Whose Miracle?

The story everyone tells

On September 11, 1973, the Chilean armed forces overthrew Salvador Allende. The Allende years had ended in hyperinflation, rationing and a budget deficit of about 20% of GDP. Augusto Pinochet then ruled for seventeen years. His regime privatized state firms, opened trade, made the central bank independent and created private retirement accounts. It also held and tortured 38,254 political prisoners, according to the Valech Report.

Chile then grew faster than its neighbors, and many people concluded that the dictatorship produced the growth:

"…free market reforms instituted by the administration of General Augusto Pinochet from 1973 to 1989."Robert Barro (2000), on the source of Chile's performance

Behind this is a broader idea: a benevolent autocrat can commit credibly to good policy and deliver growth that a democracy cannot. Pinochet is the textbook test case. But a test needs a comparison. What would Chile have looked like without him?

A miracle depends on when you start counting

Chile's growth record looks very different depending on the window you pick. Choose a start and end year to see Chile's average annual growth in income per person. The second number is the same figure for synthetic Chile, the counterfactual built in Chapter III.

Start the clock in 1985 and Chile looks like a miracle. Gary Becker put Chile's growth in income per person from 1985 to 1996 at 5% a year, far above the regional average. Start it on the day of the coup and the regime spent its first fifteen years catching up to where Chile would plausibly have been anyway. Milton Friedman praised average growth of 8% in 1976–80, but that boom came right after a collapse, and the comparison countries were growing too.

Build your own Chile

No single country resembles pre-1973 Chile closely enough to serve as a twin. The synthetic control method (Abadie and Gardeazabal 2003; Abadie, Diamond and Hainmueller 2010) builds one. It takes a weighted average of donor countries, choosing the weights so that the blend tracks Chile's income, population growth, trade openness and schooling before the coup. If the blend matches Chile for thirteen years beforehand, its path after 1973 shows what Chile would have done without Pinochet.

Try it. Mix the donor countries and watch how closely your synthetic Chile follows the real one in 1960–72.

When the algorithm searched the donor pool (Latin American countries plus Spain and Portugal), it gave positive weight to only three:

Before 1973, the blend matches Chile closely: average income of $4,218 vs. $4,037 and population growth of 1.97% vs. 2.10%. Schooling is close as well (Table 3 of the paper).

Fifteen years below the line

Here is the result. The two lines move together until 1973 and then separate. Hover or tap the chart to read each year's values.

By 1975 Chilean income per person was about $1,100 below its counterfactual. The gap stayed near $1,000 through 1983, then narrowed. Chile finally passed synthetic Chile in 1988, fifteen years after the coup and the same year voters said "No" to eight more years of Pinochet.

The finding does not depend on one dataset. Switching to the Penn World Tables, extending the pre-coup period back to 1951, and adding copper producers such as the United States, Canada, Australia and South Africa to the donor pool produces the same pattern with a larger gap: more than $2,500 by 1975, reaching about $3,000 in later years.

Could it be noise?

Any synthetic control will miss by something. To check whether Chile's miss is unusual, the paper runs placebo tests. It pretends each donor country had its own "Pinochet" in 1973, builds a synthetic version of that country, and measures its gap. If Chile's gap is larger than nearly all the fake ones, the effect is unlikely to be chance.

The p-value for each year is the share of countries whose placebo gap is at least as large as Chile's. The smallest value reported, 0.06, means no placebo country had a gap as large as Chile's.

In 1973, 1975, 1976 and 1978 (p ≈ 0.06), no placebo country had a gap as large as Chile's. The losses during the first decade are clearly distinguishable from noise. From the mid-1980s on, the effect fades and is no longer significant. The regime is associated with a significant early loss and, at best, a later return to where Chile would have been anyway.

Two recessions and a promise

Two downturns, both on the regime's watch, account for most of the gap.

What about Allende? Chile was already falling in 1972–73, and the first year or two of decline belong partly to the previous government. But inherited chaos is hard to stretch into fifteen years of underperformance. The 1975 and 1984 reform booms show up in the synthetic control too, so they were not unique to Chile.

What actually changed in 1988

The paper ends with a hypothesis that follows North and Weingast's (1989) account of England after the Glorious Revolution. Chile may have taken off because of a credible commitment rather than the general. The 1980 constitution scheduled the 1988 plebiscite. When "No" won, Pinochet accepted the result, and before the 1989 election the regime and the opposition agreed on fifty-four constitutional amendments that expanded political rights and kept the economic reforms. A transition that locked in the rules may have mattered more than the man who wrote them.

The lesson

Chile's remarkable growth in 1985–1997 did not depend on Pinochet's autocracy.

The study does not separate every policy, and it does not claim the market reforms were irrelevant. Its claim is narrower and harder to argue with: against a careful counterfactual, the regime itself shows no positive effect on income per person. It shows a large early loss.

That fits a wider literature. Easterly and Pennings (2018) find little evidence that individual leaders drive growth. Carden and Lawson (2010) find that human rights abuses slow economic liberalization rather than speed it up. Studies of Chile's return to democracy find a larger growth effect. In the paper's words:

"If society can remove obstacles to the entrepreneurial spirit and unleash the market process, the results are similar most of the time. The presence of autocrats appears to have a secondary role."