My paper was signalling the whole time - Robin Hanson wins again

post by rockthecasbah · 2020-08-04T21:13:16.016Z · score: 48 (30 votes) · LW · GW · 19 comments
Robin Hanson Meme

I recently wrote a paper on how the Jordanian monarchy decided who to give water to and who to take water away from. As I near completion, I am realizing that signalling theory from Robin Hanson gives a pretty compelling explanation of my results, probably better than my explanation. I will give a brief summary.

So in the 1990s and 2000's poor neighbourhoods of East Amman periodically had water shortages. Whole neighbourhoods would go without pumped water for a month and people would riot. The Government of Jordan (GoJ) does not like riots and were motivated to stop this.

Jordan has two relevant water sources they could use to make up the shortfall of 100-150 million cubic meters MCM. The Northern Highlands are close to the capital Amman and have a few profitable farms and a lot of smaller, unprofitable "prestige" farms owned by Jordanians. The southern desert is a good 600 km away and has four large profitable farms. The farms are owned by rich, politically connected Jordanian families and operated almost entirely by Egyptian migrant laborers.

The World Bank for twenty years suggested taxing the farms in the Northern Highlands to close the unprofitable ones then redirecting that water from the capital. Since Amman sits on the Northern Highlands the costs of transporting the water are trivial.

Instead the Jordanians paid about a billion dollars to build a massive pipeline to the southern farms, then shut them down instead. They don't publish the data I could use to compare how much more expensive the Disi pipeline solution was, but capital costs were about a billion USD and the energy costs are likely double the cost of other sources (of order 1 dollar per cubic meter). The water sectors cost recovery ration dropped by 30% the year they finished the project, financed by public debt until a 2018 fee increase forced by the WB.

The Jordanians have justified their decision for two reasons. The first is that closing the farms in the north would require negotiating with hundreds of farmers with diverse motives and finances, which was beyond the governments capacity. In the south they had only to negotiate with a small number of elites. This argument is strong.

The second argument is that closing the northern farms would have created unemployment which would create instability. They worried that farmers would lose their jobs and head to Amman to burden the social security system. The southern farms are worked almost completely by Egyptian farmers.

This argument is really weak if you think about it. Firstly, these unprofitable farms are using an average of 200 km^3, so to make up the difference they had to close 500 farms. The closed farms in the north also mostly hire Egyptians, so the lost jobs are like 2-10 per farm. So they spent a billion dollars to save 5,000 jobs. Assuming 10 Jordanian jobs lost per farm closed, they paid 200,000 USD per job saved. In a country with a GDP per capita of 5,000 USD. Assuming they protected those jobs for ten years (the aquifers will collapse eventually anyway), they could have just paid the farmers the money and saved 75%. This is a conservative estimate, since many of those farms have no Jordanians on them.

If you had a billion dollars to spend on Jordanian unemployment, paying to substitute water to keep unprofitable farms afloat is the last thing you would do. Honestly you could have just cut the Egyptian farm worker visa program and killed two birds with one stone by increasing Jordanian employment, cutting the implicit subsidy to the farms, and they would have spent 0 dollars. It is possible the Jordanians just didn't think of this, although the World Bank never got tired of pointing it out.

I slightly prefer the explanation that the GoJ was signalling loyalty to these social groups. The farmers in the Northern Highlands are inside the ruling coalition the royal court has to signal that they get special priviliges. And failing to supply East Amman is a clear signal to the masses that the King doesn't care about their lives, which they can't do. If politics is really about loyalty signalling (not unemployment), the GoJ's actions are more instrumentally rational.

Also Jordanians do perceive the water transfers as loyalty signals, based on interviews from anthropologists in donor areas.

But the paper is almost accepted so no time to change it now.

19 comments

Comments sorted by top scores.

comment by romeostevensit · 2020-08-05T01:59:22.352Z · score: 6 (4 votes) · LW(p) · GW(p)

My understanding is that it is generally politically impossible to just cut checks directly to impacted individuals without funneling it through some sort of narrative laundering organizations.

comment by rockthecasbah · 2020-08-05T02:21:21.115Z · score: 5 (3 votes) · LW(p) · GW(p)

You've raised a great question. Two comments.

Their stated goal is reducing urban unemployment

The most efficient way to reduce unemployment is neither to cut checks nor to subsidize water in the desert. The thing is, the Jordanian policymakers did not say "we want to keep everyone having the same amount of income", which is the problem you are getting at, and one other countries have attempted (with mixed results, see below). What the Jordanians said is "we want to keep urban unemployment down". So subsidizing farms is a way to keep people from migrating, but the water provision is a terribly inneficient subsidy. Furthermore, from a "cause prioritization" standpoint its also really inneficient. Paying to keep unprofitable businesses alive is a less efficient way to reduce unemployment than, say, investing in infrastructure or even just subsidizing profitable business sectors.

Basically, picking winners is hard but its better than picking an existing loser and buying them an input good at 4 times its market value. The most efficient way to reduce So their stated argument "we want to reduce urban unemployment" is either not their real goal or they are very biased.

Some polities can cut checks reasonably efficiently, but the Jordanians cannot

Some political systems do a better job of cutting checks than others. The conventional wisdom is that most European countries have successfully preserved pastoralism through cutting checks without having to distort their economy with across-the-board subsidies. The conventional wisdom is also that the US attempt to cut checks for export job loss has failed. blah blah first-past-the-post blah blah pork barrel.

The Jordanians are probably very very bad at cutting checks. The best piece of evidence is that the GoJ expects to fail at cutting checks, and they would know. The second best piece of evidence is what happens when they open municipalities [LW · GW] (patronage). I suspect USAID gets laughed out of the room when they say "the Danes did it, why can't you". So you are correct about that.

comment by JayMon · 2020-08-14T17:50:57.120Z · score: 1 (1 votes) · LW(p) · GW(p)

Question, if you know. What's the Jordanian construction sector like? While building a pipeline may only weakly protect the jobs of the Amman farmers, are local companies/contractors involved in the pipeline's construction?

Does the pipeline go to any other areas which may be exploitable in the future? Do you know what the relations were like with the Southern Desert families? This could be an attempt to push out a faction and install another.

comment by rockthecasbah · 2020-08-14T20:48:43.600Z · score: 1 (1 votes) · LW(p) · GW(p)

Great questions! You questions highlight a few minor contributing factors.

While building a pipeline may only weakly protect the jobs of the Amman farmers, are local companies/contractors involved in the pipeline's construction?

They actually contracted it to a Turkish company, GAMA, because they lacked the capacity.

Does the pipeline go to any other areas which may be exploitable in the future?

Not really. In theory it could be linked to western Disi but they left the farms there. They would probably just build a new pipeline to the Gulf of Aqaba to decrease energy cost bc Pythagoras.

Do you know what the relations were like with the Southern Desert families?

The GoJ exempted locally owned tribes, but it did become less popular locally. The central desert families north of Disi (Ma'an) are considered a strategic tribe and get special treatment, but the Disi tribes don't. They get to flirt with a lot of tourist girls tho.

This could be an attempt to push out a faction and install another.

That was part of it. The farms which were closed were mostly owned by Palestinian elite families. So the rival transjordanian elites in the military were happy to see them go. But they would never pay a billion for that pleasure. See Keulertz.

comment by habryka (habryka4) · 2020-08-04T22:23:24.621Z · score: 4 (2 votes) · LW(p) · GW(p)

Mod note: Replaced the link to the image with the actual image, since I assigned substantial probability to that being your intention, but our editor image handling being too confusing and you giving up. But feel free to revert it if it was intentional.

comment by rockthecasbah · 2020-08-05T03:43:38.751Z · score: 1 (1 votes) · LW(p) · GW(p)

This is perfect! Thank you Habryka for the help!

comment by Pattern · 2020-08-05T15:19:16.841Z · score: 3 (2 votes) · LW(p) · GW(p)

What was your argument in the paper?

comment by rockthecasbah · 2020-08-05T16:42:10.919Z · score: 11 (7 votes) · LW(p) · GW(p)

Basically, cities want secure stores of groundwater for the long term. But farmers will use up all the water if they can. Right now most countries let total anarchy reign because the farmers are hard to stop (they’re everywhere and they constantly dig wells and they hide and bribe and shoot at regulators. So our primary question was “if we look at a country with really severe urban deficit, does that motivate the government to go out and reduce overuse? Or are the challenges and perverse incentives impossible to overcome?

We found that the Jordanians took advantage of the fact that barriers to regulation are unevenly distributed between aquifers. So you can find aquifers that are cheap to enforce and have few people capable of rioting/couping (the Jordanians fear revolt more than coup, but usually expect the reverse). The current preferred approach is to tax all the farmers in every basin (aquifer) to close a few less productive farms in each basin. But then you’re spending your scarce enforcement and “pissing people off” budget in low yield basins. Instead the Jordanians targeted one area as “preserve” for future urban use and successfully shut it down completely (way lower enforcement costs). Then you concentrate your resources on protecting one area.

Ooo I can make an analogy to wildlife preserves!

By coincidence the easiest area to make preserve was 600 km south and 1 km below Amman. So they paid probably an extra 2billion usd (1B Capital + 100 M/year energy) over 10 years to pump that water. They always framed the “pissing people off” budget as about “unemployment”, but it wasn’t about unemployment because they could have used 2 billion to reduce unemployment more efficiently with almost any other policy!

comment by rockthecasbah · 2020-08-05T03:43:06.051Z · score: 3 (2 votes) · LW(p) · GW(p)

Can someone explain why this got frontpage? While I stand by my core claims, I put zero effort into organizing these ideas. Was it the meme at the top?

Did I naturally organize the argument well by chance?

Do you all really like signalling when applied to politics?

comment by Ben Pace (Benito) · 2020-08-05T08:05:51.490Z · score: 4 (2 votes) · LW(p) · GW(p)

The frontpage/personal distinction is many things, but it is primarily an on-topic/off-topic distinction, and your post is on-topic for LessWrong (whatever that weird, amorphous topic is exactly). It's an interesting post.

(The meme is great, and is one of the reasons I upvoted the post, but wasn't relevant to frontpage/personal.)

comment by oooo · 2020-08-05T04:42:13.511Z · score: 4 (3 votes) · LW(p) · GW(p)

Two words, one meme: Robin Hanson.

comment by rockthecasbah · 2020-08-05T04:58:29.452Z · score: 1 (1 votes) · LW(p) · GW(p)

hahaha

comment by ryan_b · 2020-08-05T17:04:05.215Z · score: 2 (1 votes) · LW(p) · GW(p)

I am highly confident Robin Hanson addresses this somewhere, but it strikes me that in this case signalling loyalty and being loyal are identical.

I think the signalling framework is radically superior to the utility-you-give-up framework for establishing what social things people really value.

comment by rockthecasbah · 2020-08-10T18:11:29.043Z · score: 1 (1 votes) · LW(p) · GW(p)

You’ll have to explain the “utility you give up” framework.

Signaling loyalty and actually being loyal are pretty much the same here. We need other details to see if they diverge.

Arguably letting the farms close and replacing them with competitive industries for Mafraq would be cheaper and more loyal, but would signal less loyalty. But the Jordanians do not expect to succeed at such development projects anyway.

comment by ryan_b · 2020-08-10T22:28:09.846Z · score: 2 (1 votes) · LW(p) · GW(p)

The method I am most familiar with for estimating the value of things that are intangible (like loyalty) is comparing the price paid vs. the one that could have been paid if the intangible value had not been a factor. The intuition is like branding in marketing, wherein the simple calculation is something like:

(price of Apple computer) - (price of non-Apple computer with same specs) = (value of Apple brand on the computer)

So in the water transfer example, it looks like:

$1B pipeline to prevent unemployment - $0 to cut Egyptian visas = $1B is the value of Northern Highland farmer loyalty.

The structure is intuitive, but it suffers from the usual problem of taking what people say their goals are seriously. Further, it isn't at all clear (at least to me) whether this means that is the price overall or just right now, whether this is payment for loyalty rendered or an investment in future loyalty, etc. And this just gets worse the more factors I try to weigh because the same ambiguity applies to each other factor.

Granted, I might just be very bad at the method; separating signals out from regular transactions help make social questions easier to work with for me.

comment by ChristianKl · 2020-08-11T18:33:30.168Z · score: 2 (1 votes) · LW(p) · GW(p)

While I don't like MacOS very much, it seems to me strange to argue that the average value of it to people is zero.

comment by ryan_b · 2020-08-11T19:44:26.119Z · score: 2 (1 votes) · LW(p) · GW(p)

The datedness of the example reveals how old my last exposure to the literature was. I imagine people making current examples would at least switch to iPhone or something. Possibly one of those clothing brands that have long lines for about six weeks before the trend moves.

comment by betulaster (raman-malykhin) · 2020-08-14T11:58:15.471Z · score: 1 (1 votes) · LW(p) · GW(p)

So... it's possible that there is something about Middle Eastern politics that I don't understand, and it would be cool if you could clarify. If I understand you correctly, you write that farms in the South are owned by rich people. At the same time, you write that farms in the North are somehow connected to the ruling coalition, and because of this the government had to signal loyalty to them.

I was under the impression that in monarchic/autocratic countries it was near-impossible to be rich while not being connected to the ruling group (= not being the kind of agent the ruling group would need to signal loyalty to). The farmers in the South contradict that. How does this work?

comment by rockthecasbah · 2020-08-17T12:16:38.268Z · score: 3 (2 votes) · LW(p) · GW(p)

Great question.

The truth in your model. Both the Saudi and Jordanian regimes give "import licenses" to families (Corleoni-style extended families). They basically say "only this family can import (nitrogen or automobiles or washer-dyers). The Jordanians have one on tomatoe paste which is why their tomato paste tastes crappy. Particularly when a autocrat dies, some companies or families are dispossessed if the ruler suspects they will be disloyal. Mohammad Bin Salman did this recently.

What your model is missing. Factions are rarely just the top leadership. The most stable faction type is basically a pyramid scheme. You have some leader at the top - a tribal elder, a warlord, a colonel, a utilities company owner etc. who has a direct relationship to the autocrat usually. Then beneath him are other elders, and beneath them are heads of nuclear families or clans, and beneath them are usually prosperous peasants. As you go down the rewards decrease, but the reward-responsibility combination continues. IIRC, the Mafraq tribes people saved the Jordanian monarchy as recently as Black September and continue to serve in the army in high numbers. If you just recruited from every tribe equally there could be a revolution! Can't have that.

TLDR: Even though the tribes people are not elites, this tribe supports the Jordanian army.

Fun fact: Arab tribes elect their sheyoukh, so are a more "egalitarian" faction type than is typical.