Re: Guest Post - GDP and lights

1

The conclusion, that authoritarian countries lie about GDP, is completely plausible

I thought in authoritarian countries GDP was real but has no relation to people's lives because it all goes into a few Swiss bank accounts.


Posted by: Cryptic ned | Link to this comment | 05-21-18 7:43 AM
horizontal rule
2

GDP TRUTHERISM


Posted by: Minivet | Link to this comment | 05-21-18 8:18 AM
horizontal rule
3

Only starting on the paper, but I wonder if it's measuring accuracy of national stats, as opposed to manipulation of them. eg, countries with lots of tax evasion (like, stereotypically, Mediterranean ones) might show similar issues, butin the opposite direction; authoritarian places with big SOE sectors will have national stats built on possibly inflated returns from those sectors; etc.


Posted by: Mossy Character | Link to this comment | 05-21-18 8:24 AM
horizontal rule
4

I mean, I don't have the expertise to evaluate the study's predictive power, but it's simply a fact that China is fudging its GDP, and I'd be amazed if Russia wasn't. Frankly most countries do around the margin, it's just tends to be bigger scale and more blatant in authoritarian countries, because they don't have to use tools like PPPs, quasi-public entities and off-balance sheet derivatives, they can just make up numbers.


Posted by: Ginger Yellow | Link to this comment | 05-21-18 8:25 AM
horizontal rule
5

I once worked with a guy whose job is was to calculate GDP numbers for a certain small country and he described the figure that was produced as highly speculative and also noted that his predecessor had been fucking up the calculation methodology for years.


Posted by: Spike | Link to this comment | 05-21-18 9:25 AM
horizontal rule
6

Finding out how the guy before you was fucking up the calculations is one of the great joys in life.


Posted by: Moby Hick | Link to this comment | 05-21-18 9:29 AM
horizontal rule
7

Particularly since I've never actually enjoyed the lamentations of women.


Posted by: Moby Hick | Link to this comment | 05-21-18 9:30 AM
horizontal rule
8

#notallmen


Posted by: Mossy Character | Link to this comment | 05-21-18 9:31 AM
horizontal rule
9

It's right there in the name "lament". It's not really supposed to be enjoyable.


Posted by: Moby Hick | Link to this comment | 05-21-18 9:33 AM
horizontal rule
10

Conan was definitely of the Germanic interpretation.


Posted by: Mossy Character | Link to this comment | 05-21-18 9:35 AM
horizontal rule
11

they don't have to use tools like PPPs, quasi-public entities and off-balance sheet derivatives, they can just make up numbers.

These tools would all affect reported government debt and spending, but I can't see how they would affect GDP.


Posted by: ajay | Link to this comment | 05-21-18 9:36 AM
horizontal rule
12

6: This may be one of the great differences between research and development programmers.


Posted by: dalriata | Link to this comment | 05-21-18 9:38 AM
horizontal rule
13

Relevant tweet from a seasoned China observer.


Posted by: Barry Freed | Link to this comment | 05-21-18 9:42 AM
horizontal rule
14

Note to self: Development programmers like to hear women cry.


Posted by: Moby Hick | Link to this comment | 05-21-18 9:56 AM
horizontal rule
15

(ahem) ... Archives.


Posted by: JP Stormcrow | Link to this comment | 05-21-18 10:43 AM
horizontal rule
16

I read something good about dictatorships fudging GDP growth numbers -- they clearly are,and it has big consequences -- but now I can't find it because it was on Twitter somewhere. Value add!


Posted by: Robert Halford | Link to this comment | 05-21-18 11:02 AM
horizontal rule
17

AIPMHB, Tooze's book on German stats history is super interesting.


Posted by: Mossy Character | Link to this comment | 05-21-18 11:12 AM
horizontal rule
18

Mossy, which Tooze book in particular? I'm in the middle of _The Deluge_ right now.


Posted by: Chetan Murthy | Link to this comment | 05-21-18 11:17 AM
horizontal rule
19

Statistics and the German State, 1900-1945


Posted by: Mossy Character | Link to this comment | 05-21-18 11:19 AM
horizontal rule
20

Furniture test for GDP fakery. Please compare sale and struck-through original price. Once a great country....


Posted by: lw | Link to this comment | 05-21-18 12:12 PM
horizontal rule
21

|| Dear lord but this is an astonishing story, they're coming fast and furious now.


Posted by: Barry Freed | Link to this comment | 05-21-18 9:20 PM
horizontal rule
22

21: Wow. I am only half keeping up, but the scope here is incredible.


Posted by: ydnew | Link to this comment | 05-22-18 4:00 AM
horizontal rule
23

We're a banana republic now, and not in the "understated clothing for professional women" way.


Posted by: Moby Hick | Link to this comment | 05-22-18 4:04 AM
horizontal rule
24

At least we still stand against profanity. Or academic achievement.


Posted by: Moby Hick | Link to this comment | 05-22-18 4:08 AM
horizontal rule
25

A little further into the OP, the n actually appears quite large:

After combining the Henderson et al. (2012) data on GDP and lights with the Freedom House data on democracy, I am left with 2,914 observations for 179 countries between 1992 and 2008.


Posted by: Mossy Character | Link to this comment | 05-22-18 4:12 AM
horizontal rule
26

It's pretty easy to get big N when you have all the countries in the world multiplied by all the years you are looking at. But if there's not much variance (because, for example, Sweden and Denmark are basically the same except Sweden has more alcoholics and trees, and because 1996 was a great deal like 1995), your analysis might still be driven by outliers.


Posted by: Moby Hick | Link to this comment | 05-22-18 4:18 AM
horizontal rule
27

Full disclosure: I haven't read the paper because it contained a lot of words.


Posted by: Moby Hick | Link to this comment | 05-22-18 4:21 AM
horizontal rule
28

It's the whole world. There's huge variation in lighting by country, and major change over time, considering poor-world growth over that period (not just China, most places.)


Posted by: Mossy Character | Link to this comment | 05-22-18 4:24 AM
horizontal rule
29

I was thinking that if nearly all the change in lighting is in the poor countries, then the inclusion of other countries is clouding findings.


Posted by: Moby Hick | Link to this comment | 05-22-18 4:35 AM
horizontal rule
30

All the countries have some degree of growth. And the poor countries have a mix of varying degrees of autocracy.


Posted by: Mossy Character | Link to this comment | 05-22-18 4:45 AM
horizontal rule
31

21. I'm obviously a bad person but I can't help but think the Z in "MBZ" is "Zakalwe."

(In truth it's Zayed, which is so boring.)


Posted by: DaveLMA | Link to this comment | 05-22-18 4:45 AM
horizontal rule
32

30: And richer countries are now mixing it up better.


Posted by: Moby Hick | Link to this comment | 05-22-18 4:50 AM
horizontal rule
33

||
Corona sucks.
|>


Posted by: Mossy Character | Link to this comment | 05-22-18 4:57 AM
horizontal rule
34

Have you tried putting a lime wedge in it?


Posted by: Moby Hick | Link to this comment | 05-22-18 4:59 AM
horizontal rule
35

What the fuck kind of beer needs a mixer?


Posted by: Mossy Character | Link to this comment | 05-22-18 5:02 AM
horizontal rule
36

That's why I've never tried it myself. I like Modelo if I'm drinking Mexican beer.


Posted by: Moby Hick | Link to this comment | 05-22-18 5:05 AM
horizontal rule
37

Fine. Tell them to export here.


Posted by: Mossy Character | Link to this comment | 05-22-18 5:07 AM
horizontal rule
38

They're probably lurking.


Posted by: Moby Hick | Link to this comment | 05-22-18 5:09 AM
horizontal rule
39

It's a bit early for Mexico.


Posted by: Mossy Character | Link to this comment | 05-22-18 5:12 AM
horizontal rule
40

OP paper also compares a bunch of different rankings of autocraticness. So the measure won't be perfect, but won't be arbitrary either.


Posted by: Mossy Character | Link to this comment | 05-22-18 5:51 AM
horizontal rule
41

1 appears actually to be the explanation.

The specific components for which we observe an autocracy gradient are also highly informative about governments' ability to manipulate official statistics. The components of investment and government spending have the shared characteristic of being highly dependent on government-reported information. Naturally, the government itself is the primary source of information on government expenditure, while the estimate for investment incorporates sub-estimates for investment by various levels of government and by public enterprises (Lequiller and Blades, 2014). On the other hand, the estimates for private consumption rely to a large extent on sources such as household surveys, which are more difficult to manipulate.
Public monies go into official spending stats but not into the local economy.


Posted by: Mossy Character | Link to this comment | 05-22-18 9:20 AM
horizontal rule
42
the results in column 5 indicate that the autocracy gradient is more than twice as large among countries with a history of communism.
[...]
Civilian dictatorships, which have even higher average FWI than presidential democracies, have the largest elasticity, with an implied exaggeration bias of 23%.Also worth noting is the fact that the elasticity is significantly lower for military and royal dictatorships, despite them having even higher average FWI values. One explanation for the observed difference among types of dictatorship is that royal and military dictatorships can deal with the threat of political turnover by means other than manipulation of information. In the case of royal dictatorships, these are mostly oil-rich `rentier' states with low or no taxation and extensive patronage networks

Posted by: Mossy Character | Link to this comment | 05-22-18 9:34 AM
horizontal rule
43

[FWI=autocracy index]


Posted by: Mossy Character | Link to this comment | 05-22-18 9:35 AM
horizontal rule
44

Oh hey, that's nice, they tried three other measures of freedom/autocracy and found the same association in I think all of them. (And also when making them binary variables.)


Posted by: Minivet | Link to this comment | 05-22-18 9:51 AM
horizontal rule
45

excess elasticity [between reported GDP growth and lighting growth] at times of low growth more than doubles in the most authoritarian regimes.
Which the author assigns to manipulation for publicity, but is also consistent with the embezzlement explanation. Leaner times, more skimming. OTOH,
estimates with the more flexible democracy categories in column 4 show a large and significant increase in the elasticity for the `not free' group in the year before elections. The excess elasticity for this group increases by more than 50% in the year before elections, lending support to the hypothesis that highly authoritarian regimes exaggerate economic performance even more in the run-up to elections.
Makes good sense, especially given the size of the spike.


Posted by: Mossy Character | Link to this comment | 05-22-18 9:56 AM
horizontal rule
46

I think Corona can be a good beer on a hot day, but it skunks easily because of the clear glass bottles.


Posted by: Spike | Link to this comment | 05-22-18 10:48 AM
horizontal rule
47

It was, I assure you, a hot day.


Posted by: Mossy Character | Link to this comment | 05-22-18 10:50 AM
horizontal rule
48

Also Dani Rodrik spent a while going into why it was obvious on inspection that Erdogan was manipulating the GDP stats to increase his popularity for a long time. (Or someone he retweeted did, I don't remember, he's my only source for Turkish politics.)


Posted by: Minivet | Link to this comment | 05-22-18 10:50 AM
horizontal rule
49

||

I just found out that a student of mine passed away. This was someone I liked - he came and sat in on my calculus class a few months ago, just because his friend was in the class, and he worked for the campus mail, so I saw him regularly even though he hasn't been my student for two years.

I hate not knowing how he died and what happened. If it's not available from online sleuthing, is the right thing just to resign yourself to not knowing? I suppose I'm just sad. Should I just assume it's a suicide if no one is saying? I think I'd be able to find it if it were a car crash, and I assume they'd have hinted in the email if it were medical.

|>


Posted by: heebie-geebie | Link to this comment | 05-22-18 10:56 AM
horizontal rule
50

So anyway, the paper appears solid to my stats-illiterate eyes, but appears to me in most cases to be assigning to manipulation what must in large part be explained by bad bookkeeping. It's also intensely boring.


Posted by: Mossy Character | Link to this comment | 05-22-18 12:56 PM
horizontal rule
51

This is wonderfully old school. There are all sorts of ways of estimating GDP or industrial output using proxy measures. They were much more commonly used during the Cold War when there were so many "denied" areas. Using lighting as a proxy is pretty clever and probably suggested by those photos of North and South Korea at night. You can actually see the DMZ. It's where the lights go out.

I looked at the paper. It seemed pretty solid given that no one has invented a reliable freedom-o-meter.

Thanks for the links.


Posted by: Kaleberg | Link to this comment | 05-22-18 9:21 PM
horizontal rule
52

All the countries have some degree of growth, huge variation in lighting by country.


Posted by: Luxury Furniture | Link to this comment | 07- 5-18 11:32 PM
horizontal rule