B	29.64%

For every period, pretax income minus income-tax expense reconciles exactly with netIncomeFromContinuingOperations. Continuing-operations net income also equals total net income in all six periods, with no discontinued-operations variance.

The small-pretax safeguard was not triggered. Pretax income ranged from 29.67% to 31.89% of revenue for Apple and from 35.16% to 42.96% for Meta, well above the illustrative 1% minimum.

Table 2A: Tax-Rate Normalization and Volatility

Company	Three-Year ETR Trend	Normalized ETR	ETR Range	Most Deviant Year	Tax-Rate Volatility
AAPL	14.72% → 24.09% → 15.61%	15.61%	9.37 pp	FY2024 (+8.49 pp)	High
META	17.56% → 11.75% → 29.64%	17.56%	17.89 pp	FY2025 (+12.08 pp)	High

Table 2B: Estimated Earnings Impact

Company	Earnings Direction	Estimated Normalized Earnings Impact	Confidence
AAPL	Reported Earnings Reduced	-$10.47B	High
META	Reported Earnings Reduced	-$10.38B	High

Table 2C: Supporting Evidence and Analyst Follow-Up

Company	Supporting Evidence	Analyst Follow-Up
AAPL	FY2025 10-K, filed October 31, 2025, confirms the FY2024 State Aid-related tax effect and reports a $10.7B year-over-year decrease in the FY2025 tax provision related to the State Aid Decision.	Assess the underlying FY2025 tax profile without the FY2024 State Aid-related distortion before selecting a forward normalization assumption.
META	FY2025 10-K, filed January 29, 2026, discloses a $15.93B one-time non-cash tax charge, including a $14.03B valuation allowance against U.S. federal deferred tax assets.	Separate the one-time FY2025 charge from potentially recurring CAMT and share-based-compensation tax effects when forming a forward tax-rate assumption.

This is an earnings-quality screening exercise using illustrative, author-defined thresholds. It is not tax advice, tax planning, or an investment recommendation. Financial data were retrieved on August 29, 2026.

Apple: High Volatility Driven by an Elevated FY2024 ETR

Apple's ETR moved from 14.72% in FY2023 to 24.09% in FY2024 before declining to 15.61% in FY2025. The three-year range is 9.37 percentage points, which by itself falls in the Moderate band.

However, FY2024 was 8.49 percentage points above Apple's 15.61% normalized ETR. Because the methodology applies deviations symmetrically, this exceeds the 5-point absolute-deviation threshold and results in High Tax-Rate Volatility.

Applying the normalized ETR to FY2024 pretax income produces an estimated normalized earnings impact of approximately -$10.47 billion. The negative result indicates that the elevated reported tax rate reduced continuing-operations earnings relative to the screening baseline.

Apple's FY2025 10-K confirms the FY2024 State Aid-related tax effect. The filing reports that the FY2025 income-tax provision declined by approximately $10.7 billion year over year due to the State Aid Decision. The analyst question is what tax-rate baseline is appropriate once the FY2024 distortion is separated from the underlying tax profile.

Meta: High Volatility with a Large FY2025 Tax Charge

Meta's ETR moved from 17.56% in FY2023 to 11.75% in FY2024 and then increased to 29.64% in FY2025. Its 17.89 percentage-point range exceeds the 10-point High-volatility threshold, while FY2025 also sits 12.08 percentage points above the normalized ETR.

Applying the 17.56% normalized rate to FY2025 produces an estimated normalized earnings impact of approximately -$10.38 billion. The elevated FY2025 ETR therefore reduced reported continuing-operations earnings relative to the normalization baseline.

Meta's FY2025 10-K separates the source of this result from its potential future tax-rate effects. The filing identifies a $15.93 billion one-time non-cash tax charge in Q3 2025, including a $14.03 billion valuation allowance against U.S. federal deferred tax assets.

That one-time charge should not itself be treated as a recurring tax expense. Separately, Meta's disclosures indicate that CAMT and share-based-compensation-related tax effects may influence future effective tax rates. Analysts should evaluate those potential ongoing effects separately from the FY2025 charge.

Together, the two cases demonstrate why tax-rate volatility and earnings direction should be reported separately. Both Apple and Meta qualify as High Tax-Rate Volatility, but that classification describes the magnitude of the tax-rate variation. The normalized earnings calculation separately shows whether a particular tax year supported or reduced reported earnings.

Monitoring Tax Normalization Across Future Reporting Periods

Once the baseline is established, the same workflow can be refreshed when a new completed annual period becomes available.

The update should not simply append the latest effective tax rate. It should recalculate the rolling three-year median ETR, the full ETR range, each year's absolute deviation from the normalized rate, and the normalized earnings impact using net income from continuing operations.

A recurring review can:

1. Retrieve the latest completed annual income statement.
2. Verify the tax bridge against net income from continuing operations.
3. Check that pretax income does not trigger the small-denominator safeguard.
4. Update the rolling three-year normalized ETR and volatility classification.
5. Recalculate whether the latest tax effect supported or reduced reported earnings.
6. Identify the fiscal year with the largest absolute ETR deviation.
7. Retrieve targeted filing evidence only when the volatility classification warrants further review.

Because the analysis relies primarily on completed annual tax data, there is little value in rerunning the full screen continuously. For a broader research universe, companies can instead be refreshed in small batches after annual filings become available. A consistent post-earnings data refresh process helps preserve the same calculation, validation, and Review Required rules across each update.

This creates a repeatable earnings-quality check without requiring analysts to reconstruct effective-tax-rate history manually for every company.

When an Analyst Needs to Step In

The tax-normalization screen can identify unusual effective-tax-rate patterns, but it should not decide automatically whether a tax effect is temporary, recurring, or structurally important.

Analyst review is required when:

• Pretax income is zero or negative.
• Pretax income is positive but less than 1% of revenue, creating a small denominator that can distort the effective tax rate.
• Net income from continuing operations is unavailable or differs materially from total net income.
• The tax bridge does not reconcile reliably.
• Income-tax expense is negative and represents a tax benefit.
• The effective tax rate moves sharply but the filing does not clearly explain the cause.
• A disclosed one-time charge or benefit needs to be separated from potentially recurring tax effects.
• Valuation allowances, tax-law changes, jurisdictional mix, tax credits, CAMT, or share-based-compensation effects may alter the appropriate forward tax-rate assumption.
• The three-year median no longer appears representative of the company's current tax profile.

Apple and Meta illustrate different follow-up questions.

For Apple, the FY2025 10-K confirms that the elevated FY2024 tax rate was affected by the State Aid-related charge. The analyst therefore needs to determine what underlying tax-rate assumption is appropriate once that distortion is separated from the historical series.

For Meta, the FY2025 10-K identifies the $15.93 billion Q3 2025 tax charge as one-time and non-cash. That charge should not itself be treated as a recurring expense. The analyst should separately evaluate whether CAMT, valuation-allowance treatment, share-based-compensation tax effects, or other disclosed factors could continue to influence Meta's effective tax rate.

Claude can identify where tax-rate volatility is unusually high and whether a particular tax year supported or reduced continuing-operations earnings. The analyst still determines which effects should be excluded, retained, or modeled separately in subsequent research.

From Tax Volatility to Cleaner Earnings Interpretation

Effective tax-rate volatility can materially change reported continuing-operations earnings even when the underlying pretax business trend is more stable.

Using FMP financial data through Claude MCP, this workflow establishes a company-specific normalized ETR, measures the magnitude and direction of annual tax-rate deviations, and estimates whether those tax effects supported or reduced continuing-operations earnings.

For Moderate- or High-volatility cases, targeted filing evidence helps distinguish disclosed one-time effects from factors that may remain relevant to future tax rates. When the underlying financial data, continuing-operations comparison, or pretax denominator is unreliable, the analysis returns Review Required rather than forcing a normalization conclusion.

The result is not a forecast of future tax rates. It is a repeatable review process that helps analysts identify when reported earnings may require additional tax normalization analysis before being carried into broader financial research. Teams that want to extend the method across a larger universe can apply the same controls when moving from an individual API test to a repeatable financial-data workflow.