On business cycle forecasting

Frontiers of Business Research in China

Table 3 Out-of-sample estimation results: relative QPS and LPS

Forecast horizon (months)^a	1	3	6	9	12
Models	Relative QPS
Model with channels (a), (b), and (c) simultaneously^b
AR-Logit-Factor-MIDAS	0.485 (0.000)	0.528 (0.000)	0.489 (0.000)	0.696 (0.004)	0.664 (0.001)
Models with either channel (a) or (b)
ST-Probit-MCF	0.612 (0.005)	0.631 (0.005)	0.657 (0.002)	0.725 (0.019)	0.686 (0.006)
AR-Probit-YS	0.981 (0.672)	0.949 (0.224)	0.944 (0.356)	0.877 (0.001)	0.867 (0.000)
Models without any channel (a), (b) and (c)
ST-Probit-YS-EI	0.889 (0.042)	0.889 (0.070)	0.962 (0.612)	0.912 (0.188)	0.933 (0.314)
ST-Probit-YS	1	1	1	1	1
Models	Relative LPS
Model with channels (a), (b), and (c) simultaneously
AR-Logit-Factor-MIDAS	0.703 (0.006)	0.635 (0.028)	0.530 (0.000)	0.734 (0.009)	0.783 (0.002)
Models with either channel (a) or (b)
ST-Probit-MCF	0.784 (0.341)	0.691 (0.044)	0.701 (0.012)	0.790 (0.130)	0.874 (0.509)
AR-Probit-YS	0.923 (0.051)	0.887 (0.000)	0.904 (0.036)	0.856 (0.000)	0.846 (0.000)
Models without any channel (a), (b) and (c)
ST-Probit-YS-EI	0.869 (0.020)	0.872 (0.019)	0.931 (0.231)	0.893 (0.080)	0.903 (0.112)
ST-Probit-YS	1	1	1	1	1

^a For each forecast horizon N (where N = 1, 3, 6, 9, 12), we evaluate the ability of each model to forecast the probability of a recession within the next N months
^bChannel (a): using a flexible function form by including a lagged recession probability into a dynamic Logit or Probit model of recession forecasting; Channel (b): employing a dynamic factor model to extract common factors from many monthly or weekly economic and financial variables; Channel (c): applying MIDAS to incorporate the mixed-frequency common factors in the dynamic Logit framework