It is strange that there should be so little reading in the world, and so much writing. People in general do not willingly read, if they can have any thing else to amuse them.

Dr Johnson, Boswell’s life of Samuel Johnson, Vol IV May 1, 1783, p. 513. He also said:

I never desire to converse with a man who has written more than he has read.

So, what is the explanation? Well I’ve reading a lot and writing little. Or, perhaps more accurately, I’m reading a lot of work and revising my own. Both are fairly time consuming and so I’ve cut back on these occasion thoughts. But my idle thoughts will be posted here when the mood strikes me.

A Nice little clip from Yada Yada Yada Econ
(I don’t seem to be able to embed this, so go to the site to watch it)

If you ignored the video, then the unfunny version is that Jerry Seinfeld is night guy – getting up early in the morning after a late night is morning guys problem. This is a dual-self model, as talked about by a whole host of economists. To borrow Jerry Seinfeld’s terminology, the donor that agrees the aid contract is night guy, and the morning guy is the donor that has to enforce the aid contract. However, it is emotionally quite difficult to withhold aid from poor countries – but hey that’s morning guys problem. Night guy doesn’t predict this difficulty, and so carries on making tough contracts that don’t get acted upon. Unfortunately, I really can’t embed the video in the revised version of the paper, so it will only get this far.

Another piece which won’t make it into the revised paper is a host of quotes from the Lord’s select committee on aid. Select committees in the UK parliamentary system are very topical at the minute (see the judgement that Rupert Murdoch being unfit to run a company), but they are very dry affairs. So I was surprised to find such good quotes in them, from some key people, that illustrate the difficulty in enforcing aid conditions. A case in point:

In a period of 35 years, we really should have learned to read the signs and to act on them, and I think we have a very, very short term view, unfortunately, in our foreign policy and our aid policy.”

This is Sir Edward Clay, famous for taking the Kenyan government to task on corruption while he was the British High commissioner there. His point is that being tough is difficult in the short term but it buys you more power in the long run. But typically donors think short-term and don’t punish recipients.

Then we have an interesting little exchange about the mechanics of how a decision is made regarding how to change aid in the face of poor governance/corruption/etc

Lord Tugendhat: What is the procedure in the department whereby you would decide that country X is not cutting the mustard on a sufficient number of these and that you should suspend, withhold or at any rate not dish out in the normal way your aid? Would it be the Secretary of State who decides? How would it be done?

Rachel Turner (Director, International Finance Division, Department for International Development): It would be the Secretary of State. We tend to articulate this as a four-step process if we are concerned about performance on one of these partnership principles. The four steps are fairly straightforward. The first is that we signal concerns to the Government and we intensify dialogue around them. We might delay all or part of a disbursement. In our annual report each year, we publish any specific delays to disbursement because of concerns about a partnership principle. We may change the way in which we deliver aid. We may take aid away from the Government and route it through NGOs or other sorts of partnerships. Finally, we may stop aid to the Government and/or to the country. So we have that four-step process.

So, a four step process with the last step being withholding aid, and the decision is made by the secretary of state. This is interesting because while certain statements may be made about how and when aid would be cut off, it is unclear from most documents who makes the final decision.

Here is another interesting exchange about whether donors can actually withhold aid, which could be taken straight from the theoretical models of the difficulty of withholding aid:

Baroness Kingsmill: And could we withhold aid if that was not forthcoming?
Max Lawson (acting Head of Advocacy, Oxfam): My personal feeling is that, in almost every instance, withholding aid achieves the opposite of what you want to do. I am watching with horror at the slow-motion car crash that we are seeing in Malawi at the moment , where the President, who had been doing a great job, has basically, as far as we can see, gone a little bit mad and is very sick. We are seeing a situation where he is getting very, very intransigent and he is looking across the borders at Zimbabwe and is thinking and acting in a very similar way. We are withdrawing our aid, as the UK, and all this is succeeding in doing is making him more angry, so that he is impinging swingeing cuts on his population to pay for that. I really worry that quite often a kneejerk reaction, and the removal of aid entirely, only hurts the poorest people… It is about weighing the different rights up against each other…

And yet another interesting exchange illustrating the same thing.

Lord Lipsey: It is more than just complicated, isn’t it? If you withdraw the aid, you may have a dual effect. Yes, you punish the person who has been ripping off the money but you also punish the people who were benefiting from the project. If, out of the money going to an education project, 80% was getting through and 20% was going into the pocket of the civil servant responsible for that bit of education, you would be making the beneficiaries of  the 80% suffer if you stopped the aid.
Laurence Cockcroft ( Member of  the Board of Trustees, Transparency International UK) : That is part of a broader argument. There is frankly not a ready answer to that question. In informal discussions with many people in the countries that we are concerned with, I found an extraordinary willingness to be tough with those at the top who are milking the system, but that is a slightly different point.

It is rare to get such a good insight into the thought processes of the people whose decisions you are trying to model, and so these quotes are really useful background. Unfortunately I don’t have the space for all of them, or the guts to include Jerry Seinfeld in the paper.

• The Guardian have been very supportive of the whole MVP idea, and their write up could have been written by Sachs himself such is its support for the paper.
• A View From The Cave quotes a skeptical article in Nature, and has a long summary of the points, but he says “This all not to suggest that the program was not successful. We have to take the gains at face value.” Hmm.
• Lawrence Haddad is not convinced, and doesn’t think this tells us enough about MVPs to ‘prove’ whether they are good or bad.
• I agree with Matt at Aid Thoughts when he says “Enter the Lancet, a reputable medical journal which has a worrying tendency to publish really disreputable social science research”. He also puts the study in the context of widely falling rates of infant mortality across Africa i.e. the MVP result is not exceptional, and a properly chosen counter factual would tell you that.
• Lee Crawford is… even less impressed. He goes through some of the non-touted results, showing the paper cherry picks its headline results somewhat.

Personally, I’m waiting to see Michael Clemens take the paper apart in his admirably detached and sensible fashion. Three points from me:

1. It seem incredible that the topic of value for money isn’t engaged with. At all.
2. To spend so much money with so little regard for results (and defensible methodology) is, well, shocking.
3. Claiming success based on one of many indicators is choosing your target after you’ve shot.

• William Easterly, doing an odd ‘stand-up’ routine. He also makes points about learning by failure (entrepreneurial) vs learning by experts (RCTs). It was more of the same point he hammers away at (searchers > planners) but in the RCT context.
• Michael Clemens, giving a really great talk on MVPs, and more broadly the post-MDG world where either an ‘evaluation’ or ‘goals’ mindset will take the lead. He was passionate and level-headed the whole way through.
• Bernadette Wanjala, presenting her paper (which I heard at the CSAE), which is the best evidence of which I’m aware regarding the impact of MVPs. She basically finds that the massive investment had little/no effect on those in the MVP.
• Stewart Paperin, saying that the Soros foundation helped fund the MVP without thinking it would solve poverty for the treated villages. This is despite the per person spend of several thousand dollars (492 dollars per person for 5 years, i.e. 8 times income). He completely missed  Michael Clemens’ point, and downplayed the usefulness of knowing what effect this massive investment had. I find it amazing that you can spend so much money with so little regard for the effect it has on the lives of those you claim to be trying to help.

DfID under the Conservatives

Lawrence Haddad has a nice little reflective piece with comments from several interesting people on DfID’s performance under the coalition government. In general their is praise for the 0.7% target and a good spread of different criticisms, which either means DfID are doing well (not missing anything big & obvious) or awfully (they could improve everywhere!). It definitely appears to be the former.

There is an interview with DfID’s Secretary of State, Andrew Mitchell. It isn’t particularly revealing, but it is worth a read.

The 0.7% target

The Lords start this off, with a good quote:

“We believe that development aid should be judged by the criteria of effectiveness and value for money, not by whether a specific arbitrary spending target is reached.”

There is some reaction from John Hillary in the guardian, where he states “we must now move beyond aid in any discussion of social and economic justice.” Alice Evans (head of ODI), in the new statesman, is a lot more thoughtful and worth reading. At the other end of the spectrum is Ian Birrell, in the Spectator about aid, with a fairly familiar disdain towards the whole ‘aid industry’. He makes some cheap points, and talks about the great oxfam pool debate, without giving any context or fairness to it. He gives some examples of waste – the usual stuff.

This post shows how far the aid debate in the UK has come, and how far it needs to go. The public does accept that aid is an important part of the UK government, but a mature debate about aid doesn’t consist of each side pretending aid is uniformly good or bad. When we accept that aid can be either, we can work on improving it, bit by bit, without fear that ‘aid is bad’ stories undermine the public support for aid. And we can celebrate successes with more credibility.

This is a nice example of confirmation bias (where you listen most to evidence that confirms what you already think) but also publication bias (i.e. boring papers aren’t published). Because the prevailing attitude of funders has been that clean stoves are important, the null result is interesting and the alternative result isn’t, but in time the more interesting result would be the alternative result. Unfortunately, boring papers are quiet useful – they confirm things we already know, and that is a very good thing.

One thing struck me while reading the post – publication bias and confirmation bias will often work in opposite directions. Confirmation bias will mean that something we all think is more likely to be listened to, whereas publication bias means it is less likely to be published. As usual, my go to recomendation on this kind of thing is Bad Science.

Jakob Svensson’s call for micro research on institutions, which drew my attention to this paper, forthcoming in the QJE. It looks at strong/lite participation in community monitoring of health provision and is a nice example of the kind of literature which he was pushing.

Martin Ravallion spoke about the effect of poverty on growth, and claims growth economists have mistakenly attributed this effect to inequality. You can read his thoughts in a blog post or in the forthcoming AER article.

The last panel had little structure, but had a few intersting discussions. It was nice to hear Ernest Aryeety’s optimism over lessons learnt. Ravi Kanbur pointed out that young nations need time to foster a sense of national unity – a non-linear process. It was also interesting to hear a certain member of the panel be quite so openly negative about Wade’s decision to stand in the Senegalese elections – but I’m sure he would be very happy to see Wade make the decision to stand down, as soon as it became crystal clear he would lose the election.

A bit of an eclectic mix, but there should be something worth looking at.

Basically, they wanted to look at whether RCTs as normally done are missing the point. Standard RCTs in medicine have ‘true’ placebos, with the control group receiving an injection or sugar pill, whereas in economics the control group normally has no ‘intervention’. Medicine is advanced in looking at placebo effects: an injection has a bigger effect than a pill, and a coloured pill has a bigger effect than a white pill. The effect is not just psychological: it really does effect how the body works (see Bad Science for a readable account). So, what has this got to do with development? Basically the idea is that the treated group may change their behaviour because they know they are being treated. Bulte et al found an ingenius way of testing for this.

They took two seeds. One was a ‘miracle new seed’, the other a standard seed. First, they ran a standard RCT whereby they gave the standard seed to a control group, and the new seed to the treated group (with random assignment). If you just looked at the results on this, you’d conclude it was a great success. But…

They also ran a second part of the experiment. This second group of farmers was also divided into two, and they got either the new seed or the old seed. However, they didn’t know which one they got. If you compared the yields of these farmers, there was no difference – the miracle seed didn’t work. The mechanism seems to be that treated farmers (who knew they were being treated) increased the amount of land they gave over to the miracle seed. The great thing about the paper is that it allows you to get a handle on the size of these effects. We can’t run sham clinics or sham micro-finance groups, but the indistinguishable seeds allow us to distinguish between the effect of the treatment on the treated, and the effect of knowing you’re being treated on the treated. This sheds some doubt on results in areas where we can’t easily control for this effect.

I have one question for you, readers. They call this effect (the effect of knowing you’re being treated on the treated) the pseudo-placebo effect. I thought the Hawthorne effect covered this but Erwin thinks this is just the narrow effect of working harder. I suspect we’ve had a case of semantic narrowing in economics, whereas in medicine the Hawthorne effect is much broader. Any ideas?

The CSAE conference: I am in a post-conference fog have listened to a vast number of presentations in a few days. I will post links to interesting papers at some point (once I recover) but until then I’d recommend having a look at the plenary sessions which are available to watch. I’d suggest Martin Ravallion’s would be a good place to start, as he argues poverty (not inequality) restrains growth. It was a very interesting talk.

1. Own some stocks,
2. Own a copy of Stata, and
3. Want to use 2 to check on 1

… then this post may be very useful for you. I used the stata command stockquote so you’ll have to install that first (type: findit stockquote) if you haven’t already. That command does all the heavy lifting, I’ve just written 3 small programs to mean you can use it to see what your portfolio is worth, with minimal fuss. I wrote it for 2 reasons: first to practise using dates in stata, and second to solve a problem. I got fed up with using google finance to check my (very modest) investments as it seems to add or subtract cash from the portfolio for no apparent reason. Stata does the job well, but as soon as others use it I’m sure you’ll find a bug or two: let me know if you do at paulclist -at- gmail.com.

Step 1: copy and past the hundred-ish lines of code at the end of this post (after the fold) into a do file. Run it as an individual do file, save as an ado or just stick it at the top of a do file. If several people find it useful I may write it up properly, so it can be installed from stata and so forth, but for now I’m using a low tech solution.

Step 2: Write a do file with your investments using three commands, giving information on the starting date and funds (stockstart), purchases (stockbuy) and sales (stocksell).

Step 3: Press do. You’ll get the daily price of your portfolio and the equivalent performance of the ftse.

So, lets do a little example of step 2 (included at the end of the code below too):

stockstart 2011 01 03 1000
stockbuy 10/01/2011 RR.L 152 1000
stocksell 10/01/2012 RR.L 152 1169
tsset Date, daily
tsline ftsecompare total

The first command  tells stata that you started investing on the 3rd of January 2011 with 1000 (I’ll use pounds here but the currency is determined by the currency of the stock you buy). The second command says that you bought 152 shares in Rolls Royce (you need to use Yahoo Finances ticker symbols) on the 10th of Jan 2011 at the total cost of 1000. The third command tells Stata you sold the whole stock on the 10th of Jan 2012, for a total of 1169. The fourth and fifth commands give you a graph of your performance versus the alternative of putting all you money into the ftse on the day you started investing. This is the graph it makes:

Stata for Stocks: the example graph

As long as you follow the rules of how to out the data in, it should work ok. Do let me know if you pick up any bugs, or if you find it useful!

The main code, with a 4 line example:

***********define stockstart
capture program drop stockstart
program define stockstart
args year month day cash
display "year " `year' " month " `month' " day " `day' " cash " `cash'
local sysdate = c(current_date)
display "`sysdate'"
local Day = substr("`sysdate'", 1,2)
display `Day'
local Year = substr("`sysdate'", 8,.)
display `Year'
local today=date("`sysdate'", "DMY")
display "`today'"
local Month = month(`today')
display "`Month'"
stockquote ^FTSE, fm(`month') fd(`day') fy(`year') lm(`Month') ld(`Day') ly(`Year') frequency(d)
keep date close
rename close ftse
gen cash=`cash'
gen portfolio = 0
gen total =`cash'
gen Date = date(date, "YMD")
gen ftsebase=ftse if _n==1
sum ftsebase
local ftsebase=r(mean)
gen ftseunits=`cash'/`ftsebase'
drop ftsebase
gen ftsecompare=ftseunits*ftse
save "stockFTSE.dta", replace
end
*********************************define stockbuy
capture program drop stockbuy
program define stockbuy
args date ticker units cost
local Purchasedate=date("`date'", "DMY")
local Pday=day(`Purchasedate')
local Pmonth=month(`Purchasedate')
local Pyear=year(`Purchasedate')
local sysdate = c(current_date)
local Day = substr("`sysdate'", 1,2)
local Year = substr("`sysdate'", 8,.)
local today=date("`sysdate'", "DMY")
local Month = month(`today')
stockquote `ticker', fm(`Pmonth') fd(`Pday') fy(`Pyear') lm(`Month') ld(`Day') ly(`Year') frequency(d)
keep date close
local ticker = subinstr("`ticker'",".","",.)
display "`ticker'"
rename close `ticker'
gen Date = date(date, "YMD")
joinby using "stockFTSE.dta", unmatched(both) _merge(_`ticker')
sort Date
display `Purchasedate'
capture gen `ticker'number =0
replace portfolio=portfolio-(`ticker'number*`ticker'/100) if `Purchasedate'<=Date
replace `ticker'number =`ticker'number + `units' if `Purchasedate'<=Date
replace portfolio=portfolio+(`ticker'number*`ticker'/100) if `Purchasedate'<=Date
replace cash=cash-`cost' if `Purchasedate'<=Date
drop _*
replace total=cash+portfolio
order date cash portfolio total Date ftsecompare ftse
save "stockFTSE.dta", replace
end
****************************************define stocksell
capture program drop stocksell
program define stocksell
args date ticker units cost
local Purchasedate=date("`date'", "DMY")
local Pday=day(`Purchasedate')
local Pmonth=month(`Purchasedate')
local Pyear=year(`Purchasedate')
use "stockFTSE.dta", clear
local ticker = subinstr("`ticker'",".","",.)
display "`ticker'"
replace `ticker'number =`ticker'number - `units' if `Purchasedate'<=Date
replace portfolio=portfolio-(`units'*`ticker'/100) if `Purchasedate'<=Date
replace cash=cash+`cost' if `Purchasedate'<=Date
replace total=cash+portfolio
order date cash portfolio total ftsecompare ftse Date
save "stockFTSE.dta", replace
end

***********************************an example
stockstart 2011 01 10 1000
stockbuy 11/01/2011 RR.L 152 1000
stocksell 11/01/2012 RR.L 152 1169
tsset Date, daily
tsline ftsecompare total