Review of Kelsey et al.’s “Economic Impacts of Marcellus Shale in Pennsylvania: Employment and Income in 2009” by Matthew Rousu and Zach Zoller

Written By: Timothy W. Kelsey (Penn State), Martin Shields (Colorado State), James R. Ladlee (Penn State), and Melissa Ward (Penn State), in cooperation with Tracy L. Brundage (Penn College), Jeffrey F. Lorson (Penn College), Larry L. Michael (Penn College), and Thomas B. Murphy (Penn State)

Overview:

The authors use surveys of businesses to estimate the economic impact of Marcellus Shale in Pennsylvania in 2009 where they find an economic impact of $3.2 billion.

This study is one of the most transparent we’ve seen, and overall the methods seem quite solid. One thing different from Considine’s study is the authors spend a lot of time trying to determine how much of the wages go to Pennsylvanian workers vs. out-of-state workers, and how much of the out-of-state wages would be spent in Pennsylvania.

The authors do not attempt to estimate all the impacts from Marcellus shale, and they openly admit that they hope future research will address effects on environment and health, for example. We do have concerns that the economic multiplier doesn’t seem to differ in areas that are near the border of other states relative to areas where in the center of the state. Areas closer to the state will have more “leakage” (a term the authors use), meaning more money is leaked outside of Pennsylvania. This would cause a lower economic multiplier. Given that nothing on this is noted in the study, we have to assume the same multiplier was used throughout the state, which we think is incorrect.

If graded on our best practices – how well does this study do?

What the study does well:

2) The study clearly defines the geographic region of the study as it examines the entire state of Pennsylvania.

4) The authors only count expenditures once.

5) The study has been reviewed Kathryn J. Brasier (Penn State), Steven C. Deller (University of Wisconsin), David L. Kay (Cornell University), Thomas Knapp (Penn State), and Stephen Smith (Penn State).

6) The authors are very transparent with their methodology. When the make assumptions (e.g., on page 36 they discuss the percentage of money out-of-state workers spend in Pennsylvania), they usually state them and sometimes run alternative specifications.

7) The study does compare its results, specifically comparing its economic impact results to that of Considine et al. (2010).

8) The authors state that this study was funded by the Pennsylvania Economic Development Financing Authority.

What the study does not do well:

1) This study states the economic impact in terms of both jobs and monetary impact. This could confuse some readers into thinking, for example, that in 2009 Marcellus shale led to a $3.15 billion economic impact and expanded employment by 23,600 jobs.

3) I don’t think the authors clearly define the counterfactual. They want to estimate the economic gain to Pennsylvania. Because of that, they are cutting out some “leakage” by estimating how much of the wages that go to non-Pennsylvanians are being spent in the state. However, this is really the “middle road”.

If the authors wanted to estimate the economic impact on Pennsylvania now, they should have not worried about out-of-state workers as the multiplier should correctly account for this issue. This would result in a higher estimated impact. 

If instead the authors wanted to estimate the economic impact on those living in Pennsylvania prior to the boom, the wages paid to the out-of-state workers should be removed entirely, as those workers were not here prior to the boom. Further, any workers that wouldn’t be located in the state (even if they permanently moved here) shouldn’t be included in the analysis. This would result in a lower economic impact.  By including new (permanent) Pennsylvanians, it seems the authors take the middle road on this issue.

Overall

· This study is one of the better studies assessing the economic impact of Marcellus shale. Their methods are transparent, issues are well thought out and discussed, and the reader can get a clear sense of how the authors arrived at the economic impact. One could argue with assumptions made, as in any economic impact study, but they are so clearly laid out that others could then compute an alternative estimate for the economic impact.

· The estimated economic impact, being far less than that estimated in Considine et al. (2010), makes this study’s estimated impact seem more credible. This is especially true given the transparency of the methods.