💡NEW: Updated vintage of GFC Factor until 2019💡NEW: Updated Dataset for `US Monetary Policy and the Global Financial Cycle'💡NEW: Replication Code for `Uncertain Kingdom: Nowcasting GDP and its Revisions'Feb 1, 2020
C O D E & D A T A
This section collects data and MatLab codes that I have used in my papers. The content is organized per topic. Replication material for individual papers is marked within each topic.INSTRUMENTS FOR MONETARY POLICY SHOCKS
High-Frequency
Informationally robust monetary policy surprises at monthly and at FOMC meeting frequency for the US as used in Miranda-Agrippino & Ricco (2016, 2018). All series control for central bank information effects; monthly series also control for autocorrelation in monetary surprises. Details on the construction of the series are in Miranda-Agrippino & Ricco (2016) “The Transmission of Monetary Policy Shocks”
🔸download MPI instrument
Monthly monetary policy surprises corrected for Central Bank internal forecasts for the US and the UK as used in Miranda-Agrippino (2016). Details on the construction of the series are in Miranda-Agrippino (2016) “Unsurprising Shocks: Information, Premia and the Monetary Transmission”
🔸download orthogonal surprises
🔸download Online Appendix
Narrative (extension of Romer & Romer 2004)
Monthly and quarterly monetary policy surprises for the US constructed using narrative accounts of FOMC meetings since 1960. Series end at onset of ZLB. The folder contains data and documentation for the construction, and comparison with the original R&R series. Used in Miranda-Agrippino and Rey (2015), Miranda-Agrippino & Ricco (2016), Miranda-Agrippino, Hacioglu-Hoke & Bluwstein (2018)
🔸download narrative series & documentation
IMPULSE RESPONSE FUNCTIONS: BVAR, LP, & BLP
The folder contains replication files for Miranda-Agrippino & Ricco (2016) “The Transmission of Monetary Policy Shocks”. The code estimates impulse response functions for a Bayesian Vector Autoregression with standard NIW priors, Local Projections, and Bayesian Local Projections. Identification can be one of either Cholesky or External Instruments (BSVAR-IV & BLP-IV)
🔸download BLP replication files (1.1 MB)
PROXY SVAR / SVAR-IV
The folder contains MatLab code for estimation of impulse response functions (IRFs) in structural OLS VARs identified with external instruments and compares it with standard Cholesky ordering. In the SVAR-IV the two-step estimation and identification algorithm allows for different samples in the VAR innovations and selected IV. Standard errors are calculated using a Wild Bootstrap. The code replicates Figure IV of Miranda-Agrippino (2016) “Unsurprising Shocks: Information, Premia and the Monetary Transmission”
🔸download MatLab code (450 KB)
GLOBAL FINANCIAL CYCLE
>> UPDATED Global Factor <<
Global common factor estimated from world-wide cross section of risky asset prices. The new estimate is obtained by applying the methodology in Miranda-Agrippino & Rey (2015) to an updated panel of asset prices that counts 1004 monthly series from 1980 to 2019. Details are in Miranda-Agrippino, Nenova & Rey (2020)
🔸download updated GFC Factor data
Global Factor In World Risky Asset Prices
Global common factor estimated from world-wide cross section of risky asset prices. The data sheet contains two versions of the factor: 1) one estimated from 858 price series from 1990 to 2012; 2) one estimated from 3 from 1975 to 2010. All data are monthly. Details are in Miranda-Agrippino & Rey (2020) “US Monetary Policy and the Global Financial Cycle”, RESTUD
🔸download GFC Factor data
🔸code for DFM with blocks available in FACTOR MODELS
GFC VAR
The data sheets contain the data used in the VARs estimated in Miranda-Agrippino & Rey (2020). There are two versions of it. One (heavy) includes details on the construction of each series and interpolation of quarterly GFC variables into monthly equivalents. A lite version only includes data already transformed as they enter the GFC VARs. Details on composition and data construction are in the data appendix of Miranda-Agrippino & Rey (2020)
🔸download GFC dataset (Lite, 380 KB)
🔸download GFC dataset (Full, 1.6 MB)
FACTOR MODELS
BASICS
The folder contains MatLab code for estimation of factor models under different modelling assumptions. 1) Static and Exact with spherical idiosyncratic variance (PC); 2) Static with diagonal idiosyncratic variance (EM algorithm); 3) Dynamic Factor Model (EM algorithm). The code is associated with the Lecture Slides on Factor Models prepared for a guest lecture at the University of Surrey
🔸download MatLab code
🔸download Factor Models lecture slides
DFM with Block Structure
The folder contains MatLab code for estimation of a dynamic factor model with loading restrictions for estimation of block-specific factors. The code is associated with Miranda-Agrippino & Rey (2020) “US Monetary Policy and the Global Financial Cycle”. Due to restrictions on the distribution of asset price data, the demo code reads US macro variables instead
🔸download MatLab code (43 KB)
NOWCASTING WITH DATA REVISIONS: RA-DFM
The RA-DFM introduces a flexible way to model and forecast revisions to early releases of GDP in an otherwise standard mixed-frequency DFM. The folder contains MatLab code and a basic dataset on which it runs. The model is developed in Anesti, Galvao & Miranda-Agrippino (2018) “Uncertain Kingdom: Nowcasting GDP and its Revisions”
🔸download MatLab code (850 KB)
REAL-TIME MIXED-FREQUENCY DATASET FOR THE UK ECONOMY
The data sheet contains the real-time mixed-frequency dataset for the UK used in Anesti, Galvao & Miranda-Agrippino (2018) “Uncertain Kingdom: Nowcasting GDP and its Revisions”. Original ONS vintages come from the archives of the BoE. Release dates/times are either gathered from official sources or from Bloomberg’s ECO Calendar