The Spending Wave
This article is the first in a series showing how the ideas proposed by the futurist Harry Dent (when carefully developed) support the contention that the secular bear market predicted by Dent to begin in the 2007-2010 time frame actually should have begun in the 1999-2001 time frame. This, of course, implies that the stock market peak last year marked the beginning of the long-term bear market forecast in my book Stock Cycles. Harry S. Dent developed the concept of the spending wave in his 1993 book The Great Boom Ahead. Dent proposed that economic boom and bust times are determined by demographic factors. Individual spending tends to peak in the mid-forties. Thus, aggregate spending should correlate with the size of the mid-forties population. If we plot the estimated size of the mid-forties age cohort with year we obtain a generally rising trend that shows a number of peaks and troughs due to past variations in birth and immigration rates. For example fewer babies were born during the Great Depression than either before or afterward. Thus, we would expect that 45 years later (during the 1970's and early 1980's) there would be a corresponding drop in the number of middle age people. Dent proposes that economic boom times are associated with increasing size of the mid-forties population and that bust times are associated with a decreasing size of this population. He calls these oscillations in the number of middle-agers (at their peak-spending years) the spending wave.
In The Great Boom Ahead, Dent shows a graph that shows a close correlation between the Standard and Poor 500 index in constant dollars and the birth rate 44-46 years in the past. Since birth data have only been available since 1909, his graph begins in 1953. In his subsequent The Roaring 2000's, Dent adds on the effect of immigration. Since the average age of a new immigrant is around 30, Dent figures the impact of immigration should show up on spending with a 14-16 year lag. So the modified spending wave is obtained as the sum of the birth rate lagged 44-46 years and the immigration rate lagged 14-16 years.
I extended Dent's graph backwards in time using two methods. The first method employed a crude estimate for birthrate before 1909 obtained from census data and immigration data obtained from the Immigration and Naturalization Service. It was calculated using Dent's method except I employed a fixed 45 year lag for births and 15 year lag for immigration. For the second method I assumed that the number of people in their mid-forties would reflect the numbers of twenty-year olds entering the work force some 25 years earlier. That is, the size of the mid-forties cohort in 1870 should reflect the size of the approximately age 20 cohorts entering the work force around 1845. I further assumed that the rate of growth in the labor force was proportional to the numbers of people entering the work force (that is the numbers of older workers leaving the workforce through death or retirement was small compared to the numbers of young people entering the work force). Thus, I used the rate of labor force growth over 1840 to 1850 as an estimate for the size of the cohorts entering the work force in 1845 and hence for the size of the middle-aged cohorts in 1870. I calculated the average rate of labor force increase for the periods 1840-1850, 1841-1851, 1842-1852 etc. to obtain estimates for the spending wave for 1870, 1871, 1872 etc. The two sets of estimates were combined to give the estimated spending wave for the period before 1953. The result is plotted in Figure 1 as the bold line.
Figure 1. The S&P 500 index (crosses) and numbers of age-45 consumers (bold) over time.
In Figure 1, the spending wave is compared to the inflation-adjusted Standard & Poor 500 Index. Figure 1 shows that the post-war and roaring twenties booms do appear to be associated with rising numbers of midlifers and that the Great Depression and 1970's busts do seem to be associated with declining numbers of midlifers. The spending wave is the key concept in Dent's book The Great Boom Ahead and is the reason for its title. Using his spending wave model Dent predicts a continuation in the present economic and stock market boom until around 2007 or later. Figure 1 shows 2006 since I use a fixed age of 45, whereas Dent uses 46 for the baby boomers (and 44 for earlier generations) and so gets 2007. The shape of the post-1953 spending wave in Figure 1 is the same as what Dent shows in his books.
Figure 1 provides compelling evidence for the existence and importance of demographic cycles on the stock market and, by extension, the economy. Based on the spending wave, Dent predicts that the current economic boom should last until 2007. So intriguing is this correspondence that one wonders why demographers and economists haven't seen it before. Dent's peak in 2007 does not reflect the 1957 peak in births (which would give a spending wave peak in 2003), but rather, a peak in immigration in the early 1990's. Immigration was over one million in 1990-91, up from half of that in the 1980's. This 1990-91 upsurge in immigration then produces a spending wave peak some 15 or 16 years later. Similarly, immigration averaged over one million annually in the decade before 1915 versus less than half of that after1915, producing a drop in the spending wave 15 years later (i.e. around 1929). The way Dent formulates his spending wave, it is the absolute changes in immigration (or birth) rate that affect the economy. The absolute size of the change in total immigration in 1915 and 1990 were about the same. But the first occurred in a population 2.5 times smaller than the second. It seems logical that we should focus on birth and immigration rates relative to the population, rather than absolute numbers
To explore this idea further, Figure 2 shows a smoothed population-relative spending wave in bold. The raw spending wave was summed over a running seven year period and divided by the average population over than seven year period. In this way the spending wave can be interpreted as the approximate fraction of the population in their maximum spending years. Peaks and valleys are seen, but the sharp 2006 peak is now spread out into a 2002-2006 period, suggesting a somewhat earlier termination of the bull market that what Dent projected. The detailed correspondence with the stock index is not so good, however. First we notice that the general trend is downward, reflecting the increasing share of the population occupied by the elderly. What this suggests is that an increasing share of consumer spending will be tied to elder spending that is based on mostly fixed incomes. This trend argues for smaller economic fluctuations resulting from demographic shifts. The second big difference is the virtual absence of a demographically-determined post-war spending wave boom. What this means is that increased numbers of middle-aged people per se can not explain the post-war economic boom. Here we can see why demographers and economists have missed seeing Dent's spending wave. It wasn't a factor in the much-studied postwar economic boom.
Figure 2. The modified spending wave
What can explain the post-war boom were the strongly rising wages during this period, reflecting strong productivity growth. Two things are required for there to be a rise in spending. One is the desire to spend and the other is the means to spend. The first ought to be influenced by the population-relative spending wave in much the way that Dent hypothesizes. The second depends on growth in people's earnings and the employment level. We can construct a "wage-adjusted" spending wave by multiplying the smoothed spending wave by the product of the median real wage and the percentage of workers who are employed (see Figure 2). This adjusted spending wave correlates well with the post-war stock boom. Clearly there was a spending wave in the 1950's and 1960's that arose not so much from an increase in the propensity to spend (demographics), but rather, from an increase in the ability to spend (wages & employment).
The adjusted spending wave shows a peak in 1969 and then a long decline until the early 1980's, reflecting the combined effects the rapid fall-off in the unadjusted spending wave after 1969 and stagnant wages after 1973. The rise of the wage-adjusted spending wave in the 1980's and 1990's reflects the strong growth in the middle-aged population after 1982, even though wages have remained stagnant. If the economy were to fall into recession in 2001, as seems increasingly likely, wages would likely fall in real terms and unemployment would certainly rise. We can use the relative changes in wages and employment during the 1990 recession as a template for predicting how the wage-adjusted spending wave might change after the year 2000 if we were to have a recession. This projection is also shown in Figure 2. We find that, assuming a recession beginning this year, the wage-adjusted spending wave will have peaked in 2000, suggesting that the stock market peak in 2000 (and not some later peak around 2007) was the end of the secular bull market.