Week 11

Interaction Terms

Interaction Terms

Part of: Intro Econometrics | Key concepts: Interaction Term, Chow Test, Chow-Fisher Test, Hedonic Pricing, Dummy Variable, F-test, Omitted Variable Bias


Where this lecture sits

The thread from last week

Last week (Lec_10-Dummy Variables) showed how a dummy variable lets each group have its own intercept — but it forced every group to share the same slope (parallel lines). That is a strong, often wrong, assumption: the marginal effect of education on wage, or of alcohol content on price, may genuinely differ across groups. This lecture removes that straitjacket. The interaction term lets the slope itself vary across groups, and the Chow test gives a formal way to ask "are these two (or more) groups described by the same regression, or by genuinely different ones?"

The lecture has three strands:

  1. Hedonic pricing (a worked review of Omitted Variable Bias): how adding product-characteristic controls flips the sign of the time trend in computer prices — and what "controlling for quality" really buys you. Needed for HW 7.
  2. Interaction terms: dummy × continuous (different slopes), and dummy × dummy (interaction between two categories).
  3. Chow / Chow-Fisher tests: two equivalent ways (pooled-with-dummies, or separate regressions) to test whether groups share a common regression function, including how to count the restrictions qq.

Part 0 — Review: dummies and categorical variables

Quick recap of the machinery this lecture builds on.

A binary dummy is.Aiis.A_i (=1 if observation ii is in category A, else 0) enters linearly:

Yi=β0+β1Xi+β2 is.Ai+uiY_i = \beta_0 + \beta_1 X_i + \beta_2\, is.A_i + u_i

Splitting by group, the slope β1\beta_1 is shared and only the intercept moves by β2\beta_2:

Group Model becomes Intercept
Not A Yi=β0+β1Xi+uiY_i = \beta_0 + \beta_1 X_i + u_i β0\beta_0
A Yi=(β0+β2)+β1Xi+uiY_i = (\beta_0 + \beta_2) + \beta_1 X_i + u_i β0+β2\beta_0 + \beta_2

For a variable with KK categories we create K−1K-1 dummies; the omitted group is the base category and every coefficient is read relative to it. To test whether the categorical variable matters as a whole, restrict all its dummies to zero jointly and use an F-test (H0:β2=β3=0H_0:\beta_2=\beta_3=0); for a single binary dummy a t-test suffices.

The limitation that motivates today

Dummies give different intercepts but a common slope — parallel lines. The whole point of an interaction term is to break the parallelism and let each group's line tilt differently.


Part 1 — Hedonic pricing (omitted-variable bias in action)

The question

Based on the paper "Price Indexes for Microcomputers", we want to measure how PC prices changed over 1980–1988, the years personal computers became mainstream. The dependent variable is the (log) list price; the running variable is YEAR.

The naïve regression — a misleading non-result

ln_price ~ YEAR
R output
Coefficients: Estimate Std. Error t value Pr(>|t|)
(Intercept)   5.99242    1.33905   4.475  9.4e-06 ***
YEAR          0.01779    0.01565   1.137  0.256
Multiple R-squared: 0.0024   F-statistic: 1.293 on 1 and 528 DF

The time trend is positive and insignificant (p=0.26p = 0.26, R2≈0R^2 \approx 0). Read literally: "PC prices barely moved, if anything rose slightly, over the decade." That is the opposite of the truth.

Why the naïve coefficient is wrong: the product changed

A 1988 computer is not the same product as a 1980 one — it has more RAM, a faster processor, a bigger disk. The raw YEAR coefficient mixes two things: the price change of a fixed computer and the price change from buying a better computer. Quality is an omitted variable correlated with YEAR, so β^YEAR\hat\beta_{YEAR} is biased toward zero (even positive). This is exactly the Omitted Variable Bias story from Lec_10-Dummy Variables.

Adding controls — the sign flips

Add processor-type dummies (PROC16, PROC32; 8-bit is the base), a black-and-white screen dummy (DBW) and a portable dummy (DPORT):

ln_price ~ YEAR + PROC16 + PROC32 + DBW + DPORT
R output
Coefficients: Estimate Std. Error t value Pr(>|t|)
(Intercept) 16.56651   1.31025   12.644  <2e-16 ***
YEAR        -0.11315   0.01556   -7.270  1.3e-12 ***
PROC16       0.85908   0.07327   11.724  <2e-16 ***
PROC32       1.87927   0.11994   15.668  <2e-16 ***
DBW          0.16282   0.06376    2.554  0.0109 *
DPORT       -0.17039   0.08616   -1.978  0.0485 *
Multiple R-squared: 0.363

Now YEAR is negative and highly significant: holding processor and screen type fixed, prices fell ≈11% per year. Once we stop comparing apples (cheap old machines) to oranges (powerful new ones), the genuine downward price trend appears.

Full specification — the hedonic model

Add the logged quantitative characteristics (ln_ram, ln_mhz, ln_hardisc, ln_slots, ln_numflp):

Full model — ln_price ~ YEAR + processor/screen dummies + logged characteristics
R output
Coefficients: Estimate Std. Error t value Pr(>|t|)
(Intercept) 30.43992   1.18295   25.732  <2e-16 ***
YEAR        -0.30154   0.01487  -20.277  <2e-16 ***
PROC16       0.27929   0.05751    4.856  1.6e-06 ***
PROC32       0.67969   0.10863    6.257  8.4e-10 ***
DBW          0.12001   0.04471    2.684  0.0075 **
DPORT        0.44401   0.06478    6.854  2.1e-11 ***
ln_ram       0.20183   0.02839    7.110  4.0e-12 ***
ln_mhz       0.32726   0.05338    6.131  1.8e-09 ***
ln_hardisc   0.19372   0.01543   12.556  <2e-16 ***
ln_slots     0.18755   0.03039    6.171  1.4e-09 ***
ln_numflp    0.60119   0.07045    8.534  <2e-16 ***
Multiple R-squared: 0.726

With full quality controls, YEAR =−0.30= -0.30: a quality-constant computer got ≈30% cheaper each year, and R2R^2 jumps from 0.36 to 0.73. The more characteristics we hold fixed, the more of the apparent price stability is revealed to be quality improvement.

What hedonic pricing means

Hedonic pricing decomposes the price of a heterogeneous good into the implicit prices of its characteristics. The coefficient on YEAR then measures the price change of a "computer unit" of constant quality — what you would actually pay over time for the same bundle of features. It is the standard method for building quality-adjusted price indexes for goods with lots of variation: PCs, real estate, cars.

Two modelling rules from this example
  • Dummy variables always enter linearly (you cannot take a log of a 0/1 variable).
  • Take logs of the quantitative characteristics — this is a demand/price function, so a log-log form gives coefficients that read as elasticities and tends to fit better.

Part 2 — Interaction terms: letting slopes differ

The motivating gap

Dummies move the intercept but keep the lines parallel. What if the regression lines for two groups have different slopes — e.g. an extra year of schooling raises men's wages by more (or less) than women's? We need a term that lets the marginal effect of a continuous variable depend on group membership. That term is the interaction term.

Constructing the interaction

Interact the dummy male with the continuous variable schooly:

Yi=β0+β1 schoolyi+β2 malei+β3 (schoolyi⋅malei)+uiY_i = \beta_0 + \beta_1\, schooly_i + \beta_2\, male_i + \beta_3\,(schooly_i \cdot male_i) + u_i

The interaction term is literally the product of the two variables. For women it switches off; for men it equals schooly:

schoolyi⋅malei={schoolyi⋅0=0if femaleschoolyi⋅1=schoolyiif maleschooly_i \cdot male_i = \begin{cases} schooly_i \cdot 0 = 0 & \text{if female} \\ schooly_i \cdot 1 = schooly_i & \text{if male} \end{cases}

What each group's line looks like

Group maleimale_i Model becomes Intercept Slope
Women 00 Yi=β0+β1schoolyi+uiY_i = \beta_0 + \beta_1 schooly_i + u_i β0\beta_0 β1\beta_1
Men 11 Yi=(β0+β2)+(β1+β3) schoolyi+uiY_i = (\beta_0+\beta_2) + (\beta_1+\beta_3)\,schooly_i + u_i β0+β2\beta_0+\beta_2 β1+β3\beta_1+\beta_3
Interpreting the four coefficients
  • β0\beta_0 = intercept for the base group (women).
  • β1\beta_1 = slope (return to schooling) for the base group (women).
  • β2\beta_2 = the intercept gap between men and women (the pure dummy effect).
  • β3\beta_3 = the slope gap: how much more (or less) a year of schooling is worth to men than to women.

So men's return to schooling is β1+β3\beta_1 + \beta_3. With β2>0\beta_2 > 0 and β3<0\beta_3 < 0, men start higher but their line is flatter — the lines can even cross. The two groups now differ in both slope and intercept.

The boxed takeaway

dummy⇒different intercept;dummy×continuous⇒different slope\boxed{\text{dummy} \Rightarrow \text{different intercept};\quad \text{dummy} \times \text{continuous} \Rightarrow \text{different slope}}

Worked R example — does the return to schooling differ by sex?

ln_wage ~ schooly + age_g + agesq + origin_ussr + married + divorced + male + schooly_male
R output
Coefficients:    Estimate    Std. Error t value Pr(>|t|)
(Intercept)  -0.63033498  0.08722324  -7.227  5.3e-13 ***
schooly       0.09289182  0.00330030  28.146  <2e-16 ***
...
male          0.12722976  0.06557115   1.940  0.0524 .
schooly_male -0.00025413  0.00457271  -0.056  0.9557

The interaction schooly_male is tiny and insignificant (p=0.96p = 0.96): the return to a year of schooling is essentially the same for men and women. The intercept gap male is significant only at ≈10%.

Interactions inflate collinearity

male and schooly_male are highly correlated by construction (one is the other times schooly), so adding the interaction term drains significance from male — its t-value fell once schooly_male entered. A jointly-relevant pair can look individually weak; reach for a joint F-test when in doubt (same lesson as multicollinearity in Lec_08-Goodness of Fit & Multicollinearity).

Interactions between two dummies

You can also interact a dummy with another dummy — useful for asking whether a penalty is compounded for people in two categories at once.

Are USSR-born women penalised more than additive effects predict?

Build ussr_male = origin_ussr × male and estimate ln_wage ~ ... + origin_ussr + male + ussr_male:

R output
origin_ussr -0.15648770  0.02660990  -5.881  4.2e-09 ***
male         0.11306168  0.01512553   7.475  8.4e-14 ***
ussr_male    0.07107255  0.03873032   1.835  0.0665 .

The four group intercepts (assume everyone single) are built by adding the relevant coefficients:

Group Intercept
Women, Israel (base) −0.627-0.627
Men, Israel −0.627+0.113-0.627 + 0.113
Women, USSR −0.627−0.156-0.627 - 0.156
Men, USSR −0.627+0.113−0.156+0.071-0.627 + 0.113 - 0.156 + 0.071

The interaction ussr_male =+0.071= +0.071 is only marginally significant (p=0.067p = 0.067): USSR-born men recover a little of the immigrant penalty relative to what the two separate dummies alone would predict, but the evidence is weak.

Why the interaction term is the deviation from additivity

Without ussr_male, the model forces the USSR penalty to be identical for men and women (and the male premium identical across origins). The interaction coefficient measures exactly the departure from that additive assumption — the "extra" effect of being in both categories at once.


Part 3 — Interaction terms vs. separate regressions

You can let every coefficient differ by group in two equivalent ways:

  1. One pooled regression with the group dummy interacted with every regressor.
  2. Separate regressions, one per group.
When to prefer separate regressions

Interacting the dummy with many regressors gets unwieldy fast (one extra term per variable). Running separate regressions for men and women is cleaner and — importantly — allows the error variance to differ between groups, which improves efficiency if the groups really are heteroskedastic relative to each other. The pooled-with-interactions model implicitly assumes a common error variance.

Estimating ln_wage ~ schooly + age_g + agesq + origin_ussr + married + divorced separately on women and on men gives visibly different coefficient vectors (e.g. women's intercept −0.268-0.268 vs men's −0.884-0.884, women's return to schooling 0.0960.096 vs men's 0.0900.090). The formal question — are these differences real, or sampling noise? — is answered by the Chow test.


Part 4 — Chow tests: do two groups share one regression?

The Chow test

A Chow test asks whether two (or more) groups are described by the same regression function (same intercept and same slopes) or by different ones. It is just an F-test of a joint restriction, with a specific name. There are two equivalent implementations.

Version A — using dummy variables (pooled)

Start from Yi=β0+β1xi+uiY_i = \beta_0 + \beta_1 x_i + u_i and allow the group man to differ in both intercept and slope by adding the dummy and its interaction:

Yi=β0+β1xi+β2 mani+β3 (xi⋅mani)+ui⏟unrestricted\underbrace{Y_i = \beta_0 + \beta_1 x_i + \beta_2\, man_i + \beta_3\,(x_i \cdot man_i) + u_i}_{\text{unrestricted}}
  • Men: Yi=(β0+β2)+(β1+β3)xi+uiY_i = (\beta_0+\beta_2) + (\beta_1+\beta_3)x_i + u_i
  • Women: Yi=β0+β1xi+uiY_i = \beta_0 + \beta_1 x_i + u_i

The null "men and women share one regression" is

H0: β2=β3=0.H_0:\ \beta_2 = \beta_3 = 0.

The restricted model is the pooled Yi=β0+β1xi+uiY_i = \beta_0 + \beta_1 x_i + u_i. Test with the usual F-statistic:

F=(RSSR−RSSUR)/qRSSUR/(N−M)∼F(q, N−M)\boxed{F = \frac{(\text{RSS}_R - \text{RSS}_{UR})/q}{\text{RSS}_{UR}/(N-M)} \sim F(q,\,N-M)}

We reject if the intercept differs (β2≠0\beta_2 \neq 0), the slope differs (β3≠0\beta_3 \neq 0), or both.

Wage regressions for men vs women
R output
> lfit   <- lm(ln_wage ~ schooly + male + schooly_male, mydata)
> lfit_r <- lm(ln_wage ~ schooly, mydata)
> anova(lfit, lfit_r)
  Res.Df    RSS Df Sum of Sq      F      Pr(>F)
1  10169 6809.3
2  10171 6847.5 -2   -38.245 28.558  4.3e-13 ***

F=28.56F = 28.56 on (2, 10169)(2,\,10169) with p≈4×10−13p \approx 4\times10^{-13}: reject H0H_0. Men and women do not share the same wage regression.

With many regressors the restriction count grows

If the base model has several explanatory variables, allowing all of them to differ by group means interacting the dummy with each one. For a model with intercept + 4 slopes, letting every coefficient vary adds 5 terms, so H0:β5=β6=β7=β8=β9=0H_0:\beta_5=\beta_6=\beta_7=\beta_8=\beta_9=0 — an F-test with 5 restrictions.

Version B — using separate regressions

Equivalently, fit the model separately on each group and combine the residual sums of squares. For men Yi=β0m+β1mxi+uimY_i = \beta_0^m + \beta_1^m x_i + u_i^m and women Yi=β0f+β1fxi+uifY_i = \beta_0^f + \beta_1^f x_i + u_i^f, the null is

H0: β0m=β0f,  β1m=β1fH_0:\ \beta_0^m = \beta_0^f,\ \ \beta_1^m = \beta_1^f

and the statistic uses the pooled RSS as restricted, and the sum of the group RSSs as unrestricted:

F=(RSSR−(RSSm+RSSf))/q(RSSm+RSSf)/(N−M)∼F(q, N−M)F = \frac{\big(\text{RSS}_R - (\text{RSS}_m + \text{RSS}_f)\big)/q}{(\text{RSS}_m + \text{RSS}_f)/(N-M)} \sim F(q,\,N-M)
What NN and MM mean here
  • NN = total number of observations (both groups pooled).
  • MM = total number of parameters in the unrestricted model = parameters in the men's model plus parameters in the women's model.
  • q=dfR−dfURq = df_R - df_{UR} = the number of restrictions.
Worked separate-regression Chow test (6-regressor wage model)

With RSSm=1375.8\text{RSS}_m = 1375.8, RSSf=1164.8\text{RSS}_f = 1164.8, pooled RSSR=2602.2\text{RSS}_R = 2602.2, N=8628N = 8628, and M=14M = 14 (7 parameters ×\times 2 groups), q=7q = 7: F=(2602.2−(1375.8+1164.8))/7(1375.8+1164.8)/(8628−14)=29.8  >  F0.05(7, 8614)≈2.F = \frac{(2602.2 - (1375.8 + 1164.8))/7}{(1375.8 + 1164.8)/(8628 - 14)} = 29.8 \;>\; F_{0.05}(7,\,8614) \approx 2. Reject: men and women have different wage regressions. (Same conclusion as Version A — the two versions are equivalent.)

Comparison of the two versions

Version A — dummies (pooled) Version B — separate regressions
What you estimate one regression with dummy + interactions one regression per group
Restricted RSS RSS of pooled model (no dummies) RSS of pooled model (no dummies)
Unrestricted RSS RSS of the dummy-interacted model RSSm+RSSf\text{RSS}_m + \text{RSS}_f (sum across groups)
Error variance assumed common across groups allowed to differ across groups
Result identical F and decision ... ... to Version A
Best when few regressors; want common-variance efficiency many regressors; worried about group heteroskedasticity

Counting the restrictions qq

q=dfR−dfURq = df_R - df_{UR} — count parameters, subtract

For y=β0+β1x1+β2x2+uy = \beta_0 + \beta_1 x_1 + \beta_2 x_2 + u (3 parameters):

Groups Unrestricted df Restricted df qq
2 groups (A, W) N−6N-6 (3 params ×\times 2) N−3N-3 (N−3)−(N−6)=3(N-3)-(N-6) = \mathbf{3}
3 groups (A, W, F) N−9N-9 (3 params ×\times 3) N−3N-3 (N−3)−(N−9)=6(N-3)-(N-9) = \mathbf{6}

Shortcut: q=(params per group)×(groups−1)q = (\text{params per group}) \times (\text{groups} - 1) — the number of coefficient equalities the null imposes.

Chow tests over time

If the "group" is a time period (e.g. before vs after a policy), the Chow test becomes a structural-stability / structural-break test: define a dummy = 1 in one period, build interaction terms, and F-test whether the relationship is stable across periods.


Part 5 — The Chow-Fisher test (three or more groups)

The Chow-Fisher test generalises the separate-regression Chow test to more than two groups — here, the demand for alcoholic beverages split into beer, wine, and spirits.

Setup — demand for alcohol (British Columbia 2012)

Log-log demand function (logs because it is a demand model): ln⁡(pricei)=β0+β1ln⁡abvi+β2ln⁡volumei+β3ln⁡unitsi+γ′Countryi+ui\ln(price_i) = \beta_0 + \beta_1 \ln abv_i + \beta_2 \ln volume_i + \beta_3 \ln units_i + \boldsymbol{\gamma'}\mathbf{Country}_i + u_i where abv = % alcohol by volume.

Interactions reveal group-specific elasticities

Add interaction terms between the product-type dummies and ln abv (spirits is the base):

ln⁡(pricei)=⋯+β4 beeri+β5 winei+β11(beeri⋅ln⁡abvi)+β12(winei⋅ln⁡abvi)+ui\ln(price_i) = \dots + \beta_4\, beer_i + \beta_5\, wine_i + \beta_{11}(beer_i \cdot \ln abv_i) + \beta_{12}(wine_i \cdot \ln abv_i) + u_i

Because the outcome and abv are both logged, β1\beta_1 (and β1\beta_1 + each interaction) is a price elasticity with respect to abv:

Price elasticity of abv by product type

With β1=1.522\beta_1 = 1.522, β11=−0.97545\beta_{11} = -0.97545 (beer), β12=+0.51056\beta_{12} = +0.51056 (wine):

  • Spirits (base): 1.5221.522
  • Wine: 1.522+0.51056=2.0331.522 + 0.51056 = \mathbf{2.033}
  • Beer: 1.522−0.97545=0.5471.522 - 0.97545 = \mathbf{0.547}

The price sensitivity to alcohol content is strongest for wine, weakest for beer — exactly the kind of group-varying slope a single common coefficient would hide.

The Chow-Fisher F-statistic

Run separate regressions for beer, wine, and spirits and combine their RSSs against the pooled (restricted) RSS:

F=(RSSR−(RSSbeer+RSSwine+RSSspirits))/q(RSSbeer+RSSwine+RSSspirits)/(N−M)F = \frac{\big(\text{RSS}_R - (\text{RSS}_{beer} + \text{RSS}_{wine} + \text{RSS}_{spirits})\big)/q}{(\text{RSS}_{beer} + \text{RSS}_{wine} + \text{RSS}_{spirits})/(N-M)}
Activity 11, part 4 — the null and the test statistic

The null says every coefficient is equal across the three product types: H0: β0beer=β0wine=β0spirits, …, β3beer=β3wine=β3spirits, γ1beer=γ1wine=γ1spirits,…H_0:\ \beta_0^{beer}=\beta_0^{wine}=\beta_0^{spirits},\ \dots,\ \beta_3^{beer}=\beta_3^{wine}=\beta_3^{spirits},\ \gamma_1^{beer}=\gamma_1^{wine}=\gamma_1^{spirits},\dots With 9 parameters per group:

  • q=9×2=18q = 9 \times 2 = 18 (each of the 9 coefficients set equal across the two extra groups).
  • M=9×3=27M = 9 \times 3 = 27 (parameters in all three group regressions).
  • N=5844N = 5844. F=25.47  >  F0.975(18, 5844−27)=1.754  ⇒  reject H0.F = 25.47 \;>\; F_{0.975}(18,\,5844-27) = 1.754 \;\Rightarrow\; \textbf{reject } H_0. Beer, wine, and spirits do not follow a common demand regression.
Don't confuse qq and MM

MM counts all parameters in the unrestricted (all-groups-separate) model; qq counts the equality restrictions the null imposes, which is (params per group) ×\times (groups −1-1). For 3 groups and 9 params: M=27M = 27 but q=18q = 18, not 27.


Practice exam — adding X2X^2, model choice, and OVB

Setup

Researchers believed XX and ZZ alone explain YY: (1)ln⁡Y=γ0+γ1X+γ2Z+u\text{(1)}\quad \ln Y = \gamma_0 + \gamma_1 X + \gamma_2 Z + u A researcher suspects X2X^2 also matters and estimates: (2)ln⁡Y=β0+β1X+β2Z+β3X2+u\text{(2)}\quad \ln Y = \beta_0 + \beta_1 X + \beta_2 Z + \beta_3 X^2 + u (3)ln⁡Y=β0+β1X+β2Z+β3ln⁡X2+u\text{(3)}\quad \ln Y = \beta_0 + \beta_1 X + \beta_2 Z + \beta_3 \ln X^2 + u (All of X,Y,ZX,Y,Z are strictly positive.)

1(a) — How did the researcher conclude X2X^2 matters? From the p-value of β^3\hat\beta_3, here 5.18×10−75.18\times10^{-7}. It is far below 5%, so the coefficient on X2X^2 (in model 3) is significant ⟹ the variable is relevant.

1(b) — Why is model (3) better than model (2)? Compare R2R^2 (or equivalently the RSSs): R(3)2=0.7778>R(2)2=0.7713R^2_{(3)} = 0.7778 > R^2_{(2)} = 0.7713. Model (3) explains more variation, so the ln⁡X2\ln X^2 specification fits better than the raw X2X^2 one.

2 — Which assumption does model (1) violate? (assume model 3 is the truth)

If model (3) is the true DGF, then model (1) omits the relevant regressor ln⁡X2\ln X^2, which is absorbed into the error uu. If ZZ and ln⁡X2\ln X^2 are correlated, then in model (1) the error is correlated with the regressor ZZ: cov⁡(u,Z)≠0\operatorname{cov}(u, Z)\neq 0. This violates assumption A2, E[ui∣x1i,…,xKi]=0\mathbb{E}[u_i \mid x_{1i},\dots,x_{Ki}] = 0 (mean independence of the error from the regressors) — answer (ii). The other classical assumptions (zero mean, no autocorrelation, homoskedasticity) are not what fails.

3 — If ZZ and ln⁡X2\ln X^2 are uncorrelated, how do γ^2\hat\gamma_2 and β2\beta_2 compare?

Omitting ln⁡X2\ln X^2 from model (1) generally causes omitted-variable bias in the other coefficients, so we'd normally expect E[γ^2]≠β2\mathbb{E}[\hat\gamma_2] \neq \beta_2. But OVB only transmits through correlation between the omitted variable and the included one. If Z⊥ln⁡X2Z \perp \ln X^2, omitting ln⁡X2\ln X^2 does not bias the coefficient on ZZ: E[γ^2]=β2\mathbb{E}[\hat\gamma_2] = \beta_2 — answer (iii). This is the precise condition under which OVB vanishes: no correlation between the omitted and included regressor.


Summary

  1. Hedonic pricing / OVB review: regressing PC ln_price on YEAR alone gives a positive, insignificant trend because newer machines are better machines — quality is an omitted variable correlated with time. Adding characteristic controls flips YEAR to strongly negative (≈−30%/yr with full controls). The YEAR coefficient then measures the price of a constant-quality "computer unit". Rules: dummies enter linearly, quantitative characteristics enter as logs (demand function).
  2. A dummy variable gives different intercepts but a common slope (parallel lines). An interaction term — the product of a dummy and a continuous variable — gives different slopes. In Y=β0+β1x+β2D+β3(xD)+uY=\beta_0+\beta_1 x+\beta_2 D+\beta_3(xD)+u: β2\beta_2 is the intercept gap, β3\beta_3 is the slope gap, and the value-1 group's slope is β1+β3\beta_1+\beta_3.
  3. Dummy × dummy interactions measure the departure from additivity — the extra effect of being in two categories at once (e.g. USSR-born and male). Interactions are highly collinear with their parents, so they drain individual significance; lean on joint F-tests.
  4. Letting every coefficient differ by group can be done by interacting the dummy with every regressor (pooled) or by separate regressions. Separate regressions are cleaner with many regressors and allow group-specific error variances.
  5. A Chow test is an F-test of "do the groups share one regression?". Version A (dummies): H0:H_0: all dummy + interaction coefficients =0=0, unrestricted RSS from the interacted model. Version B (separate): unrestricted RSS =RSS1+RSS2=\text{RSS}_1+\text{RSS}_2, restricted RSS from the pooled model. The two are equivalent.
  6. Counting restrictions: q=(params per group)×(groups−1)q = (\text{params per group})\times(\text{groups}-1); M=(params per group)×(groups)M = (\text{params per group})\times(\text{groups}); NN = pooled sample size. For the 3-group alcohol model with 9 params: q=18q=18, M=27M=27.
  7. The Chow-Fisher test extends the separate-regression Chow test to 3+ groups (beer/wine/spirits). Interactions with ln abv reveal different elasticities per type (wine 2.03, spirits 1.52, beer 0.55), and the F-test rejects a common demand regression.
  8. Exam OVB rule: omitting a relevant variable biases an included coefficient only if the two are correlated. Correlated ⟹ A2 violated, E[u∣X]≠0\mathbb{E}[u\mid X]\neq 0, bias. Uncorrelated ⟹ that coefficient stays unbiased (E[γ^2]=β2\mathbb{E}[\hat\gamma_2]=\beta_2).