Background
The Likelihood Ratio Test (LRT) is a statistical method used to compare the fit of two nested models—one of which is a special case of the other. Specifically, it evaluates the maximized value of the likelihood function under the null hypothesis and an alternative hypothesis.
Historical Context
The origin of the likelihood ratio test dates back to the early 20th century, with significant contributions from pioneers such as Sir Ronald A. Fisher and Jerzy Neyman. It has since become a cornerstone in statistical inference, particularly in the fields of econometrics and biostatistics.
Definitions and Concepts
In statistical inference, the likelihood ratio test is one of the three major classical tests for evaluating restrictions on an unknown parameter or a vector of unknown parameters (θ), the other two being the Lagrange Multiplier (LM) test and the Wald test. The test is based on maximum likelihood estimation (MLE) of the parameter(s) in question.
Key Definitions:
- Likelihood Functions (L): Functions representing the probability of data given parameters.
- θ̂R and θ̂U: Maximum likelihood estimators of the parameters under restriction (θ̂R) and without restriction (θ̂U), respectively.
- Likelihood Ratio (λ): The ratio of the likelihood functions evaluated at θ̂R over θ̂U, i.e., λ = L(θ̂R)/L(θ̂U).
- Asymptotic Test Statistic: For large sample sizes, -2ln(λ) follows an asymptotic chi-square distribution with degrees of freedom equal to the number of restrictions.
Major Analytical Frameworks
Classical Economics
Likelihood Ratio Tests are utilized in the context of model comparisons, but do not organically fit into classical economic theories which are more concerned with market behaviors than statistical methods.
Neoclassical Economics
In empirical studies evaluating consumer behavior or production functions, likelihood ratio tests can help compare restricted models (consistent with neoclassical assumptions) against more general alternative models.
Keynesian Economics
When modeling macroeconomic indicators, such as GDP or unemployment rates subject to specific policy restrictions, LRT may be used to test whether additional parameters improve model fit.
Marxian Economics
While less commonly used directly, likelihood ratio tests may still apply in empirical validation of models testing the impacts of labor theories of value or exploitation rates.
Institutional Economics
Statistical validation, including LRT, may ensure structural models encompassing institutional factors hold under sample data.
Behavioral Economics
LRT can be critical in verifying whether inclusion of behavioral variables significantly enhances the prediction of economic outcomes over rational-choice models.
Post-Keynesian Economics
In testing models for potential non-linearity and multiple equilibrium scenarios, LRT might be employed to validate more complex model iterations against simpler, constrained variants.
Austrian Economics
Generally focused on qualitative models, Austrian economics applies less frequent application of LRT.
Development Economics
Comparative evaluation of growth models or policy interventions may task LRT to ascertain improvements when additional explanatory variables are introduced.
Monetarism
Within models analyzing the effects of monetary aggregates, LRT serves to confirm parameter restrictions’ consistency with theoretical expectations dictated by monetarist principles.
Comparative Analysis
In contrast to Wald and LM tests, LRT fundamentally compares likelihoods, thereby potentially being more robust under certain conditions. Wald tests analyze the parameters directly, while LM evaluates constraints’ consistency from the unrestricted model’s perspective.
Case Studies
Examples of LRT applications in econometrics include:
- Testing the additional contribution of structural variables in extended models of economic productivity.
- Evaluating the inclusion of different monetary policy instruments in predictive macroeconomic models.
Suggested Books for Further Studies
- “Econometric Analysis” by William H. Greene
- “Econometric Theory and Methods” by Russell Davidson and James G. MacKinnon
- “Statistical Methods for the Social Sciences” by Alan Agresti and Barbara Finlay
Related Terms with Definitions
- Maximum Likelihood Estimation (MLE): A method used to estimate the parameters of a statistical model by maximizing a likelihood function.
- Lagrange Multiplier Test: A test used to assess the constraints on parameters within an unrestricted model without estimating the restricted model.
- Wald Test: A statistical test that evaluates the significance of individual estimated coefficients in a fitted model.
By thoroughly understanding the likelihood ratio test, its applications can be appropriately contextualized within broader econometric and statistical methodologies.