Gary Feinman’s proposal that trends in economic inequality have a strong effect on who benefits from the technological innovations has resulted in a very lively discussion. Most comments seem to agree that there is definitely something to it, but question the details. In particular, when did the surge in labor-saving and convenience-creating technologies began? And how well does it coincide with the Great Compression, the period during the 20th century when incomes and wealth of common people were converging with those of the elites?
“The Great Compression” was coined by the Harvard economist Claudia Goldin and her co-author Robert Margo. They date it to the post-war period, late 1940s, 1950s and 1960s. However, as I argue in my forthcoming book on the structural-demographic analysis of American history, the great majority of indicators point to an earlier start – during the Progressive Era, and certainly no later than the New Deal.
Here’s the graph showing how the 1 percent did in terms of incomes and wealth:
Clearly, after the crash of 1929 and into the 1970s the trend was down, down. However, there are reasons to believe that the peak of 1929 is an artificial one (due to the bull market of the Roaring Twenties) and the actual peak was right around 1900, or at the very end of the Gilded Age (the last three decades of the nineteenth century). Unfortunately, good quantitative data for incomes is available only from 1913 and for wealth from 1922, so I will not insist on this interpretation here and now.
Various indicators of well-being, however, clearly had their turning points around 1900-1910:
Of particular interest is the curve of relative wages (wages divided by GDP per capita). This statistic tells us what proportion of the fruits of economic growth goes to workers, as opposed to employers. Relative wages began increasing during the Progressive Era (the first two decades of the twentieth century). The overall curve of well-being (averaged over different proxies) also began growing during the Progressive Era. So the start of the Great Compression should be dated as c.1910 (or, if you want a range, between 1900 and 1920). Interestingly, the Great Depression had a short-lasting or even no effect on various well-being indicators, which continued to grow during the 1930s.
In light of this re-periodization of the Great Compression, let’s look at the automobile data. The Historical Statistics of the United States provides several data sets relevant to this issue, but only one extends all the way to 1900, the number of motor vehicle registrations. Here’s what the data look like, when scaled by the US population:
We see that the curve takes off after 1910 and that during the 1920s the automobile truly becomes an item of mass consumption: 2 registrations for 10 Americans (babies included). There are a couple of setbacks associated with the Great Depression and WWII, and then the curve resumes its surge.
Curiously, it hits a ceiling right after 1980s. Perhaps that indicates the saturation of the market, but if so, how do we explain the decline after 1990? Unfortunately the data stops in 1995, so we don’t know what happened after that.
I conclude that the growth in automobile ownership fits the Great Compression very well. So the Feinman thesis (modified to account for an earlier start of the Great Compression) is strongly supported.
(As an additional note, the same pattern seems to hold for electricity and appliances – see a comment by O.Voron on the previous blog).
It’s risky to focus on automobile sales as a proxy for wealth … my millennial kids and many of their friends can afford cars but choose not to buy them, for a wide variety of reasons, ranging from aversion to debt to hyper-awareness of the environmental costs … we’ll need to find another metric … sales of new homes would also be misleading, since many people in the millennial generation apparently feel the same way about buying homes as they do about buying cars!
My approach is to look at multiple proxies, and see whether they agree, or not. I haven’t done it, but my guess is that if we add together different labor-saving devices I listed in the previous blog, we will see that the improvement dates to the first half of the twentieth century, not to the post-war era.
That first graph though shows that inequality was more or less constant from 1910 (with lots of volatility, mostly due to valuation effects and noise in the data – that big drop in the mid 1970’s should also make one suspicious about what it really is the graph is illustrating) until 1940. It was WW2 which was the “Great Equalizer” which if I remember correctly, is the major thesis advanced… if not by Claudia herself, at least by some other historians who work in that area.
Taking the second graph at face value – in that these indicators measure the well being of a typical person – what it shows is that life improvements took place *before* inequality started falling. Of course the caveat here is that wage/GDP ratio, which measures labor’s share in income. I’m not sure if this is a good measure of inequality. For example, Piketty, has the share of 1% in national income follow the trend in graph 1, not the wage/GDP ratio in graph 2. For US, his wage/GDP ratio goes back only to 1929 but it shows it as high in 1929 “on the eve of the Great Depression” then take plunge during the GD, then recover, then fall again during and right after WW2. In other words, it fits more closely with the picture of inequality dynamics in graph 1 rather than the wage/GDP ratio in graph 2.
Also in terms of popular perceptions, for what it’s worth – in literature etc. – the first two decades of the 20th century are generally seen as a period of increasing inequality. a la Great Gatsby.
The caveat I would introduce here is that what matters is *why* inequality is increasing. Inequality can go up because the rich get richer while the poor get poorer (or stagnate) or it can increase when both rich and poor are getting richer but one group is getting richer faster than the other. What characterized the period from 1890 until 1930 is the second kind of rising inequality. What has characterized the period from 1970’s to present is the former kind. I think this is a crucial difference.
Thanks, Radek. Lots of ideas there, and disentangling them will take a serious study, not just a blog post, or even a series of them. I think that there is something to this idea, but clearly the evidentiary basis is not enough to publish in a serious journal. Just enough for a blog. Also, we only have two periods to compare: the Great Compression and the Second Gilded Age. We need more. That’s what, I hope, we will be able to do with Seshat – to see whether there is a statistical pattern that holds over many time periods and geographic regions. But we first need to get the project funded…
There seems to be a lag in inequality results compared to well-being indicators. That is, well-being has a turning point around 1910, income/wealth around 1929; then well-being has a turning point around 1960, income/wealth around 1975-9. So, close to 30 year lag in both cases. Which suggests inequality is consequence rather than cause.
Or, both are a consequence of another factor, responding with varied time lags. Things can get complicated in dynamical systems affected by nonlinear feedbacks.
Indeed. A near 30 year lag suggests something generational. Maybe some demographer with an historical bent should look into that 🙂
On that point I’m somewhat intrigued by the path for stature. Stature is an indicator of well being but with a lag, essentially a generation, as it’s mostly determined by nutrition (hence well being) during childhood. So if it is capturing improvements in well being you would expect it to follow the rest but with a delay (life expectancy works because it incorporates child and infant mortality, which would be contemporaneously affected by well being).
(Note: I just looked at the other blog post where this is brought up in the comments. Is this stature that people born in year X achieved during their lifetime? If so, then this does line up)