Figure 1, titled “Components of the Bad-Environment Good Environment Distribution,” has three panels. All three panel plot different probability densities. The top panel, labeled “Good Component” shows a density that is peaked in the middle, but skewed to the right, having a longer right tail. The middle panel, labeled “Bad Component,” shows a density that is peaked in the middle but skewed to the left. The bottom panel, labeled “Sum,” is peaked in the middle, but with symmetric tails.
Figure 2, titled “Time-varying Shape Parameters of Bad Environment – Good Environment Distribution,” has two panels. The top panel, labeled “Large pt – Good environment,” shows a distribution that is skewed to the right. The bottom panel, labeled “Large nt – Bad Environment,” shows a distribution that is skewed to the left.
Figure 3, titled “Filtered Quarterly Demand and Supply Shocks,” has two panels. The top panel, labeled “Demand Shocks,” has time on the horizontal axis, with year labels running from 1965 to 2015. The vertical axis runs from -6 to +4. Vertical gray shading denotes periods of NBER-defined recession. A blue line, the demand shocks, is plotted over the full time span. The series shows little persistence. It is generally more volatile and tends to have large negative spikes during recessions that occurred around 1980, 1982, and 2009. The bottom panel is labeled “Supply Shocks.” The axes labels are the same as for the top panel. The paper also shows gray shading to denote NBER recessions. A blue line plots supply shocks. The series shows little persistence, and it also tends to spike down during recessions, particularly during those in the early 1970s, and again in 2009.
Figure 4 is titled “Filtered Quarterly Demand and Supply Variances.” It has three panels. All three panels show time on the horizontal axes, with year labels running from 1965-2015. The vertical axes run from 0 to 4. NBER recession shading is shown using gray vertical bars. The top panel, labeled “Panel A: Demand Variances” shows two lines corresponding to good variance (in green) and bad variance (in red). The green line shows a large hump, rising from around 1970, peaking around 1980, and then subsiding to near zero by 1990. The red line shows a different pattern. It generally stays at a low level except that is spikes up during some recessions. In particular, it spikes up to about 1.5 during the recession of 1982, to about 1 around the time of the 2000 recession, it to about 3½ during the recession in 2009. The middle panel is titled “Panel B: Supply variances.” It also has a green line for good variance and a red line for bad variance. The prominent pattern for the green line is a smooth secular decline from a level of about 1 around 1970 to less than one-half by 2000. The red line usually stays at a low level but it spikes up during some recessions, in particular those of 1975, 1980, 1990, and 2009. The bottom panel is labeled “Demand and Supply Variances.” Good variance, the green line, shows a large hump around 1980, and then generally declines until the 2009 recession, when it spikes up. The red line generally maintains a low level, but it spikes up during most recessions, with the largest spikes in the recessions of 1975 and 2009.
Figure 5 is titled “Quarterly Conditional skewness of Macroeconomic Variables.” It has four panels. All four panels show time on the horizontal axes, with year labels running from 1965-2015. The vertical axes run from -0.5 to 0. NBER recession shading is shown using gray vertical bars. The top panel is titled “Conditional real GDP growth skewness.” It shows a horizontal line at 0 for reference. A blue line plots the estimated conditional skewness for real gdp growth. The line takes on both positive and negative values, but it spends more time in negative territory. The line spikes deep into negative territory during some recessions. In particular, it spikes down during the recessions of 1982, 1990, 2000, and 2009. It remains negative at the end of the sample period. The second panel is titled “Conditional unemployment gap skewness.” It shows a horizontal line at 0 for reference. A blue line plots the estimated conditional skewness for the unemployment rate gap. The line takes on both positive and negative values, but it spends more time in positive territory. The line spikes deep into positive territory during some recessions. In particular, it spikes up during the recessions of 1982, 1990, 2000, and 2009. It remains positive at the end of the sample period, and the average level in the final two decades is higher than earlier in the sample. The third panel is titled “Conditional inflation skewness.” It shows a horizontal line at 0 for reference. A blue line plots the estimated conditional skewness for inflation. The line takes on both positive and negative values, but it spends more time in negative territory. The line spikes deep into negative territory during some recessions. In particular, it spikes up during the recessions of 1982, 1990, 2000, and 2009. It remains negative at the end of the sample period, and the average level in the final two decades is more negative than earlier in the sample. The fourth panel is titled “Conditional core inflation skewness.” It shows a horizontal line at 0 for reference. A blue line plots the estimated conditional skewness for core inflation. The line takes on both positive and negative values, but it spends more time in negative territory. The line spikes deep into negative territory during some recessions. In particular, it spikes up during the recessions of 1982, 1990, 2000, and 2009. It remains negative at the end of the sample period, and the average level in the final two decades is more negative than earlier in the sample.
Figure 6 is titled, “Quarterly Conditional Covariance among macroeconomic variables.” Is has two panels. Both panels show time on the horizontal axes, with year labels running from 1965-2015. NBER recession shading is shown using gray vertical bars. The top panel is titled “Real growth – Inflation Covariance.” The vertical axes run from about -0.2 to 0.6 for the top panel. A horizontal line at zero is included for reference. There are five lines plotted showing the four components of the covariance under the model, and the total covariance. The line for total covariance, in blue, is generally positive. It shows a large positive hump in the 1970s and 1980s before subsiding to levels near zero by around 2000. It spikes back up into positive territory during the 2009 recession. The main drivers for the overall covariance are “good demand variance” plotted in green, which is responsible for the hum shape in the 1980s, and the “bad demand component” plotted in purple, which is responsible for the spike in the 2009 recession. These two components are uniformly positive. The two lines labeled “good supply variance” and “bad supply variance” are not as volatile as the lines for demand variance. The contributions for supply variance are uniformly negative. The bottom panel is titled “Unemployment gap – Core Inflation Covariance.” The vertical axes run from about -0.15 to 0.1 for the bottom panel. A horizontal line at zero is included for reference. There are five lines plotted showing the four components of the covariance under the model, and the total covariance. The line for total covariance, in blue, is generally negative. It shows a large negative hump in the 1970s and 1980s before subsiding to levels near zero by around 2000. It spikes back down into negative territory during the 2009 recession. The main drivers for the overall covariance are “good demand variance” plotted in green, which is responsible for the hum shape in the 1980s, and the “bad demand component” plotted in purple, which is responsible for the spike in the 2009 recession. These two components are uniformly negative. The two lines labeled “good supply variance” and “bad supply variance” are not as volatile as the lines for demand variance. The contributions for supply variance are uniformly positive.
Figure 7 is titled, “Explaining Realized 10-yeasr Bond Return variance.” It has one panel. The panel shows time on the horizontal axis, with year labels running from 1965-2015. The vertical axis runs from -1e-3 to +8e-3. Three time series are plotted. A blue line shows actual realized volatility, labeled “Data.” It is relatively volatile, with large positive spikes, the two largest of which span from 1980-1983, and again from 2008-2010. A red line labeled “Macro factors+macro risks” plots the fitted values from that model. It fits the general pattern for actual realized quite well, particularly the two large spikes in 1980-1983 and 2008-2010. A green line is labeled “Macro level factors+macro risk factors+financial factors” and it plots the fitted values for that model. It lies virtually on top of the green line, indicating that the two models fit the data about equally well.