Tom Pepinsky

Category: Economics

  • Good Things Work

    Over lunch today, some economist colleagues and I discussed the recent turn towards randomized clinical trials (RCTs) in development economics. For those who don’t know, this is the move to implement large scale medical-style trials of economic policies in poor countries. Things like, if we give microfinance to people, does it work? Study this by randomly dividing the population into a group that receives it, and one that does not. Chart the difference in the two groups. (With the proper assumptions, of course) you therefore have a great way to tell if microfinance works, and if so, how much.

    These RCTs are a big deal, and they are really expensive. They are probably a fad–although one that I’ve kinda taken part in myself–and I do  not have any penetrating analysis beyond what much smarter people have already said about it. Our main conclusion over lunch is that above and beyond what existing critiques have said, RCTs aren’t likely to tell us much about what we care about for making poor people better off. These are two economists who care a lot about specific poor countries in Southeast Asia–take, for example, Burma or Cambodia–but who draw very few actionable conclusion from RCTs about how one might improving these people’s lives.

    The main conclusions of magisterial studies like this are that you should probably let women participate in decisions and you should vaccinate children against malaria and you should let credit flow to productive growth sectors. That’s hardly radical, but it’s nice to be absolutely positive that Good Thing do Work, and that not always in way that we think that they do. It would useful to know if they did not, and it’s good to be sure that we’re not just cherry-picking the evidence to confirm our biases. (I am sure there are people who might disagree that governments should do Good Things, and it’s definitely important that the evidence on the effectiveness of what many of us used to think were Good Things is pretty weak.) But my lunch partners and I–again, these two are old school development economists with deep country expertise, something rare in the U.S. these days–concluded that the main problems facing poor countries are in implementation.

    Here’s what I mean. RCTs are great for confirming that Good Things that we think Work actually do. But the main problem facing most poor countries is not that we don’t know that, but how to get the policies in place.

    I can see six possible reasons why obviously pro-poor policies (and I don’t mean income redistribution, I mean things like making sure children aren’t drinking filthy water) aren’t provided by politicians.

    1. politicians don’t know if it’s a good idea to provide them

    2. politicians don’t know how to provide them

    3. politicians have budget constraints and so they can’t afford to provide them

    4. politicians have political incentives that make them direct their energies elsewhere

    5. the bureaucracy responsible for making these things happen is by necessity so confusing and complicated that it’s just really easy for well-meaning politicians and bureaucrats to fail and it’s no one person’s fault

    6. politicians are evil and don’t care about the poor (this is not a joke, I think it aptly describes people like Than Shwe)

    RCTs solve problem (1). But I am convinced that (2-6) are the actual barriers to allowing Good Things to Work, and these are precisely the problems whose solutions are unlikely to be found through RCTs. It’s also probably the case that when you try to solve one of these problems you create another one (like, removing budget constraints [massive aid] creates the wrong political incentives [massive corruption]).

    So how do we get Good Things to Work? No idea. I am increasingly convinced that the solutions are not to be found in institutional engineering or technocratic competence, but I don’t know where to go from there. I would tell Burma’s government to do Good Things and that they will definitely Work, but there’s no way to get them to listen, and there’s no secret why not.

  • How Does Burma Lie with Statistics?

    Today I (TP) attended a talk that you could not easily find in the U.S.: a presentation on Burma’s macroeconomy and how it works, by an economist at Macquarie University. Poor Burma has a horrible economy. But for me, the striking thing about any research on Burma is always how hard it is to find good information about the country, even for people who speak Burmese and who do research there. Today, the presentation highlighted the fact that Burma’s government produces national statistics that are obviously false.

    Now, it is common to complain about the statistics produced by all countries’ governments. But examples of out-and-out fabrication of national economic statistics are comparatively rare. In most cases, people who complain about fabricated economic data either (a) do not understand what the statistics mean, (b) don’t understand how the statistics are calculated, and or (c) disagree with the conclusions that the data suggest. To take one case, it is fairly common now for certain types of people to complain that someone in the U.S. government is cooking the core inflation data. The reasons are probably that they want there to be high inflation (so they can blame it on the president) and they don’t understand that inflation is not the same as loose monetary policy or high gas prices.

    In Burma, though, economic data are transparently false. For example, the Burmese government reports that Burma’s economy has grown by about 12% per year on average (in real terms) for the past decade. If that were true then Burma, not China, would be the world’s fastest growing economy. In reality, the economy has probably grown about 3% per year, which is OK for a rich country but extra bad for a poor country like Burma.

    The question that interests me is not why the Burmese government lies with statistics, but how it lies. When you think about it, it’s not hard at all to tell that the figures are false. Data on the components of GDP seem accurate, but there’s no way that a country that reports that 50% of its GDP is in agriculture and that it has no noticeable increase in energy use has been growing so rapidly for so long. But the problem is, since it’s not hard to figure this out, how could they be cooking the books?

    One possibility is that they are just making up numbers. It’s possible, but this raises the question (to me at least) of why bother fabricated data that everyone knows is fabricated. Remember, no one is fooled by the government’s figures. An interesting alternative is that it comes down to inflation rates. When you calculate GDP, you calculate it in nominal terms and then apply a correction to adjust for yearly inflation. If you are the Burmese government, though, you persistently underestimate inflation. They do this because inflation is a signal of some pretty rotten stuff going on in terms of the government using the printing presses to finance its own consumption. So you claim that inflation (which is hard to measure anyway) is a lot lower than it is. The consequence is that Burma has nominal GDP growth, but is unwilling to correct properly for price changes. That will mean that its official figures will dramatically overstate real GDP growth.

    Exchange rates are also interesting. The Burmese government officially says that you trade 6 kyat to the U.S. dollar (more or less). But in reality, the market says that the rate is around 900 kyat to the dollar. This means that when government-owned enterprises report their profits (much of which are in U.S. dollars for things like natural gas exports), they are only required to submit 6/900 = .67% (less than a penny on the dollar) of it to the government’s accounts. The rest can be used as slush money.

    If this all sounds pretty close to money laundering, it is. But it’s always more interesting to me to figure out how one launders money than simply to observe that it’s being laundered.