13 August 2012

Muskrat Falls Cost Estimates: the Skinner Numbers #nlpoli

Former natural resources minister Shawn Skinner said this past weekend that he expected the next cost estimate for Muskrat Falls will be around $8.0 to $8.5 billion. [video; Skinner comments are at about 14:00]

Assuming that is for the dam, line to St. John’s, and the line to Nova Scotia, Skinner’s estimate would mean that Nalcor’s cost estimate in 2010 was between 29% and 37% out.

Sadly for proponents of the Muskrat Falls megaproject, those cost increases won’t be the end of it.

Skinner’s Projections

In his comments on CBC’s On Point, Skinner notes that just this past week that he has heard estimates of up to $9 and $10 billion for the revised project cost.  He dismisses those numbers but doesn’t say why.  Then he settles on $8.0 to $8.5 billion but doesn’t explain why he believes those numbers are more likely to turn up.

The original estimate for the Muskrat Falls project as announced in 2010 was $6.2 billion.  The project included a dam, a line to Soldier’s Pond (the infeed), and a line to Nova Scotia.  According to information presented at the public utilities board hearings, Nalcor approved the project at Decision Gate 2 in the expectation that the figures could be as much as 50% under what they would turn out to be.

Skinner’s low-end estimate of DG 3 numbers ($8.0 billion) would be 29% more than the DG 2 numbers.  His second number ($8.5 billion) is 37% above the estimate. Those are within the range of accuracy at DG 2. Skinner also mentioned $9.0 billion and $10 billion.  Those would be respectively 45% (still within the DG 2 accuracy range) and 61% more than the DG 2 estimate.  That would be 11% over the upper end of the accuracy range deemed acceptable at DG 2 by Nalcor.

While Skinner dismisses those figures without explanation,  research by Brent Flyvberg and others into megaprojects and cost over-runs indicates that cost over-runs of 50% are common on megaprojects.  A cost over-run of 70% wouldn’t be unusual. 

Flyvberg and his colleagues also found that megaproject supporters – like Skinner – tend to underestimate project costs for a variety of reasons.  Knowing Skinner’s source for that $10 billion estimate would be useful but, as we’ll see below, the number is within the range we should be expecting. And, don’t forget, it is within the range that Nalcor and the provincial government accepted in November 2010.

Comparing Different Estimates

The following table shows three project configurations, the base costs for each and the possible increases.  

The first row is the project that Skinner was apparently talking about, namely the original Muskrat Falls announcement.  The second row is the Manitoba Hydro International description from the public utilities board review. It includes Nalcor’s plan to add more thermal generation over the lifespan of the project than the island current has at Holyrood.  MHI pegged the cost of the dam, infeed line to the island and the extra thermal at $6.6 billion. The third row covers the Muskrat Falls proposal the Muskrat Falls proposal, plus the additional thermal generation identified by MHI as well as $1.1 billion in interest costs that some suggested should be included.  More information on these is available in an earlier post.

As for the percentage increases, the first two represent the percentage over-runs Shawn Skinner talked about in his interview.  The third is the 50% suggested by Flyvberg’s research.

Table 1





+ 50%


November 2010

Dam, island and Nova Scotia lines



$6.2 billion


$8.0 billion


$8.5 billion


$9.3 billion


Nalcor plans, to 2067 (MHI)

Dam, island line, additional thermal



$6.6 billion


$8.5 billion


$9.042 billion


$9.9 billion


The Whole Enchilada

Dam, island line, Nova Scotia line, planned thermal, plus interest


$8.9 billion

$11.25 billion

$12.2 billion

$13.35 billion

While those other two configurations are potentially useful for later discussion, let’s focus on the project as announced in 2010.  Let us also use project supporter Skinner as being a useful guide to what we can expect. 

After all, he was a cabinet minister at the time the project passed through Decision Gate 2.  He was briefed subsequently on the project in detail when he took over as natural resources minister.  Skinner knows the project better than most commentators outside government. Given all that, Skinner is prepared to accept that those initial numbers (DG 2) were off by a substantial amount.  

Skinner’s current projection is well within the likely range of cost estimates based on the DG 2 range of accuracy, even if his preference for $8.0 billion is optimistic. But still, it has taken a project that was touted as costing $6.2 billion and increased the projected cost by 30% or more.

As a crude comparison, we can check Skinner’s figures against the 1998 estimates of the dam and infeed costs to see where they fit.

1998 Plus Inflation as a Check 

There are some differences between the design of the Muskrat Falls dam and infeed line considered in 1998 and proposed in August 2010.  Note that date, incidentally.  It shows the very short time frame between the date Nalcor first looked at feeding Muskrat Falls power to the island (August 2010) and when Nalcor and the provincial government passed the project through DG 2 and signed a term sheet with Emera (November 2010).

Still, for whatever technical differences  between the 1998 and 2010 concepts, the November 2010 numbers seemed low compared to the 1998 estimates.  In 1998, Nalcor’s predecessor  - Newfoundland and Labrador Hydro - forecast that the dam and infeed would each cost $2.1 billion for a total of $4.2 billion.  Nalcor’s 2010 forecast  was $2.9 billion for the dam and $2.1 billion for the infeed, for a total of $5.0 billion.

Using Inflation alone,  we would have expected that the $4.2 billion dam and infeed projection in 1998 to reach $5.34 billion in 2010.  That’s about eight percent above the Nalcor 2010 number. Eight percent might not seem like much when the figures are this large but just look at it:  that eight percent presents $340 million.

The table below shows what you could expect if you took the 1998 forecast and allowed for inflation plus the same projected cost over-runs from the table above.  The second line takes the Nova Scotia line forecast  from 2010 and adds the anticipated over-run percentages from Skinner’s interview plus the “typical” over-run.  The third line totals each estimate.

What you wind up with in Table 2 is another way of estimating the potential cost of the project described in 2010.  it serves as a useful check of the other comments you will see and hear in the weeks ahead.

Table 2


2010 (inflat)

2010 +29%

2010 +37%

2010 +50%

Dam and infeed

$4.2 billion

$5.34 billion

$6.88 billion

$7.32 billion

$8.0 billion

NS Line


$1.2 billion

$1.55 billion

$1.64 billion

$1.8 billion


$4.2 billion

$6.54 billion

$8.43 billion

$8.96 billion

$9.8 billion

All things considered, it is interesting that extending the 1998 figures gives you costs that are consistent with the new project despite apparent technical differences. Skinner’s projections of $8.0 billion would seem to be light by quite a bit. The low-end of his cost projections (an additional 29%) would give you about $8.5 billion for the whole project. As others have  said, megaproject supporters tend to underestimate costs.

A more “typical” expected over-run of 50% for a megaproject of this type would put the cost of the whole project at about $10 billion. 

According to this table, the higher number Skinner was personally expecting ($8.5 billion) actually winds up being in the middle of the range of outcomes from the above assessment. It seems like a reasonable number using this “back of the envelope” assessment.

However, Skinner’s rejection of the higher numbers isn’t warranted.  Recall that according to information presented at the public utilities board review, Nalcor will allow their Decision Gate 3 number to be as much as 30% below the final result.   Based on that range of accuracy,  we could see this project finish at $9.8 billion, and we’d actually have a better result than Nalcor accepted at DG3. Using Nalcor’s own criteria, they would sanction this project with a DG 3 estimate of about $8.5 billion and allow that the final cost could be $11 billion.

The Political Implication of Cost Increases 

Earlier this year, economist Wade Locke said that if the cost of the project costs climbed to $8.0 billion or more at this stage, then Nalcor and the provincial government should rethink their plan.

And once you have thought about that for a second, consider something else Shawn Skinner said this past weekend when he mentioned the $9 and $10 billion figure: “I don't know if the government would move forward if it were to go to that kind of a number."

Given the numbers Shawn talked about, he’s just as easily talking about the nine and 10 billion he doesn’t want to think about.  They are  within the range of possible outcomes the planners would apparently accept based on Shawn’s own expected cost estimates.


In short:

  • All things considered, no one should be surprised if Nalcor shows up with revised cost estimate of $8.5 billion at this stage.
  • Given Nalcor’s track record of estimating on this project, the range of accuracy the company has already set for DG3, experience in other provincial public sector projects, and international research on megaprojects, we should reasonably expect that this project would cost $10 billion or more no matter what Nalcor brings forward as a Decision gate 3 estimated project cost.