Does Appalachia have a coal curse?

Natural resources are generally thought of as beneficial for economic development. Yet, many of the regions with abundant resources do not perform well economically, that is, there seems to be a resource curse. This might be driven by disincentives to education, or crowding out other sectors and making the economy depend too much on the boom/bust cycle of resources.

This post investigates whether there is a coal curse present in Appalachia. Could it be that the per capita income of coal-producing counties in Appalachia is lower because of a resource curse?

Douglas and Walker (2014) examine this question in a recent working paper. They take a look at 409 Appalachian counties, roughly half of which are coal-producing (as we all know, coal is to Appalachia as corn is to the Midwest). Comparing Appalachian counties within themselves ensures that the sample is as culturally homogeneous as possible. The period covered is relatively long, it goes from 1970 to 2010.

The first thing we can see is that the income per capita of coal-producing counties tends to get closer to that of non-coal counties when coal prices are on the rise. In periods with decreasing prices, the difference once again widens. One can clearly see this happening in the boom of the 70s, then the bust of the 80s and early 90s, and then again in the boom of the 2000s.

Coal prices and growth in Appalachia

So clearly, coal county income per capita fluctuates with the coal boom/bust cycle. But what are the net effects? Do the positive effects of coal booms outweigh the negative effects of coal busts in the long run? This is the main question of the paper.

The authors examine the effects of resource abundance (as measured by coal revenue as a percentage of total personal income net of transfers in a given county) on long-term (decade-long) average annual growth rates in personal income per capita (net of transfers). Several control variables are also used, the most important of which include the initial level of per capita income (i.e. at the beginning of the decade), whether the county is rural or part of a metro area and energy prices.

There is a double causality problem with this model, as resource abundance could not only affect growth, but growth can affect the extent of resource extraction as well. For this reason, the authors instrument for resource abundance using a dummy indicating whether the county ever produced coal (correlated with coal revenues but not with current growth rate), and the percentage of slaves in the county in 1860 (negatively correlated with coal production as there were no/few plantations in coal-rich counties, but uncorrelated with current growth).

The main findings are that resource abundance does have a significant negative effect on growth rates. In particular, a one standard deviation (0.5) increase in coal revenue as a % of total personal income results in a 0.7 percentage point drop in the annual average growth rate. This can accumulate over the years: for instance, over the course of 40 years, this can mean a 32% difference in personal income. These results are robust to using nine alternative measures of resource abundance.

So, we’ve established that coal hurts growth in Appalachia. But through what channels does this happen? Douglas and Walker (2014) test whether disincentives to education are to blame. The idea is that coal mining creates these disincentives (because jobs in the coal industry rarely require college or even high school education), and low educational attainment then negatively affects growth.

This question is answered by estimating two equations. First, trying to explain growth using the same variables as above plus educational attainment. Once again, to avoid double causality between education and growth, the authors use an instrument: in this case it is the percentage of people employed in educational services in a county. This is correlated with educational attainment because the proximity of educational institutions stimulates human capital formation. But it is not directly correlated with growth. The second equation of the model then tries to explain educational attainment as a function of resource abundance and various control variables.

Estimating these two equations yields the following results. First, education positively affects growth as expected. Second, resource abundance has a negative effect on educational attainment. More specifically, a one standard deviation (0.5) increase in coal revenues as a percentage of total personal income increases the number of high school drop-outs by 3 percentage points, and decreases the number of college educated people by 2.2 percentage points. Third, putting the results from the two equations together, we can estimate that a one standard deviation increase in resource abundance decreases annual growth by 0.18 percentage points via increasing the number of high school drop-outs, and by 0.09 percentage points via decreasing the number of college graduates.

Depending on to what extent these two measures (high school drop-outs and college graduates) are correlated, this can mean that overall higher resource abundance decreases annual growth by anywhere between 0.09 and 0.27 (=0.09+0.18) percentage points through educational disincentives. This is roughly between 13% and 39% of the total effect of resource abundance on growth.

Therefore, high reliance on coal decreases growth to a significant degree by providing disincentives to education. Nevertheless, disincentives to education do not explain the majority of the growth-retarding effects of coal in Appalachia. The coal curse must act through other channels as well.

These other channels may include poor (corrupt) institutions, or coal crowding out other industries leading to low economic diversification, which then makes the counties susceptible to fluctuations in the coal industry.

It must be mentioned that this whole story does not necessarily need to apply to other regions, time periods or resources. The authors for instance quote a paper that finds evidence that the oil industry had positive growth effects in the South Central United States.

So to sum up, it does appear that Appalachia has a coal curse. Coal-producing counties do have lower growth rates on average. It also seems to be the case that this disadvantage of coal counties is partially driven by the coal industry creating disincentives to education. About 15% to 40% of the coal curse can explained this way.

Advertisements

One thought on “Does Appalachia have a coal curse?

  1. Pingback: ZeeConomics | The Dutch disease and the distribution of resource revenues

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s