6 Phone Calls Can Predict US Investment Without B________

Millions of dollars are being spent to persuade Ajit Pai FCC decisions on pole attachments, broadband paperwork, local regulations, neutrality and backhaul prices will “incent investments and more broadband deployment. I’ve listened to CTO’s, CFO’s and CEO’s explaining investment strategy for nearly two decades. I’m convinced none of his policy proposals will have significant impact on actual investment or deployment. Pai can determine whether I’m right by making six phone calls and asking one question on each. Just call the CEOs or CFOs of AT&T, Charter, Comcast, Sprint, T-Mobile, & Verizon. They control 90% of mobile and 65% of landlines. 

Ask them, “If I make these changes, how much will you increase your capital spending and rural deployment?” Verizon made a point of telling Wall Street that despite the 5G build, they wouldn’t increase capex in the coming years. Charter/Time Warner has said they will cut and I believe AT&T similar. Sprint has said they will not come close to their fiscal 2016 capex. Most of the remaining companies – Century, Frontier, Altice – are financially weaker and even less likely to invest.

Capital spending is mostly driven by technology change, competition, market demand, underlying financial problems, and equipment cost.

Some technology examples:

Verizon in 2008 realized they were about to fall behind AT&T in the 3.5G generation. AT&T’s GSM network could easily be upgraded to about 4x the speed/capacity. Verizon’s CDMA technology was falling behind. Jumping to 4G LTE was the best solution. They bit the bullet and built the world’s first large LTE network.  

Ambani at Reliance Jio realized the new LTE technology could give them an enormous edge over the other Indian networks. He invested $20B in building one of the best networks in the world and has 89M active customers in less than nine months.   

Competition can be a driving force:  

Vodafone and Bharti, to match Jio, have both raised their capital spending by about $6B.  

Similarly, AT&T raised capex from ~$18B to ~$22B. Verizon LTE was killing them and they had to catch up. When they got to 97% LTE coverage, they reduced capex. A year after AT&T told Wall Street about the reductions, the FCC made a surprise move on neutrality. Some blockheads in D.C. claimed the neutrality decision drove the capex drop. That would be true iff Randall Stephenson had a time machine and could jump ahead a year before he made the capex decision.

Verizon CEO Ivan Seidenberg told me ~2013 he was going to build Fios because, “We have to get cable out of the house.” They were losing too many customers. Two years later, folks in D.C. gave Verizon some relief from competition. 

Comcast CEO Brian Roberts explained why he moved quickly to DOCSIS 3.0. “I looked Ivan in the eyes when we shared a stage. I could see he was going to build Fios and we would have to do something. When I got back to the office, I told my people to do what we had to do to keep up.” What became DOCSIS 3 was then just some proposals from engineers like John Chapman. Suddenly, Comcast wanted it, funded CableLabs to advance the research, and pushed suppliers, like Chapman’s Cisco, to deliver the equipment ASAP.

Market Demand

AT&T was overwhelmed by the success of the iPhone and the network couldn’t handle it. Meeting that demand was the second reason they raised capex from ~$18B to ~$22B for a few years.

Underlying financial problems

Charter actually went broke a few years ago. Before and after the bankruptcy, they cut capex drastically. When Malone took over, they were able to again invest and soon were beating the DSL competition.

Deutsche Telekom has not made enough profit to cover its dividends the last few years. So they held off upgrading their DSL despite losing many customer to the cablecos that cover 2/4rds of Germany.

(Would an increase in profits mostly go to investment? Probably not. In most cases, including DT & AT&T, the companies have pledged they would use the money to increase dividends. That is the primary fallacy in “investment incentives,” a euphemism for higher profits. The telco lobbyists and those who believe them assume most of the increased profits would be reinvested. Some will be, but it’s usually just a small fraction. The policy goal is “investment,” not “investment incentives.” Unless you have reason to believe the “incentives” would actually be invested, you been bamboozled.)

Equipment cost

DOCSIS 3 upgrades cost as little as $20/home, so 80-90% of cable lines in the U.S. were upgraded within a few years. The increased speeds (from ~ 10 megabits down to ~50 megabits down) allowed that repaid the costs many times over. Europe wasn’t far behind. The economics were so favorable that most of   

(The $20 estimate comes from a NY TImes interview by Saul Hansell of  Liberty CTO Balan Nair. I think Balan’s cost was  lower than others because he had planned ahead, but even using my higher figures the upgrade cost was minimal.)

———————————

In D.C., even the smartest and best informed severely overestimate the regulatory impact on investment. Increasing telcos profits does give them some “incentive,” but I’ve never seen evidence it’s been more than minimal. For example, Michael Pai used a deep,y flawed study to claim that U.S. “lighter regulation” produced higher speeds than European “stronger regulation.” A smart scholar I think honorable made that claim based on the datum U.S. landlines were somewhat faster.

In fact, the U.S. networks would have been faster if the FCC Chairs were Democratic, Republican, or even trained monkeys. SInce ~1997 or so, the U.S. has had 95% cable coverage, built for video. Nearly all of that was upgraded to DOCSIS 3.0 by the time C. did the study a few years ago. Conversely, the major countries in Europe have much lower cable availability. Italy has almost none, France and Spain about 35%, England ~50% and Germany ~65%. Cable is so much faster than most DSLs, the difference in cable penetration explains much or all of performance difference. 

The study was wrong because it wasn’t adjusted for “confounding variables.” The end of AT&T’s LTE rollout and the near insolvency of Sprint two years ago had such a large effect any claim about NN was swamped.   

Anyone who doesn’t realize how few of the telecom policy analyses are worthwhile needs to read John Ioannides’ historic paper, Why Most Published Research Findings Are False. In 2005, Stanford Professor Ioannidis took a sample of papers from the leading medical journals and discovered about half of them fell apart because of sloppy statistics. Few believed him at first but his findings have been repeatedly confirmed. The policy papers are far worse than the medical ones, many of which included a double-blind control group and much larger samples. Ignoring confounding variables was one of the common problems.  

Read the paper closely if you haven’t. I bet if the papers submitted to the FCC or presented at TPRC were similarly examined, 90% would be found inadequate.  

B______ is nearly universal in the statistics presented by both sides in telecom policy.

—————

Note: No new company has gained even a 2% market share in the last 15 years. I strongly support community networks and small entrepreneurs, but it’s highly unlikely any will have enough coverage to make a difference in the next decade. If we want to have an impact on the service and costs of 90%, policy needs to influence these very few dominating companies.

Nobel Laureate Gary Becker, in a 2009 Verizon-funded paper, pointed out there were circumstances where modest competition or even the threat of competition could have an impact. He drew on the work of University of Chicago colleagues George Stigler and Milton Friedman, who also won Nobels. Since essentially none of the smaller companies have grown to significant size in fifteen years, that shouldn’t apply here. Becker suggested Clearwire and Open Range as possible factors. Both failed miserably in the interim. 

From Leichtman, recent broadband totals. They all have different amounts of video, wireless, etc. but broadband figures are a fairly robust proxy