Sirota’s Statistics Fail to Vindicate New Deal

Here I continue my critique of David Sirota’s case for the New Deal. Sirota wrote a piece for the January 2 Denver Post. I wrote a reply on January 4 as well as a short letter for the Post, published January 7. (I’ve also written several other articles on the matter.) Sirota completely ignores my critique of his case in his January 6 follow-up for the Post’s Politics West. I continue my case here in the hopes that eventually Sirota will attempt to engage the debate and respond to his critics.

1. Sirota’s Smear of Shlaes

The worst aspect of Sirota’s follow-up his his unjust smearing of Amity Shlaes, the respected economic historian who wrote The Forgotten Man. Reading Sirota’s claims, I doubt that he has read the book. Whether or not he did, he misrepresents its contents.

Sirota claims that Shlaes’s books has been “discredited,” that she “wholly omits some relevant data and deviously manipulates other numbers,” and that she is dishonest.

To back up this claim, Sirota quotes a Slate article by historian Eric Rauchway, which states, “She has unemployment at 20 percent in the 1937-38 recession. … A third of the people Shlaes counts as unemployed had a job that the New Deal gave them through its relief programs.”

Notably, Rauchway does not claim to “discredit” Shlaes’s book, though he argues she is pushing an ideology (as is he), and his case is considerably more nuanced than is Sirota’s. For example, Rauchway points out a couple of FDR’s virtues: he supported freer world trade and he helped repeal Prohibition. Indeed, I recently praised FDR for opposing Prohibition. What Rauchway does not do is demonstrate that the New Deal as a package improved the economy. For example, he praises FDR’s union laws without examining the economic case that those laws contributed considerably to unemployment at the time.

To Sirota, Shlaes is dishonest because she uses an unemployment figure of 20 percent during the 1937-38 recession. But Sirota uses practically the same figures to make his case. Sirota writes:

So to end this historical revisionism once and for all — to compare apples to apples, rather than apples to conservatives’ fuzzy math — let’s go to the great equalizer, the Census Data, and specifically Census document HS-29 (available in PDF or Excel formats). Quoting directly from Census data, here are the unemployment rates and total number of official unemployed at the beginning and end of the presidential terms since the Great Depression…

The unemployment rate that this Census data provides for 1938? Nineteen percent. So, if Shlaes uses this figure, she’s dishonest, but if Sirota uses the same figure, he’s invoking the “great equalizer” of data. Oops.

Notice that Sirota admits that there was a renewed recession in 1937-38. In his first article, he claimed — without any evidence or economic argument — that pulling back on the New Deal caused this. As I review, FDR’s new wage controls in a period of monetary contraction primarily caused this increased unemployment.

If Sirota would check Shlaes’s book, he would find that she neither omits nor manipulates the data. (I quote from the 2008 paperback.) On pages 402-403, Shlaes writes a detailed note about her unemployment figures. She writes:

During the 1920s and the Depression, Washington did not keep the sort of systematic unemployment data that it collects today. … Still, there were some national numbers to talk about. The Labor Department collected figures, as did the Census Bureau [Sirota’s “great equalizer”], the Commerce Department, [etc.] …

For the 1930s, I have gone with month-by-month figures calculated by Richard K. Vedder and Lowell E. Gallaway, in Out of Work: Unemployment and Government in Twentieth Century America (San Francisco: Independent Institute; New York: Holmes & Meier, 1993 [p. 77]). These authors use [scholar Stanley] Lebergott and government numbers as their basis. Economists on the right such as Michael Darby, Harry Scherman Fellow at the National Bureau of Economic Research, argued later that Lebergott and the BLS both overestimated the number of unemployed by counting as unemployed people who actually had full- or part-time work in make-work programs such as the WPA. But I have gone with the traditional numbers.

So Sirota can argue that Shlaes does not use the best numbers, but he cannot fairly claim that she omitted or manipulated the data.

Neither Sirota nor Rauchway offer month-by-month, or even year-by-year, unemployment figures that they think are properly adjusted to account for federal make-work. Amazingly, Colorado Media Matters (CMM) provides a good citation for this. According to the Darby figures, unemployment dropped steadily from 22.9 percent in 1932 to 9.1 percent in 1937, then bumped back up to 12.5 percent in 1938.

Shlaes does stray from the “traditional numbers” in her “Afterword to the Paperback Edition.” On page 393 she writes:

In the very best years of Roosevelt’s first two terms, unemployment still stood above 9 percent. Nine percent is better than horrendous, but it hardly is a figure that induces hope.

Indeed, as I’ve noted, Obama’s own economic advisors believe the current recession will peak at 9 percent unemployment without their intervention (even though I think their intervention will only make matters worse). Today we all think that employment approaching 9 percent is a disaster, yet the defenders of the New Deal praise FDR because, using the Darby figures, FDR almost managed to get unemployment down to 9 percent before it spiked back up to double-digits.

Perhaps Shlaes should have explained the Darby figures in more detail and earlier in her book. However, the statistical matter is not nearly as clear-cut as CMM would have us believe. While CMM accuses its opponents of cherry picking the data, CMM itself cherry-picks its sources. The Vedder and Gallaway book came out in 1993, the same year as the paper by Robert Margo cited by CMM. However, Vedder and Gallaway certainly were aware of Darby’s work, and they ran his figures through their regressions (see pages 42-43). They write:

Questions have been raised about the basic accuracy of the government unemployment-rate data… Both the [Robert] Coen and the Darby estimates place unemployment at lower levels in the early thirties than the official Bureau of Labor Statistics estimate derived by Stanley Lebergott. We are inclinded to agree with Gene Smiley that the Lebergott/BLS estimates are probably as good as any. (pages 42, 79-80, footnotes omitted)

What is interesting about this is that the reference to Smiley is a 1983 article in the Journal of Economic History — the same article that CMM cites here and here to advocate use of the lower statistics.

Smiley influenced Vedder and Gallaway, who in turn influenced Shlaes. Smiley, Vedder, and Gallaway are intimately familiar with Darby’s figures. Yet Sirota and Colorado Media Matters want us to believe that Shlaes’s use of statistics is sinister rather than merely a case of making a reasonable use of limited data.

Even granting the lower figures, Sirota’s entire case boils down to the claim that FDR’s New Deal worked wonders when it couldn’t manage to get unemployment below double digits for most of FDR’s first eight years in office. That’s a pretty lousy case.

2. Sirota Confuses Correlation with Causation

Sirota’s argument is that “the pre-WWII New Deal era saw the single largest drop [in unemployment] in American history.” He goes on at length about this point.

But, as I argued in my first reply, it’s pretty hard to do worse than unemployment approaching 25 percent. Hoover, with the help of the Federal Reserve, decimated the economy. The fact is that the economy continued to struggle under FDR and never really recovered till the ’40s. The question, then, is whether unemployment dropped under FDR because of or in spite of FDR’s policies.

Sirota apparently never learned the fundamental lesson of statistics that correlation does not prove causation. He sees that unemployment dropped under FDR (relative to its horrific rate under Hoover), and he concludes that FDR’s policies therefore helped reduce unemployment. That Sirota’s argument may be seductively simplistic does not change the fact that it is completely wrong.

To get the idea of the sort of logical fallacy Sirota is committing, consider the following examples:

* A patient feels ill, so his doctor bleeds him with leeches. Slowly the patient recovers. Therefore, the blood-letting caused the recovery. QED.

* A woman reporting physical ailments visits Franz Mesmer, who applies magnets to her and stares deeply into her eyes, after which the woman reports improved health. Therefore, to follow Sirota’s method, we must conclude that the magnets caused the improved health.

* Jill comes down with a nasty cold. She smokes one pack of cigarettes per day. Over a period of two weeks, Jill mostly recovers from her cold. Therefore, by Sirota’s logic, Jill’s smoking reduced the severity of her cold.

If Sirota wishes to make his case, he cannot merely claim that unemployment dropped between Hoover and FDR, a fact that everyone acknowledges. (Sirota wrongly claims that Shlaes fails to acknowledge this.) He must demonstrate, using evidence and argument, that the economy improved somewhat because of FDR’s policies, rather than in spite of them. He must tackle head-on the arguments and evidence amassed by the critics of the New Deal showing that FDR’s policies made matters worse than they otherwise would have been, not better.

3. National Product

Sirota looks mainly at unemployment, but he also mentions economic growth. Growth is tightly tied to employment: with nearly a quarter of the nation out of work, less stuff gets made. So it is not surprising that, with unemployment less than that (but still dramatically higher than normal), production went up relative to the late Hoover years.

Sirota would not be surprised that critics of the New Deal discuss this relative increase in production if he would check their works. Here’s what Shlaes has to say on the matter (page 395):

What about the oft-cited rising industrial production figure? The boom in industrial production of the 1930s did signal growth… [but a]t many points during the New Deal, net private investment was not merely low but negative. Companies were using more capital goods than they were buying.

Again the reality is considerably more complex than Sirota acknowledges.

At this point, Sirota has written two articles that utterly fail to make his case that the New Deal helped the economy. Perhaps he’ll seriously grapple with the relevant issues in a future article.

4. Sirota’s Stereotyping

This is a minor point tacked on at the end. Sirota claims that, after his first column, he received “angry email from conservatives,” and he claims to be answering “right-wingers.” However, as I pointed out in my first reply, “many critics of the New Deal are not conservatives and many conservatives praise FDR.”

In Sirota’s simplistic world of angry conservatives and enlightened liberals, where does he fit treatments of Hoover, the Republican president preceding FDR? As I’ve noted, every critic of the New Deal I’m aware of also excoriates Hoover. But Sirota lacks either the ability or the will to see beyond his ideological blinders. He leads with stereotypes and ad hominem attacks and weakly follows with arguments.

Sirota received at least one e-mail that was neither angry nor from a conservative: my own. (If Sirota wishes to call me a conservative, perhaps he will explain what he’s talking about, given that I’m an atheist who supports gay rights, the right to abortion, the re-legalization of drugs, and the total repeal of all censorship.)

On January, 4, I wrote to Sirota:

Dear Mr. Sirota,

I’ve written a criticism of your recent op-ed on the New Deal here:
[link]

I would be more than happy to publish your reply, which you are welcome to send me via e-mail.

Thanks,
Ari Armstrong

Angry conservative, indeed.