nowcast: Nowcasting of a quarterly time series using a dynamic factor model.

Description

Estimate nowcasting and forecasting for a quarterly series. For more details read the Vignettes.

Usage

nowcast(y, x, q = NULL, r = NULL, p = NULL, method = "2sq",
  blocks = NULL)

Arguments

Stationary quarterly time series.

A monthly time series matrix (mts) representing regressors variables. The series must be stationary.

Dynamic rank. Number of error terms.

number of commom factors.

AR order of factor model.

method

There are three options: "2sq": "Two stages: quarterly factors" as in Giannone et al. 2008; "2sm": "Two stages: monthly factors" as in Ba<U+0144>bura and Runstler 2011; "EM": Expected Maximization as in Ba<U+0144>bura et al. 2011.

blocks

a binary matrix Nx3 that characterizes the regressors variables in global (1st column), nominal (2nd column) and real (3rd column). If NULL, the matrix assume 1 for all cells.

Value

A list containing two elements:

yfcst

the original y series and its in-sample and out-of-sample estimations.

reg

regression model between y and the estimated factors. Not available for EM method.

factors

the estimated factors and DFM model coefficients.

xfcst

the original regressors and their out-of-sample estimations.

month_y

the monthly measure for quarterly y variable. Only available for EM method.

References

Giannone, D., Reichlin, L., & Small, D. (2008). Nowcasting: The real-time informational content of macroeconomic data. Journal of Monetary Economics, 55(4), 665-676.<doi:10.1016/j.jmoneco.2008.05.010>

Ba<U+0144>bura, M., & R<U+00FC>nstler, G. (2011). A look into the factor model black box: publication lags and the role of hard and soft data in forecasting GDP. International Journal of Forecasting, 27(2), 333-346. <doi:10.1016/j.ijforecast.2010.01.011>

Ba<U+0144>bura M., Giannone, D. & Reichlin, L. (2011). Nowcasting, in Michael P. Clements and David F. Hendry, editors, Oxford Handbook on Economic Forecasting, pages 193-224, January 2011. <doi:10.1093/oxfordhb/9780195398649.001.0001>

Examples

Run this code

# NOT RUN {
### Method 2sq (two stages: quarterly factors)
gdp <- month2qtr(x = USGDP$base[,"RGDPGR"])
gdp_position <- which(colnames(USGDP$base) == "RGDPGR")
base <- Bpanel(base = USGDP$base[,-gdp_position],
               trans = USGDP$legend$Transformation[-gdp_position],
               aggregate = TRUE)
now2sq <- nowcast(y = gdp, x = base, r = 2, p = 2, q = 2, method = '2sq')

### Method 2sm (two stages: monthly factors)
base <- Bpanel(base = USGDP$base[,-gdp_position],
               trans = USGDP$legend$Transformation[-gdp_position],
               aggregate = F)
now2sm <- nowcast(y = gdp, x = base, r = 2, p = 2, q = 2, method = '2sm')

### Method EM
# selecting and transforming y  
gdp <- month2qtr(x = USGDPshort$base[,"GDPUS"])
gdp <- ts(c(gdp,NA,NA,NA,NA), start = start(gdp), frequency = 4)
gdp_stationary <- gdp/lag(gdp, k = -1) -1
gdp_position <- which(colnames(USGDPshort$base) == "GDPUS")

# selecting and transforming x 
base <- USGDPshort$base[,-gdp_position]
trans <- USGDPshort$legend[-gdp_position,"transformation"]
stationaryBase <- cbind(base[,trans == 1]/lag(base[,trans == 1], k = -1) - 1,
                        diff(base[,trans == 2]))
colnames(stationaryBase) <- colnames(base)[c(which(trans == 1),which(trans == 2)) ]
stationaryBase <- stationaryBase[,colnames(base)]

# DFM estimation via EM
blocks <- matrix(c(1,0,1,1,0,1,1,1,0,1,1,0,1,1,0,1,1,0,1,1,0,1,1,0,1,1,
                   0,1,1,0,1,1,0,1,0,1,1,1,0,1,1,0,1,0,1,1,1,0,1,1,0,1,
                   1,0,1,0,1,1,0,1,1,0,1,1,0,1,1,1,0,1,0,1,1,0,1,1,1,0), byrow = T, ncol = 3)
nowEM <- nowcast(y = gdp_stationary, x = stationaryBase, r = 1, p = 1, q = 1,
                 method = 'EM', blocks = blocks)
# }