Skip to main content
v2026.11,986 entries · CC-BY 4.0

Tax Data: Microdata, Statistical Metrics, and Empirical Methods in Economics

An authoritative guide to analyzing administrative tax data, calculating effective tax rates, and estimating the elasticity of taxable income in economics.

Written and maintained by CASRAI Editorial Board

Published

On this page

Tax data in empirical economics refers to the structured microdata and aggregate records compiled by revenue authorities from mandatory filings, which researchers analyze to measure economic income, simulate fiscal policy reforms, and quantify behavioral responses to taxation. In quantitative research, administrative tax records provide high-precision measures of earnings and deductions across the full income distribution, serving as the empirical foundation for estimating effective tax rates, progressivity indices, and the elasticity of taxable income.

Definition of tax data in empirical economics

In quantitative economic research and public finance, administrative tax data represents the official records collected by national and regional revenue administrations—such as the Internal Revenue Service (IRS) in the United States, HM Revenue and Customs (HMRC) in the United Kingdom, or national tax registries in Nordic nations. Unlike self-reported household surveys, which frequently suffer from recall errors, non-response, and severe top-coding among high-net-worth earners, administrative tax filings capture legally binding declarations of wages, business income, investment returns, and itemized deductions. Economists utilize both population-level administrative registers and stratified microdata samples to investigate income inequality, evaluate fiscal policy, and trace intergenerational mobility.

Tax data is classified into three primary structures depending on the research design: cross-sectional microdata, longitudinal panels, and geographic aggregations. Cross-sectional microdata, exemplified by the IRS Statistics of Income (SOI) Public Use File, provides detailed return-level entries for a representative sample of filers in a single tax year. Longitudinal panels link identical taxpayers across successive years through anonymized identifiers, allowing econometricians to observe income trajectories and distinguish between transitory earnings shocks and permanent behavioral shifts. Geographic aggregations compile income components into local areas, such as ZIP codes or counties, supporting spatial economics and place-based policy evaluation. When researchers analyze observational records rather than experimental trials, tax datasets serve as a vital repository of research data that minimizes measurement error across macroeconomic cycles.

Working with administrative filings requires clear classification of empirical inputs. Researchers treat income streams, deduction amounts, and tax credits as continuous variables, whereas filing statuses, filing schedules, and audit indicators function as categorical classifications. Understanding how continuous and categorical dimensions interact is essential for model specification; researchers often contrast these data structures as demonstrated in methodological comparisons of quantitative vs. categorical variables in econometric modeling.

Mathematical formulas for tax metrics

Empirical analyses of tax data rely on four core mathematical formulations: the Effective Tax Rate (ETR), the statutory Marginal Tax Rate (MTR), the Marginal Net-of-Tax Rate, and the Elasticity of Taxable Income (ETI). Each formula formalizes a distinct economic concept regarding tax burdens and behavioral incentives.

Effective Tax Rate (ETR)

The Effective Tax Rate measures the proportion of economic income absorbed by tax liabilities. Unlike statutory bracket rates, the ETR reflects the cumulative impact of deductions, exemptions, and non-refundable tax credits. Formally, for taxpayer i in tax year t:

ETRit = (Tit / Yit) × 100%

Where:

  • Tit represents the total assessed tax liability after all allowable non-refundable tax credits.
  • Yit represents the chosen economic income base. In empirical studies, Y is defined either as Adjusted Gross Income (AGI) or as Comprehensive Expanded Income (Yexp = AGI + Tax-Exempt Interest + Non-Taxable Social Security Benefits + Excluded Foreign Income).

Marginal Tax Rate (MTR) and Net-of-Tax Rate

The Marginal Tax Rate, denoted by τ, represents the incremental tax liability incurred on the next dollar of earned income:

τit = ΔTit / ΔYit = [T(Yit + ΔY) – T(Yit)] / ΔY

Where ΔY represents a small positive income increment (conventionally $10 or $100 in microsimulation packages such as NBER TAXSIM). The Marginal Net-of-Tax Rate—also known as the net-of-tax share or keep rate—is defined as:

Keep Rate = 1 – τit

The net-of-tax rate reflects the marginal incentive to earn: an individual facing a 35% marginal tax rate retains 65 cents of every additional dollar earned.

Elasticity of Taxable Income (ETI)

The Elasticity of Taxable Income measures the percentage change in reported taxable income resulting from a one-percent change in the marginal net-of-tax share. Grounded in the public finance framework established by Martin Feldstein and synthesized by Emmanuel Saez, Joel Slemrod, and Seth Giertz (2012), the ETI serves as a sufficient statistic for the efficiency costs and deadweight loss of income taxation. The point elasticity e is defined as:

e = ∂ ln(Y) / ∂ ln(1 – τ) = [dY / Y] / [d(1 – τ) / (1 – τ)]

In empirical panel econometrics, researchers estimate the elasticity using the canonical Gruber-Saez (2002) difference specification across two periods (t and t+k):

Δ ln(Yit) = e · Δ ln(1 – τit) + η · Δ ln(Yit – T(Yit)) + Xit‘β + εit

Where:

  • Δ ln(Yit) is the change in the natural logarithm of taxable income between period t and period t+k: ln(Yi,t+k) – ln(Yi,t).
  • e is the substitution elasticity of taxable income with respect to the net-of-tax share.
  • η represents the income effect parameter, capturing changes in virtual after-tax income: ln(Yi,t+k – Ti,t+k) – ln(Yi,t – Ti,t).
  • Xit is a vector of time-varying demographic and economic covariates (e.g., age, marital status, baseline income spline controls).
  • εit represents the idiosyncratic econometric error term.

Worked example of tax rate and elasticity calculation

To illustrate how these formulas apply to administrative microdata, consider a concrete worked example using individual income tax parameters. Assume an empirical study tracks a single tax filer (Taxpayer A) before and after a statutory tax reform.

Baseline Period (t = 0)

In the baseline year, Taxpayer A reports the following financial items on Form 1040:

  • Gross Wage Income: $120,000.00
  • Standard Deduction: $14,600.00
  • Taxable Income (Y0): $120,000.00 – $14,600.00 = $105,400.00

The statutory progressive tax schedule for a single filer consists of four marginal brackets:

  • 10% on taxable income from $0 to $11,600: $11,600 × 0.10 = $1,160.00
  • 12% on taxable income from $11,601 to $47,150: ($47,150 – $11,600) × 0.12 = $35,550 × 0.12 = $4,266.00
  • 22% on taxable income from $47,151 to $100,525: ($100,525 – $47,150) × 0.22 = $53,375 × 0.22 = $11,742.50
  • 24% on taxable income from $100,526 to $105,400: ($105,400 – $100,525) × 0.24 = $4,875 × 0.24 = $1,170.00

Calculating total tax liability and baseline metrics:

Total Tax Liability (T0) = $1,160.00 + $4,266.00 + $11,742.50 + $1,170.00 = $18,338.50

Statutory Marginal Tax Rate (τ0) = 24.00% (0.24)

Baseline Net-of-Tax Rate (1 – τ0) = 1.00 – 0.24 = 0.7600 (76.00%)

Effective Tax Rate on Gross Income = ($18,338.50 / $120,000.00) × 100% = 15.28%

Effective Tax Rate on Taxable Income = ($18,338.50 / $105,400.00) × 100% = 17.40%

Policy Reform and Behavioral Response Simulation (t = 1)

Suppose a statutory tax reform reduces the fourth bracket marginal tax rate from 24.00% to 21.00% while keeping bracket boundaries and prior rates unchanged. We evaluate the predicted behavioral response using an empirical elasticity of taxable income of e = 0.25 (a standard mid-range estimate from Saez, Slemrod, and Giertz 2012):

1. New Marginal Tax Rate (τ1): 21.00% (0.21)

2. New Net-of-Tax Rate (1 – τ1): 1.00 – 0.21 = 0.7900 (79.00%)

3. Proportional Change in Net-of-Tax Rate (Δ ln(1 – τ)):

Δ ln(1 – τ) = ln(0.7900) – ln(0.7600) = -0.23572 – (-0.27444) = +0.03872 (approx. +3.87% increase in the marginal keep rate)

4. Predicted Log Change in Taxable Income:

Δ ln(Y) = e · Δ ln(1 – τ) = 0.25 × 0.03872 = +0.00968 (approx. +0.97% increase)

5. New Predicted Taxable Income (Y1):

Y1 = Y0 × exp(0.00968) = $105,400.00 × 1.00973 = $106,425.22

The model predicts that Taxpayer A will report an additional $1,025.22 in taxable income due to reduced deduction shifting, timing adjustments, and marginal labor effort. Recomputing tax liability under the reformed rules:

Brackets 1 to 3 remain unchanged ($1,160.00 + $4,266.00 + $11,742.50 = $17,168.50). Income in the reformed 21% bracket is ($106,425.22 – $100,525.00) = $5,900.22 × 0.21 = $1,239.05.

New Total Tax Liability (T1) = $17,168.50 + $1,239.05 = $18,407.55

New Effective Tax Rate on Gross Income = ($18,407.55 / $120,000.00) × 100% = 15.34%

Interpretation of tax metrics

Interpreting empirical tax metrics requires distinguishing between distributional incidence and marginal behavioral incentives. The Effective Tax Rate measures vertical equity—how tax burdens are distributed across economic strata. In contrast, the Marginal Tax Rate dictates deadweight loss, as economic theory establishes that the excess burden of taxation increases quadratically with the marginal tax rate.

The difference between the marginal tax rate and the effective tax rate is termed the progressivity wedge (MTR – ETR). In a progressive tax system, the marginal rate consistently exceeds the effective rate across all positive brackets. A wider progressivity wedge indicates steeper progression: the tax code imposes higher marginal disincentives on incremental effort while maintaining lower average tax burdens on inframarginal income.

The table below provides reading guidance for empirical tax metrics across income quintiles in standardized administrative tax datasets:

Income Strata Typical ETR Range Typical MTR Range Typical Progressivity Wedge Empirical ETI Estimate (e) Primary Behavioral Driver
Bottom Quintile (0-20%) -10.0% to 3.5% 0.0% to 15.0% Negative to +5.0% 0.05 to 0.15 Extensive margin labor supply (EITC phase-in/phase-out)
Second Quintile (20-40%) 3.0% to 8.5% 10.0% to 15.0% +4.0% to +8.0% 0.08 to 0.18 Secondary earner labor participation
Middle Quintile (40-60%) 8.0% to 13.0% 12.0% to 22.0% +4.0% to +10.0% 0.10 to 0.20 Standard vs itemized deduction thresholds
Fourth Quintile (60-80%) 12.5% to 18.0% 22.0% to 24.0% +6.0% to +9.5% 0.15 to 0.28 Retirement contributions and timing adjustments
Top Decile (90-99%) 20.0% to 26.0% 32.0% to 35.0% +9.0% to +14.0% 0.25 to 0.45 Charitable deductions, pass-through structuring
Top 1 Percent (99-100%) 24.0% to 31.0% 37.0% to 40.8% +10.0% to +16.0% 0.40 to 0.80 Capital gain realizations, income shifting across tax bases

As shown in the reading guidance, empirical elasticities rise markedly at the top of the income distribution. Lower- and middle-income filers possess limited scope to alter their tax base because third-party information reporting (W-2 forms) verifies wage income. High-income earners, conversely, derive substantial income from capital gains, partnerships, and S-corporations, where legal avoidance strategies and intertemporal income shifting generate larger observed elasticities.

Assumptions and common empirical mistakes

Researchers working with administrative tax data must navigate several methodological pitfalls that threaten econometric validity. Five common mistakes frequently compromise empirical tax studies:

  1. Conflating statutory rates with effective tax burdens: A frequent error is attributing changes in economic behavior solely to statutory headline rates without accounting for the tax base. Broadening the tax base through capping deductions can increase effective tax rates even while statutory rates fall. Empirical models must always compute effective rates based on standardized economic income.
  2. Attributing taxable income changes purely to labor supply: Early public finance studies interpreted high ETI estimates as proof that high taxes sharply reduce work effort. As established by Saez, Slemrod, and Giertz (2012), changes in taxable income reflect a composite response: real labor supply, administrative tax evasion, and legal tax avoidance (e.g., deferring capital gains or shifting corporate profits). Conflating avoidance with labor supply leads to flawed policy recommendations regarding economic productivity.
  3. Neglecting sampling weights in public-use microdata: In files like the IRS SOI Public Use File, high-income filers are heavily oversampled to ensure statistical reliability. Analyzing unweighted microdata introduces severe selection bias, skewing sample means and regression coefficients toward high-wealth behaviors. Econometricians must apply sampling weights (such as XTOTWT) in all population-level estimations.
  4. Ignoring non-filer selection and sample attrition: Administrative tax registers omit individuals whose gross income falls below mandatory filing thresholds. In longitudinal panels, individuals who experience severe unemployment or exit the formal labor force drop out of the tax register, generating non-random attrition. Researchers studying poverty or low-income programs must combine tax microdata with survey records or administrative social assistance files to address sample truncation.
  5. Failing to address bracket endogeneity: In a progressive tax system, taxable income and the marginal tax rate are simultaneously determined: an unexpected positive income shock pushes a taxpayer into a higher marginal bracket (bracket creep), introducing mechanical positive correlation between income and tax rates. To resolve this endogeneity, researchers construct simulated instruments—such as calculating the marginal tax rate using baseline income inflated by general economic growth rather than actual realized income—using microsimulation engines like TAXSIM.

Understanding these assumptions is essential when researchers evaluate large administrative cohorts. In observational research designs, such as those evaluated in comparative analyses of RCTs vs. observational studies, controlling for endogenous policy exposure through instrumental variables is vital for causal inference.

How to report tax metrics in empirical research

Empirical papers reporting tax metrics must follow rigorous reporting standards aligned with the American Psychological Association (APA 7th edition) guidelines and standard econometric practice. Authors must explicitly report the administrative data source, the sample restriction rules, the exact definitions of income and tax liability, the estimation methodology, and complete test statistics.

Below is a standard APA-style reporting template for an instrumental variables regression estimating the elasticity of taxable income:

“An instrumental variables regression using the Gruber-Saez two-stage panel specification was estimated on a longitudinal sample of administrative tax filers (N = 14,822) observed across the policy reform window. To address bracket endogeneity, the log change in the marginal net-of-tax share was instrumented using a simulated policy instrument evaluated at baseline income. The estimated substitution elasticity of taxable income with respect to the net-of-tax rate was statistically significant, e = 0.24, SE = 0.06, t(14820) = 4.00, p ≤ 0.001, 95% CI [0.12, 0.36]. The income effect parameter was small and statistically indistinguishable from zero, η = -0.02, SE = 0.03, t(14820) = -0.67, p = .503. These results indicate that a 10% increase in the marginal keep rate generates a 2.4% expansion in reported taxable income, driven primarily by deduction adjustments rather than changes in broad labor earnings.”

When presenting summary tables of effective tax rates across cohorts, authors should present means alongside weighted standard errors, sample counts, and exact quartile boundaries to allow cross-study comparisons.

How to calculate tax metrics in Excel, R, and Python

The following verified code snippets illustrate how to calculate the Effective Tax Rate, Marginal Tax Rate, and behavioral elasticity response in Microsoft Excel, R, and Python.

Excel Formula Implementation

Suppose an Excel worksheet contains gross income in cell A2 ($120,000) and assessed tax liability in cell B2 ($18,338.50):

To calculate the Effective Tax Rate (cell C2):

=B2 / A2

Format cell C2 as a percentage with two decimal places (displays 15.28%).

To calculate the behavioral percentage change in taxable income given an elasticity in cell D2 (0.25), baseline net-of-tax rate in E2 (0.76), and reformed net-of-tax rate in F2 (0.79):

=D2 * (LN(F2) - LN(E2))

Format as a percentage (displays 0.97%).

R Implementation

The following R script calculates effective tax rates, computes the progressivity wedge, and estimates the predicted change in taxable income across a sample data frame:

# R Script: Tax Data Metrics Calculation
library(dplyr)

# Sample microdata frame
tax_data = data.frame(
  taxpayer_id = 101:105,
  gross_income = c(45000, 78000, 120000, 250000, 850000),
  taxable_income = c(30400, 63400, 105400, 235400, 835400),
  tax_liability = c(3408, 9317, 18338.5, 52835, 275835),
  marginal_rate = c(0.12, 0.22, 0.24, 0.35, 0.37)
)

# Calculate ETR, Net-of-Tax Rate, and Progressivity Wedge
tax_metrics = tax_data |>
  mutate(
    etr_gross = (tax_liability / gross_income) * 100,
    etr_taxable = (tax_liability / taxable_income) * 100,
    net_of_tax_rate = 1 - marginal_rate,
    progressivity_wedge = (marginal_rate * 100) - etr_gross
  )

# Simulate behavioral response for e = 0.25 under a 2 percentage point MTR reduction
eti = 0.25
tax_metrics = tax_metrics |>
  mutate(
    reformed_mtr = marginal_rate - 0.02,
    reformed_notr = 1 - reformed_mtr,
    d_ln_notr = log(reformed_notr) - log(net_of_tax_rate),
    pct_change_income = eti * d_ln_notr * 100,
    simulated_taxable_income = taxable_income * exp(eti * d_ln_notr)
  )

print(tax_metrics[, c("taxpayer_id", "etr_gross", "marginal_rate", "pct_change_income")])

Python Implementation

The following Python script utilizes pandas and numpy to compute tax metrics and simulate elasticity adjustments from microdata arrays:

# Python Script: Tax Microdata Calculation
import numpy as np
import pandas as pd

data = {
    'filer_id': [101, 102, 103, 104, 105],
    'gross_income': [45000.0, 78000.0, 120000.0, 250000.0, 850000.0],
    'taxable_income': [30400.0, 63400.0, 105400.0, 235400.0, 835400.0],
    'tax_liability': [3408.0, 9317.0, 18338.5, 52835.0, 275835.0],
    'marginal_rate': [0.12, 0.22, 0.24, 0.35, 0.37]
}

df = pd.DataFrame(data)

# Compute Effective Tax Rates and Net-of-Tax Share
df['etr_gross'] = (df['tax_liability'] / df['gross_income']) * 100.0
df['etr_taxable'] = (df['tax_liability'] / df['taxable_income']) * 100.0
df['net_of_tax_rate'] = 1.0 - df['marginal_rate']
df['progressivity_wedge'] = (df['marginal_rate'] * 100.0) - df['etr_gross']

# Behavioral simulation: ETI = 0.25 under a 2 percentage point MTR tax cut
eti = 0.25
df['reformed_mtr'] = df['marginal_rate'] - 0.02
df['reformed_notr'] = 1.0 - df['reformed_mtr']
df['d_ln_notr'] = np.log(df['reformed_notr']) - np.log(df['net_of_tax_rate'])
df['pred_pct_change_income'] = eti * df['d_ln_notr'] * 100.0
df['simulated_taxable_income'] = df['taxable_income'] * np.exp(eti * df['d_ln_notr'])

for idx, row in df.iterrows():
    print(f"Filer {int(row['filer_id'])}: Gross ETR={row['etr_gross']:.2f}%, MTR={row['marginal_rate']*100:.1f}%, Predicted Income Change=+{row['pred_pct_change_income']:.2f}%")

Frequently asked questions

What is the difference between tax data and survey income data?

Administrative tax data consists of legally mandated filings submitted to revenue authorities, whereas survey income data relies on voluntary household responses to questionnaires. Tax data offers near-complete coverage, exact dollar entries, and robust sampling at the top of the income distribution, but lacks detailed demographic variables (such as education and race). Survey data provides richer sociological characteristics but suffers from attrition, non-response, and severe under-reporting of high-income capital assets.

How do researchers access administrative tax microdata?

Access pathways depend on the confidentiality tier of the data. De-identified public-use files (such as the IRS SOI Public Use File) can be downloaded directly from agency repositories or university data archives. Confidential, population-level microdata requires formal research proposals, background clearances, and secure execution within Federal Statistical Research Data Centers (FSRDC) or agency partnerships like the IRS Joint Statistical Research Program. To prevent taxpayer re-identification, public-use records undergo extensive masking, as described in methodological standards for digitally anonymised data.

Why is NBER TAXSIM so widely used in empirical economics?

NBER TAXSIM is a standardized microsimulation engine that calculates federal and state income tax liabilities and marginal tax rates from microdata inputs dating back to 1960. Researchers use TAXSIM because calculating marginal rates by hand across 50 state tax systems and decades of legislative changes is error-prone. TAXSIM enables econometricians to construct simulated policy instruments that isolate exogenous legislative tax shifts from endogenous taxpayer behavior.

How do economists use tax data to measure intergenerational mobility?

Longitudinal tax microdata allows researchers to link children to their parents via dependent claims on tax returns over multiple decades. By comparing the adult earnings percentile of children to the earnings percentile of their parents at a similar age, economists calculate the rank-rank slope of relative mobility and evaluate neighborhood opportunity, as detailed in comprehensive analyses of intergenerational mobility using IRS records.

What does the elasticity of taxable income measure?

The Elasticity of Taxable Income measures the percentage change in reported taxable income induced by a one-percent increase in the marginal net-of-tax rate (1 – τ). It captures the total behavioral response to tax changes—combining labor supply adjustments, changes in business form, intertemporal timing shifts, and tax deduction avoidance. In public finance, the ETI serves as the central parameter for calculating deadweight loss and optimal top income tax rates.

What is the progressivity wedge?

The progressivity wedge is the numerical difference between a taxpayer’s marginal tax rate and their effective tax rate (MTR – ETR). In a progressive tax system, this wedge is positive because earlier dollars of income are taxed at lower marginal bracket rates or sheltered by standard deductions. A wider wedge indicates a more steeply progressive schedule, generating stronger marginal behavioral disincentives relative to the average tax burden.

Sources

  • Internal Revenue Service. (2024). Statistics of Income: Individual Income Tax Returns Complete Report (Publication 1304). IRS Statistics of Income Division. https://www.irs.gov/statistics/soi-tax-stats-individual-income-tax-returns (accessed 2026-11-14).
  • Feenberg, D., and Coutts, E. (1993). An Introduction to the TAXSIM Model. Journal of Policy Analysis and Management, 12(1), 189–194. National Bureau of Economic Research. https://www.nber.org/taxsim (accessed 2026-11-14).
  • Saez, E., Slemrod, J., and Giertz, S. H. (2012). The Elasticity of Taxable Income with Respect to Marginal Tax Rates: A Critical Review. Journal of Economic Literature, 50(1), 3–50. https://www.aeaweb.org/articles?id=10.1257/jel.50.1.3 (accessed 2026-11-14).
  • Gruber, J., and Saez, E. (2002). The Elasticity of Taxable Income: Evidence and Implications. Journal of Public Economics, 84(1), 1–32. https://doi.org/10.1016/S0047-2727(01)00085-8 (accessed 2026-11-14).
  • Piketty, T., Saez, E., and Zucman, G. (2018). Distributional National Accounts: Methods and Estimates for the United States. Quarterly Journal of Economics, 133(2), 553–609. https://academic.oup.com/qje/article/133/2/553/4430651 (accessed 2026-11-14).
  • Organisation for Economic Co-operation and Development. (2024). OECD Tax Database: Explanatory Annex and Methodology. OECD Centre for Tax Policy and Administration. https://www.oecd.org/tax/tax-policy/tax-database/ (accessed 2026-11-14).

Data: Major Administrative Tax Microdata Resources for Empirical Economic Research

A comparative cross-compilation of major administrative tax datasets and microsimulation systems used in empirical economics, detailing administering bodies, population coverage, sampling structure, access tiers, and primary empirical use cases.

Period covered: 1960/2026. Coverage: United States and OECD Economies.

Major Administrative Tax Microdata Resources for Empirical Economic Research
Dataset SystemAdministering BodyPopulation CoverageSampling DesignAccess GovernancePrimary Econometric Applications
IRS Statistics of Income Public Use File (SOI PUF)Internal Revenue Service (IRS)United States individual tax filers (Form 1040 series)Stratified probability sample with oversampling of high-income and high-wealth returnsDe-identified public use file (subsampled, blurred continuous variables, top-coded)Tax liability distribution, deduction utilization, and cross-sectional tax incidence
NBER TAXSIM Microdata ModelNational Bureau of Economic Research (NBER)United States federal and 50 state tax jurisdictions (1960-present)User-supplied microdata input files or survey samples (CPS, SIPP, PSID)Open academic web service, Stata ado package, and local binary executablesMarginal tax rate calculation, policy counterfactuals, and labor supply elasticity instruments
Opportunity Insights Tax PanelOpportunity Insights / Harvard UniversityDe-identified U.S. population tax records matched across generations (1989-present)Near-universe longitudinal population register of tax filers and dependentsAggregated public statistics and restricted FSRDC project accessIntergenerational income mobility, neighborhood causal effects, and higher education mobility
Longitudinal Administrative Databank (LAD)Statistics CanadaCanadian tax filers and non-filing spouses/dependents (T1 Family File)20% longitudinal random sample linked annually across years (1982-present)Restricted Research Data Centres (RDC) under Statistics Act vettingLife-cycle earnings dynamics, immigrant economic integration, and retirement savings
HMRC Datalab Administrative MicrodataHM Revenue and Customs (United Kingdom)United Kingdom personal (Self Assessment, PAYE) and corporate tax recordsAdministrative universe of tax records with anonymized research extractionsSecure physical and virtual datalab enclaves for approved researchersFirm investment behavior, corporate tax avoidance, and wealth distribution
OECD Taxing Wages Harmonized DatasetOrganisation for Economic Co-operation and Development (OECD)38 OECD member nations and partner economiesHarmonized microsimulation models across standardized family archetypesPublicly downloadable online database and SDMX API feedsInternational tax wedge comparisons, average vs marginal rate benchmarking, and cross-country progressivity

Follow CASRAI

Research-administration guidance, standards updates and independent tool reviews.

Ask CASRAI · free to try

Ask about Tax Data: Microdata, Statistical Metrics, and Empirical Methods in Economics

2 free questions a day, no account.

An AI assistant for research: funder rules, compliance, methods and publishing. It links the sources behind its answers and labels answers that come from a web search.

Answers draw on CASRAI's guides and dictionary plus the federal and funder documents we index: Federal Register, Grants.gov, Regulations.gov and UKRI.

Works on this site and inside Claude, Cursor and the AI tools you already use.

Everything CASRAI publishes — this page, the dictionary, the guides and the news — stays open to read, with no account and no card.

Referenced across the research world

University of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logoUniversity of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logo
  • University of Cambridge logo
  • Columbia University logo
  • Crossref logo
  • University of Edinburgh logo
  • Harvard University logo
  • University of Oxford logo
  • Princeton University logo
  • Stanford School of Medicine logo
  • University College London logo
  • ORCID logo

View CASRAI adoption →