Dynamic Response Models

Routing Summary

Dynamic models add time structure: carryover, lags, competitive reaction, and time-varying parameters.

Concept Map

ConceptNoteTypeDepends OnKey Result
Carryover & distributed lagsCarryover Effects and Distributed LagsconceptFunctional Forms in Marketing, Markets Data and Sales Drivers, Design of Static Response ModelsTemporal carryover via geometric lags, PDL, feedback, and bias correction
Competitive reaction functionsReaction Functions and Competitive DynamicsconceptCarryover Effects and Distributed Lags, Market Share ModelsCompetitor reaction functions via Cournot, Bertrand, Stackelberg, Sweezy
Shape of the response functionShape of the Marketing Response FunctionconceptFunctional Forms in Marketing, Carryover Effects and Distributed LagsResponse shape (concave/S/convex) drives optimal budget allocation
Dynamic model designDesign of Dynamic Response ModelsconceptDesign of Static Response Models, Carryover Effects and Distributed Lags, Reaction Functions and Competitive DynamicsExtends static to dynamic via time, lags, and short/long-run effects

Notes

  • Carryover Effects and Distributed Lags — CONTAINS: Koyck/geometric distributed lag, long-run multiplier theorem, geometric lag with purchase feedback (GLPF), Almon polynomial distributed lag (PDL), autoregressive distributed lag (ADL), Doyle-Saunders lead-lag taxonomy (6 cases), return-to-normality time-varying-parameter model, ratchet (asymmetric) models, Clarke (1976) aggregation-bias theorem, and the Bass-Leone and Weiss-Weinberg-Windal recovery procedures.
  • Reaction Functions and Competitive Dynamics — CONTAINS: Cournot, Bertrand, and Stackelberg (leader-follower) reaction functions, Sweezy kinked demand curve, absolute-change and relative (log-log) reaction models, the generalized reaction matrix (VAR system), Tobit model for censored reactions, and distributed reaction lags.
  • Shape of the Marketing Response Function — CONTAINS: prior-knowledge levels (0/1/2), concave (diminishing returns) theorem, S-shaped (convex-concave) response with threshold, convex (increasing returns) response, threshold-effect kinked-linear model, hysteresis and path dependence via unit root, pulsing-strategy table by shape, and ratchet (historical-maximum) models.
  • Design of Dynamic Response Models — CONTAINS: ADL framework, direct-lag identification strategy, short-run vs long-run effects (impact vs total multiplier) theorem, model-order selection (AIC, BIC, adjusted , Ljung-Box Q), simultaneity/endogeneity assessment, and the ARIMA/ARMAX / transfer-function connection.

Sources

See Also