As former British Prime Minister Benjamin Disraeli famously said, “There are three kinds of lies: lies, damned lies, and statistics.” Statistics are particularly concerning when they’re taken out of context or used to prop up policies which the body of the larger economic analysis would not support. When it comes to the Commission’s analysis of the Business Data Services market, basing policies on a handful of outlier statistical results while ignoring the weight of the evidence and the data’s deep-seated flaws would result in a disastrous outcome for broadband investment.
Over the past several months, the FCC – first through its hired economist, and later through Staff – has released over 100 regressions that purport to analyze the data the Commission has collected about the Business Data Services market. Each time, the FCC announced that the regressions show that ILECs retain market power for legacy DS1 and DS3 services. Each time, economists, including those the FCC asked to conduct peer reviews of the FCC regressions, observed that the regressions suffer from significant flaws that render them unreliable, including the severe correlation/causation problem that economists refer to as “endogeneity,” incomplete and incorrect data on pricing and the number of competitors, mismatches in the pricing and competitor data, and incorrect methods for computing the statistical significance of the results. And each time, we noted that some of the most significant of these flaws are not fixable because of the limitations of the data available to the FCC’s economists.
Undeterred, FCC Staff released yet another set of regressions in late August. These regressions address one of the problems with the earlier regressions – the flawed method for determining statistical significance. As a result of that correction, many of the prior results purportedly demonstrating the presence of market power in legacy DS1s and DS3s fell by the wayside. Those results now show nothing of the sort. But, no less important, as explained in a white paper filed last week by Drs. Israel, Rubinfeld, and Woroch, the revised regressions do not address the core, foundational flaws that continue to plague the entire exercise. Because of these deep-seated flaws, the regressions continue to produce wildly inconsistent and often anomalous results that in many cases conflict with basic economics. These erratic results confirm that the endogeneity and other data-related flaws are dominating the regressions, and thus they cannot be used to support any conclusions about ILEC market power for DS1s and DS3s.
That the results are unreliable is evident on their face. How else to explain a finding that ILECs lower their prices in response to competition in areas where they lack the pricing flexibility to do so? Or that they lower their prices when one or four competitors enter the market but not when two or three competitors enter the market? With anomalous results like these, the only explanation is that the underlying data are so flawed that they’re driving irrational results. The Commission Staff has, in fact, recognized that some of the regression results are “difficult to believe,” and they have dismissed a limited number of those results. But the anomalies are far more pervasive than Staff acknowledges.
Moreover, in purporting to find support for the Commission’s re-regulation agenda, Staff leans on the handful of results that would buoy that outcome, while ignoring or dismissing those that don’t. But the problem is this: most of the results, including the ones from the more rigorous regressions, don’t support a finding of market power. The majority show no statistically significant price effects from additional competition. Thus, even if one were to ignore the flaws in the regressions and accept their results at face value, they still don’t get the Commission where it evidently wants to go.
For example, the latest round of analyses includes six sets of regressions for DS3s. Four of the six – including the most rigorously tailored of the regressions – do not find evidence of market power. They are flatly inconsistent with the two results that do. How can the minority of results trump the majority, particularly when the majority were the result of a more rigorous regressions? And the data similarly suggest no market power in DS1s. In fact, the few regressions the Commission Staff points to show price effects of only 3-4% for DS1s – hardly evidence of a problem in a need of a regulatory solution. (And it should be noted that analysis of the Ethernet market produced no regressions that would support a finding of market power for Ethernet at any speed.)
Notably, none of the regressions conducted in Phase I pricing flexibility areas, where price caps still constrain rates, show evidence of market power in the pricing of any service. If the Commission is going to rely on these regressions, therefore, it has no choice but to conclude that price cap regulation must be driving prices to competitive levels; otherwise, one would expect to see price reductions in response to competition in these areas.
Of course, the economic testimony in the record already debunked arguments for a reset of price cap indices and an adjustment of the productivity factor going forward. But if the Commission is going to rely on these flawed regressions as evidence of anything, it cannot have it both ways: it cannot credit the regressions for Phase II areas while discarding the very same regressions for Phase I areas.
Without reliable evidence of significant market power, there is simply no data-driven basis for new heavy-handed rate regulation of BDS services. Chairman Wheeler astutely declared soon after joining the FCC that “[i]ncentivizing competition is a job for governments at every level. We must build on and expand the creative thinking that has gone into facilitating advanced broadband builds around the country…Working together, we can implement policies at the federal, state, and local level that serve consumers by facilitating construction and encouraging competition in the broadband marketplace.” But, as the record in this proceeding makes clear, government rate regulation will not “facilitate advanced broadband builds around the country.” Instead, it will do exactly the opposite – discourage facilities-based entry by limiting the returns, particularly in rural areas, available to those willing to risk investing and punishing those that already have taken that risk.