econometrics

econometrics
Name	Econometrics
Founded	Early 20th century
Key people	Ragnar Frisch, Jan Tinbergen, Lawrence Klein, Tjalling Koopmans
Fields	Economics, Statistics

econometrics is the application of statistical and mathematical methods to economic data in order to give empirical content to economic relationships and to test or develop hypotheses. It serves as a bridge between abstract economic models and the real-world economic phenomena they seek to explain, providing tools for estimation, inference, and forecasting. The field is fundamental to both academic research in economics and practical decision-making in institutions like the Federal Reserve, World Bank, and private financial institutions.

Overview

The primary goal is to quantify economic relationships, moving beyond qualitative theory to provide numerical estimates of parameters like price elasticity or the marginal propensity to consume. It relies heavily on probability theory and statistical inference to draw conclusions from often imperfect and observational economic data. Pioneers such as Ragnar Frisch, who coined the term, and Jan Tinbergen laid its foundations, with later advancements coming from figures like Lawrence Klein and Tjalling Koopmans. The discipline is central to the work of major economic bodies, including the International Monetary Fund and the European Central Bank, and is used to evaluate policies from the New Deal to modern monetary policy.

Methodology

The standard methodology, formalized by pioneers like Tjalling Koopmans and debated by Milton Friedman, often follows a structure of model specification, estimation, and validation. Specification involves defining a mathematical model based on economic theory, such as a supply and demand system. Estimation, frequently performed using software like Stata or R (programming language), employs techniques like ordinary least squares to calculate parameter values from datasets, which may come from sources like the Bureau of Labor Statistics or the World Bank. Validation involves testing the model's assumptions using diagnostic checks for issues like heteroskedasticity or autocorrelation, concepts rigorously explored by scholars like David Hendry.

Models and techniques

A core set of models and techniques has been developed to handle various data structures and research questions. The linear regression model is the fundamental workhorse, extended into frameworks like generalized linear models. For analyzing time-ordered data, time series analysis methods, including autoregressive integrated moving average models, are essential, with foundational work by George Box and Gwilym Jenkins. For data where variables are determined simultaneously, simultaneous equations models and methods like two-stage least squares, developed by Henri Theil and others, are used. More recent computational advances have popularized techniques like machine learning, applied in institutions from Goldman Sachs to the Bank of England.

Applications

Applications are vast and influence both macroeconomic policy and microeconomic analysis. In macroeconomics, it is used to build large-scale models for forecasting GDP and informing the decisions of the Federal Reserve or the European Central Bank, a tradition started by Lawrence Klein's Project LINK. In finance, it underpins models for asset pricing and risk management used by JPMorgan Chase and analyzed through events like the 1987 crash. In labor economics, it assesses the impact of policies like the Earned Income Tax Credit or minimum wage laws. It also plays a crucial role in evaluating development programs for the World Bank and in industrial organization studies for entities like the Federal Trade Commission.

Assumptions and limitations

Reliability depends on a set of stringent statistical assumptions, violations of which can lead to biased or inconsistent results. A primary concern is endogeneity, often arising from omitted variable bias or simultaneity, addressed by methods pioneered by James Heckman and others. The assumption of exogeneity is critical for causal inference. Other common issues include multicollinearity, heteroskedasticity (addressed by the White test), and autocorrelation in time series data. Furthermore, the heavy reliance on observational data, unlike controlled experiments in fields like biology, makes establishing true causality challenging, a point emphasized by critics like Edward Leamer.

History and development

The field emerged in the early 20th century, with the term itself coined by Ragnar Frisch, who, along with Jan Tinbergen, developed the first formal models, earning them the inaugural Nobel Memorial Prize in Economic Sciences. The Cowles Commission, with figures like Tjalling Koopmans and Trygve Haavelmo, provided a rigorous probabilistic foundation in the 1940s and 1950s. The following decades saw the rise of computational economics, enabling large-scale models like those of Lawrence Klein. The 1980s and 1990s brought a focus on time series econometrics and methods for causal inference, advanced by Clive Granger, Robert Engle, and James Heckman. Today, the field continues to evolve with the integration of big data and machine learning techniques.

Category:Econometrics Category:Applied statistics