Selected Topics in Econometrics: Dummy Variables and the Chow Test

A vocabulary flashcard deck covering core econometric principles including dummy variable definitions, ANOVA and ANCOVA models, regression geometries, and the Chow test for structural change.

Dummy Variable

An artificial, binary, or indicator variable used to quantify qualitative attributes (such as sex, race, color, or marital status) by assigning a value of $1$ for the presence of the attribute and $0$ for its absence.

Base Category

The benchmark or omitted category assigned the value of $0$, serving as the reference point against which other categories are compared in a regression model.

$(m-1)$ Rule

The rule stating that if a qualitative variable has $m$ categories, exactly $m - 1$ dummy variables must be introduced in the model to avoid perfect collinearity.

ANOVA Model

An Analysis of Variance regression model in which all explanatory variables (regressors) are exclusively qualitative or dummy variables, written as $Y_i = \beta_0 + \beta_1 D_1 + \beta_2 D_2 + \dots + \mu$.

ANCOVA Model

An Analysis of Covariance regression model that includes a mixture of quantitative regressors ($X$) and qualitative dummy regressors ($D$), written as $Y_i = \beta_0 + \beta_1 X_1 + \beta_2 X_2 + \beta_3 D_1 + \beta_4 D_2 + \dots + \mu$.

Intercept Dummy

An additive dummy variable coefficient ($\beta_2 D$) that measures the differential intercept, shifting the vertical intercept between different qualitative groups while keeping the slope constant.

Slope Dummy

A multiplicative dummy variable interaction ($\beta_3 [D \times X]$) that measures the differential slope coefficient, allowing the slope with respect to a quantitative variable to differ between groups.

Parallel Regressions

The graphical outcome (Case 1) in a dummy model when only the intercept dummy ($\beta_2$) is statistically significant, producing parallel regression lines that have different intercepts ($\beta_0$ versus $\beta_0 + \beta_2$) but identical slopes ($\beta_1$).

Concurrent Regressions

The graphical outcome (Case 2) in a dummy model when only the slope dummy ($\beta_3$) is statistically significant, resulting in regression lines that share the same intercept ($\beta_0$) but have different slopes ($\beta_1$ versus $\beta_1 + \beta_3$).

Dissimilar Regressions

The graphical outcome (Case 3) in a dummy model when both the differential intercept dummy ($\beta_2$) and the differential slope dummy ($\beta_3$) are statistically significant, resulting in regression lines with different intercepts and different slopes.

Coincident Regressions

The graphical outcome (Case 4) occurring when both the differential intercept ($\beta_2$) and differential slope ($\beta_3$) dummies are non-significant, indicating that a single regression line applies to all categories.

Seasonal Dummy Variables

Dichotomous indicator variables incorporated into time-series regressions to account for periodic seasonal effects (such as using $D_2$, $D_3$, and $D_4$ for the second, third, and fourth quarters relative to the first quarter base).

Structural Change

A situation where the values of the parameters in a regression model do not remain constant across different time periods due to external forces, policy shifts, or government actions.

Chow Test

An $F$-test used to determine whether a structural break or change has occurred across distinct time periods by comparing the residual sum of squares of sub-period regressions to that of a pooled regression.

Chow Test Statistic

The test statistic defined as $\frac{[RSS - (RSS_1 + RSS_2)]/k}{(RSS_1 + RSS_2)/(n - 2k)}$, where $RSS$ is the residual sum of squares of the pooled model, $RSS_1$ and $RSS_2$ are the sub-period residual sums of squares, $k$ is the number of estimated parameters, and $n$ is the total number of observations.

Dummy Variable

A qualitative explanatory variable taking the value of $1$ to indicate the presence of an attribute and $0$ to indicate the absence of the attribute.

Base Category

The category assigned the value $0$, serving as the omitted category or benchmark against which other categories are compared.

$(m - 1)$ Dummy Rule

The principle stating that if a qualitative variable has $m$ distinct categories, exactly $m - 1$ dummy variables must be introduced to avoid perfect collinearity.

ANOVA Model

An Analysis of Variance regression model in which all explanatory variables (regressors) are dummy variables, formulated as $Y_i = \beta_0 + \beta_1 D_1 + \beta_2 D_2 + \beta_3 D_3 + \nu$.

ANCOVA Model

An Analysis of Covariance regression model containing a mixture of quantitative regressors and qualitative dummy variables, formulated as $Y_i = \beta_0 + \beta_1 X_1 + \beta_2 X_2 + \beta_3 D_1 + \beta_4 D_2 + \nu$.

Differential Intercept Coefficient

The coefficient of an additive dummy variable ($\beta_2 D$) that indicates how much the intercept of the category with $D = 1$ differs from that of the base category.

Differential Slope Coefficient

The coefficient of an interaction or multiplicative dummy term ($\beta_3 (D \times X)$) that indicates the change in slope between the category with $D = 1$ and the benchmark category.

Slope-Intercept Dummy Model

A specification combining additive and multiplicative dummy terms to allow both intercepts and slopes to differ between groups: $Y = \beta_0 + \beta_1 X + \beta_2 D + \beta_3 (D \times X) + \nu$.

Parallel Regressions

A dummy variable regression scenario (Case 1) where only the differential intercept dummy ($\beta_2$) is statistically significant, resulting in lines with different intercepts ($\beta_0$ and $\beta_0 + \beta_2$) but identical slopes ($\beta_1$).

Concurrent Regressions

A dummy variable regression scenario (Case 2) where only the differential slope dummy ($\beta_3$) is statistically significant, producing lines that share the same intercept ($\beta_0$) but have different slopes ($\beta_1$ versus $\beta_1 + \beta_3$).

Dissimilar Regressions

A regression scenario (Case 3) where both the intercept dummy ($\beta_2$) and the slope dummy ($\beta_3$) are statistically significant, resulting in different intercepts and different slopes across groups.

Coincident Regressions

A regression scenario (Case 4) where both the intercept dummy ($\beta_2$) and the slope dummy ($\beta_3$) are non-significant, showing that a single common regression line applies across groups.

Seasonal Analysis (Dummy Variables)

A technique using quarterly or period dummy variables (e.g., $D_2, D_3, D_4$ with quarter 1 as base) to isolate and account for seasonal shifts in economic relationships.

Structural Change

A shift where the parameter values of an econometric model do not remain constant throughout the entire sample period due to external forces, policy interventions, or government actions.

Chow Test

An $F$-test used to examine whether a structural break has occurred across time periods by evaluating the residual sums of squares ($RSS$) of sub-sample regressions relative to the pooled sample.

Chow Test Statistic

The formula used to compute the Chow test: $\text{CHOW} = \frac{[RSS - (RSS_1 + RSS_2)]/k}{(RSS_1 + RSS_2)/(n - 2k)}$, where $k$ represents the number of estimated parameters and $n$ is the total sample size.