| Constructor/destructor of SimDesign | |
| SimDesign | Constructor: sets client object and variable type identifiers. |
| ~SimDesign | Destructor. |
| Public member functions of SimDesign | |
| AddExtrasToReference | Override this virtual function to add additional terms to the reference model |
| AddImpulseBreakToRefPar | Adds a break to a parameter vector formulated in terms of impulse dummies that are in the client database. |
| CreateDatabase | Creates database for the DGP in the client object |
| CreateReference | Creates the design, which is the reference model for the DGP and GUM |
| DummySaturation | Saturates the client database with the specified dummies. |
| FormulateDGP | Formulates the DGP model in the client object. |
| FormulateModel | Formulates the model in the client object. |
| FormulateModelInReference | Formulates a model in the client object with respect to the reference model. |
| FormulateReferenceModel | Formulates the reference model in the client object. |
| Generate | Generates X, u, y, Dy, stores them in the client database. |
| GenerateX | Generates X. |
| GetARParInDGP | |
| GetDGPno | Gets the DGP number. |
| GetDgpParCount | |
| GetObsCount | |
| GetObsInitCount | |
| GetRefConstantLabel | |
| GetRefModelObject | Gets the client object reference |
| GetRefParDGP | |
| GetRefParNames | |
| GetRefUParCount | |
| GetSigmaInDGP | |
| GetTitle | Gets the title |
| IsModelDeltaY | Returns TRUE if using Dy as dependent variable in FormulateModel, FALSE otherwise (default). |
| PreGenerate | To be called after setting the initial seed (optional) for the experiment, but before starting the Monte Carlo loop. |
| RemoveFromModel | Removes variables by name from the model. |
| SetDGP | Sets the DGP parameters in terms of the reference model. |
| SetDGPError | Sets the DGP error properties. |
| SetDGPno | Sets the DGP number. |
| SetModel | Sets the DGP parameters in terms of the reference model. |
| SetRenewX | Set TRUE if data on X is to be renewed for every replication, FALSE otherwise. |
| SetTitle | Sets the title |
| SetXAutoCorrelation | |
| SetXCorrelation | Creates a correlation matrix for X by setting rho. |
| TransformX1 | X transformation immediately after drawing the N[0,1] terms. |
| TransformX2 | X transformation after applying correlation (if any) to X. |
| UndoFormulateModelDeltaY | Reformulates the model in the client object from Dy to y, and re-estimates. |
| UseModelDeltaY | Set TRUE to use Dy as dependent variable in FormulateModel, FALSE otherwise. |
| Public enums of SimDesign | |
| enum1 | USE_CONSTANT_NONE, USE_CONSTANT_U, USE_CONSTANT_X |
| enum2 | USE_XCOR_MATCONST, USE_XCOR_MATPOWIJ, USE_XCOR_AUTO |
| enum3 | SAT_NONE, SAT_IIS, SAT_SIS, SAT_IIS_SIS |
| Protected data members of SimDesign | |
| m_asDummy | names of dummy saturation terms added to the database |
| m_asEps | names of y, residual and regressors X created in m_oRefModel |
| m_asPar_RM | names of the regressors of the reference model |
| m_asStep | optional step dummies added by Create(), with names Step[#:#] |
| m_asX | names of y, residual and regressors X created in m_oRefModel |
| m_asY | names of y, residual and regressors X created in m_oRefModel |
| m_bHaveX | TRUE if m_oRefModel has data on X |
| m_bModelDeltaY | TRUE: using Dy as dependent variable in FormulateModel |
| m_bRenewX | TRUE if data on X is to be renewed for every replication |
| m_bTrend_RM | TRUE: Trend in GUM |
| m_cStep_RM | no of step dummies in reference model (taken from m_asStep[0:]) |
| m_cTdiscard | no of observations to discard after generation |
| m_cTest | no of observations in estimation sample |
| m_cU_RM | max no of X and U variables (at lag 0) in reference model |
| m_cXlag_RM | lag-length of X's in reference model |
| m_cX_RM | max no of X and U variables (at lag 0) in reference model |
| m_cYlag_RM | max Y lag in reference model |
| m_dARParDGP | optional autoregressive error |
| m_dSigmaDGP | DGP uses N[0,sigma^2] |
| m_dXcorrelation | settings for X correlation |
| m_iConstant_RM | specifies constant in reference model, one of USE_CONSTANT_... |
| m_iDGP | DGP number (if any) |
| m_iPeriod1est | sample to use in estimation |
| m_iPeriod2est | sample to use in estimation |
| m_iUvar | the regressor type identifiers |
| m_iXcorrelationMode | settings for X correlation |
| m_iXvar | the regressor type identifiers |
| m_iYear1est | sample to use in estimation |
| m_iYear2est | sample to use in estimation |
| m_iYvar | the regressor type identifiers |
| m_oRefModel | the client Modelbase object |
| m_sTitle | title (string) |
| m_vParDGP_RM | DGP coefficients of the reference model (0: excluded) |
| m_vParModelNonZero_RM | model formulation in terms of the reference model |
| obj | reference to Modelbase derived client object. NB SimDesign takes ownership of this object, and will delete it when the destructor is called. |
| iYvar | identifies endogenous variables |
| iXvar | identifies free regressors |
| iUvar | identifies unrerstricted (fixed) regressors |
| vRefPar | reference parameter vector to add break to |
| dFraction >=2 | break of size dSize in every dFraction'th observation >=0 break of size dSize in first round(dFraction * m_cTest) observations >=-1 break of size dSize in last round(-dFraction * m_cTest) observations >=-100 break of size dSize in 5 blocks of round(-dFraction/100 * m_cTest / 5) obs evenly spread as: [bbbb .... bbbb .... bbbb .... bbbb .... bbbb] |
| dSize | scalar specifying size for each break (== 0: do nothing), or vector of break sizes (one entry for each break, first entry is for first break value, etc.). |
| sFile | if string: file to load |
| cTmax | no of observations in the created database |
| cXmax | no of regressors k in the created database |
| vSteps | empty or Sx2 matrix with step indices. This creates i=0,..,S-1 steps, wich are one
for observations vSteps[i][0] : vSteps[i][1], and zero otherwise.
First the database is created (from file or empty) with cTmax observations. Deterministic(2) is called to add Constant, Trend and Seasonal. Then y, Dy and u are appended, followed by x1,..,xk. Finally, if vSteps is not <> then it specifies the step dummies to add to the DGP, with names Step[#:#] where # is the observation index. Remember that an Ox program starts with the default seed. I normally use the convention to run each experiment of the default seed (so related experiments use the same data). However, this could cause problems if random data is created here, because it can be identical to that in the first replication. |
| cYlag | no of lags for dependent variable |
| cX | no of regressors |
| cXlag | lag length for regressors |
| iConstantMode | treatment of Constant: USE_CONSTANT_NONE, USE_CONSTANT_U, USE_CONSTANT_X |
| bTrend | TRUE for a Trend in the GUM |
| cStep | number of steps in the GUM: added to the database by CreateDatabase, and selected in that order. |
| cTdiscard | no of observations to discard. The actual number discarded is max(cTdiscard,cYlag,cXlag). |
| cTest | no of observations to use for estimation
Calls FormulateReferenceModel to formulate the reference model in the client object. Calls SetModel to make the model the same as the reference model (this can be changed later). |
| sType | type: "I:", "DI:", "S1:", "S". |
| cTdiscard | no of leading dummy terms to discard. |
| oClient | the client object. |
| oClient | the client object. |
| oClient | target client object |
| vParInRef | parameter values, non-zero selects the corresponding term |
| iConstantMode | mode for the Constant (if included) |
| bUseDeltaY | if TRUE, the model is formulated with Dy instead of y The order of variables is: 1) y or Dy 1) Constant (if any: Xvar or Uvar) 2) Trend (if any) as Xvar 2) lags of y: y_1,..,y_p as Yvar 3) X regressors: x1,..x1_r, x2,..x2_r, ... 4) Step breaks created in CreateDatabase as Xvar 5) Dummies as Xvar 6) AddExtrasToReference So the constant is always the last term. |
| oClient | target client object |
| oClient | the object to generate data into. This should be a clone of the reference object (so that the reference model is already formulated). |
| iRep | replication index (0 is first) |
| iRep | replication index (0 is first) |
| cT | sample size |
| asPar | array of names to remove (give zero parameter in reference model). |
| vPar | parameters for the DGP, corresponding to the order in the reference model. |
| dESE | standard error (default 1) |
| dARPar | autoregressive parameter (default 0) |
| vPar | parameters for the DGP, corresponding to the order in the reference model. A non-zero value indicates that the variable is in the model. |
| dCorrelation | rho |
| bConstant | USE_XCOR_MATCONST: one on diagonal, rho otherwise; USE_XCOR_MATPOWIJ: rho^|i-j|; USE_XCOR_AUTO: autocorrelated regressors, scaled by sqrt(1 - rho^2) NB With USE_XCOR_AUTO the first observation will be zero. Ideaaly this is taken into account by incrementing no of discards when calling CreateReference. |
| mX | generated X |
| mX | correlated X |
| oClient | the client object. |