Liberalization Increases Growth

In Assessing Economic Liberalization Episodes: A Synthetic Control Approach (wp version), Billmeier and Nannicni evaluate economic liberalizations, defined as “comprehensive reforms that extend the scope of the market, and in particular of international markets,” over the period 1963-2005. The authors compare real per-capita GDP in treatment countries with that in a synthetic control, a weighted average of similar countries with the weights optimized so that the synthetic control matches as well as possible a variety of pre-treatment characteristics including secondary school enrollment, population growth, and the investment share as well as GDP. The results are often impressive, for example.

[In Indonesia] the average income over the years before liberalization is literally identical to that of the synthetic control, which consists of Bangladesh (41%), India
(23%), Nepal (23%), and Papua New Guinea (13%). After the economic liberalization in 1970, however, Indonesian GDP per capita takes off and is 40% higher than the estimated counterfactual after only five years and 76% higher after ten years.

Here is a selection of figures (see the paper for more). In each case the solid line is the liberalizing country and the dashed line the synthetic control.


To be sure, not all liberalizations are successful. In particular, liberalizations in Africa especially after 1991 appear to be less successful. Whether this is because these liberalizations were half-hearted, were not combined with other institutional improvements or because countries that liberalized earlier were better able to link to global supply chains, is unclear. See the paper for more discussion.

These results bolster similar earlier findings from Warciarg and Welch (2008) who, using different methods, examined the average effect of liberalizations on growth from a wide variety of countries over a 50 year period. They found that on average growth increased after liberalization by a remarkable 1.5 percentage points. Here is the key figure.


See Development and Trade: The Empirical Evidence, a video from MRU, for more on the WW study as well as other types of evidence.


Every country can be explained by special factors aside from liberalization, which of course is a necessary condition.

Re S. Korea: US aid, exports to USA; re Indonesia: at a loss to explain, possibly it's a low base so anything would work; re Chile: Dr. Copper at work, note their economy rises and falls according to copper; re Botswana: diamonds; re The Gambia - almost the same as the control seems to me, not much change; re Mexico-- debt forgiveness and NAFTA helped.

're Indonesia:'

Oil - Indonesia joined OPEC in 1962 (what a wonderful coincidence to pick 1963 to 2005, Indonesia leaving OPEC in 2009, after becoming a net oil importer in 2004).

But some people find oil to be far too simplistic an explanation for economic, much less world affairs. After all, it is just a mystery why the Japanese invaded the Dutch East Indies -

'Preference in obtaining concession had been given to Dutch firms and overwhelmingly to Royal Dutch in an attempt to continue the Dutch monopoly over all country’s resources. However, US companies had also been interested in obtaining concession. Finally, after threats of counter action against Dutch companies in the United States, restriction were relaxed. Thus, Caltex and Stanvac were firmly established in Indonesia as producers and refiners well before the Second World War. In 1939, production of crude oil had reached 170,000 barrels per day and total refining capacity is approximately 180,000 barrels per day. Indonesia then dominated the oil industry in the Far East where over 75% of the crude oil produced came from Indonesian wells.'

And note how the U.S. so consistently presses for liberalizing oil markets world wide. Unless the market is the U.S., the company is Unocal, and the intended buyer is Chinese, that is.

Wouldn't exports to USA fall under trade when discussing trade liberalization?

The question is, as always, liberalization of what markets. Especially relevant now is financial liberalization--and let's be honest here, the empirical evidence is wildly mixed (at best). There are also real concerns with trade liberalization in some markets, and national aggregation might miss the differential effect on different segments of the population in liberalizing countries.

But you've heard all that before.

Yes, having the government pick and choose who to finance is best. You seem smart.

I wonder how much of it is also the fall of birth rates and societies with falling birth rates can better afford lower wages and fitting into the global supply chain.

I would think there's a problem in that we identify specific countries & dates of liberalization already knowing the outcome of them. Still surprising to see Chile doing worse than the synthetic control, since its known as a successful reformer. I guess most of its growth occurred after the years examined..

It'd be more interesting to study why liberalization didn't work in Africa than to pick a country where it did work to make a general statement. And by "work", why is GDP growth the measure to study? Wouldn't it be more interesting to measure the impact on the HDI or GINI coefficient or both, or more? If I'm a dictator ( ) setting the path for my nation with the aim to turn my people into 1st world citizens in, say 30 years, would I pick liberalization? Why?

He didn't cherry pick a success story and make a generalization. See the 2nd figure. It's surprising, though, that Chile is the only one shown to have performed worse than its 'synthetic control'. Where's the so called economic miracle?

Chile never liberalized its copper industry, even under Pinochet, providing an easy explanation for those with a nagging feeling of confirmation bias -

'Codelco's history begins with Law 11,828, of May 5, 1955, that created the Copper Office (Departamento del Cobre) of the Chilean government, approved under President Carlos Ibáñez del Campo. During the administration of President Eduardo Frei Montalva, Congress sanctioned Law 16,425, on January 25, 1966, and transformed the Copper Office into the Copper Corporation of Chile (Codelco).

With the constitutional reform that nationalized copper (Law 17,450 July 11, 1971), during President Salvador Allende's government, full ownership of all copper mines and copper fields in the country were transferred to Codelco. The creation of the Corporación Nacional del Cobre de Chile, as it is currently known, was formalized by decree of April 1, 1976, under the Augusto Pinochet dictatorship. A study by Goldman Sachs of January 2006 estimated the current value of the company between US $24.5 and $27.5 billion.'

And for those unfamiliar with how Chile decided its sovereignty was more important than non-Chilean company profits, there is always this link -

I think the graph from Chile has an important fact - the gap between actual performance (the solid line) and the control happened before the 1976 liberalization. Chile's collapse from the control prior to 1976 is probably because of the disastrous policies of the Communist Salvador Allende. Presumably this change is not reflected on the control line (what exactly consittutes the control is an important question here). Pinochet did not originally push liberalization until some time after he overthrew Allende. The economic miracle comes from the sharp recovery after 1976. If the Chilean graph was extended to past 1985, we'd see much stronger imrpovement from the baseline especially after 1990.

Not having read the paper, I can't say for sure my analysis is correct, but it makes sense when you combined history and the graph.

The biggest source of error in libertarian economics, and a substantial source for the discipline as a whole, is over-reductionism. Mathematical models treat every human and every unit of GDP the same when we know that they exist in very different qualitative conditions.

For example, before Western colonization, many parts of Africa hadn't yet invented the wheel. But libertarians believe that we can give them cars, roads, and electricity, and then if they just cut taxes they will become Germans. And we are surprised when they do not.

Why oh why can't we find one ahistorical simplistic economic model to describe all of human kind? This is the question that the data screams at libertarian economists. Perhaps if we keep adding mathematical terms we can overfit a model that will give us a moment of peace.

Spoken like someone who has never read libertarian economists. Can you even point to one who has told Africans that the solution to all their problems is tax cuts?

Also, to be puckish, I'll just leave this here:

I haven't looked at the full paper, but I don't see how you look at that collection of graphs as anything other than a mixed bag. Only 2 of 6 show a clear benefit to liberalization. Botswana makes it look like liberalization is a symptom, not a cause.

My prior is still quite pro-liberalization, but I get the nagging confirmation bias feeling looking at those charts.

This is a site where a nagging confirmation bias feeling is considered a failure - mainly because this place is all about ensuring certain biases are always confirmed.

Synthetic controls produces excessively confident graphs. Matching and weighting is still matching and weighting and thinking that because you have matched trends in observables that you have also matched trends in unobservables is wrong, wrong, wrong.

Isn't China the poster child for economic liberalization? Once they allowed market reforms under Deng Xioping the big takeoff started. Conversely, the US began the "Great Stagnation" at the same time as market reforms began under Carter. Curious isn't it?

paper is gated -- how many actual controls did they run?they should apply the same synthetic control methodology to some actual countries that didn't liberalize but that had significant changes for better or for worse in their growth rates. it would also be interesting to see this synthetic control get constructed five years before the liberalization event or five years after.

I have a nagging suspicion that a synthetic control applied to any country which exceeded its previous growth trend, with the synthetic control's fitting window chosen to match a policy change, would demonstrate that the policy change was responsible for the improved growth.

second link is ungated.

Apart from the obvious question of what constitutes utility, economic growth per se, or some imore "equal" distribution of goods, serivces, and power, I am reminded of a comment from a British consultant I met in Botswana in 2003: he averred that there was little or no corruption in that peacable, albeit HIV-infested, kingdom because civil service wages were on a par with those in London. In short, distributing the benefits of economic growth wisely, at least among the civil service, is a nett positive.

I don't know that one factor can explain a whole society. If pay were enough to stop corruption, politicians and CEOs would be the least corrupt individuals.

Good comments here, we could organize a whole symposium over this topic, and then some. I will post a quote from the excellent book by David S. Landes, "The Wealth and Poverty of Nations", p. 266: "The fact remains that history's strongest advocates of free trade - Victorian Britain, post-Would War II United States-- were strongly protectionist during their own growing stage. Don't do as I did; do as I can afford to do now." And BTW Landes is a free-marketeer. It's the classic "Infant industries" argument, coupled with the pioneering work done by Paul Krugman and others that shows network effects from (often random) initial conditions create long term growth. "Build it (initially with subsidies) and they will come (and it will prosper, if free trade exists)".

nee nee...
nee nee nee nee...
nee nee nee nee nee nee nee nee nee nee nee OMITTED VARIABLE BIAS!

Here is a useful discussion of evaluating regulatory reforms using synthetic controls:

obviously like your website but you have to check the spelling on quite a few of your posts.
A number of them are rife with spelling problems and I in finding
it very bothersome to tell the reality on the other hand I will definitely come again again.

Greetings! I know this is somewhat off topic but I was wondering if
you knew where I could locate a captcha plugin for my comment form?
I'm using the same blog platform as yours and I'm having difficulty finding one?
Thanks a lot!

Comments for this post are closed