Markov-Switching Dynamic Factor Models in Real Time

Abstract

We extend the Markov-switching dynamic factor model to account for some of the speci…cities of the day-to-day monitoring of economic developments from macroeconomic indicators, such as mixed-sampling frequency and ragged-edge data. First, we evaluate the theoretical gains of using promptly available data to compute probabilities of recession in real time. Second, we show how to estimate the model that deals with unbalanced panels of data and mixed frequencies and examine the bene…ts of this extension through several Monte Carlo simulations. Finally, we assess its empirical reliability to compute real-time inferences of the US business cycle and compare it with the alternative method of forecasting the probabilities of recession from balanced panels.Keywords: Business Cycles, Output Growth, Time Series. JEL Classi…cation: E32, C22, E27We a re in d e b te d to M a rc e lle C h a u ve t fo r k in d ly sh a rin g p a rt o f th e re a l-tim e d a ta v inta g e s u se d in th e e m p iric a l a p p lic a tio n . We th a n k th e e d ito r, th e a sso c ia te e d ito r a n d tw o a n o ny m o u s re v ie w e rs fo r th e ir c o m m e nts. P a rt o f th is p a p e r w a s w ritte n w h ile th e th ird a u th o r w a s v isitin g th e B a n k o f S p a in . F in a n c ia l su p p o rt fro m th e S p a n ish g ove rn m e nt, c o ntra c t g ra nts E C O 2 0 1 5 -7 0 3 3 1 -C 2 -1 -R a n d E C O 2 0 1 6 -7 6 1 7 8 -P (M IN E C O / F E D E R ), a n d 1 9 8 8 4 / G E R M / 1 5 (G ro u p s o f E x c e lle n c e , Fu n d a c ió n S é n e c a , a n d S c ie n c e a n d Te ch n o lo g y A g e n c y ), is g ra te fu lly a ck n ow le d g e d . A ny e rro rs a re o u r re sp o n sib ility. T h e v ie w s in th is p a p e r a re th o se o f th e a u th o rs a n d d o n o t re p re se nt th e v ie w s o f th e E u ro p e a n C o m m issio n , th e B a n k o f S p a in o r th e E u ro sy ste m . C o d e s a n d d a ta th a t re p lic a te th e re su lts c a n b e d ow n lo a d e d fro m th e a u th o rs' w e b site s.

Green shoots and double dips in the euro area: A real time measure

“…8 To examine this assumption, we estimated the common factors from both linear and Markov-switching dynamic factor models from the same dataset and we obtain that the sample correlation between both factors was 0.97. 9 This result is not surprising since Camacho and Perez-Quiros (2007) show that the persistence in a time series whose dynamics track the business cycle can be alternatively captured both by Markov-switching models and by linear autoregressive methods.…”

The Role of Indicator Selection in Nowcasting Euro Area GDP in Pseudo Real Time

“…The majority of studies simply convert all the data at the lower available frequency by taking quarterly averages of monthly indicators, and the ragged-(or jagged-) edge nature of the data requires that missing monthly observations for the quarter to be forecast are predicted usually with univariate autoregressive models; on this, see McGuckin et al (2007). 3 Camacho and Perez-Quiros (2010), Camacho et al (2012), Ferrara et al (2010), Giannone et al (2009, Kuzin et al (2009) are notable exceptions, as they respectively use approximate Kalman filter models, Markov-switching dynamic factors, non parametric methods, mixedfrequency VARs, and MIDAS regressions of Clements and Galvão (2008).…”

Monitoring the World Business Cycle

“…To overcome this drawback, we propose an extension of the Markov-switching DFM (MS-DFM) suggested in Camacho et al (2012). As in their proposal, the MS-DFM advocated by Kim and Yoo (1995), Chauvet (1998) and Kim and Nelson (1998) is enabled to deal with mixing frequencies, publication delays and di¤erent starting dates in the economic indicators.…”

“…3 Apart from GDP (loading factor of 0:22), the economic indicators with the largest loading factors are industrial production (loading factor of 0:47) and the VIX (loading factor of -0:22). As expected, the loading factors for all of the indicators except the VIX are positive and statistically signi…cant, indicating that these series are procyclical, i.e., positively correlated with the common factor that represents the world overall economic activity.…”

“…To check whether the business cycle information 3 The lag lengths used in the empirical exercise were always set to 2. However, we performed several exercises to check that our results were robust to other reasonable choices of the lag lengths.…”