Emil Kirkegaard's self-citations

I recently looked at every paper Kirkegaard lists on his Google Scholar profile as having 1 or more citations. I aimed to determine how many of his 404 total listed citations were self-citations (i.e. citations in papers he authored/co-authored). The answer was remarkably high: 316 (78%). The spreadsheet in which I compiled these results is copied and pasted below, with each paper numbered according to its position when all papers are ranked in descending order of citation count. I lumped the six top ones together because I looked at their self-citation numbers before looking at the rest of them.

	Total cites	Self-cites
Top 6	164	112
7	13	12
8	11	8
9	11	10
10	10	2
11	10	10
12	10	10
13	9	8
14	9	9
15	9	7
16	9	6
17	8	7
18	8	8
19	7	7
20	7	5
21	6	3
22	6	5
23	6	6
24	6	6
25	6	6
26	6	6
27	5	4
28	5	5
29	4	4
30	4	3
31	4	4
32	4	4
33	4	4
34	4	4
35	4	4
36	4	4
37	3	3
38	3	3
39	3	1
40	2	2
41	2	2
42	2	2
43	2	1
44	2	0
45	1	1
46	1	1
47	1	1
48	1	1
49	1	1
50	1	0
51	1	0
52	1	1
53	1	0
54	1	1
55	1	1
56	1	1

The "Trump economy": fact or fiction?

Has President Trump actually improved the state of the U.S. economy above and beyond what would be expected if the trends that occurred under President Obama simply continued unchanged? He regularly claims as much (including in yesterday's State of the Union), as do many of his supporters, so it seems useful to determine to what extent this claim is true. This post aims to answer this question by focusing on the national unemployment rate.

The right-wing hashtag #jobsnotmobs, which was widely shared on Twitter during the elections last year, seems to suggest that Trump's policies have benefited American workers by reducing the unemployment rate. This claim can be tested empirically using the "interrupted time series" design. This method involves looking at trends before an intervention, then comparing them to the trends after an intervention, to see what effect (if any) the intervention had on the outcome being measured.

Monthly national unemployment data were downloaded from the Bureau of Labor Statistics website. The pre-Trump trend in the unemployment rate, because Obama is typically credited by Trump's critics for the economic improvements that have taken place since Trump took office, will be defined based on the start and end of the Obama administration. For this part, I will be generous to Trump and give him credit for all economic trends that happened from Nov. 2016 onward, even though he didn't take office until January 2017 and Nov. 2016 includes a week before Trump won the election. This yields two possible definitions of the "pre-Trump" period:

1. Jan. 2009 - Oct. 2016 (inclusive)
2. Feb. 2009 - Oct. 2016 (inclusive)

What if you wanted to include all months before Trump's first month in office (Jan. 2017)? Then you get these definitions (also inclusive):

3. Jan. 2009 - Dec. 2016

4. Feb. 2009 - Dec. 2016

For definition 1, the monthly pre-Trump decline in unemployment rates is -0.031%, and for definition 2, it is -0.037%. For definitions 3 and 4, the decline is -0.033% and -0.038%, respectively. These values yield an average pre-Trump monthly unemployment decline (MUD) of -0.035%.

Now what has happened since Trump took office? Has the unemployment rate started plummeting at a much, much faster rate than back when that evil socialist Obama was president? No, not at all: If you measure starting from November 2016, unemployment rates have declined at a monthly rate of -0.027%, from January 2017, -0.029%, and from February 2017 (his first full month in office), -0.03%, for an average post-Trump MUD of -0.029%. (Note that "post-Trump" means "post-election of Donald Trump", i.e. the time since it became clear he would be the next president of the USA.) These results, thus, provide no support whatsoever for the hypothesis that Trump has "reduced unemployment", that is, decreased the national unemployment rate below what would be expected if nothing changed from Obama's presidency. 

Next, I examined trends in the number of jobs created, also based on a BLS dataset that I downloaded. The pre-Trump trend (Jan '09-Oct. '16) in this number is +10,100/month, and that from Feb. '09 to Oct. '16 is +9,800/month. The pre-Trump trend from Jan. '09-Dec. '16 is +10,500, and that from Feb. '09 to Dec. '16 is +10,200/month. These values yield an average pre-Trump trend of +10,000 jobs/month. The post-Trump trend (Nov. '16-Jan. '19, "definition 1", including both Jan. '19 and Dec. '18, which both only have "preliminary" data according to the BLS) is +5,200 jobs/month, or if you start on Jan '17 (definition 2), +2,200, or if you start on Feb. '17 (definition 3), +7,100, for an average post-Trump trend of +4,800 jobs/month. Clearly, then, we have not seen the job creation rate increase on average since, much less as a result of, the election or inauguration of Donald Trump. 

Meanwhile, given that this time series is highly variable from one month to the next, I calculated the average numbers of jobs created/month for the four definitions of the "pre-Trump" period listed above. For definitions 1, 2, 3, and 4, the average number of jobs created/month in the pre-Trump period were 109,000, 118,000, 110,000, and 120,000, respectively, for an average of +114,000 jobs/month. In the post-Trump period, the averages for the three definitions are 204,000, 205,000, and 203,000, respectively, for a total average of +204,000 jobs/month added in the Trump presidency. This looks like clear evidence that Trump increased the number of jobs, but much of this large discrepancy is clearly driven by the Great Recession that Obama inherited in the first year or so of his presidency.