Finding accurate statistics for pre-WWII unemployment rates in the United States is a surprisingly difficult task. Read any work about the Great Depression from the intellectual giants of the field - Barry Eichengreen, Christina Romer, Ben Bernanke, Peter Temin, and Charles Kindleberger come to mind - and you'll find a wealth of data, charts, and graphs on currency values, discounting, foreign lending, commodity prices, industrial output, etc. But unemployment? The Bureau of Labor Statistics (BLS) doesn't have reliable unemployment data for this era, despite the fact that the Knights of Labor had been calling for a BLS as early as the late-19th century. Did unemployment really crest at 25%? These were catastrophic levels, I thought. No wonder the Great Depression, like the Civil War, posed a potentially mortal threat to the republican experiment known as the United States Constitution, toppling many liberal democracies in Europe before the damage was done. Consider that the unemployment rate at the time of this writing - 4.9% - still masks a lot of underlying problems in the economy, including the weakened middle class, stagnant salaries, high CEO pay while workers' wages have fell behind productivity, high levels of private debt (think of student loans, credit cards, and mortgages) and the large gap between the rich and poor. Most of us, particularly historians in academia, don't feel like we've fully recovered from the Great Recession of 2008 and economic dissatisfaction, especially among the white working class, must be included in any discussion of the rise of the bombastic and xenophobic Donald Trump. So was unemployment in the 1930s really five times as great? It turns out the answer to this question is really complex and one can easily get lost in the details.
The very first time I heard about the Great Depression's 25% unemployment rate was during my doctoral program. The financial crisis of 2008 happened during my second year of coursework at UCSB - not only were there tons of comparisons to the Great Depression being floated in the the news all the time, but I had picked a research project and developed a broad interest in economic history. I wanted to know more. Serving as a teaching assistant for Professor Laura Kalman's survey of US History from 1920 to the present, I saw a variant of this famous graph, posted below, included in a powerpoint slide. An Enduring Vision by Boyer et al., the textbook I read assiduously during my junior year of high school while taking AP US History, includes this graph.
The very first time I heard about the Great Depression's 25% unemployment rate was during my doctoral program. The financial crisis of 2008 happened during my second year of coursework at UCSB - not only were there tons of comparisons to the Great Depression being floated in the the news all the time, but I had picked a research project and developed a broad interest in economic history. I wanted to know more. Serving as a teaching assistant for Professor Laura Kalman's survey of US History from 1920 to the present, I saw a variant of this famous graph, posted below, included in a powerpoint slide. An Enduring Vision by Boyer et al., the textbook I read assiduously during my junior year of high school while taking AP US History, includes this graph.
The Historical Statistics of the United States (HSUS) published this graph, which was based on the work of Stanley Lebergott in the post-WWII years. Lebergott, whose numbers are listed here, estimated unemployment rates based on, among other things, falling output in various sectors of the economy. If the graph is to be believed, then yes, unemployment did, indeed, reach 25%. But why couldn't I verify it with data from the Bureau of Labor Statistics, or almost anywhere else for that matter? Some insights into the economics profession are in order.
The 1930s was a watershed decade in the development of the modern field of economics. Not only was John Maynard Keynes crafting his famous and groundbreaking theories about implementing deficit spending to revive aggregate demand in a depressed economy, but New Deal public works programs, particularly the Works Progress Administration (WPA), were dispatching teams of social scientists into the field to collect valuable information. Historians are indebted to the WPA for valuable oral history interviews of ex-slaves who recalled their life experiences living under slavery from seven decades previously.
Note that discussions of all of the pre-Great-Depression financial panics - think of 1837, 1873, 1893, 1907 - rarely mention unemployment rates, and if they do, they are rough estimates, often around the same 20-25%. There are numerous reasons for this, but one worth underscoring is the pre-1929 economic orthodoxy that is, sadly, still prevalent among many libertarian circles and the Republican Party donor class. That is, balanced budgets, free trade, fixed exchange rates, adherence to the gold standard, "limited government" (with the important exception that government could be activist if it protected the wealthy and undermined the disadvantaged), and strong currencies were to be maintained at all costs. Recessions occurred because of "bad investments" and the prescribed antidote was almost always the same: let "the market" work itself out naturally; "purge the rot" from the system as Herbert Hoover proclaimed. On an individual level, if people lost their jobs, this was invariably the result of some personal failing such as a poor work ethic. If prices and employment gyrated abruptly in wild swings, well...that was just the unfortunate and "natural" consequence of restoring the presumably durable and self-regulating gold standard. It was not the responsibility of governments, according to these archaic and outmoded assumptions, to care for the unemployed, so it is not surprising that we lack accurate unemployment statistics. If it is not abundantly clear already, the satire you detect in my prose, directed toward this myopic fealty to the "free market" and "rugged individualism," is entirely necessary - not only is this mythological worldview at odds with the historical reality, but its over-emphasis on individual choice ignores how individuals are constantly interacting with larger forces that are beyond their control. I digress...
For my lectures on the 1930s in my undergraduate survey classes, I wanted accurate graphs of unemployment rates. Quick Google searches revealed this graph, presented by the St. Louis branch of the Federal Reserve and based on a National Bureau of Economic Research (NBER) study:
The 1930s was a watershed decade in the development of the modern field of economics. Not only was John Maynard Keynes crafting his famous and groundbreaking theories about implementing deficit spending to revive aggregate demand in a depressed economy, but New Deal public works programs, particularly the Works Progress Administration (WPA), were dispatching teams of social scientists into the field to collect valuable information. Historians are indebted to the WPA for valuable oral history interviews of ex-slaves who recalled their life experiences living under slavery from seven decades previously.
Note that discussions of all of the pre-Great-Depression financial panics - think of 1837, 1873, 1893, 1907 - rarely mention unemployment rates, and if they do, they are rough estimates, often around the same 20-25%. There are numerous reasons for this, but one worth underscoring is the pre-1929 economic orthodoxy that is, sadly, still prevalent among many libertarian circles and the Republican Party donor class. That is, balanced budgets, free trade, fixed exchange rates, adherence to the gold standard, "limited government" (with the important exception that government could be activist if it protected the wealthy and undermined the disadvantaged), and strong currencies were to be maintained at all costs. Recessions occurred because of "bad investments" and the prescribed antidote was almost always the same: let "the market" work itself out naturally; "purge the rot" from the system as Herbert Hoover proclaimed. On an individual level, if people lost their jobs, this was invariably the result of some personal failing such as a poor work ethic. If prices and employment gyrated abruptly in wild swings, well...that was just the unfortunate and "natural" consequence of restoring the presumably durable and self-regulating gold standard. It was not the responsibility of governments, according to these archaic and outmoded assumptions, to care for the unemployed, so it is not surprising that we lack accurate unemployment statistics. If it is not abundantly clear already, the satire you detect in my prose, directed toward this myopic fealty to the "free market" and "rugged individualism," is entirely necessary - not only is this mythological worldview at odds with the historical reality, but its over-emphasis on individual choice ignores how individuals are constantly interacting with larger forces that are beyond their control. I digress...
For my lectures on the 1930s in my undergraduate survey classes, I wanted accurate graphs of unemployment rates. Quick Google searches revealed this graph, presented by the St. Louis branch of the Federal Reserve and based on a National Bureau of Economic Research (NBER) study:
| What I like about this graph is that the gray areas show recessions and increasing unemployment while the white areas show growth, recovery, and a reduction in unemployment. But this graph was also problematic. Was unemployment really close to zero percent in 1929? Was it around the same in 1942? This would be impossible, even in a wartime economy, so perhaps this was only one subset of the workforce. Perhaps this only showed industrial employment with the assumption that agricultural employment - around 20% of the workforce back then compared to less than 5% today - was more difficult to determine. It would have been nice for the graph to indicate as such.  |  Robert Coen's estimates of unemployment from a 1973 article The next order of business was to find accurate comparisons of unemployment rates between the 1930s and today. Just as comparing the value of the dollar in the 1830s to today is an inexact science - there was no Federal Reserve in the 1830s let alone Federal Reserve notes in our wallets and purses - comparing unemployment between the Great Depression and today would be imprecise. See the graph on the left. |
Right away it became clear that I'd need to do some time-consuming research and graph-making on my own, or at the very least, conduct a very focused search in academic search engines. It was important to get this right, I thought. A quick Google search showed the graph above, but there were at least three problems. Why was it starting in 1931, the year of the destructive and worldwide bank runs, and comparing it to 2007? Wouldn't a better comparison show 1929 and 2008 side by side? 1929, like 2008, witnessed a pernicious and calamitous stock market crash, the start of a recession that completely paralyzed the financial sector, and the beginnings of long slog of inordinately high unemployment. The most recent recession, defined by the NBER as two straight quarters of negative GDP, technically started in December 2007, but its most deleterious effects were not felt, perceived, or reported on widely until the fall of 2008, coinciding with the fall of Lehman Brothers. And unemployment, as a lagging indicator, did not rise precipitously until late-2008 to early-2009. The other problems I saw with the graph above were that it relied on Lebergott's data and that the red line was incomplete, stopping in approximately 2011. I knew I could do better.
As the graph above shows, many studies comparing the Great Depression to the Great Recession still rely on Lebergott's series. But in the scholarly literature, Robert Coen (1973), Michael Darby (1976) and Christina Romer (1986) have added important contributions. Much of the corrections relate to the various ways in which we categorize the unemployed. Darby found that Lebergott's series was flawed because it did not count "emergency workers," many of whom worked for the WPA, among the ranks of the employed. The correction resulted in a fairly significant reduction in the peak unemployment rate. 2-3.5 million workers, or 4 to 7 percent of the workforce, according to Darby, were mistakenly counted as unemployed in Lebergott's study. If Darby was correct, then the peak unemployment rate in the 1930s could not have been 25%, but closer to 20-22%.
As the graph above shows, many studies comparing the Great Depression to the Great Recession still rely on Lebergott's series. But in the scholarly literature, Robert Coen (1973), Michael Darby (1976) and Christina Romer (1986) have added important contributions. Much of the corrections relate to the various ways in which we categorize the unemployed. Darby found that Lebergott's series was flawed because it did not count "emergency workers," many of whom worked for the WPA, among the ranks of the employed. The correction resulted in a fairly significant reduction in the peak unemployment rate. 2-3.5 million workers, or 4 to 7 percent of the workforce, according to Darby, were mistakenly counted as unemployed in Lebergott's study. If Darby was correct, then the peak unemployment rate in the 1930s could not have been 25%, but closer to 20-22%.
On the first Friday of every month - a date I'm not ashamed to say that I look forward to - the BLS releases detailed employment statistics for the previous month, including the official unemployment rate (U3) and the number of jobs added or lost. It's always interesting to see how different the BLS report is from all of the estimates of leading economists, which is usually based on weekly jobless reports (generally below 300,000 is good) and ADP non-farm payrolls. Sometimes the estimates are very close, within a few thousand of the actual total, and at other times, I've seen them be wildly off the mark, as in, say, a difference of 100,000. Then there are the necessary adjustments that come in subsequent months, often because holidays or other random events can skew the numbers a bit. As the graph below illustrates, there are at least six different ways to measure unemployment, or what the BLS calls "labor underutilization." The narrowest measure is U1, the broadest measure is U6, currently less than 10%, which is fairly good, and the "official statistic" of unemployment is U3. Over the past two decades, we've seen U6 about 5-10 points higher than U3.
For unemployment data in the 1930s, frustratingly, we have only one data point per year. The BLS, in contrast, releases figures once per month. So that the graph I was about to make had some congruency, I took the recent BLS statistics for each month, added them up and then divided by twelve to get a yearly average. This would allow me to compare the two recessions with the same data points. Without any further adieu, here is the graph I now show in my lecture on the Great Depression.
To arrive at the blue line, I averaged the data from Lebergott, Coen, and Darby - adding them up and dividing by three. I recognize this might be apostasy among social scientists and would welcome their input. My alibi is that I am trained as a historian, a field, I've always held, that lies more in the humanities than in the social sciences. Importantly, my own research expertise is concerned more with narrative and textual evidence rather than statistical analysis and complex econometrics.
What do we make of this graph? The restoration of "full employment," if we can call it that, has taken a very long time and a lot of people have suffered in the meantime. As I've written elsewhere, monetary policy was up to the challenge, but fiscal policy was ambiguous at best, and at worst, counterproductive and contractionary.
For those interested in the lessons of history and the implications for constructing wise policy, there is a positive take-away. The graph naturally begs us to ask: why did unemployment never get above 10%, or less than half the total of the 1930s? In no particular order, I've identified these four reasons: 1) Deposit insurance in the form of FDIC prevents the devastating bank runs that occurred in the 1930s. So much for government regulation stifling business, right? 2) A social safety net: today we have Social Security and Medicare, which prevent a lot of the elderly from falling into poverty. Receiving monthly checks even in the face of reduced hiring in the private sector does much to prevent demand from collapsing completely; 3) improvement in FED policy. Friedman and Schwartz famously argued that the FED raised interest rates in 1930 when it should have lowered them. FED chairpersons Ben Bernanke and Janet Yellen heeded these lessons and the unprecedented post-2008 lowering of interest rates through open-market operations surely prevented the economy from falling off a cliff; and, related to this, 4) the absence of a gold standard, an imbalanced and asymmetric system with no agreed-upon "rules of the game" that forced countries to raise interest rates in the face of declining gold reserves, but had no accompanying requirement for countries to increase lending when they hoarded gold, as the United States and France did in the 1930s. It may seem counterintuitive, but devaluation was actually key to recovery in the United States and other countries. Zero academic economists want to return to the gold standard. Prices and employment, the domain of the FED, are much more stable without it. All of this is to say that a stable economy requires management and regulation, not the presumably "natural" laws of the laissez-faire "free market."
There is lots more that I could say, but this blog entry cannot get too long. The U3 unemployment rate is only one statistical indicator of the health of an economy. There are many caveats, including the labor force participation rate, to say nothing of peoples' sense of well-being. However, the trajectory of U3 since 2008 does show a substantial and significant recovery and further, incontrovertible testimony that the Great Recession was not as bad as the Great Depression. If 2008 was like 1929, then our current year, 2016, was like 1937. At this point, New Deal programs had reduced unemployment significantly, but Roosevelt's return to economic orthodoxy in the form of balanced budgets and reduced public spending, combined with some taxes that were going into effect and trends in the business cycle, contributed to the "Roosevelt Recession" of 1937-1938 where unemployment shot up significantly. This was a recession within a depression. Unemployment would not return to normal levels until World War II, due in large part to government spending, directly on the military and indirectly in the form of government contracts to private companies (a phenomenon that is conveniently left out by New Deal critics who, in spite of evidence, wish to show the inefficacy of government spending). In any case, we are unlikely to return to the unemployment levels of the Roosevelt Recession, which testifies to improved public policy over time.
The graph I've put together comparing unemployment during the Great Depression and Great Recession seems simple enough, but believe me, it took a long time to put together. It is for my students. It is also for me. When we evaluate professors, we tend to distinguish between good researchers and good teachers. This is a false choice. It does not have to be one or the other. I knew plenty of professors at UCSB who were brilliant scholars and amazing teachers at the same time. It's important to go the extra mile and give our students the best research in the classroom. They deserve as much.
What do we make of this graph? The restoration of "full employment," if we can call it that, has taken a very long time and a lot of people have suffered in the meantime. As I've written elsewhere, monetary policy was up to the challenge, but fiscal policy was ambiguous at best, and at worst, counterproductive and contractionary.
For those interested in the lessons of history and the implications for constructing wise policy, there is a positive take-away. The graph naturally begs us to ask: why did unemployment never get above 10%, or less than half the total of the 1930s? In no particular order, I've identified these four reasons: 1) Deposit insurance in the form of FDIC prevents the devastating bank runs that occurred in the 1930s. So much for government regulation stifling business, right? 2) A social safety net: today we have Social Security and Medicare, which prevent a lot of the elderly from falling into poverty. Receiving monthly checks even in the face of reduced hiring in the private sector does much to prevent demand from collapsing completely; 3) improvement in FED policy. Friedman and Schwartz famously argued that the FED raised interest rates in 1930 when it should have lowered them. FED chairpersons Ben Bernanke and Janet Yellen heeded these lessons and the unprecedented post-2008 lowering of interest rates through open-market operations surely prevented the economy from falling off a cliff; and, related to this, 4) the absence of a gold standard, an imbalanced and asymmetric system with no agreed-upon "rules of the game" that forced countries to raise interest rates in the face of declining gold reserves, but had no accompanying requirement for countries to increase lending when they hoarded gold, as the United States and France did in the 1930s. It may seem counterintuitive, but devaluation was actually key to recovery in the United States and other countries. Zero academic economists want to return to the gold standard. Prices and employment, the domain of the FED, are much more stable without it. All of this is to say that a stable economy requires management and regulation, not the presumably "natural" laws of the laissez-faire "free market."
There is lots more that I could say, but this blog entry cannot get too long. The U3 unemployment rate is only one statistical indicator of the health of an economy. There are many caveats, including the labor force participation rate, to say nothing of peoples' sense of well-being. However, the trajectory of U3 since 2008 does show a substantial and significant recovery and further, incontrovertible testimony that the Great Recession was not as bad as the Great Depression. If 2008 was like 1929, then our current year, 2016, was like 1937. At this point, New Deal programs had reduced unemployment significantly, but Roosevelt's return to economic orthodoxy in the form of balanced budgets and reduced public spending, combined with some taxes that were going into effect and trends in the business cycle, contributed to the "Roosevelt Recession" of 1937-1938 where unemployment shot up significantly. This was a recession within a depression. Unemployment would not return to normal levels until World War II, due in large part to government spending, directly on the military and indirectly in the form of government contracts to private companies (a phenomenon that is conveniently left out by New Deal critics who, in spite of evidence, wish to show the inefficacy of government spending). In any case, we are unlikely to return to the unemployment levels of the Roosevelt Recession, which testifies to improved public policy over time.
The graph I've put together comparing unemployment during the Great Depression and Great Recession seems simple enough, but believe me, it took a long time to put together. It is for my students. It is also for me. When we evaluate professors, we tend to distinguish between good researchers and good teachers. This is a false choice. It does not have to be one or the other. I knew plenty of professors at UCSB who were brilliant scholars and amazing teachers at the same time. It's important to go the extra mile and give our students the best research in the classroom. They deserve as much.
RSS Feed
