> Munipalities don't do it because they have no money.
Why isn't it self-funding? Issue a bond, pay for the network, lease it wholesale to competing ISPs, use the lease money to pay back the bond over time. Zero tax dollars.
Why isn't it self-funding? Issue a bond, pay for the network, lease it wholesale to competing ISPs, use the lease money to pay back the bond over time. Zero tax dollars.
Then there is the politics. It's a political non-starter to issue city bonds to build a network you lease out to private ISPs to sell $70/month internet service that poor people and the elderly can't afford. So some sort of cross-subsidy has to be baked into the deal, and that involves charging higher-income customers more than $70/month. But that makes it harder to hit the 40% uptake rate. And if you can't hit that, you're really screwed because the uptake rate drives your per-household cost.
And then there is logistics. Who do you pay to maintain said fiber network? Cities have no experience or staff capable of doing it. So you have to create a new department, and staff it (at least in Baltimore) a relatively expensive unionized workforce (which makes it harder to hit your 40% operating margin above).
[1], [2] Roughly cribbed from the financials of EBP's network in Chattanooga.
[3] The FiOS network is only 10 years old, and Verizon is about to spend a bunch of money to upgrade from GPON to NGPON-2.
No on the economics, yes on the politics and again no on the logistics.
> Say you build out to the whole city of Baltimore, and get 40% to subscribe at a cost of $5,0000 per subscriber.[1]
How did you get that $5,000 number? It's far more than Verizon paid for FiOS and I can't get it to line up with EPB's numbers either.
> Say you charge $70 for service, and 60% of your revenue goes out the door in maintenance and customer service.
60% is far in excess of what maintenance and customer service costs. A brand new fiber network does not require significant maintenance and customer service is a small fraction of the monthly cost.
> 30-year bonds may be practical for a school building, but not so much for a telecom network where you'll need to make ongoing capital expenditures to upgrade the network.
30 year bonds work just fine for fiber networks. Upgrading active equipment is a minor effort, covered by your operating budget, unlike building the network which is CAPEX heavy.
> It's a political non-starter to issue city bonds to build a network you lease out to private ISPs to sell $70/month internet service that poor people and the elderly can't afford.
Yes, politics is a problem, but the above isn't what's stopping builds.
> And then there is logistics. Who do you pay to maintain said fiber network?
You outsource it like everybody else, unless it becomes cheaper to do it in-house. This is a solved problem.
The network cost $330 million,[1] and has 71,000 subscribers.[2]
> 60% is far in excess of what maintenance and customer service costs.
http://static.epb.com/annual-reports/2016/media/EPB_2016_Fin... ($122.6 million in fiber optic revenues, $78.5 million in fiber optic operating expenses). That is 64% of revenues going to operating expenses.
[1] https://www.washingtonpost.com/news/the-switch/wp/2013/09/17...
[2] http://www.timesfreepress.com/news/business/aroundregion/sto...
Both figures are inaccurate. The $330M includes $111.5M in federal grants which was used for EPB's smart grid. Thus only $220M can be attributed to the fiber network.
EPB fiber also had 83,000 subscribers at the end of FY2016, as per their annual report. 87,000 projected by end of FY2017.
Taken together this gives us $2650 per connected subscriber, almost half of what you claimed. Even with the smart grid grant, it would be less than $4000. Obviously this figure goes down with the number of subscribers.
> $122.6 million in fiber optic revenues, $78.5 million in fiber optic operating expenses). That is 64% of revenues going to operating expenses.
That's not correct. The annual report clearly states on page 11 that operating expenses were $41M, which is equal to 31% of revenues. Furthermore, it should be noted that operating expenses do not equal maintenance and customer service only.
Page 11 shows $41 million in "operation expenses" but there is also other line items for "cost of services" and administrative overhead. Those are "operating expenses" in the accounting sense. The cash flow statement on Page 19 shows that the total "operating expenses" is $78 million.
As to the smart grid project, that was a separate $232M project. https://www.smartgrid.gov/project/epb_smart_grid_project.htm...
Of those smart grid investments, only the Backhaul Communications part could conceivably (in part) be attributed to the fiber network buildout costs. In no way can 100% of the federal smart grid grant be seen as a subsidy to the fiber network. Even if you subscribe to conspiracy theories about the power side secretly subsidizing the fiber side, that 100% subsidy theory just does not survive the collision with reality. Given how litigated this issue has been, there's no way the opponents would let a hundred million dollar misuse of federal grant money slide.
Either way, no matter what dollar cost you peg the fiber network buildout costs at, your $5,000 cost per subscriber is plain wrong. The correct figure is closer to $2650, but it's still not even close to $5,000, even if you slap on an additional hundred million dollars in costs.
This obviously also makes your payback times wrong.
To give some context to EPB fiber's costs, ten years ago Verizon paid $850 per premise passed and an additional $880 per premise connected. These are in line with EPB's costs if you assume a 50% take rate for FiOS.
Just to rub it in on how wrong your estimate of $5,000 per subscriber is, the very source your cite yourself states that the average cost to connect a new subscriber to EPB fiber is $1,200.
But, wait, there's more!
Your analysis on the economics of fiber networks just crumble when you look at the annual report of EPB. Here's the punchline:
- operating expenses include debt service!
EPB fiber pays EPB $10M per year for the use of the fiber network. This is enough to cover the costs of the $220M bond. In fact it covers half of EPB's whole debt service. Incidentally EPB fiber's operating income would more than covers the remaining debt service with change to spare.
In other words your assumption of 60% in operating expenses excluding debt service is just plain wrong and all your calculations are wrong.
And that's even before factoring in the fact that service at $70 per month usually does not include cable tv. This in turn means that a large chunk of operating expenses for that service can be excluded, as cable tv rights are a major cost item.
All in all, even assuming debt service is included, 60% operating expenses for a fiber network is not a given.
Right, but sources say the $111.5M was used to build the fiber network that was a part of both the Smart Grid and broadband projects: "In 2009, a $111 million federal stimulus grant offered the opportunity to expedite construction of a long-planned fiber-optic network, said David Wade, chief operating officer for the power company. (EPB also had to borrow $219 million of the network’s $330 million cost.) Mr. Wade said it quickly became apparent that customers would be willing to pay for the one-gigabit connection offered over the network." (https://www.nytimes.com/2014/02/04/technology/fast-internet-...)
> Of those smart grid investments, only the Backhaul Communications part could conceivably (in part) be attributed to the fiber network buildout costs.
Yes, but that's the fiber backbone underlying the residential internet service.
> Either way, no matter what dollar cost you peg the fiber network buildout costs at, your $5,000 cost per subscriber is plain wrong.
Sure, with the newer 83,000 subscribers figue, it's more like $4,000 rather than $5,000.
> To give some context to EPB fiber's costs, ten years ago Verizon paid $850 per premise passed and an additional $880 per premise connected. These are in line with EPB's costs if you assume a 50% take rate for FiOS.
The FiOS uptake rate is under 40% (and during the first decade was under 1/3). Even with your numbers that's $850 / 0.4 + $880 = $3,000 per customer. And those numbers are 2006 projections at the onset of the project: https://www.sec.gov/Archives/edgar/data/732712/0001193125061.... Through 2010, Verizon spent $23 billion on FiOS, and since then spent at least another $3 billion wiring up New York. It has about 7 million subscribers, or about $3,700 per subscriber. See also https://techcrunch.com/2013/04/08/google-fiber-cost-estimate ("As a point of comparison, it was estimated that it cost Verizon, before it halted FiOS buildout, about $4,000 per home to connect it to its fiber network.").
> Just to rub it in on how wrong your estimate of $5,000 per subscriber is, the very source your cite yourself states that the average cost to connect a new subscriber to EPB fiber is $1,200.
It says that's the "Average cost of hooking up new fiber optic customer for EPB." I don't take that to be the cost including construction of the network in the first place.
> - operating expenses include debt service!
Does it?
"Fiber optics revenues are recognized on the accrual basis at the time services are provided. Operating revenues include service sales net of bad debt expense and miscellaneous revenue related to ber optics operations. This miscellaneous revenue includes ad revenue, late payment fees, and rental income. Operating expenses include those expenses that result from the ongoing operation of the ber optics systems. Non-operating expenses consist of interest expense on indebtedness and various miscellaneous expenses."
> EPB fiber pays EPB $10M per year for the use of the fiber network.
Is that for debt service? That could be for shared O&M too.
> And that's even before factoring in the fact that service at $70 per month usually does not include cable tv. This in turn means that a large chunk of operating expenses for that service can be excluded, as cable tv rights are a major cost item.
Television packages are also a major revenue source. In computing the operating margin, if you exclude TV from costs you also need to exclude TV from revenues.
True, but that still does not mean you can attribute 100% of the federal smart meter grant to the fiber buildout. Even if you chose to ignore all the other investments made into the smart metering infrastructure, fiber routes built to service the electric grid (such as routes to substations) cannot be used for providing Internet service. As such it is not reasonable to attribute 100% of the $111.5M grant to EPB fiber's network.
> Sure, with the newer 83,000 subscribers figue, it's more like $4,000 rather than $5,000.
This again assumes 100% of the federal grant being used to subsidize EPB fiber. I don't think that argument has any merit, as both the Feds and the incumbents would have words if that were the case.
> The FiOS uptake rate is under 40% (and during the first decade was under 1/3).
Yes, but we aren't talking about Verizon here. EPB's take rate is somewhere between 50% and 66% (83k subs out of 120k households, but part of those 83k are businesses), so my point still stands that costs are equivalent, given the actual take rate.
> It says that's the "Average cost of hooking up new fiber optic customer for EPB." I don't take that to be the cost including construction of the network in the first place.
No, but it includes the new construction needed to hook up customers in new developments.
> Is that for debt service? That could be for shared O&M too.
"$10 million: Fiber optic system access fees and rents paid to electric system in 2014"
> Television packages are also a major revenue source. In computing the operating margin, if you exclude TV from costs you also need to exclude TV from revenues.
Yes, but TV services are usually a loss leader and a break even proposition at best. Margins would probably improve by dropping TV.
"EPB issued $229 million in revenue bonds in 2008. Of this amount, $162 million was used to build the fiber optic network, which is owned by the EPB's Electric Division and is used for both the smart grid and telecommunications services."
"To date, EPB's Fiber Optic Division has borrowed approximately $50 million from the Electric Division to finance the costs of adapting the broadband network to provide telecommunications services to its customers."
For simplicity, let's ignore the fact that a part of these costs should be attributed to the power company, as the fiber network is also used for smart metering.
At a cost of $212M for 83 000 subscribers, this gives us a cost per subscriber of $2554.
If you add $28M to account for the first three years interest and other financing costs, it becomes $240M for 83 000 subscribers, resulting in $2892 per subscriber.
So there. A far cry from $5000 per subscriber.
https://www.cga.ct.gov/2012/rpt/2012-R-0515.htm
At 4% interest, 83 000 subscribers and attributing none of the costs of the fiber network to the smart meter project the payback period is 10 years and 7 months. A far attribution of costs to the smart meter project would bring the payback period to under 10 years.