Background
Durbin’s test is a statistical method used primarily in econometrics to detect the presence of first-order serial correlation in the error terms when a lagged dependent variable is included in the regression model. Serial correlation, or autocorrelation, occurs when residuals (errors) in a time series model are correlated with each other.
Historical Context
The test is named after James Durbin, who introduced this methodology to address situations where the existing Durbin–Watson test becomes invalid due to the presence of a lagged dependent variable in the model. This contribution is significant to time-series analysis and econometrics since the common Durbin–Watson test cannot be employed in such cases.
Definitions and Concepts
- First-order Serial Correlation: When the error term of one period is correlated with the error term of the immediately previous period.
- Lagged Dependent Variable: A dependent variable that is lagged by one or more time periods in regression analysis.
- Ordinary Least Squares (OLS) Residuals: The residuals obtained from fitting a regression model using the Ordinary Least Squares method.
Major Analytical Frameworks
Classical Economics
In Classical economics, time series data might not be widely considered since earlier models often assumed static conditions.
Neoclassical Economics
Neoclassical models can involve dynamic conditions where time series data and therefore methods to check for serial correlations, like Durbin’s test, become crucial.
Keynesian Economic
Being dynamic by nature, Keynesian analysis might require applying Durbin’s test when models involve time-lagged relationships between macroeconomic variables.
Marxian Economics
Though not focusing extensively on statistical tests for models, analyses of historical and economic cycles might benefit indirectly from methods to check correlation in temporal economic data.
Institutional Economics
Understanding institutional behaviors over time could apply time series analysis where tests for serial correlation, such as Durbin’s test, are valuable.
Behavioral Economics
Behavioral models may include dynamics and autocorrelation checks in understanding how past behaviors predict current economic decisions.
Post-Keynesian Economics
Given its focus on historical time-series data, Post-Keynesian Economics often deals with lagged relationships wherein tests like the Durbin’s test become applicable.
Austrian Economics
Temporal sequences and evolving economic processes in Austrian economics have situational overlap with the necessity to check for serial autocorrelations.
Development Economics
Analyzing time trends in development indicators frequently necessitates tests for serial correlation especially when incorporating lagged dependent variables.
Monetarism
In analyzing monetarist models where money supply affects future periods, Durbin’s test helps identify autocorrelations in time-lagged money flow variables.
Comparative Analysis
Durbin’s test fills a significant gap left by the Durbin–Watson test in dealing with models specifically containing lagged dependent variables. While it provides an asymptotically standard normal distribution under the null hypothesis, the specificity in addressing correlation with lagged variables sets it apart from more generic methods like LM tests or Breusch-Godfrey tests.
Case Studies
Case studies involving economic growth models with lagged GDP measure, dynamic consumption models, or investment functions over time might report results of Durbin’s test to validate model assumptions.
Indicators:
- Nicoletti, Nicola, and Lampiris models, examining lagged GDP in growth regression analysis
- Various econometric software applications that showcase examples like STATA, EViews, or R programming implementation of time series models.
Suggested Books for Further Studies
- “Introductory Econometrics: A Modern Approach” by Jeffrey M. Wooldridge
- “Econometric Analysis” by William H. Greene
- “Applied Econometrics with R” by Christian Kleiber and Achim Zeileis
Related Terms with Definitions
- Serial Correlation: The occurrence of patterns in the error terms that persist over time periods.
- Durbin–Watson Test: Another test for detecting autocorrelation which however becomes invalid or unreliable when there’s a lagged dependent variable.
- Autoregressive Model: A model where the current value of a series is regressed on its previous values.
