The golden age was a period of strong, but sometimes erratic, economic growth. To illustrate this, consider the quarterly rate of change since 1947 in gross domestic product (GDP.) GDP is an aggregate measure of all the goods and service produced in the country. It’s not perfect, but it is a reasonable barometer of the state of the economy, and growth rates tell you where it is headed. The Bureau of Economic Analysis (BEA) provides relevant data in many forms. Graph one is a data plot of quarterly percent change from the preceding period in real (inflation adjusted) GDP, seasonally adjusted at annual rates, from BEA. This plot is color-coded by presidential administration, blue for Democrats, and Red for Republicans. As you can see, this number jumps around a lot, but less so in recent decades. Also shown are the average values for all data from 1947 through 1980, 3.73%, in green, and the average from 1981 through the first quarter of 2011, 2.81%, in red. Note that since 1981, average GDP growth has been almost a full percentage point lower than it was during the Golden Age.
Also shown is an 8 year moving average in yellow. I chose an 8 year span for the moving average to smooth out over the length of a two-term presidency, but any long moving average will tell the same story. Until recently, the lowest dip in that (moving average) line bottoms out at 2.35% in 1982, and the highest bottoms out 2.71% in 1986, with no particular pattern across time. The lowest bottom of all occurred in the recent Great Recession, bottoming out at 1.54% in the second quarter of 2009. The dramatic change is in the tops. After a broad double peak from 1966 to 1969, topping out at 5.39% and 5.29%, the remaining peaks are between 3.99% and 4.07% - scarcely above the golden age total period average. The other notable feature is the steady decline in quarterly data peaks throughout the G. W. Bush administration and the corresponding slide in the eight year average, culminating in the crash of 2007.
During the Great Stagnation, the reduced volatility in the quarterly GDP growth numbers appears in the graph as generally less severe bottoms, and even more dramatically as less expansive tops. The degree of volatility can be precisely determined using the statistical technique of standard deviation. The next graph shows how the standard deviation has changed since WW II.
The blue curve is the standard deviation (Std Dev), based on the previous 34 quarters. We’ll get to the red line later. Sure enough, the Std Dev falls over time, as the blue best-fit trend line confirms, but it does not fall monotonically.
How can we account for this? In the 50's, there was economic turmoil, as the economy restabilized during the first wave of the baby boom, and about 7 million soldiers reentered the work force following WWII - which came hard on the heels of the Great Depression. This was a dynamic time, with wild growth peaks in 1950, '52, and '55, along with recessions in 1953, '58, and '59-60. Then came the First Little Moderation - a decade with steadier GDP growth and – most significantly, no recessions until late in 1969. The 70's brought stagflation, the end of the Viet Nam war, and another series of wild gyrations in GDP growth - though, except for one spike in 1978, and the recession of 1980 - not as wild as the '50's. Reagan's high spending and deficits held recessions at bay after 1982, and volatility began to decline a few years later. After a modest peak in late 1987, GDP growth declined as well, and the resulting recession contributed to denying George H.W. Bush a second term. The Clinton administration gave us relative prosperity, along with the Second Little Moderation – another multi-year span without a recession. Volatility remained steady at a low level throughout his two terms, before dropping further to an all-time low in 1999. Despite this, his policies were too conservative to rekindle anything like a golden age. By 2000, the wheels were about to come off the Clinton boom anyway, but the grotesque economic policies of the George W. Bush administration threw us farther over a deeper cliff than was necessary. His profligate spending made Reagan look prudent, and gave us an anemic and declining series of GDP mini-peaks, culminating in the crash of 2007. Barack Obama’s administration has not been able to foster significant growth and a robust recovery while dealing with a Congress dominated by blue dog Democrats and overtly obstructionist Republicans. It’s particularly interesting that the recent Great Recession barely registers as a blip on the blue line.
Which brings us to the red line in Fig 2: relative standard deviation (RSD), determined by dividing the standard deviation by the average of the same 34 data points, here multiplied by 3.5 to put it on the same scale as Std Dev. Part of the reason that the Std Dev became smaller over time is that the numbers used to calculate it were smaller. To this extent that this is operating, the Great Moderation is a meaningless data artifact. However, if this were strictly true, the best fit line for RSD would be exactly horizontal. It’s not, but it does have a smaller slope than does the best fit line for Std Dev, so the idea is not totally without merit. Let’s have a closer look.
By the end of the first Little Moderation, RSD dropped to a level below that of most of the Great Moderation. It then rose slightly, was stable from ’71 to ’73, and took a big jump with the recession of ’74. After slipping over the rest of the decade, RSD shot to an all time high during the recessions of ’80 and ’82. The big drop occurred between ’87 and ’90, and the low level was maintained until the Great Recession.
Volatility is the intensity of variability. Big changes in short time spans register as volatility, causing Std Dev and RSD to increase. As figure 1 shows, the biggest variations come from recessions and quick, strong recoveries. There were three recessions in the 50’s so RSD never had a chance to decline. The 60’s and the 90’s were both recession-free, so RSD could decline in the one case, and stay low in the other. The recessions of the 70’s were deep, but the recoveries were strong, so volatility climbed. The back-to-back recessions of ’80 and ’82, with sharp recoveries in ’81 and ’83 took RSD to an all-time high. The Great Recession propelled RSD back up to a level not seen since the 80’s. Simply, all of the volatility jumps can be explained in terms of recessions. Avoid recessions, and Std. Dev. will be low.
How, then, do we explain the two Bush administrations, with their recessions in 1990 and 2001, but no jump in volatility? In each case the fall into recession was not sudden – it followed several months of low or declining GDP growth. Similarly, in each case, the climb back out of recession was slow, faltering, and failed to generate even a single quarter where GDP growth topped 7%. In contrast, from WWII until 1983, top recovery quarters typically exceeded 10%.
What this indicates is that the Great Moderation really is a data artifact – though not quite in the way I expected. Reduced GDP growth numbers play a part, but the real key is understanding how recessions contribute to observed Std Dev. Recession-free times have low Std Dev values, and tepid recoveries from recessions that occur in a low growth context will also have low values. While the Great Moderation is real, the standard explanation is inadequate, and comes from failing to look at the data with a critical eye.
This was intended to be a chapter in a multi-author book project. The project fell through for reasons not related to the project itself.