Are Enbridge’s job numbers credible?

Putting aside the impact of the proposed Enbridge pipeline on GHG emissions or spills on land and at sea, the case in favour of the pipeline rests on creating jobs. Personally, I think industry and government use “jobs” as a euphemism for “profits” as that is where the lion’s share of revenues go. But for the moment let’s consider the possibility that trading the environment for the economy is a bet we are willing to take. So how many jobs are we talking here?

The pipeline is a $5 billion investment, according to Enbridge (technically $5.5 billion but $500 million is a contingency reserve). The number thrown around by pipeline proponents is 63,000 person-years of employment, which sounds pretty good (National Energy Board document here). Careless journalists and cynical politicians sometimes blur a person-year of employment with a job, in which case I’ve been at my current job for 13 person-years.

In terms of hard numbers, though, Enbridge figures there will be an average of 1,850 workers building the pipeline over the three-year construction period, peaking at 3,029 in the third quarter of 2015. They note that all of these jobs may not be filled locally, and in construction of large projects workers typically move around to where the action is. It is also worth noting that these workers would not all be unemployed if the pipeline did not go a head; the vast majority would likely be working somewhere else. There is some case to be made that these jobs would reduce unemployment based on the current stagnation in the economy, though by 2014-15 this may no longer be the case, and we may have excess demand for skilled construction workers.

So how do we get from an average of 1,850 workers for three-years to 63,000 person-years of employment (construction only)? To answer this question we have to understand input-output models, which use GDP data to proxy the flow of income through the economy. Modellers “shock” the I-O model to estimate an increase in economic activity. The important pieces are (a) that direct expenditures on the pipeline also lead to employment in upstream industries that provide the goods and services that are inputs to construction and operations (called “indirect employment”); and (b) income to workers, whether direct or indirect, support jobs in the local economy on food, housing, cars, entertainment and so forth (called “induced employment”).

Neither of these impacts is particularly controversial, and the early description of such in the application to the NEB is consistent with normal practice. But when the numbers are presented a massive over-estimate of job creation results. The main juice for indirect jobs comes from sourcing the manufactured pipe domestically through Canadian steel mills – this is assumed to happen, and should if the project goes ahead, but I have yet to see a firm commitment that this will be the case. Anyway, the purported impact of purchasing inputs is misleadingly labelled “direct input purchases” and is counted as “direct employment” rather than indirect employment. This employment from “direct input purchases” is estimated to about three times the person-years as “on-site employment” (17,227 vs 5,537 person-years), which seems overstated to me given how labour-intensive construction is.

The problem comes in what is normally considered “induced employment” but is labelled as “indirect employment” leading to 13,251 person-years of employment (figures in Table 4-9). This can be seen in Table 4-10, which purports to show direct and indirect employment effects by industry category, but clearly includes induced employment in the form of, for example, 137 person-years of arts, entertainment and recreation employment, and 401 person-years in information and cultural industries, 95 in crop and animal production, and 423 person-years in government (as if the latter is going to happen federally, in BC or Alberta where public service cuts have been on the order books!).

Finally, a second run of the model is undertaken using all of this new employment income as the shock to come up with what they are calling induced employment, adding 26,679 person-years of employment. This leads to a stated estimate of 39,930 person-years of “spin-off employment” by adding 26,679 and 13,251. The total number represents 64% of the claimed employment gain of project construction. Such a number is truly incredible as it suggests that for every job created in actually building the pipeline and its inputs leads to two spin-off jobs; whereas in typical I-O models you get one induced job for every new direct or indirect job.

One problem with the induced numbers is, as stated above, we cannot assume that those workers would have been unemployed, so some portion of the income driving induced employment may be defensible but not all. Also it is assumed that tax revenues going to governments get spent on public services, which a lovely thought but pretty far from reality federally, Alberta or BC. So the induced numbers should be heavily discounted, because they are as if 36,000 new skilled workers just showed up out of nowhere, earning $68,000 per year, and paying taxes to governments that were interested in supporting public services.

Also, those numbers are just the construction phase. On a permanent basis, the number of jobs is quite small, estimated by Enbridge at 217 workers on the pipeline and terminal facilities. The vast majority of employment for the project comes during the construction phase, which is consistent with an industry that is one of the most highly capital intensive. They then estimate that the 217 permanent jobs increase five-fold to 1,146 total jobs per year in Table 4-14, seemingly with the same estimation issues as with construction. Total jobs are likely to be about half of what Enbridge claims, based on standard I-O multipliers for pipeline transportation.

The danger in all of this is that a number gets established, and then becomes pasted into various summaries, briefing notes and so forth, and few people go and check the math, and any assumptions and caveats associated with the number are left out of the summary.

[AFTERNOON UPDATE] OK, I had a call with Statscan and they are not able to release the study they did for Enbridge to me, so I will have to wait on my request from Enbridge. I was able to talk through multipliers for direct, indirect and induced broadly for the Oil and Gas Construction Industry sector, that provide a rough estimate. They lead to about 3,000 fewer direct, 11,000 more indirect, offsetting 11,000 fewer induced person-years. The same caveats about induced jobs would still apply, though, even with the lower number.

Still, while the total jobs is about 3,000 person-years fewer than the Enbridge, they are in the same ballpark. So I am scratching my head a bit, in particular as it relates to direct jobs and how all of those reported indirect jobs could include such large numbers of in far-flung industry categories. There is some kind of flaw in how this is being modelled but without deeper information I cannot get at it. It could be that Oil and Gas Construction Industry [code 2300D0] in the I-O Model is broader than pipeline building (in the NAICS, 23712, Oil and Gas Pipeline and Related Structures Construction).