API Reference

Complete API documentation for all classes and functions in the NARDL-Fourier library.

NARDL

Standard Nonlinear ARDL model (Shin, Yu & Greenwood-Nimmo, 2014).


    
    Python

class NARDL(data, depvar, exog_vars, decomp_vars, maxlag=4, ic='AIC', case=3)

Parameters

Parameter	Type	Description
`data`	pd.DataFrame	Time series data
`depvar`	str	Dependent variable name
`exog_vars`	list	Exogenous/control variables (linear)
`decomp_vars`	list	Variables to decompose (asymmetric)
`maxlag`	int	Maximum lag order (default: 4)
`ic`	str	Information criterion: 'AIC', 'BIC', 'HQIC'
`case`	int	Bounds test case (1-5, default: 3)

Attributes

Attribute	Type	Description
`model`	OLSResults	Fitted statsmodels OLS model
`best_lags`	dict	Optimal lag structure {p, q, r}
`best_ic`	float	Best IC value
`long_run`	dict	Long-run multipliers per variable
`short_run`	dict	Short-run coefficients
`ect`	dict	Error correction term info
`bounds_test`	dict	PSS bounds test results
`wald`	dict	Wald asymmetry tests
`diagnostics`	dict	Diagnostic test results

Methods

Method	Returns	Description
`summary()`	str	Full model summary

FourierNARDL

Fourier NARDL with smooth structural breaks (Zaghdoudi et al., 2023).


    
    Python

class FourierNARDL(data, depvar, exog_vars, decomp_vars, maxlag=4, max_freq=3, ic='AIC', case=3)

Additional Parameters

Parameter	Type	Description
`max_freq`	int	Maximum Fourier frequency to test (default: 3)

Additional Attributes

Attribute	Type	Description
`best_freq`	float	Optimal Fourier frequency k*
`fourier_test`	dict	Joint significance test for sin/cos terms

BootstrapNARDL

Bootstrap Fourier NARDL with bootstrap cointegration tests (Bertelli et al., 2022).


    
    Python

class BootstrapNARDL(data, depvar, exog_vars, decomp_vars, maxlag=4, max_freq=3,
                     ic='AIC', case=3, n_bootstrap=1000, bootstrap_case='III',
                     random_state=42)

Additional Parameters

Parameter	Type	Description
`n_bootstrap`	int	Number of bootstrap replications (default: 1000)
`bootstrap_case`	str	Case: 'II' or 'III' (default: 'III')
`random_state`	int	Random seed for reproducibility

Additional Attributes

Attribute	Type	Description
`bootstrap_test`	BootstrapCointegrationTest	Bootstrap test object
`bootstrap_results`	dict	Bootstrap test results

Additional Methods

Method	Returns	Description
`cointegration_decision()`	str	Decision with interpretation
`plot_bootstrap_distributions(figsize)`	Figure	Bootstrap distribution plots

PSSBoundsTest

PSS Bounds Test for cointegration.


    
    Python

class PSSBoundsTest(model, k, case=3)

Parameters

Parameter	Type	Description
`model`	OLSResults	Fitted NARDL model
`k`	int	Number of regressors
`case`	int	Case (1-5)

BootstrapCointegrationTest

Bootstrap cointegration tests following Bertelli et al. (2022).


    
    Python

class BootstrapCointegrationTest(model, data, depvar, decomp_vars, exog_vars,
                                  best_lags, n_bootstrap=1000, case='III',
                                  random_state=None)

Methods

Method	Returns	Description
`get_results()`	dict	Complete test results
`summary_table()`	pd.DataFrame	Formatted summary table

ResultsTable

Generate publication-ready tables.


    
    Python

class ResultsTable(model)

Methods

Method	Parameters	Description
`regression_table(output)`	output: 'dataframe', 'latex', 'html', 'markdown'	Regression results table
`long_run_table(output)`	output: format string	Long-run multipliers table
`short_run_table(output)`	output: format string	Short-run coefficients table
`diagnostics_table(output)`	output: format string	Diagnostic tests table
`bounds_test_table(output)`	output: format string	Bounds test table
`wald_test_table(output)`	output: format string	Wald asymmetry tests table
`to_latex(filename, tables)`	filename: str, tables: list or 'all'	Export to LaTeX file

NARDLPlots

Visualization tools for NARDL models.


    
    Python

class NARDLPlots(model)

Methods

Method	Parameters	Description
`dynamic_multipliers(var, horizon, ci)`	var: str, horizon: int, ci: float	Dynamic multiplier plot
`cusum(alpha)`	alpha: significance level	CUSUM plot
`cusumsq(alpha)`	alpha: significance level	CUSUM of squares plot
`residual_diagnostics()`	-	Residual diagnostic plots

Utility Functions


    
    Python

from nardl_fourier.utils import (
    partial_sum_decomposition,
    check_stationarity,
    compute_fourier_terms
)

Function	Description
`partial_sum_decomposition(x)`	Decompose series into positive/negative partial sums
`check_stationarity(x, test, alpha)`	ADF/PP/KPSS stationarity tests
`compute_fourier_terms(T, k)`	Compute sin/cos terms for frequency k

Diagnostic Functions


    
    Python

from nardl_fourier.diagnostics import (
    jarque_bera_test,
    breusch_godfrey_test,
    breusch_pagan_test,
    cusum_test,
    cusumsq_test,
    run_all_diagnostics
)

Function	Description
`jarque_bera_test(resid)`	Normality test
`breusch_godfrey_test(model, nlags)`	Serial correlation test
`breusch_pagan_test(model)`	Heteroskedasticity test
`cusum_test(resid, alpha)`	CUSUM stability test
`cusumsq_test(resid, alpha)`	CUSUM of squares test
`run_all_diagnostics(model)`	Run all diagnostics at once