Written and maintained by CASRAI Editorial Board
Last updated
Econometrics is the field that applies statistical and mathematical methods to economic
data in order to test economic theories, estimate the size of economic relationships, and
forecast economic outcomes. It sits directly between two parent disciplines: it borrows its
core toolkit — regression, estimation theory, hypothesis testing — from statistics,
and it applies that toolkit to the substantive questions of economics — how
much a minimum-wage increase changes employment, how responsive consumer demand is to price,
whether a training program actually raised participants’ earnings. Almost every empirical claim
in modern economics — a published estimate, a policy-impact number, a forecast — rests on an
econometric method somewhere underneath it.
This guide answers “what is econometrics” in real depth: its core questions and methods, how
it relates to neighboring disciplines, its major subfields, and — because this page is published
by CASRAI, a research-administration standards body — the
funding landscape, research tools, and career pathways that matter to anyone conducting or
administering econometrics research specifically.
What Is Econometrics? A Working Definition
Econometrics is usually split into two complementary halves:
- Theoretical econometrics develops and evaluates the statistical methods
themselves — proving that an estimator is consistent, deriving the correct standard errors for a
given data structure, or working out how an estimator behaves when a modeling assumption fails. - Applied econometrics uses established methods to answer a substantive
empirical question — estimating a demand elasticity, testing whether a policy changed an
outcome, or forecasting next quarter’s inflation rate.
What distinguishes econometrics from general statistics is less the mathematics itself than
the character of the data it has to work with. Economic data is almost always
non-experimental — researchers cannot randomly assign a country a monetary
policy or randomly assign a worker a wage the way a lab scientist can randomize a treatment.
That single constraint drives most of the field’s distinctive methodology: because researchers
usually cannot run a controlled experiment, econometrics has developed an extensive set of tools
for extracting causal relationships from observational data — instrumental variables, panel-data
methods, and quasi-experimental designs like difference-in-differences
and regression discontinuity design — and
for diagnosing when a straightforward regression will mislead, through concepts like
endogeneity, omitted-variable bias, and
simultaneity (where two variables determine each other jointly, such as price and quantity in a
market). Economic data also tends to have structure statistics more generally treats as a special
case rather than the default: it usually arrives as a time series (the same
variable observed over successive periods, with values correlated across time) or as
panel data (the same units — people, firms, countries — observed repeatedly over
time), both of which violate the independence assumptions that much of introductory statistical
theory is built on.
How Econometrics Relates to Neighboring Disciplines
- Economics. Econometrics is economics’ primary empirical toolkit and, in
most universities, its home department — the large majority of econometrics faculty and PhD
training happens inside economics departments, not standalone econometrics programs. See
CASRAI’s guide to economics for the broader discipline
econometrics serves. - Statistics. Econometrics shares its mathematical foundation — regression,
maximum-likelihood estimation, hypothesis testing, Bayesian inference — with general statistics,
and many methods (robust standard errors, generalized method of moments) were developed by
researchers moving fluidly between the two fields. Econometrics extends that shared foundation
with methods purpose-built for non-experimental, time-dependent, and panel-structured economic
data. See CASRAI’s guide to statistics. - Mathematics. Econometric theory is built on linear algebra (matrix
regression, projection), calculus and real analysis (asymptotic theory), and probability theory
(distribution theory for estimators and test statistics) — a PhD-level econometrics sequence
leans on all three. - Game theory. A specific branch of econometrics — structural econometrics —
estimates the parameters of explicit game-theoretic models of behavior (how firms set prices
under competition, how bidders behave in an auction) directly from observed data, rather than
just testing a game-theoretic prediction’s qualitative direction. See CASRAI’s guide to
game theory for the theoretical models this branch of
econometrics estimates. - Actuarial science. Actuarial science and econometrics both build
statistical models of risk and uncertainty over time — survival/hazard models, time-series
forecasting — from overlapping mathematical foundations, applied to different domains (insurance
and pension risk versus economic relationships more broadly). See CASRAI’s guide to
actuarial science. - Data science and computer science. Modern applied econometrics
increasingly overlaps with machine learning — using flexible, algorithmic prediction methods
alongside, or as an input into, causal-inference designs (sometimes called “causal machine
learning”). See CASRAI’s guide to data science.
Major Subfields and Branches of Econometrics
Econometrics is commonly organized along a few overlapping axes — by the structure of the
data, by the estimation framework, and by the purpose of the analysis:
- Microeconometrics — analyzes individual-, household-, or firm-level data,
often using discrete-choice models (for outcomes like “buy or don’t buy,” “employed or not”) and
treatment-effect methods to evaluate specific interventions or policies. - Time-series econometrics (macroeconometrics) — analyzes data observed
repeatedly over time (GDP, inflation, exchange rates), modeling how a variable depends on its own
past and forecasting future values. See CASRAI’s guide to
ARIMA models for research time
series. - Panel-data econometrics — combines cross-sectional and time-series data
(the same units observed over multiple periods), using fixed- or random-effects estimators to
control for unobserved, time-invariant differences across units. See CASRAI’s guides to the
Hausman test for fixed vs. random
effects and mixed-effects
models. - Financial econometrics — models asset prices, returns, and volatility over
time, an area with close ties to quantitative finance. - Spatial econometrics — extends standard methods to account for spatial
dependence, where outcomes in one location are correlated with outcomes in neighboring
locations. - Bayesian econometrics — estimates economic models within a Bayesian
inference framework (combining prior beliefs with observed data) rather than the classical
frequentist framework most introductory econometrics teaches first. - Causal inference / program evaluation — the applied toolkit for estimating
the causal effect of a specific policy or intervention from non-experimental data: randomized
controlled trials where feasible, and quasi-experimental designs — instrumental variables,
difference-in-differences,
regression discontinuity design, and its
fuzzy variant — where it is not. - Structural econometrics — estimates the underlying parameters of an
explicit economic (often game-theoretic) model directly, rather than a reduced-form statistical
relationship, so the estimated model can be used to simulate counterfactual policies.
Who Funds Econometrics Research
Econometrics research is funded through a mix of federal science agencies, mission-driven
institutions, and a smaller number of private foundations — understanding this landscape matters
for research administrators supporting econometrics faculty and PhD students:
- National Science Foundation (NSF). The Economics Program,
one of the core programs within NSF’s Division of Social and Economic Sciences (SES) — part of
the Directorate for Social, Behavioral and Economic Sciences (SBE) — is the primary federal
funder of basic econometric research tied to economic applications. SES’s
Methodology, Measurement and Statistics (MMS) program funds the development of
statistical and measurement methods used across the social and behavioral sciences, including
econometric methodology specifically, and NSF’s Division of Mathematical Sciences (DMS)
Statistics program, in NSF’s Mathematical and Physical Sciences directorate, funds some
theoretical work at the boundary between statistics and econometrics. - The Federal Reserve System. The Federal Reserve Board and the twelve
regional Federal Reserve Banks employ economists who produce a substantial share of applied
macroeconometric and monetary-economics research in the US, published through in-house
working-paper series rather than a competitive external grant program. - Research organizations. The National Bureau of Economic Research
(NBER), a private nonprofit, is the discipline’s central hub for disseminating
econometric and economic working papers and organizing collaborative research programs; it is
not primarily a grant-making body itself, though some NBER-affiliated research programs
administer specific funded initiatives. - Private foundations. The Alfred P. Sloan Foundation
maintains a dedicated Economics program area focused on the economics of science, technology,
and labor markets, which includes methodological econometrics work in that space. Foundation
priorities and eligible topics shift over time — research offices should confirm current scope
and deadlines directly on each funder’s own site before advising a PI, rather than relying on a
general overview like this one.
Research Methods and Tools in Econometrics
At a general level, econometric research relies on:
- Regression-based estimation — ordinary least squares (OLS) as the baseline
method, extended with heteroscedasticity-robust
standard errors and cluster-robust standard
errors to keep inference valid when the classical assumptions of constant error variance or
independent observations don’t hold, and diagnosed for heteroscedasticity with tests such as the
Breusch-Pagan test. - Instrumental-variables (IV) estimation — used when a regressor is
suspected to be endogenous (correlated with the error term), isolating variation in that
regressor that comes from a source unrelated to the outcome except through the regressor
itself. - Panel-data estimators — fixed-effects and random-effects models for data
that repeatedly observes the same units, with the choice between them typically guided by a
Hausman test. - Time-series methods — autoregressive and moving-average models (ARIMA),
vector autoregressions (VAR), and tests for stationarity, used to model and forecast variables
observed sequentially over time. - Discrete-choice and limited-dependent-variable models — probit, logit, and
related models for outcomes that are binary, ordered, or censored rather than continuous. - Software. Applied econometricians commonly work in Stata, R (with packages
such as plm and fixest for panel data), Python (statsmodels, linearmodels), and
EViews or MATLAB for specialized time-series and structural work. - Data sources. Government administrative records and household surveys,
firm- and market-level microdata, and financial-market data are the typical raw material,
increasingly supplemented by proprietary data from private firms and digital platforms under
data-use agreements.
Career and Training Pathways
Econometrics training is usually embedded within, rather than separate from, broader economics
graduate education:
- Undergraduate. A bachelor’s degree in economics or a joint economics/
statistics program typically includes an introductory econometrics course covering regression and
basic inference, alongside calculus and probability/statistics prerequisites. - Doctoral training. A PhD with an econometrics specialization is typically
earned inside an economics department: the first one to two years cover a core econometrics
theory sequence (alongside microeconomic and macroeconomic theory) culminating in qualifying
exams, followed by original dissertation research. Some students instead pursue econometric
methodology through a statistics or biostatistics PhD program, particularly for theoretical or
methods-focused work. - Professional societies. The Econometric Society, an
international academic society founded in 1930, is the discipline’s principal professional home,
publishing the flagship journal Econometrica and focused specifically on the
intersection of economic theory with statistics and mathematics. Econometricians are also
frequently active in the broader American Economic Association (AEA), given how
closely the two fields are institutionally linked. - Career destinations. Econometricians work across academic economics and
statistics departments, central banks and finance ministries (e.g., the Federal Reserve System,
the U.S. Treasury), international organizations (the World Bank, the IMF), government statistical
agencies, and — increasingly — the private sector, particularly quantitative finance, technology
companies’ causal-inference and experimentation teams, and economic-consulting firms.
Frequently Asked Questions
What is the difference between econometrics and statistics?
Econometrics applies the general mathematical toolkit of statistics — regression, estimation,
hypothesis testing — specifically to economic data, and has developed its own specialized methods
(instrumental variables, panel-data estimators, time-series techniques) to handle the
non-experimental, time-dependent nature of that data. Statistics is the broader parent
discipline; econometrics is its economics-focused branch.
What is the difference between econometrics and economics?
Economics is the substantive social science that studies how resources are allocated and how
markets, institutions, and policy shape economic outcomes; econometrics is the statistical
methodology economists use to test those theories and estimate the size of economic relationships
from real-world data. Almost every empirical economics paper uses econometric methods, but
econometrics itself is a distinct methodological field with its own theory and, often, its own
graduate coursework.
Do you need a PhD to work in econometrics?
Applied econometric roles in industry (finance, technology, economic consulting) are often
open to strong master’s-level candidates with solid regression and causal-inference skills; a PhD
is typically required for academic research positions, roles focused on developing new
econometric methodology, and many senior research-economist positions at central banks and
international organizations.
What software do econometricians use?
Stata and R are the most common tools in academic econometrics; Python has grown quickly for
applied and industry work, particularly where econometric methods are combined with machine
learning; EViews and MATLAB remain common for specialized time-series and structural modeling
work.
What math is required for econometrics?
Undergraduate econometrics typically requires calculus, linear algebra, and introductory
probability/statistics; PhD-level econometric theory adds real analysis, matrix algebra, and
asymptotic theory, since the proofs behind most estimators’ properties (consistency, efficiency,
asymptotic normality) are built on that mathematical foundation.
Related CASRAI Resources
This guide is part of CASRAI’s Branches of Science
series, which maps the major academic and scientific disciplines along with the
research-administration context — funding, methods, and career pathways — that a general
encyclopedia entry typically omits. Econometrics sits directly between
economics and
statistics, and overlaps closely with
game theory and
actuarial science. For the specific quantitative
methods covered in more depth elsewhere on CASRAI, see the guides on
difference-in-differences,
regression discontinuity design,
endogeneity, and the
Hausman test.
