Word is that Stephen Moore is in the mix for NEC staff. I think he would fit in perfectly in the Trump White House (hence the reference to Rosabeth Moss Kantor’s concept of “homosocial reproduction”). After all, NEC Chair Kudlow just said the budget deficit is shrinking. Now, consider these instances of Mr. Moore’s sheer mendacity (or, I admit, it could be statistical incompetence):
May 31, 2017:
I just heard Stephen Moore of Heritage Foundation saying we had a trillion dollar budget deficit, in a debate on CNN about the Paris Accord (1:42PM CST).
The FY 2017 budget deficit is $603 billion, according to the just released budget (see Table S-1). Add his serial lying to the statistical atrocity that is the ALEC-Laffer-Moore-Williams index, and all I can say is Mr. Moore should not be paid as a “CNN economic analyst”.
Update, 7:14 PM Pacific: I’d forgotten this tabulation of errors by Mr. Moore. The man has no shame.
July 21, 2017:
Readers will recall that, in response to CNN bringing Stephen Moore on board as an “economics analyst”, I noted that Stephen Moore is a liar. Now I continue with the Kansas edition.
For those who forgot, in the original Wichita Eagle piece, Moore wrote:
No-income-tax Texas gained 1 million jobs over the past five years; California, with its 13 percent tax rate, managed to lose jobs. Oops. Florida gained hundreds of thousands of jobs while New York lost jobs. Oops. Illinois raised taxes more than any other state over the past five years, and its credit rating is the second lowest of all the states, below that of Kansas.
For those who don’t believe all of this, here are the numbers for job creation of the four largest states over the past 20 years: Texas, up 49 percent; Florida, up 35 percent; California, up 19 percent; New York, up 4 percent.
As the Wichita Eagle itself recounted (as it decided to no longer publish pieces by Moore):
Correction: The 5-year time period mentioned in this piece was December 2007 to December 2012. Over that time, Texas gained 497,400 jobs, California lost 491,200, Florida lost 461,500 and New York gained 75,900. The totals in this piece are incorrect.
This is old news (by the way, the version posted on the Heritage Foundation makes no mention of this egregious error). But it behooves us to remember this episode because it highlights the extent of the man’s mendacity. Now as the Brownback experiment has ended, it’s useful to see exactly how disastrous the Moore prescription has been for the Kansas economy.
From June 7th post:
From May 27th post:
Figure 2: Log nonfarm payroll employment for Missouri (blue), Kansas (red), and US (black), seasonally adjusted, 2011M01=0. Source: BLS, and author’s calculations.
And from September 2016, an assessment of what a counterfactual with no government spending reductions cum tax cuts would have implied for Kansas (it would’ve done better than under the Brownback program):
Figure 3: Log Kansas GDP (black), fitted values without drought (red), and without government output reductions (blue). Drought dates as suggested by Political Calculations (light brown). Source: BEA, and author’s calculations.
Watch for more chapters of “Stephen Moore Is a Liar” as long as CNN continues to employ him as an “economics analyst”
January 1, 2018:
Since reader Rick Stryker is still busily trying to defend lying in the cause of a deeper truth, let me document fully the mendacity of Stephen Moore. On May 31st on CNN, he states the budget deficit is $1 trillion.
Here are actual data on actual budget deficit as reported by the Office of Management and Budget (bold blue).
Figure 1: Federal budget balance as reported by OMB (bold blue), and CBO projection (dark blue), and implied budget balance under HR 1 Tax Cuts and Jobs Act (pink), in millions of dollars by fiscal year. Source: OMB via FRED, CBO.
Rick Stryker today leaps to Mr. Moore’s defense to write:
Menzie accused Steven Moore of being a liar or statistically incompetent (or both) since Moore said that the last time he checked the deficit was $1 trillion but in fact the budget deficit is on the order of $600 billion. Here, Menzie was making the incorrect assumption that the budget deficit is always an accurate estimate of the change in borrowing actually performed by the Treasury and that therefore the budget deficit is the only legitimate way to measure that change.
Since I was listening to CNN, I think I know what I heard (as opposed to mind reading Mr. Moore, and asserting he meant Federal borrowing) — he said budget deficit. For this, there is an official definition, one that OMB reports, CBO uses, etc. But if you doubt me, here is the exact quote from a transcript:
MOORE: Because, the reason is that the rest of the world wants our money. And that is all about financing a climate change industrial complex around the world, and we’re the ones who are going the fund it. And last time I checked, we have a trillion-dollar budget deficit and we don’t have the money to send to all of the countries.
Apparently Mr. Moore had not checked for some four years, if we are to take this statement at face value. (Of course, Rick Stryker will assert this transcript is “fake news” since it came from CNN). Do note, he says “budget deficit”, not Federal borrowing…if we are to believe CNN (and my recollection).
But let’s say borrowing is what he meant. Well, as of 2017Q3, the one year change in Federal debt held by the public (so matching FY2017) was $504 billion — still short of a trillion dollars. Let’s give Mr. Moore a break, and remember he made the comment on May 31st. As of the end of 1st quarter, the one year change was $445 billion. For me (I don’t know what kind of math other people use), that’s a lot less than $1 trillion.
So, prepare for more alternative facts, alternative definitions, mind-reading, general dissemblement and other mental gyrations in defense of lying in the new year. It’s already started.
Addendum: Rick Stryker writes I should apologize to Stephen Moore for misrepresenting his comments. He writes:
…I feel compelled to point out that on this New Year’s Day that a good New Year’s resolution for you would be to put up a (very long) special post apologizing to everyone you attacked with mistaken analysis over the previous year. You could start the post by apologizing to Stephen Moore. …
Let’s take a poll — how many agree with Rick Stryker?
January 2, 2018:
On July 9th, 2014, the Wichita Eagle published an op-ed by Mr. Stephen Moore:
No-income-tax Texas gained 1 million jobs over the past five years; California, with its 13 percent tax rate, managed to lose jobs. Oops. Florida gained hundreds of thousands of jobs while New York lost jobs. Oops. Illinois raised taxes more than any other state over the past five years, and its credit rating is the second lowest of all the states, below that of Kansas. [emphasis added MDC]
Now, as of July 9th, one would know employment figures for May 2014. Let’s plot the data that he should have been looking at.
The astute observer will note that this graph does not match with Stephen Moore’s characterization. Yael Abouhalkah determines that one reason for the disjuncture is that Moore actually used calculations for December 2007 to December 2012. Why would Mr. Moore use a sample ending in 2012 for an article in mid-2014? Maybe because the picture is much more amenable to his thesis.
Notice, even then, the pattern of changes does not match up with what Mr. Moore described. That’s because he added up the numbers incorrectly. Moreover, the household survey, upon which the civilian employment series is based upon, imprecisely measures employment at the state level, given the relatively small samples used (see this graphical depiction). The latest vintage of these series differs from those Mr. Moore knew of in July 2014.
Why did he use this imprecisely measured series? This question takes on a heightened importance given Mr. Moore’s emphasis on “jobs” in the op-ed, since the establishment series tabulates jobs, while the household series tabulates employed individuals. What does Figure 1 look like using nonfarm payroll numbers?
Figure 3: Log nonfarm payroll employment for California (blue), Florida (red), New York (green), and Texas (black), all normalized to 2009M05. Source: BLS May 2014 release via ALFRED and author’s calculations.
Note that, while Texas still leads, Florida and California tie, hence casting doubt on the thesis that low income taxes necessarily lead to faster employment growth. Of course, anybody with a rudimentary knowledge of the (absence of) correlation between the Moore-Laffer-ALEC economic outlook ranking and economic growth would know any relationship found by Mr. Moore would be fragile.
This episode prompted the Kansas City Star to ban Mr. Moore from their publication.
July 11, 2015:
In addition, Stephen Moore is collaborator with Arthur Laffer and John Williams on the ALEC-sponsored Rich States, Poor States economic outlook index, which apparently has pretty much zero correlation with subsequent economic performance.
From an Econometric Assessment of the World according to the American Legislative Exchange Council (ALEC)
For eight years, the American Legislative Exchange Council has been producing a ranking that purports to measure the competitiveness of individual states. From ALEC, Rich States, Poor States, 2015:
The Economic Outlook Ranking is a forecast based on a state’s current standing in 15 state policy variables. Each of these factors is influenced directly by state lawmakers through the legislative process. Generally speaking, states that spend less—especially on income transfer programs, and states that tax less—particularly on productive activities such as working or investing—experience higher growth rates than states that tax and spend more.
Hence I think it is reasonable to ask whether the ALEC-Laffer-Moore-Williams “economic outlook” ranking (hereafter “ALEC ranking”) actually has any predictive power, above and beyond other demographic and geographic indicators. This analysis follows up on a more ad hoc analysis in this post, and confirms conclusions arrived at in Fisher w/LeRoy and Mattera (2012). In addition, this analysis serves as a rejoinder to ALEC’s rebuttal asserting that critiques did not incorporate sufficient controls and allow for differing time horizons, to wit:
Moreover, rigorous statistical methods show that a higher economic outlook ranking in Rich States, Poor States does indeed correlate with a stronger state economy.
H/t Michael Hiltzik, who has done a tremendous job tracking the ALEC ranking. Let’s examine the available data.
Figure 1 depicts the one year growth in real Gross State Product and the ALEC ranking, lagged one year, for growth over the 2008-2014 period. Should the ALEC thesis be correct, one should see a negative sloped relationship.
Figure 1: Real Gross State Product growth (log first difference) and lagged ALEC ranking for Economic Outlook. Nearest neighbor fit line (red), window = 0.3. Source: BEA and ALEC, Rich States, Poor States, 2015, and author’s calculations.
There is no obvious correlation between annual growth rates and the ALEC rankings. Note that lagging the ranking so that, for instance, the growth rate between 2013 and 2014 is related to the ALEC ranking in 2012 (instead of 2013) does not change the results much. (The ALEC 2013 ranking pertains to data reported in 2012, as far as I can tell). This outcome is unsurprising because the ALEC rankings are highly persistent. If one estimates a panel autoregressive model for the ALEC rankings, the AR coefficient is 0.95, and the adjusted R2 is 0.90.
Defenders of the ALEC rankings have argued that the rankings are aimed at predicting growth at a longer time horizon than annual. Given that the perspective of the RSPS methodology is supply-side, this argument is prima facie sensible. In order to accommodate that argument, I display in Figure 2 the data for three year (nonoverlapping) horizons (for 2014, and 2011).
Figure 2: Nonoverlapping average three year real Gross State Product growth (log difference) and three year lagged ALEC ranking for Economic Outlook. Nearest neighbor fit line (red), window = 0.3. Source: BEA and ALEC, Rich States, Poor States, 2015, and author’s calculations.
Figure 3 provides analogous information, at the six year horizon (the maximum possible given the span of ALEC rankings).
Figure 3: Average six year real Gross State Product growth (log difference) and six year lagged ALEC ranking for Economic Outlook. Nearest neighbor fit line (red), window = 0.3. Source: BEA and ALEC, Rich States, Poor States, 2015, and author’s calculations.
The nonparametric fitted lines indicate a slightly negative slope overall, suggesting some content to the argument that lower ranked states grow slower. In both Figures 2 and 3, North Dakota and Texas constitute outliers along the y-dimension. In order to discern how much these two observations drive the results, I omit them and replot in Figure 4.
Figure 4: Average six year real Gross State Product growth (log difference) and six year lagged ALEC ranking for Economic Outlook, excluding North Dakota and Texas. Nearest neighbor fit line (red), window = 0.3. Source: BEA and ALEC, Rich States, Poor States, 2015, and author’s calculations.
No obvious pattern emerges from this last plot. In order to move beyond simple graphics, I now implement a series of regressions. First to a simple examination of the a bivariate relationship, encompassing the lower 48 states, one obtains at the annual frequency:
Δyi,t = -0.00003ALECi,t-1
Adj-R2 = -0.00, SER = 0.027, N=288. bold indicates significance at 10% msl, using heteroscedasticity and serial correlation corrected standard errors.
(I use the lower 48 as I do not have demographic and geographic data for the Alaska and Hawaii.)
Essentially, there is no explanatory power for the ALEC-Laffer indices in terms of year on year GSP real growth. Allowing for individual state-fixed effects, one obtains:
Δyi,t = 0.0004ALECi,t-1
Adj-R2 = -0.00, SER = 0.027, N=288. bold indicates significance at 11% msl, using heteroscedasticity and serial correlation corrected standard errors.
The coefficient is positive, indicating that lower ranked states, and hence states that have a less business friendly environment according the ALEC criterion, exhibit higher economic growth. The coefficient is borderline statistically significant. I would say that the use of state-level fixed effects is a fairly blunt way to account for state-specific factors. A better approach controls for state factors that economic theory suggests might be important for growth, such as geographic and demographic factors. Here we include log population density (LDENSITY, to account for urbanization), dryness (measured as inverse, WET), mildness of weather (MILD) and proximity to navigable waterways (DISTANCE); these four variables are defined such that positive coefficients are expected. These variables are time-invariant; hence one cannot estimate a fixed effects regression incorporating these variables. I also include log real price of oil (LRPOIL) to control for oil producing states, allowing the coefficient to vary across states.
Δyi,t = 0.0002ALECi,t-1-0.01LDENSITYi + 0.019WETi + 0.001MILDi + 0.0001DISTANCEi + LRPOILi,t
Adj-R2 = 0.36, SER = 0.021, N= 288. bold indicates significance at 10% msl, using heteroscedasticity and serial correlation corrected standard errors.
The ALEC ranking has no statistical significance, while a drier and more mild climate is associated with faster growth, with statistical significance. Proximity to navigable water is also a positive factor.
What if we estimate a comparable regression, looking at 6 year growth rates (but omitting oil prices which are the same for all states)? Then one obtains:
Δyi,t = -0.018 -0.0001ALECi,t-6 +0.001LDENSITYi – 0.0007WETi – 0.0003MILDi -0.00001DISTANCEi
Adj-R2 = -0.02, SER = 0.007, N= 48. bold indicates significance at 10% msl, using heteroscedasticity and serial correlation corrected standard errors.
Notice that the ALEC coefficient is not statistically significant. Thus far, the one case it has shown up as borderline significant, it goes the opposite of the ALEC-Laffer-Moore-Williams thesis.
Finally,Professor Ed (“no recession”) Lazear recently asserted that states with fast growing employment have low tax rates and right-to-work laws. He also argues that one needs to incorporate the depth of the drop in employment in 2008-09, in order to explain the growth in employment (so, a sort of version of the bounceback thesis he forwarded in the Economic Report of the President, 2009). I can’t find the working paper that provides the basis for his assertion, but I can estimate a comparable regression. In order to make the proposition testable, I examine 5 year growth rates in output, and refer to the ALEC rankings in 2009. I add the change in output from 2008 to 2009 (so TROUGHDEPTH takes on a value of -0.05 for observation i if output dropped 5% going into 2009).
Δyi,t = 0.065 -0.0010ALECi,t-5 +0.0046LDENSITYi + 0.0042WETi – 0.0014MILDi +0.00005DISTANCEi – 0.3413TROUGHDEPTHi
Adj-R2 = -0.04, SER = 0.037, N= 48. bold indicates significance at 10% msl, using heteroscedasticity corrected standard errors.
The ALEC variable does not exhibit statistical significance. Omitting ND and TX would not change this basic results. Moreover, the trough variable coefficient, while exhibiting the right sign, is not statistically significant.
Bottom line: The ALEC ranking, which purports to measure business-friendly policies, is not correlated with real GSP growth, either short term or medium term. This is true if one controls for additional variables.
I don’t think this necessarily means that policies such as right to work, or tax rates, and so forth do not have an impact on growth. For instance, Kolko, Neumark, and Cuellar Mejia, J.Reg.Stud. (2013) conclude that what’s important is “less spending on welfare and transfer payments; and more uniform and simpler corporate tax
structures.” Hence, the ALEC economic outlook ranking is, in my assessment, a manifestation of faith based economics.
Addendum: As Wisconsin has risen in ALEC rankings (33rd in 2008 to 17th in 2014), Wisconsin’s growth rate has declined relative to the US.
Figure 5: Wisconsin-US real growth differential (log first difference) and lagged ALEC ranking for Economic Outlook for Wisconsin (down is a “better” ranking). Light green shaded area pertains to GDP growth rates during the Walker era. Source: BEA and ALEC, Rich States, Poor States, 2015, and author’s calculations.
If it is not obvious, this is not the direction that ALEC-Laffer-Moore-Williams posit.
Note: Thanks to Professor Neumark, who kindly provided the data set used in his J.Reg.Stud. paper for use by my students in my PA819 statistics course (all students had to use his data set to analyze whether business conditions as measured by various indices (but not the ALEC ranking) influenced state level growth). The demographic and regional variables are drawn from that data set.
Bottom Line:Stephen Moore is a liar. He will fit in perfectly at the Trump NEC.