How to do econometrics with R

Cran task view: econometrics

Linear regression model (Linear regression models)

The z linear model can be fitted with OLS using the LM () function in the stats package, which also has various test methods for comparing models such as summary () and ANOVA ().

The Coeftest () and Waldtest () functions in the Žlmtest package are similar functions that also support asymptotic testing (for example: Z-Test instead of test, chi-square test instead of F-test).

The Linear.hypothesis () in the Žcar bag can test the more general linear hypothesis.

These features of the ŽHC and HAC covariance matrices can be implemented in the sandwich package.

The Žcar and Lmtest kits also provide a number of methods for regression diagnosis and diagnostic testing.

The z tool variable regression (two-phase least squares) is provided by the Ivreg () in the AER package, and the other one implements the Tsls () in the SEM package.

Micro-econometrics (microeconometrics)

Z Many microeconomic models belong to the generalized linear model, which can be fitted by the GLM () function of the stats package. Includes the Logit and Probit models for selecting class data (choice), a Poisson model for counting class data (count data). The values of these model regression elements can be obtained and visualized using effects.

The z negative two generalized linear model can be implemented by the GLM.NB () of the mass package. The AOD package provides another implementation of the negative two-item model and contains other models of over-dispersed data.

The z Edge (zero-inflated) and hurdle count models can be provided by the PSCL package.

Z Multiple responses (multinomial response): A specific ALADI variable (individual-specific covariates) Multiple models can only be provided by the nnet () function in the Multinom package. The Mlogit package implementation includes specific individuals and specific selection (choice-specific) variables. Generalized additive models of multiple responses can be fitted by Vgam packets. The Bayesian approach to multiple probit models is provided by the MNP package, and various Bayesian multiple models (including Logit and Probit) are available in the BAYESM package.

Z Sequential Response (Ordered response): The proportional advantage regression of sequential responses is implemented by the POLR () function in the mass package. Package ordinal provides a cumulative link model (models link cumulative) that includes a proportional advantage model (propotional odds models) and a more general specification for sequential data (ordered). The Bayesian sequential Probit model is provided by package Bayesm.

Z Censored Response (censored response): The basic censored regression model (for example, the Tobit model) can be fitted by the survival () function in the Suevreg package, a convenient interface Tobit () in the AER package. A more in-depth censored regression model, including a model of panel data, is provided by the Censreg package, and the sample selection model is available in the Sampleselection package.

Z Miscellaneous: A further refinement of the microeconomic tools is provided by the Micecon family package: Cobb-douglas analysis, Translog, two functions in micecon; scale elasticity unchanged (Constant elasticity of Scales, CES) function in miceconces; symmetric return one or two profit (symmetric normalized quadratic profit,snqp) function in MICECONSNQP; almost ideal demand function model system (almost Ideal Demand System, AIDS) function in Miceconaids package; stochastic Frontier Analysis (Stochastic Frontier analyses) in Frontier package The BAYESM package performs a Bayesian approach in micro-metrology and marketing (marketing); the relative distribution is inferred in the packet reldist.

Other regression models (further regression models)

Z Nonlinear least squares regression modeling can be implemented using the NLS () in the stats package.

Z (Quantile Regression): Quantreg (including linear, non-linear, censored, local polynomial, and additive-indexable regression).

Z Linear model of panel data: PLM. A Space Panel model package (SPLM) is being developed r-forge.

Z Generalized momentum methods (generalized method of moments,gmm) and generalized empirical likelihood (generalized empirical Likelihood,gel): GMM.

Z Linear structural Equation model: SEM, including two-stage least squares.

Z Simultaneous equations estimation: Systemfit.

Z Non-parametric kernel method: NP.

Žbeta regression: Betareg and GAMLSS

Z Intercept (Gaussian) regression: Truncreg.

Z Nonlinear mixed-effect model: Nlme and Lme4.

Z Generalized additive models: MGCV, GAM, GAMLSS and Vgam.

Z Miscellaneous: Package Vgam, design, and HMISC packages provide a number of extension tools for linear model processing (generalized), and Zelig is an easy-to-use unified interface for many regression models.

Basic Time Series architecture (basic infrastructure)

The "TS" Class of the Žstats package is the standard class for R's regular interval time series (especially for annual, quarterly, and monthly data).

The time series in the z "TS" format can be forced to be interchanged with "Zooreg" in the zoo package without losing information. Zoo package rules and irregular interval time series of the schema (the latter through the class "Zoo"), where the time information can be any class. This includes daytime sequences (typically, in the "date" time index) or intraday sequences (for example, the "POSIXCT" Time index).

The ITS, tseries, and TimeSeries (former fseries) packages, built on the "Posixt" time-date class, also provide the architecture of irregular interval time series, especially for financial analysis. Z

TimeSeries Modeling (Time series modelling)

The Žstats package has classic time-series modeling tools, Arima () functions for Arima Modeling and Box-jenkins-type analysis.

The Žstats package also provides a Structts () function to fit the structure time series.

Z can fit a linear regression model with an AR error term via OLS using the GLS () function in the Nlme package.

The filtering and decomposition of the Z time series can be stats with the decompose () and holtwinters () functions of the packet.

Z The expansion of these methods, especially prediction and model selection, in the forecast package.

There are various sequential filtering methods in the Žmfilter.

Z Estimation vector autoregressive (VAR) model, there are several methods available: The simple model can be stats in the Package AR () fitting, VARs package provides a more sophisticated model, DSE Estvarxls () and Bayesian method in Msbvar. The DYNLM package has a convenient interface to fit the dynamic regression model via OLS, and dyn implements a different method for other regression functions.

Z can fit more advanced dynamic equations with DSE.

Žtsdyn provides a variety of nonlinear Autoregressive time series models.

Z Gaussian linear state space model can be fitted using DLM (maximum likelihood, Kalman filter/smoothing, and Bayesian method).

Z Package Urca, tseries, and cadftest provide the unit root and cointegration techniques.

Z Time series Factor analysis in the TSFA package.

Z Package SDE provides simulation and inference of stochastic differential equations.

Z Asymmetric price conduction modeling in APT package.

Miscellaneous

Z Matrix operation (manipulations). As a vector and matrix language, R has a number of basic function processing matrices that complement the matrix and Sparsem packages.

Z Re-sampling (Bootstrap). In addition to the recommended boot package, there are some other conventional bootstrapping techniques in the bootstrap or Simpleboot package, and some functions are specifically designed for time series data, such as the maximum entropy bootstrap in the Meboot packet, The Tsbootstrap () function in the Tseries package.

Z Inequality (inequality). In order to measure inequalities (inequality), concentration (concentration) and poverty (poverty), the INEQ package provides some basic tools such as: roulunds curve (Lorenz curves), Pen ' s parade, Gini coefficient (Gini coefficient).

Z Structural changes (Structural change). R has a strong ability to handle structural changes and change points of parametric models, and can refer to Strucchange and segmented packages.

DataSet (data sets)

Žpackages AER and Ecdat contain a number of data sets from econometrics textbooks and magazines (applied econometrics, Business/economic statistics).

Žaer also provides a number of examples to reproduce the analysis from the textbook and literature, demonstrating various econometric methods.

Žfints is the R reference to Tsay's "analysis of Financial Time Series" (2nd ed., 2005, Wiley), containing the datasets, functions, and scripts needed to run some of these examples.

The Ždnmoney package provides Canadian currency money.

The ŽPWT package provides the Payne World table (Penn).

Z Package Expsmooth, FMA and Mcomp are forecasting with exponential smoothing:the State Space approach (Hyndman, Koehler, Ord, Snyde) R, Springer), "Forecasting:methods and Applications" (Makridakis, Wheelwright, Hyndman, 3rd ed., 1998, Wiley) and the M- Competitions "Time series Packet

The Z package erer contains the functions and datasets in the book "Empirical Research in economics:growing-with R" (Sun, forthcoming).

Source: http://blog.renren.com/blog/332766053/904482985