Econometrics and Financial Modeling Key Tests Models & Ratios Study Guide 2025/ 2026

Econometrics and Financial Modeling
Econometrics
– Key Tests,
and Financial
Models &Modeling
Ratios
Econometrics
Study
– Key
Guide.pdf
Tests,
and Financial
Models &Modeling
Ratios Study
– Key
Guide.pdf
Tests, Models & Ratios Study Guide.pdf




● Econometrics and Financial
Modeling – Key Tests, Models &
Ratios Study Guide




Guidehttps://www.stuvia.com/dashboard!@_)#*)(@$)($@*($@)($@*_
Econometrics and Financial Modeling
Econometrics
– Key Tests,
and Financial
Models &Modeling
Ratios
Econometrics
Study
– Key
Guide.pdf
Tests,
and Financial
Models &Modeling
Ratios Study
– Key
Guide.pdf
Tests, Models & Ratios Study Guide.pdf

,Econometrics and Financial Modeling_ Key Tests, Models, and Ratios.pdf Econometrics and Financial Modeling_ Key Tests, Models, and Ratios.pdf Econometrics and Financial Modeling_ Key Tests, Models, and Ratios.pdf


Terms in this set (103)



A vignette shows regression residuals plotted Conditional heteroskedasticity: Breusch-Pagan test = n*R^2 ~ chi-squared
against the independent variable forming a fan with df = number of independent variables | H0: no conditional
shape that widens as X increases — what's the issue heteroskedasticity (do NOT want to reject) | Fix: use robust (White)
test, and fix? standard errors | Trap: coefficients are unbiased — only SEs and t-stats are
wrong


A vignette shows a Durbin-Watson stat of 1.3 in a Cannot use DW when a lagged DV is a regressor (AR models). Use
regression that includes a lagged dependent Breusch-Godfrey instead: chi-squared test for lag 1, F-stat with (n-p-k-1) df
variable — what's wrong with this setup? for p>1. DW is only valid when no lagged DV present.


A regression has high R^2 and significant F-stat but Multicollinearity. Test with VIF = 1/(1-Rj^2) where Rj^2 comes from
no individual slopes are significant — what's the regressing Xj on the other independent variables. VIF > 5 warrants
likely issue? investigation, VIF > 10 is serious. Fix: drop a variable or get more data.


Adjusted R-squared formula and why it matters R-bar^2 = 1 - [(n-1)/(n-k-1)] * (1-R^2) | Penalizes additional variables unlike
R^2 which always rises or stays flat when a variable is added | Adj R^2 can
be negative; R^2 cannot | Use Adj R^2 for comparing models with
different numbers of variables

Econometrics and Financial Modeling_ Key Tests, Models, and Ratios.pdf Econometrics and Financial Modeling_ Key Tests, Models, and Ratios.pdf Econometrics and Financial Modeling_ Key Tests, Models, and Ratios.pdf

, Econometrics and Financial Modeling_ Key Tests, Models, and Ratios.pdf Econometrics and Financial Modeling_ Key Tests, Models, and Ratios.pdf Econometrics and Financial Modeling_ Key Tests, Models, and Ratios.pdf




AIC vs BIC — formulas and when to use each AIC = n*ln(SSE/n) + 2(k+1)


BIC = nln(SSE/n) + ln(n)(k+1)


Lower is better for both
AIC for forecasting/prediction
BIC for best descriptive fit (BIC penalizes complexity more heavily when n
is large)


F-test for joint significance of restrictions F = [(SSE_restricted - SSE_unrestricted)/q] / [SSE_unrestricted/(n-k-1)] | q =
number of restrictions (variables dropped)


Tests whether the dropped variables jointly matter


Dummy variable rule Use n-1 dummy variables for n categories. The omitted category becomes
the intercept/reference. If all dummies = 0, regression represents the
omitted baseline. Using n dummies creates perfect multicollinearity.




Econometrics and Financial Modeling_ Key Tests, Models, and Ratios.pdf Econometrics and Financial Modeling_ Key Tests, Models, and Ratios.pdf Econometrics and Financial Modeling_ Key Tests, Models, and Ratios.pdf