1. Introduction. Hang Qian 1 Iowa State University

Size: px
Start display at page:

Download "1. Introduction. Hang Qian 1 Iowa State University"

Transcription

1 Users Guide to the VARDAS Package Hang Qian 1 Iowa State University 1. Introduction The Vector Autoregression (VAR) model is widely used in macroeconomics. However, macroeconomic data are not always observed at the same frequency. For example, we might have quarterly GDP and monthly CPI data. Shall we average monthly data into quarterly data, or interpolate monthly data from quarterly data? Neither. With the Var(ied) Da(ta) S(ampling), or VARDAS model, we can directly use data of different frequencies as if we were running a standard VAR model for economic forecasting as well as the impulse-response analysis. The VARDAS model is built upon a stationary VAR(p) model with highest frequency data. Complete data are unavailable since some variables might be observed at a lower frequency. Note that lower frequency data (aggregated variables) are mostly formed by linear combinations of high frequency data (disaggregated variables). For instance, the quarterly data are the sum or average of the three monthly data in that quarter. On the one hand, disaggregated variables themselves exhibit auto-correlations and cross-correlations in accordance with the VAR structure. On the other hand, the observed aggregated data impose linear constraints on the interactions of disaggregated variables. That is, these disaggregated variables must sum up to or average to some known numbers, namely the realizations of the aggregated variables. The VARDAS model is estimated in a Bayesian framework, where the latent variables can be conveniently handled by the Gibbs sampler with data augmentation techniques. The basic idea of the VARDAS model is that auto-correlations and cross-correlations make latent disaggregated variables follow a multivariate normal distribution (MVN), while the observed aggregated data further make it a constrained MVN. Once the model parameters are estimated, it is ready for forecasting and studying the dynamic effects of economic shocks as long as suitable identification conditions are imposed. In other words, after the stage of parameter estimation, the VARDAS model is equivalent to a standard reduced form VAR model. The way of recovering structural VAR model from its reduced form remains unchanged. In fact, choosing the VARDAS model instead of aggregating all the data to the lowest frequency has a major advantage on structural VAR identification with zero-contemporaneous-effect constraints. Since the VARDAS model operates on regressions with the (latent) high frequency data, contemporaneity means a shorter time, which effectively relaxes the identification assumption. 1 Contact author: hqi@iastate.edu or matlabist@gmail.com

2 2. Usage The VARDAS package is written in MATLAB, available at Technical details aside, the VARDAS package is friendly to users, who only need to provide the data and the computer will do the rest as routinely as a standard VAR model. The package also comes with graphic users interfaces (GUI) that allow loading the data from an EXCEL file and running the estimation with the mouse clicking. The data can be prepared in an EXCEL file (for GUI users) or stored in a MATLAB float type variable (for MATLAB function users). Suppose we have quarterly GDP and monthly CPI data. In an EXCEL file, prepare your data similar to those in Figure 1. Figure 1: Balanced Aggregation Data in EXCEL DATE GDP CPI Figure 2: Balanced Aggregation Data in MATLAB As is seen in the spreadsheet, the quarterly data should be stored at the last entry of that quarter, namely March, June, September and December. Also make sure that the monthly data starts from the first month of a quarter and ends with the last month of a quarter. Otherwise, the computer has to truncate off the initial and/or last data points. Note that month-quarter aggregation is balanced aggregation, for there are always three months in a quarter. The computer works faster on balanced aggregation, though it does support unbalanced aggregation (see below). Once the computer reads those data into MATLAB, the data variable is shown in Figure 2, where all blank entries are replaced by the data type called. If you prefer to call MATLAB functions to run the VARDAS model, you can prepare your data similar to those in Figure 2. You are recommended to

3 add a transpose sign to your data variable (since during the estimation data are processed in that way), though the computer will do it for you if you do not. Sometimes we need to work on data of unbalanced aggregation. A leading example is that GDP data prior to 1947 does not have quarterly observations. So we encounter quarter-annual aggregation within a time series. In that case, prepare your data in EXCEL similar to those in Figure 3 or in a MATLAB variable similar to Figure 4. Figure 3: Unbalanced Aggregation Data in EXCEL Date GDP 1945Q1 1945Q2 1945Q3 1945Q Q1 1946Q2 1946Q3 1946Q Q Q2 160 Figure 4: Unbalanced Aggregation Data in MATLAB There is no need for the user to specify the aggregation structure, even for the unbalanced aggregation. By counting a run of immediately before a data point reveals the aggregation structure. (That is why an aggregated value is required to be recorded at the end of an aggregation circle.) The user only needs to tell the computer whether the aggregated data is formed by summation or averaging of the disaggregated data. Though the VARDAS package supports data of arbitrary frequencies and unbalanced aggregation in any fashion, the estimation time poses a challenge to a user s patience. The package contains several functions that differ in model generality and estimation speed. VARDAS1: This version only supports two frequencies and balanced aggregation. Some data are of low frequency and some are of high frequency Each series must stick to one frequency and cannot change its frequency halfway. For example: a VAR with monthly GDP and quarterly CPI, M1 etc. VARDAS2: This version only supports two frequencies. A time series changes its frequency once from low to high If more than one series change frequency, the change must occur at the same time For example: A time series with older data being annual and newer being quarterly VARDAS3:

4 This version supports unbalanced aggregation or mixed frequencies in any fashion Disaggregated data are sampled block by block The user specifies the size of the block The sampling speed is slow VARDAS4: This version supports unbalanced aggregation or mixed frequencies in any fashion Disaggregated data are sampled at one time from a giant MVN The sampling speed is very slow Let T be the number of observation, d be the number of component variables in the VAR system, p be the lag order of the VAR, and R be the number of draws in MCMC after the burn-in periods. The input arguments of these functions are Y: Variables in VAR (T-by-d matrix, or d-by-t matrix) nlag: number of lags in the VAR model sum_flag: specify how the latent high frequency data are grouped into low frequency data 1 = sum, 0 = average add_constant: whether to add a constant in the VAR regression 1 = add, 0 = do not add ndraws: number of draws in MCMC burn_in: number of burn-in draws in MCMC The output arguments of these functions are Phi_draws: posterior draws of phi(1),,phi(p), d-by-d-by-p-by-r 4-dimension array Sigma_draws: posterior draws of covariance matrix, d-by-d-by-r 3-dimension array constant_draws:posterior draws of the intercept (not the mean), d-by-r matrix For those who prefer to use graphic interfaces, double click setup.m and press F5 (or type setup in the command window), the GUI will pop up. Load the EXCEL data, click the relevant options, and press the button the estimate the model. In additional, the VARDAS package includes two bonus functions IRF_SHORT.m and IRF_LONG.m to compute the impulse response function of structural shocks. IRF_SHORT.m uses Cholesky decomposition, which effectively imposes one-way recursive restriction of zero contemporaneous effects. On the other hand, IRF_LONG.m identifies structural shocks by long-run neutrality constraints. It should be noted that recovering the structural model from the reduced-form VAR or VARDAS involves scientific insights on the economy and thus had better to be used with care.

5 3. Two Notes Differenced Data The VAR model in use is covariance stationary. However, many macroeconomic variables contain unit roots. It is common to put the first-differenced variables in the VAR system, though in the current model cointegration relations and error correction terms are not included. The VARDAS model with cointegration is left for future research. For differenced data, the aggregated data are not the sum or the simple average of the disaggregated data, but instead a weighed average. To see this, let *Y t + t t=1 be the latent monthly GDP series and we actually put ΔY t = Y t Y t 1 as a component variable in the VAR system. We observe the quarterly GDP series Y t,t+2 = Y t + Y t+1 + Y t+2, t = 1, 4, 7,.. Define the quarterly differenced data Δ 3 Y t,t+2 = Y t,t+2 Y t 3,t 1. The observable quarterly-differenced data and the unobservable monthly-differenced data are linked by the relation Δ 3 Y t,t+2 = 2 j=0 (Y t+j 2 = j=0 (ΔY t+j = ΔY t+2 Y t 3+j ) + 2ΔY t+1 + ΔY t 1+j + ΔY t 2+j ) + 3ΔY t + 2ΔY t 1 + ΔY t 2 To estimate a model with temporal aggregation of differenced data, use VARDAS_DIFF1.m if some data are of low frequency and some are of high frequency; use VARDAS_DIFF2.m if a time series changes its frequency once from low to high. Logarithmic Data Our model is a linear model that handles temporal aggregation by exploring the fact that normality is preserved under linear transformations. Using logarithmic variables in the VAR has many merits, but it also introduces nonlinearity in the aggregation structure. Suppose three monthly data are averaged into a quarterly data such that Y 1,3 = Y 1 + Y 2 + Y 3. If we put logarithmic data in the VAR system, we know (ln Y 1, ln Y 2, ln Y 3 ) are multivariate normal. However, ln Y 1,3 ln Y 1 + ln Y 2 + ln Y 3, which implies that conditional on observed ln Y 1,3 the disaggregated variates do not have a closed form distribution. One way to solve this problem is to redefine the disaggregated data such that ln Y 1,3 = ln Y 1 + ln Y 2 + ln Y 3, where *ln Y t + t t=1 are used as a component series in the VAR system. Note that under this definition the disaggregated data cannot be interpreted as the calendar monthly data. They only bear a statistical interpretation such that the geometric average of the latent ln Y 1, ln Y 2, ln Y 3 equals to the observed ln Y 1,3. The approximation error is almost negligible if monthly changes are small and the geometric averaging works well in practice.

6 4. Technical Details Consider a d-dimension VAR(p) system: Y t = c + Φ 1 Y t Φ p Y t 1 + ε t, where ε t ~N(0, Ω) The reference time unit is t = 1, T, which indexes the highest frequency data in the VAR system. Let Y = (Y t,, Y T ), which is unobservable since some of the component series may be observed at some lower frequencies. We have specified the way constructing the data matrix in Section 2 (Usage). Essentially, we record the aggregated data at the end of the aggregation circle, leaving other entries being. Now we add a transpose sign to this data matrix, and denote it as Y, which is a d-by-t matrix analogue to Y. Suppose Y were known, it was a standard VAR model. The package treats this system as a multi-equation regression model (also called seemingly unrelated regression, SUR), and assigns conjugate, proper and diffused priors to the parameters. Once we set {c, Φ 1,, Φ p } a normal prior and Ω an inverse Wishart prior, the posterior conditionals are normal and inverse Wishart respectively. Now conditional on all the model parameters {c, Φ 1,, Φ p, Ω} as well as our data Y, we discuss how to sample the latent Y. We use the symbol to vectorize a matrix column by column. If the initial Y 1,, Y p come from the stationary distribution of the VAR system, then Y ~N(μ, Γ), where μ = (μ 1,, μ 1 ), μ 1 = (I d Γ 0 Γ 1 Γ T 1 Γ Γ = 1 Γ 0 Γ T 2, ( Γ T 1 Γ T 2 Γ 0 ) p i=1 Φ i where the autocovariance matrix Γ j = E[(Y T μ)(y T j recursively by Γ j = i=1 Φ i Γ j i. p ) 1 c. The dt-by-dt covariance matrix is given by μ) ], j = 0,1,2, can be computed If we knew nothing about Y, the latent Y could be sampled directly from N(μ, Γ). However, our observation on the aggregated variables deepens our understanding on the disaggregated variables. In other words, the observed Y imposes linear constraints on the interaction of Y such that disaggregated variables must sum up to some known numbers, namely those realizations of aggregated data. This motivates us to construct a transformation matrix so as to link Y with Y. To this end, we examine the aggregation structure. Let E be a d-by-t logical matrix according to whether an entry in Y is or not. The transformation matrix A can be constructed in the following way. First, let the

7 main diagonal of A be ones and other elements be zeros. Second, examine each row of the logical matrix E. Suppose we are reading row i and column j of E. If the (i, j) entry is zero, skip and proceed to column j + 1 (or conclude this row). Otherwise, (i, j)entry being one implies a temporally aggregated data is observed. So we search column j 1, j 2, for a run of zeros. Suppose there are M zeros (M 0) in a row immediately before column j; we then add M ones to A. For > 0, the locations in A are row (j 1)d + i, column (j 1)d + i md, m = 1,, M. Note that M = 0 implies one-period trivial aggregation. The new series AY transforms the original series Y in such a way that for a (M + 1)-period temporal aggregation, the first M disaggregated variates are retained, while the last disaggregated variate is replaced by the sum of the disaggregated variates. As a special case, for a one-period aggregation, the variate is simply retained. Clearly, AY can be classified into two blocks: the latent disaggregated variates block and the observed variates block. The latter have their realizations contained in Y. To be exact, AY ~N(Aμ, AΓA ) and Y 0 Y ~N[η 0 + Γ 01 Γ 11 1 (Y 1 η 1 ), Γ 00 Γ 01 Γ 11 1 Γ 10 ], where Y 0 is the subvector of AY selected by the logical vector 1 E, Y 1 is the subvector of Y selected by the logical vector E η 0 and η 1 are two subvector of Aμ selected by the logical vector 1 E and E respectively. Γ 01 is the submatrix of AΓA with rows selected by 1 E and column selected by E Γ 00, Γ 11, Γ 10 are defined similarly. Lastly, Note that in the transformation we squeezed out one disaggregated variate at the end of the aggregation circle, and replaced it with an aggregated data. However, sampling the last disaggregated variates is trivial in that it must equal to the difference between the aggregated value and the sum of the rest disaggregated values. That finishes the cycle of the Gibbs sampler to the VARDAS model. The above sampling procedure is implemented in VARDAS4.m, though the sampling speed is slow due to large matrix manipulation. We can take advantage of the Markov property of the AR process to sample latent variables block by block, which increases the nodes on the MCMC chain but greatly improves the sampling speed. In the package, VARDAS1.m and VARDAS2.m sample latent variables of one aggregation circle in a block, and VARDAS3.m allows the user to choose the block size. Suppose we want to sample Y t,, Y t+j use the joint distribution of Y t p in the block. The idea of the block Gibbs sampler is that we only need to,, Y t,, Y t+j,, Y t+j+p to form the conditional normal distribution due to the Markov property. As for the two ends, we either truncate the variables in use or augment shadow data Y 0, Y 1,. VARDAS1.m and VARDAS2.m use the former method and VARDAS_DIFF1.m and VARDAS_DIFF2.m use the latter.

8 5. Concluding Remarks It is both feasible and sensible to make full use of varied frequency data in a VAR model. The VARDAS model does not involve complicated estimation techniques; it only trivially manipulates normal and conditional normal distributions. From the perspective of a user, all he needs to do is to provide the mixed data and the computer will do the rest. Comments, suggestions and bug reports are more than welcome and appreciated. Reference: Hang Qian, 2011, Vector Autoregression with Varied Frequency Data Full text available at

Bayesian Graphical Models for Structural Vector AutoregressiveMarch Processes 21, / 1

Bayesian Graphical Models for Structural Vector AutoregressiveMarch Processes 21, / 1 Bayesian Graphical Models for Structural Vector Autoregressive Processes Daniel Ahelegbey, Monica Billio, and Roberto Cassin (2014) March 21, 2015 Bayesian Graphical Models for Structural Vector AutoregressiveMarch

More information

Part 6: Multivariate Normal and Linear Models

Part 6: Multivariate Normal and Linear Models Part 6: Multivariate Normal and Linear Models 1 Multiple measurements Up until now all of our statistical models have been univariate models models for a single measurement on each member of a sample of

More information

STA 4273H: Statistical Machine Learning

STA 4273H: Statistical Machine Learning STA 4273H: Statistical Machine Learning Russ Salakhutdinov Department of Computer Science! Department of Statistical Sciences! rsalakhu@cs.toronto.edu! h0p://www.cs.utoronto.ca/~rsalakhu/ Lecture 7 Approximate

More information

Gibbs Sampling in Linear Models #2

Gibbs Sampling in Linear Models #2 Gibbs Sampling in Linear Models #2 Econ 690 Purdue University Outline 1 Linear Regression Model with a Changepoint Example with Temperature Data 2 The Seemingly Unrelated Regressions Model 3 Gibbs sampling

More information

at least 50 and preferably 100 observations should be available to build a proper model

at least 50 and preferably 100 observations should be available to build a proper model III Box-Jenkins Methods 1. Pros and Cons of ARIMA Forecasting a) need for data at least 50 and preferably 100 observations should be available to build a proper model used most frequently for hourly or

More information

Lecture 7a: Vector Autoregression (VAR)

Lecture 7a: Vector Autoregression (VAR) Lecture 7a: Vector Autoregression (VAR) 1 Big Picture We are done with univariate time series analysis Now we switch to multivariate analysis, that is, studying several time series simultaneously. VAR

More information

Lecture 7a: Vector Autoregression (VAR)

Lecture 7a: Vector Autoregression (VAR) Lecture 7a: Vector Autoregression (VAR) 1 2 Big Picture We are done with univariate time series analysis Now we switch to multivariate analysis, that is, studying several time series simultaneously. VAR

More information

TAKEHOME FINAL EXAM e iω e 2iω e iω e 2iω

TAKEHOME FINAL EXAM e iω e 2iω e iω e 2iω ECO 513 Spring 2015 TAKEHOME FINAL EXAM (1) Suppose the univariate stochastic process y is ARMA(2,2) of the following form: y t = 1.6974y t 1.9604y t 2 + ε t 1.6628ε t 1 +.9216ε t 2, (1) where ε is i.i.d.

More information

Warwick Business School Forecasting System. Summary. Ana Galvao, Anthony Garratt and James Mitchell November, 2014

Warwick Business School Forecasting System. Summary. Ana Galvao, Anthony Garratt and James Mitchell November, 2014 Warwick Business School Forecasting System Summary Ana Galvao, Anthony Garratt and James Mitchell November, 21 The main objective of the Warwick Business School Forecasting System is to provide competitive

More information

Vector Autoregression with Mixed Frequency Data

Vector Autoregression with Mixed Frequency Data MPRA Munich Personal RePEc Archive Vector Autoregression with Mixed Frequency Data Hang Qian June 2013 Online at http://mpra.ub.uni-muenchen.de/47856/ MPRA Paper No. 47856, posted 27. June 2013 04:29 UTC

More information

A Fully Nonparametric Modeling Approach to. BNP Binary Regression

A Fully Nonparametric Modeling Approach to. BNP Binary Regression A Fully Nonparametric Modeling Approach to Binary Regression Maria Department of Applied Mathematics and Statistics University of California, Santa Cruz SBIES, April 27-28, 2012 Outline 1 2 3 Simulation

More information

Identifying SVARs with Sign Restrictions and Heteroskedasticity

Identifying SVARs with Sign Restrictions and Heteroskedasticity Identifying SVARs with Sign Restrictions and Heteroskedasticity Srečko Zimic VERY PRELIMINARY AND INCOMPLETE NOT FOR DISTRIBUTION February 13, 217 Abstract This paper introduces a new method to identify

More information

Factor models. March 13, 2017

Factor models. March 13, 2017 Factor models March 13, 2017 Factor Models Macro economists have a peculiar data situation: Many data series, but usually short samples How can we utilize all this information without running into degrees

More information

Title. Description. var intro Introduction to vector autoregressive models

Title. Description. var intro Introduction to vector autoregressive models Title var intro Introduction to vector autoregressive models Description Stata has a suite of commands for fitting, forecasting, interpreting, and performing inference on vector autoregressive (VAR) models

More information

1 Teaching notes on structural VARs.

1 Teaching notes on structural VARs. Bent E. Sørensen February 22, 2007 1 Teaching notes on structural VARs. 1.1 Vector MA models: 1.1.1 Probability theory The simplest (to analyze, estimation is a different matter) time series models are

More information

Timevarying VARs. Wouter J. Den Haan London School of Economics. c Wouter J. Den Haan

Timevarying VARs. Wouter J. Den Haan London School of Economics. c Wouter J. Den Haan Timevarying VARs Wouter J. Den Haan London School of Economics c Wouter J. Den Haan Time-Varying VARs Gibbs-Sampler general idea probit regression application (Inverted Wishart distribution Drawing from

More information

Structural VARs II. February 17, 2016

Structural VARs II. February 17, 2016 Structural VARs II February 17, 216 Structural VARs Today: Long-run restrictions Two critiques of SVARs Blanchard and Quah (1989), Rudebusch (1998), Gali (1999) and Chari, Kehoe McGrattan (28). Recap:

More information

Identifying Aggregate Liquidity Shocks with Monetary Policy Shocks: An Application using UK Data

Identifying Aggregate Liquidity Shocks with Monetary Policy Shocks: An Application using UK Data Identifying Aggregate Liquidity Shocks with Monetary Policy Shocks: An Application using UK Data Michael Ellington and Costas Milas Financial Services, Liquidity and Economic Activity Bank of England May

More information

1 Linear Difference Equations

1 Linear Difference Equations ARMA Handout Jialin Yu 1 Linear Difference Equations First order systems Let {ε t } t=1 denote an input sequence and {y t} t=1 sequence generated by denote an output y t = φy t 1 + ε t t = 1, 2,... with

More information

BV4.1 Methodology and User-friendly Software for Decomposing Economic Time Series

BV4.1 Methodology and User-friendly Software for Decomposing Economic Time Series Conference on Seasonality, Seasonal Adjustment and their implications for Short-Term Analysis and Forecasting 10-12 May 2006 BV4.1 Methodology and User-friendly Software for Decomposing Economic Time Series

More information

Dynamic Factor Models and Factor Augmented Vector Autoregressions. Lawrence J. Christiano

Dynamic Factor Models and Factor Augmented Vector Autoregressions. Lawrence J. Christiano Dynamic Factor Models and Factor Augmented Vector Autoregressions Lawrence J Christiano Dynamic Factor Models and Factor Augmented Vector Autoregressions Problem: the time series dimension of data is relatively

More information

SUPPLEMENT TO MARKET ENTRY COSTS, PRODUCER HETEROGENEITY, AND EXPORT DYNAMICS (Econometrica, Vol. 75, No. 3, May 2007, )

SUPPLEMENT TO MARKET ENTRY COSTS, PRODUCER HETEROGENEITY, AND EXPORT DYNAMICS (Econometrica, Vol. 75, No. 3, May 2007, ) Econometrica Supplementary Material SUPPLEMENT TO MARKET ENTRY COSTS, PRODUCER HETEROGENEITY, AND EXPORT DYNAMICS (Econometrica, Vol. 75, No. 3, May 2007, 653 710) BY SANGHAMITRA DAS, MARK ROBERTS, AND

More information

1. Shocks. This version: February 15, Nr. 1

1. Shocks. This version: February 15, Nr. 1 1. Shocks This version: February 15, 2006 Nr. 1 1.3. Factor models What if there are more shocks than variables in the VAR? What if there are only a few underlying shocks, explaining most of fluctuations?

More information

Advanced Forecast. For MAX TM. Users Manual

Advanced Forecast. For MAX TM. Users Manual Advanced Forecast For MAX TM Users Manual www.maxtoolkit.com Revised: June 24, 2014 Contents Purpose:... 3 Installation... 3 Requirements:... 3 Installer:... 3 Setup: spreadsheet... 4 Setup: External Forecast

More information

Econometría 2: Análisis de series de Tiempo

Econometría 2: Análisis de series de Tiempo Econometría 2: Análisis de series de Tiempo Karoll GOMEZ kgomezp@unal.edu.co http://karollgomez.wordpress.com Segundo semestre 2016 IX. Vector Time Series Models VARMA Models A. 1. Motivation: The vector

More information

Factor models. May 11, 2012

Factor models. May 11, 2012 Factor models May 11, 2012 Factor Models Macro economists have a peculiar data situation: Many data series, but usually short samples How can we utilize all this information without running into degrees

More information

Vector Autoregression

Vector Autoregression Vector Autoregression Jamie Monogan University of Georgia February 27, 2018 Jamie Monogan (UGA) Vector Autoregression February 27, 2018 1 / 17 Objectives By the end of these meetings, participants should

More information

Factorization of Seperable and Patterned Covariance Matrices for Gibbs Sampling

Factorization of Seperable and Patterned Covariance Matrices for Gibbs Sampling Monte Carlo Methods Appl, Vol 6, No 3 (2000), pp 205 210 c VSP 2000 Factorization of Seperable and Patterned Covariance Matrices for Gibbs Sampling Daniel B Rowe H & SS, 228-77 California Institute of

More information

BEAR 4.2. Introducing Stochastic Volatility, Time Varying Parameters and Time Varying Trends. A. Dieppe R. Legrand B. van Roye ECB.

BEAR 4.2. Introducing Stochastic Volatility, Time Varying Parameters and Time Varying Trends. A. Dieppe R. Legrand B. van Roye ECB. BEAR 4.2 Introducing Stochastic Volatility, Time Varying Parameters and Time Varying Trends A. Dieppe R. Legrand B. van Roye ECB 25 June 2018 The views expressed in this presentation are the authors and

More information

Bayesian modelling of real GDP rate in Romania

Bayesian modelling of real GDP rate in Romania Bayesian modelling of real GDP rate in Romania Mihaela SIMIONESCU 1* 1 Romanian Academy, Institute for Economic Forecasting, Bucharest, Calea 13 Septembrie, no. 13, District 5, Bucharest 1 Abstract The

More information

Structural VAR Models and Applications

Structural VAR Models and Applications Structural VAR Models and Applications Laurent Ferrara 1 1 University of Paris Nanterre M2 Oct. 2018 SVAR: Objectives Whereas the VAR model is able to capture efficiently the interactions between the different

More information

Structural Vector Autoregressions with Markov Switching. Markku Lanne University of Helsinki. Helmut Lütkepohl European University Institute, Florence

Structural Vector Autoregressions with Markov Switching. Markku Lanne University of Helsinki. Helmut Lütkepohl European University Institute, Florence Structural Vector Autoregressions with Markov Switching Markku Lanne University of Helsinki Helmut Lütkepohl European University Institute, Florence Katarzyna Maciejowska European University Institute,

More information

Econ 424 Time Series Concepts

Econ 424 Time Series Concepts Econ 424 Time Series Concepts Eric Zivot January 20 2015 Time Series Processes Stochastic (Random) Process { 1 2 +1 } = { } = sequence of random variables indexed by time Observed time series of length

More information

Bayesian Mixed Frequency VAR s

Bayesian Mixed Frequency VAR s Bayesian Mixed Frequency VAR s Bjørn Eraker Ching Wai (Jeremy) Chiu Andrew Foerster Tae Bong Kim Hernan Seoane September 1, 28 Abstract Economic data can be collected at a variety of frequencies. Typically,

More information

Index. Pagenumbersfollowedbyf indicate figures; pagenumbersfollowedbyt indicate tables.

Index. Pagenumbersfollowedbyf indicate figures; pagenumbersfollowedbyt indicate tables. Index Pagenumbersfollowedbyf indicate figures; pagenumbersfollowedbyt indicate tables. Adaptive rejection metropolis sampling (ARMS), 98 Adaptive shrinkage, 132 Advanced Photo System (APS), 255 Aggregation

More information

Lecture 2: Univariate Time Series

Lecture 2: Univariate Time Series Lecture 2: Univariate Time Series Analysis: Conditional and Unconditional Densities, Stationarity, ARMA Processes Prof. Massimo Guidolin 20192 Financial Econometrics Spring/Winter 2017 Overview Motivation:

More information

Time Series Analysis. Smoothing Time Series. 2) assessment of/accounting for seasonality. 3) assessment of/exploiting "serial correlation"

Time Series Analysis. Smoothing Time Series. 2) assessment of/accounting for seasonality. 3) assessment of/exploiting serial correlation Time Series Analysis 2) assessment of/accounting for seasonality This (not surprisingly) concerns the analysis of data collected over time... weekly values, monthly values, quarterly values, yearly values,

More information

Computational Macroeconomics. Prof. Dr. Maik Wolters Friedrich Schiller University Jena

Computational Macroeconomics. Prof. Dr. Maik Wolters Friedrich Schiller University Jena Computational Macroeconomics Prof. Dr. Maik Wolters Friedrich Schiller University Jena Overview Objective: Learn doing empirical and applied theoretical work in monetary macroeconomics Implementing macroeconomic

More information

Chapter 6. Maximum Likelihood Analysis of Dynamic Stochastic General Equilibrium (DSGE) Models

Chapter 6. Maximum Likelihood Analysis of Dynamic Stochastic General Equilibrium (DSGE) Models Chapter 6. Maximum Likelihood Analysis of Dynamic Stochastic General Equilibrium (DSGE) Models Fall 22 Contents Introduction 2. An illustrative example........................... 2.2 Discussion...................................

More information

Variable Targeting and Reduction in High-Dimensional Vector Autoregressions

Variable Targeting and Reduction in High-Dimensional Vector Autoregressions Variable Targeting and Reduction in High-Dimensional Vector Autoregressions Tucker McElroy (U.S. Census Bureau) Frontiers in Forecasting February 21-23, 2018 1 / 22 Disclaimer This presentation is released

More information

Identification of Economic Shocks by Inequality Constraints in Bayesian Structural Vector Autoregression

Identification of Economic Shocks by Inequality Constraints in Bayesian Structural Vector Autoregression Identification of Economic Shocks by Inequality Constraints in Bayesian Structural Vector Autoregression Markku Lanne Department of Political and Economic Studies, University of Helsinki Jani Luoto Department

More information

Research Division Federal Reserve Bank of St. Louis Working Paper Series

Research Division Federal Reserve Bank of St. Louis Working Paper Series Research Division Federal Reserve Bank of St Louis Working Paper Series Kalman Filtering with Truncated Normal State Variables for Bayesian Estimation of Macroeconomic Models Michael Dueker Working Paper

More information

Econ 423 Lecture Notes: Additional Topics in Time Series 1

Econ 423 Lecture Notes: Additional Topics in Time Series 1 Econ 423 Lecture Notes: Additional Topics in Time Series 1 John C. Chao April 25, 2017 1 These notes are based in large part on Chapter 16 of Stock and Watson (2011). They are for instructional purposes

More information

STA 4273H: Statistical Machine Learning

STA 4273H: Statistical Machine Learning STA 4273H: Statistical Machine Learning Russ Salakhutdinov Department of Statistics! rsalakhu@utstat.toronto.edu! http://www.utstat.utoronto.ca/~rsalakhu/ Sidney Smith Hall, Room 6002 Lecture 11 Project

More information

FaMIDAS: A Mixed Frequency Factor Model with MIDAS structure

FaMIDAS: A Mixed Frequency Factor Model with MIDAS structure FaMIDAS: A Mixed Frequency Factor Model with MIDAS structure Frale C., Monteforte L. Computational and Financial Econometrics Limassol, October 2009 Introduction After the recent financial and economic

More information

Identifying the Monetary Policy Shock Christiano et al. (1999)

Identifying the Monetary Policy Shock Christiano et al. (1999) Identifying the Monetary Policy Shock Christiano et al. (1999) The question we are asking is: What are the consequences of a monetary policy shock a shock which is purely related to monetary conditions

More information

Labor-Supply Shifts and Economic Fluctuations. Technical Appendix

Labor-Supply Shifts and Economic Fluctuations. Technical Appendix Labor-Supply Shifts and Economic Fluctuations Technical Appendix Yongsung Chang Department of Economics University of Pennsylvania Frank Schorfheide Department of Economics University of Pennsylvania January

More information

Bayesian Inference for DSGE Models. Lawrence J. Christiano

Bayesian Inference for DSGE Models. Lawrence J. Christiano Bayesian Inference for DSGE Models Lawrence J. Christiano Outline State space-observer form. convenient for model estimation and many other things. Bayesian inference Bayes rule. Monte Carlo integation.

More information

Ch3. TRENDS. Time Series Analysis

Ch3. TRENDS. Time Series Analysis 3.1 Deterministic Versus Stochastic Trends The simulated random walk in Exhibit 2.1 shows a upward trend. However, it is caused by a strong correlation between the series at nearby time points. The true

More information

A BASE SYSTEM FOR MICRO TRAFFIC SIMULATION USING THE GEOGRAPHICAL INFORMATION DATABASE

A BASE SYSTEM FOR MICRO TRAFFIC SIMULATION USING THE GEOGRAPHICAL INFORMATION DATABASE A BASE SYSTEM FOR MICRO TRAFFIC SIMULATION USING THE GEOGRAPHICAL INFORMATION DATABASE Yan LI Ritsumeikan Asia Pacific University E-mail: yanli@apu.ac.jp 1 INTRODUCTION In the recent years, with the rapid

More information

Online appendix to On the stability of the excess sensitivity of aggregate consumption growth in the US

Online appendix to On the stability of the excess sensitivity of aggregate consumption growth in the US Online appendix to On the stability of the excess sensitivity of aggregate consumption growth in the US Gerdie Everaert 1, Lorenzo Pozzi 2, and Ruben Schoonackers 3 1 Ghent University & SHERPPA 2 Erasmus

More information

ECON 4160, Autumn term 2017 Lecture 9

ECON 4160, Autumn term 2017 Lecture 9 ECON 4160, Autumn term 2017 Lecture 9 Structural VAR (SVAR) Ragnar Nymoen Department of Economics 26 Oct 2017 1 / 14 Parameter of interest: impulse responses I Consider again a partial market equilibrium

More information

Gibbs Sampling in Latent Variable Models #1

Gibbs Sampling in Latent Variable Models #1 Gibbs Sampling in Latent Variable Models #1 Econ 690 Purdue University Outline 1 Data augmentation 2 Probit Model Probit Application A Panel Probit Panel Probit 3 The Tobit Model Example: Female Labor

More information

A primer on Structural VARs

A primer on Structural VARs A primer on Structural VARs Claudia Foroni Norges Bank 10 November 2014 Structural VARs 1/ 26 Refresh: what is a VAR? VAR (p) : where y t K 1 y t = ν + B 1 y t 1 +... + B p y t p + u t, (1) = ( y 1t...

More information

Stochastic process for macro

Stochastic process for macro Stochastic process for macro Tianxiao Zheng SAIF 1. Stochastic process The state of a system {X t } evolves probabilistically in time. The joint probability distribution is given by Pr(X t1, t 1 ; X t2,

More information

STA 414/2104: Machine Learning

STA 414/2104: Machine Learning STA 414/2104: Machine Learning Russ Salakhutdinov Department of Computer Science! Department of Statistics! rsalakhu@cs.toronto.edu! http://www.cs.toronto.edu/~rsalakhu/ Lecture 9 Sequential Data So far

More information

LECTURE 3 The Effects of Monetary Changes: Statistical Identification. September 5, 2018

LECTURE 3 The Effects of Monetary Changes: Statistical Identification. September 5, 2018 Economics 210c/236a Fall 2018 Christina Romer David Romer LECTURE 3 The Effects of Monetary Changes: Statistical Identification September 5, 2018 I. SOME BACKGROUND ON VARS A Two-Variable VAR Suppose the

More information

Cointegrated VARIMA models: specification and. simulation

Cointegrated VARIMA models: specification and. simulation Cointegrated VARIMA models: specification and simulation José L. Gallego and Carlos Díaz Universidad de Cantabria. Abstract In this note we show how specify cointegrated vector autoregressive-moving average

More information

Switching Regime Estimation

Switching Regime Estimation Switching Regime Estimation Series de Tiempo BIrkbeck March 2013 Martin Sola (FE) Markov Switching models 01/13 1 / 52 The economy (the time series) often behaves very different in periods such as booms

More information

Lecture 4: Bayesian VAR

Lecture 4: Bayesian VAR s Lecture 4: Bayesian VAR Hedibert Freitas Lopes The University of Chicago Booth School of Business 5807 South Woodlawn Avenue, Chicago, IL 60637 http://faculty.chicagobooth.edu/hedibert.lopes hlopes@chicagobooth.edu

More information

CONJUGATE DUMMY OBSERVATION PRIORS FOR VAR S

CONJUGATE DUMMY OBSERVATION PRIORS FOR VAR S ECO 513 Fall 25 C. Sims CONJUGATE DUMMY OBSERVATION PRIORS FOR VAR S 1. THE GENERAL IDEA As is documented elsewhere Sims (revised 1996, 2), there is a strong tendency for estimated time series models,

More information

NEW HOLLAND IH AUSTRALIA. Machinery Market Information and Forecasting Portal *** Dealer User Guide Released August 2013 ***

NEW HOLLAND IH AUSTRALIA. Machinery Market Information and Forecasting Portal *** Dealer User Guide Released August 2013 *** NEW HOLLAND IH AUSTRALIA Machinery Market Information and Forecasting Portal *** Dealer User Guide Released August 2013 *** www.cnhportal.agriview.com.au Contents INTRODUCTION... 5 REQUIREMENTS... 6 NAVIGATION...

More information

Y t = log (employment t )

Y t = log (employment t ) Advanced Macroeconomics, Christiano Econ 416 Homework #7 Due: November 21 1. Consider the linearized equilibrium conditions of the New Keynesian model, on the slide, The Equilibrium Conditions in the handout,

More information

A Primer on Vector Autoregressions

A Primer on Vector Autoregressions A Primer on Vector Autoregressions Ambrogio Cesa-Bianchi VAR models 1 [DISCLAIMER] These notes are meant to provide intuition on the basic mechanisms of VARs As such, most of the material covered here

More information

Forecasting with Bayesian Global Vector Autoregressive Models

Forecasting with Bayesian Global Vector Autoregressive Models Forecasting with Bayesian Global Vector Autoregressive Models A Comparison of Priors Jesús Crespo Cuaresma WU Vienna Martin Feldkircher OeNB Florian Huber WU Vienna 8th ECB Workshop on Forecasting Techniques

More information

Cointegrating Regressions with Messy Regressors: J. Isaac Miller

Cointegrating Regressions with Messy Regressors: J. Isaac Miller NASMES 2008 June 21, 2008 Carnegie Mellon U. Cointegrating Regressions with Messy Regressors: Missingness, Mixed Frequency, and Measurement Error J. Isaac Miller University of Missouri 1 Messy Data Example

More information

Bayesian Estimation of Input Output Tables for Russia

Bayesian Estimation of Input Output Tables for Russia Bayesian Estimation of Input Output Tables for Russia Oleg Lugovoy (EDF, RANE) Andrey Polbin (RANE) Vladimir Potashnikov (RANE) WIOD Conference April 24, 2012 Groningen Outline Motivation Objectives Bayesian

More information

Bayesian Dynamic Linear Modelling for. Complex Computer Models

Bayesian Dynamic Linear Modelling for. Complex Computer Models Bayesian Dynamic Linear Modelling for Complex Computer Models Fei Liu, Liang Zhang, Mike West Abstract Computer models may have functional outputs. With no loss of generality, we assume that a single computer

More information

New Information Response Functions

New Information Response Functions New Information Response Functions Caroline JARDET (1) Banque de France Alain MONFORT (2) CNAM, CREST and Banque de France Fulvio PEGORARO (3) Banque de France and CREST First version : February, 2009.

More information

Vector Autoregressive Model. Vector Autoregressions II. Estimation of Vector Autoregressions II. Estimation of Vector Autoregressions I.

Vector Autoregressive Model. Vector Autoregressions II. Estimation of Vector Autoregressions II. Estimation of Vector Autoregressions I. Vector Autoregressive Model Vector Autoregressions II Empirical Macroeconomics - Lect 2 Dr. Ana Beatriz Galvao Queen Mary University of London January 2012 A VAR(p) model of the m 1 vector of time series

More information

Using Model Selection and Prior Specification to Improve Regime-switching Asset Simulations

Using Model Selection and Prior Specification to Improve Regime-switching Asset Simulations Using Model Selection and Prior Specification to Improve Regime-switching Asset Simulations Brian M. Hartman, PhD ASA Assistant Professor of Actuarial Science University of Connecticut BYU Statistics Department

More information

ST 740: Markov Chain Monte Carlo

ST 740: Markov Chain Monte Carlo ST 740: Markov Chain Monte Carlo Alyson Wilson Department of Statistics North Carolina State University October 14, 2012 A. Wilson (NCSU Stsatistics) MCMC October 14, 2012 1 / 20 Convergence Diagnostics:

More information

7. MULTIVARATE STATIONARY PROCESSES

7. MULTIVARATE STATIONARY PROCESSES 7. MULTIVARATE STATIONARY PROCESSES 1 1 Some Preliminary Definitions and Concepts Random Vector: A vector X = (X 1,..., X n ) whose components are scalar-valued random variables on the same probability

More information

2.5 Forecasting and Impulse Response Functions

2.5 Forecasting and Impulse Response Functions 2.5 Forecasting and Impulse Response Functions Principles of forecasting Forecast based on conditional expectations Suppose we are interested in forecasting the value of y t+1 based on a set of variables

More information

Estimating and Identifying Vector Autoregressions Under Diagonality and Block Exogeneity Restrictions

Estimating and Identifying Vector Autoregressions Under Diagonality and Block Exogeneity Restrictions Estimating and Identifying Vector Autoregressions Under Diagonality and Block Exogeneity Restrictions William D. Lastrapes Department of Economics Terry College of Business University of Georgia Athens,

More information

Introduction to Probabilistic Graphical Models

Introduction to Probabilistic Graphical Models Introduction to Probabilistic Graphical Models Sargur Srihari srihari@cedar.buffalo.edu 1 Topics 1. What are probabilistic graphical models (PGMs) 2. Use of PGMs Engineering and AI 3. Directionality in

More information

Vector Auto-Regressive Models

Vector Auto-Regressive Models Vector Auto-Regressive Models Laurent Ferrara 1 1 University of Paris Nanterre M2 Oct. 2018 Overview of the presentation 1. Vector Auto-Regressions Definition Estimation Testing 2. Impulse responses functions

More information

STA 4273H: Statistical Machine Learning

STA 4273H: Statistical Machine Learning STA 4273H: Statistical Machine Learning Russ Salakhutdinov Department of Statistics! rsalakhu@utstat.toronto.edu! http://www.utstat.utoronto.ca/~rsalakhu/ Sidney Smith Hall, Room 6002 Lecture 3 Linear

More information

VAR Models and Applications

VAR Models and Applications VAR Models and Applications Laurent Ferrara 1 1 University of Paris West M2 EIPMC Oct. 2016 Overview of the presentation 1. Vector Auto-Regressions Definition Estimation Testing 2. Impulse responses functions

More information

Forecasting Levels of log Variables in Vector Autoregressions

Forecasting Levels of log Variables in Vector Autoregressions September 24, 200 Forecasting Levels of log Variables in Vector Autoregressions Gunnar Bårdsen Department of Economics, Dragvoll, NTNU, N-749 Trondheim, NORWAY email: gunnar.bardsen@svt.ntnu.no Helmut

More information

Technical Vignette 5: Understanding intrinsic Gaussian Markov random field spatial models, including intrinsic conditional autoregressive models

Technical Vignette 5: Understanding intrinsic Gaussian Markov random field spatial models, including intrinsic conditional autoregressive models Technical Vignette 5: Understanding intrinsic Gaussian Markov random field spatial models, including intrinsic conditional autoregressive models Christopher Paciorek, Department of Statistics, University

More information

1 Teaching notes on structural VARs.

1 Teaching notes on structural VARs. Bent E. Sørensen November 8, 2016 1 Teaching notes on structural VARs. 1.1 Vector MA models: 1.1.1 Probability theory The simplest to analyze, estimation is a different matter time series models are the

More information

New Introduction to Multiple Time Series Analysis

New Introduction to Multiple Time Series Analysis Helmut Lütkepohl New Introduction to Multiple Time Series Analysis With 49 Figures and 36 Tables Springer Contents 1 Introduction 1 1.1 Objectives of Analyzing Multiple Time Series 1 1.2 Some Basics 2

More information

Class 4: VAR. Macroeconometrics - Fall October 11, Jacek Suda, Banque de France

Class 4: VAR. Macroeconometrics - Fall October 11, Jacek Suda, Banque de France VAR IRF Short-run Restrictions Long-run Restrictions Granger Summary Jacek Suda, Banque de France October 11, 2013 VAR IRF Short-run Restrictions Long-run Restrictions Granger Summary Outline Outline:

More information

When Do Wold Orderings and Long-Run Recursive Identifying Restrictions Yield Identical Results?

When Do Wold Orderings and Long-Run Recursive Identifying Restrictions Yield Identical Results? Preliminary and incomplete When Do Wold Orderings and Long-Run Recursive Identifying Restrictions Yield Identical Results? John W Keating * University of Kansas Department of Economics 334 Snow Hall Lawrence,

More information

Structural Interpretation of Vector Autoregressions with Incomplete Identification: Revisiting the Role of Oil Supply and Demand Shocks

Structural Interpretation of Vector Autoregressions with Incomplete Identification: Revisiting the Role of Oil Supply and Demand Shocks Structural Interpretation of Vector Autoregressions with Incomplete Identification: Revisiting the Role of Oil Supply and Demand Shocks Christiane Baumeister, University of Notre Dame James D. Hamilton,

More information

Cross-sectional space-time modeling using ARNN(p, n) processes

Cross-sectional space-time modeling using ARNN(p, n) processes Cross-sectional space-time modeling using ARNN(p, n) processes W. Polasek K. Kakamu September, 006 Abstract We suggest a new class of cross-sectional space-time models based on local AR models and nearest

More information

Part 8: GLMs and Hierarchical LMs and GLMs

Part 8: GLMs and Hierarchical LMs and GLMs Part 8: GLMs and Hierarchical LMs and GLMs 1 Example: Song sparrow reproductive success Arcese et al., (1992) provide data on a sample from a population of 52 female song sparrows studied over the course

More information

VAR Model. (k-variate) VAR(p) model (in the Reduced Form): Y t-2. Y t-1 = A + B 1. Y t + B 2. Y t-p. + ε t. + + B p. where:

VAR Model. (k-variate) VAR(p) model (in the Reduced Form): Y t-2. Y t-1 = A + B 1. Y t + B 2. Y t-p. + ε t. + + B p. where: VAR Model (k-variate VAR(p model (in the Reduced Form: where: Y t = A + B 1 Y t-1 + B 2 Y t-2 + + B p Y t-p + ε t Y t = (y 1t, y 2t,, y kt : a (k x 1 vector of time series variables A: a (k x 1 vector

More information

Modelling trends in the ocean wave climate for dimensioning of ships

Modelling trends in the ocean wave climate for dimensioning of ships Modelling trends in the ocean wave climate for dimensioning of ships STK1100 lecture, University of Oslo Erik Vanem Motivation and background 2 Ocean waves and maritime safety Ships and other marine structures

More information

Appendix: Modeling Approach

Appendix: Modeling Approach AFFECTIVE PRIMACY IN INTRAORGANIZATIONAL TASK NETWORKS Appendix: Modeling Approach There is now a significant and developing literature on Bayesian methods in social network analysis. See, for instance,

More information

Outline. Overview of Issues. Spatial Regression. Luc Anselin

Outline. Overview of Issues. Spatial Regression. Luc Anselin Spatial Regression Luc Anselin University of Illinois, Urbana-Champaign http://www.spacestat.com Outline Overview of Issues Spatial Regression Specifications Space-Time Models Spatial Latent Variable Models

More information

CALCULATION OF FLOWS AND DISAGGREGATION OF ACCUMULATED VALUES

CALCULATION OF FLOWS AND DISAGGREGATION OF ACCUMULATED VALUES CENTRAL BANK OF ICELAND WORKING PAPERS No. 3 CALCULATION OF FLOWS AND DISAGGREGATION OF ACCUMULATED VALUES by Gudmundur Gudmundsson August 200 CENTRAL BANK OF ICELAND Economics Department Central Ban of

More information

A Introduction to Matrix Algebra and the Multivariate Normal Distribution

A Introduction to Matrix Algebra and the Multivariate Normal Distribution A Introduction to Matrix Algebra and the Multivariate Normal Distribution PRE 905: Multivariate Analysis Spring 2014 Lecture 6 PRE 905: Lecture 7 Matrix Algebra and the MVN Distribution Today s Class An

More information

1 Data Arrays and Decompositions

1 Data Arrays and Decompositions 1 Data Arrays and Decompositions 1.1 Variance Matrices and Eigenstructure Consider a p p positive definite and symmetric matrix V - a model parameter or a sample variance matrix. The eigenstructure is

More information

Topic 4 Unit Roots. Gerald P. Dwyer. February Clemson University

Topic 4 Unit Roots. Gerald P. Dwyer. February Clemson University Topic 4 Unit Roots Gerald P. Dwyer Clemson University February 2016 Outline 1 Unit Roots Introduction Trend and Difference Stationary Autocorrelations of Series That Have Deterministic or Stochastic Trends

More information

Web Appendix for The Dynamics of Reciprocity, Accountability, and Credibility

Web Appendix for The Dynamics of Reciprocity, Accountability, and Credibility Web Appendix for The Dynamics of Reciprocity, Accountability, and Credibility Patrick T. Brandt School of Economic, Political and Policy Sciences University of Texas at Dallas E-mail: pbrandt@utdallas.edu

More information

Online Appendix to: Marijuana on Main Street? Estimating Demand in Markets with Limited Access

Online Appendix to: Marijuana on Main Street? Estimating Demand in Markets with Limited Access Online Appendix to: Marijuana on Main Street? Estating Demand in Markets with Lited Access By Liana Jacobi and Michelle Sovinsky This appendix provides details on the estation methodology for various speci

More information

0.1 factor.bayes: Bayesian Factor Analysis

0.1 factor.bayes: Bayesian Factor Analysis 0.1 factor.bayes: Bayesian Factor Analysis Given some unobserved explanatory variables and observed dependent variables, the Normal theory factor analysis model estimates the latent factors. The model

More information

Nesting and Equivalence Testing

Nesting and Equivalence Testing Nesting and Equivalence Testing Tihomir Asparouhov and Bengt Muthén August 13, 2018 Abstract In this note, we discuss the nesting and equivalence testing (NET) methodology developed in Bentler and Satorra

More information