The year of wishful thinking: The market research on Network Rationalization
July 23, 2012
At the end of last week, the participants in the Advisory Opinion on Network Rationalization submitted reply briefs that address the issues and arguments in the initial briefs filed a couple of weeks ago. Much of the debate about the advisability of proceeding with the consolidation plan hinges on the market research conducted last year to determine how big the revenue losses caused by the change in service standards might be.
As those who have been following this story know well, there were two phases of quantitative research. The first round was conducted last August. When the results came in, the Postal Service estimated a total volume loss of 7.7%, a revenue loss of $5.3 billion, and a net contribution loss of $2 billion. That net loss would have nearly wiped out all of the $2.6 billion in operational savings from the consolidations.
The Postal Service decided it needed to have the research firm re-do the survey, and the results of the second phase, done in October, were presented as testimony with the Request for an Advisory Opinion in December. The Phase 2 survey showed a total volume loss of 1.7%, a revenue loss of $1.3 billion, and a net contribution loss of $500 million. (All the numbers for the two surveys are here.)
The Postal Service didn’t mention the existence of the Phase 1 research in its initial testimony, but after a number of interrogatories were posed concerning the market research, the Postal Service revealed there was an earlier phase. The Postal Service claimed the earlier round of research had been “abandoned” and left “incomplete.” This research was “flawed” and “unreliable” because the concept statement read to participants to initiate interviews had mentioned other initiatives on the table — closing post offices and eliminating Saturday delivery — and this led customers to overestimate how much less mail they would send if service standards were changed.
Several participants in the Advisory Opinion have argued that the Phase 1 research was valid, probably more valid than the Phase 2, and the Postal Service is seriously underestimating how much revenue it will lose. In its Reply Brief submitted a few days ago, the Postal Service responds to those arguments.
This section of the Reply Brief is entitled: “Opposition to reliance upon the Phase 2 quantitative market research estimates of the impact on mail volume caused by MPNR [Mail Processing Network Rationalization] is based on wishful thinking, not science: The only valid estimation of impact upon mail volume of changes in service standards derives from the Phase 2 research.”
The tone of that title pretty much captures the Postal Service’s attitude. This section of the brief is really a piece of work, and it’s worth a closer look.
The Postal Service would never do that
The Postal Service attorneys begin this section of the Reply Brief as follows: “Various Docket No. N2012-1 intervenors align themselves behind one phase or another of the market research results presented by the Postal Service without recognizing what they are doing: voting on the results believed to best serve their own positions. Quite certainly, that is not what the Postal Service itself did, as reflected in its Initial Brief.”
The Postal Service thus accuses the intervenors of picking the research that suits their purposes but denies doing the same thing itself. The Postal Service, however, did a lot more than “vote” on the results that served its position. The Postal Service didn’t choose between two sets of results: It rejected the first set because it didn’t like the numbers, and it commissioned a second round to get what it wanted.
The Postal Service’s Reply Brief then proceeds to state the following: “[The Postal Service] did not rely upon other market research, such as Phase 1, that was incomplete and which studied a quite distinct set of circumstances that could impact volume entered by its customers.”
But the Phase 1 research was not “incomplete” in any meaningful sense of the term. The market research firm gave the Postal Service the data about likely customer behavior, and the Postal Service converted the data into estimates of loss volume and revenue. The only thing left “incomplete” was that the market research firm had not yet “scrubbed” the data for some minor data entry errors.
As for the issue of what the Phase 1 research actually measured, it’s not quite accurate to say that it “studied a quite distinct set of circumstances that could impact volume entered by its customers.” The questions posed in the Phase 1 survey are all clearly focused on the change in service standards, not other circumstances. It did not “study” circumstances other than service standards. It just happened to mention them in the concept statement used to initiate the interviews.
The Reply Brief then says, “Virtually no party addresses this issue head-on by comparing the concept statement of Phase 1 of the research versus Phase 2 concept statement, although GCA [Greeting Card Association] does acknowledge that only the Phase 2 research addresses change caused by MPNR alone.”
Well, since the Postal Service seems to think that looking at the concept statements will help its argument, let’s take a look. The Phase 1 statement is here, and the entire questionnaire, here. The Phase 2 statement for commercial accounts is here (p. 100).
Addressing the concept statements, head-on
The passage at issue in the Phase 1 statement says the following:
“In the past two years, the Postal Service has had budget deficits of over $8 billion and expects to have a similar budget deficit this next year. To address the budget deficits, the Postal Service is exploring several changes, including
- Legislative reform to change government requirements to pre-pay health and pension benefits
- Eliminating Saturday mail delivery to homes and businesses
- Closing many small post offices while shifting retail access to alternative locations and channels
The Postal Service is also considering revising the service standards for First-Class Mail within the continental U.S.”
The concept statement then proceeds to describe the changes in service standards. The fact that this statement refers to legislation, Saturday delivery, and closing post offices is now at the heart of the Postal Service’s explanation for why the Phase 1 results showed such huge losses.
The Phase 2 statement deletes that passage referencing eliminating Saturday delivery and closing post offices. But it does something more. In the concept statement used for commercial accounts, there are many more details about the new service standards. The statement, for example, explains that mailers who present their mail pre-sorted by the critical entry time will still qualify for overnight delivery — something that was not mentioned in the Phase 1 concept statement.
The new statement also notes that “processing operations at the destinating plants will likely increase the frequency of the mail being available for pick-up, resulting in earlier availability of mail than is currently possible based on existing processing windows.” That too was not mentioned in the Phase 1 statement.
The Phase 2 statement also says this: "Businesses using bulk First Class, Standard, or Periodical Mail may have access to fewer locations accepting this mail and potentially result in a need to transport this mail to a location different from the one they are currently using. However, there is also the potential that these mailers may be able to achieve improved transportation efficiencies since the service areas of these facilities may be larger than they are currently. For example, if a mailer currently sends mail to two facilities which are consolidated to one, this can allow the mailer to prepare a larger mailing for deposit at the consolidated site, resulting in a reduction in the required transportation.” (Italics added)
That passage is not in the Phase 1 concept statement either. The Phase 2 concept statement thus adds numerous details that make it seem as though the change in service standards will actually improve things for commercial mailers. This helps explain why some mailers actually said they would send more mail when the change in service standards slowed down delivery — an anomaly that’s almost impossible to explain otherwise and a phenomenon so improbable that it led NALC's expert witness to dismiss the results completely.
Blaming the messenger
The Postal Service wants to blame the references to legislation, five-day delivery, and closing post offices in the Phase 1 concept statement for the huge revenue losses projected by the survey. There are two problems with this argument.
First, the survey was clearly about the change in service standards, not “other circumstances.” The questions are all variations of the following: “Assuming that the changes to First Class Mail had been in place during the past 12 months, what is the likelihood that this change would have caused your organization to modify the number of individual pieces of mail your organization sent by any means?” Why would a customer think that Saturday delivery or closing post offices was part of the question?
Granted, participants in the qualitative study expressed concern about the other changes being contemplated, particularly with respect to the cumulative effect they might have, but there’s nothing in the Phase 1 quantitative survey instrument and the questions it asks that would lead customers to think they were being asked about how they would respond to causes other than the change in service standards.
But there’s a bigger problem with the Postal Service’s argument. Even if one grants that mentioning other causes in the concept statement could have contributed to the respondents’ projections, the numbers just don’t add up in a way favorable to the Postal Service’s position. Consider the following:
The Phase 1 concept statement refers to four factors — legislation, Saturday delivery, closing post offices, and the service standards.
Legislative reform on pre-paying health and pension benefits would not lead mailers to reduce their volumes. To the contrary, such legislation would relieve the Postal Service of some of its debt and actually help mailers by reducing the pressure for rate increases and service reductions.
That leaves only two causes (aside from the change in service standards) that would affect customers’ projections of their mail volumes — closing post offices and eliminating Saturday delivery. So how much might these changes affect mailers? What portion, in other words, of the projected losses in the Phase 1 survey can be attributed to these other causes?
That question is not a hypothetical about which one can only speculate. The Postal Service has addressed both of them quite explicitly in testimony before the PRC for other Advisory Opinions.
Adding up the numbers
In the Five-Day case, Opinion Research Corporation (ORC), the same market research firm used for Network Rationalization, determined that eliminating Saturday delivery would cause total volumes to drop 0.7%, gross revenues by $428 million, and net contribution revenues by $197 million. (The PRC said the losses would be three times that amount because of the way the Postal Service manipulated the data, but the Postal Service continues to contest the PRC’s finding.)
As for closing thousands of post offices, when it submitted its Request for an Advisory Opinion on the Retail Access Optimization Initiative to close 3,700 post offices, the Postal Service did not commission any market research, and it did not include any lost revenue in its cost-savings analysis, which found the plan would save $200 million.
Similarly, in its case for POStPlan, the plan to reduce hours at 13,000 post offices, the Postal Service again did not commission any market research because it maintains that there will be no lost revenue at all. Just a few days ago, when he was cross-examined before the PRC (testimony, p. 200), the Postal Service’s witness for POStPlan reiterated several times that “the Postal Service does not anticipate that we’re going to lose revenue based on POStPlan.” When it comes to the retail network, the Postal Service assumes that if your post office is closed, you’ll go to another post office or use an alternative that keeps your business with the USPS.
The Phase 1 research, the so-called “all-causes” research, thus introduces three possibilities that might lead customers to reduce mail volumes. As noted above, this research showed a total volume loss of 7.7%, a revenue loss of $5.3 billion, and a net contribution loss of $2 billion. Eliminating Saturday delivery accounts for $200 million in contribution losses, and closing post offices accounts for none. To what can we then attribute the remaining $1.8 billion in lost revenue if not the lone remaining cause, the change in service standards?
Even if we assume for a moment that the Postal Service has underestimated losses from eliminating Saturday delivery and closing post offices, the numbers don’t look much better. Say the PRC was right and eliminating Saturday delivery will lose $600 million, and say that closing the 3,700 RAOI post offices might cause a loss of $50 million (about 20 percent of their total revenue). That adds up to $650 million. We’d still be left with nearly $1.35 billion that would need to be attributed to the change in service standards.
These calculations thus suggest that the Phase 1 survey showed that the contribution losses associated with Network Rationalization would be somewhere between $1.35 billion and $2 billion — a lot more, any way you look at it, than the $500 million projected by the Phase 2 survey.
For astute observers only
The Postal Service argues that the “broad concept statement” used in Phase 1 produced results that were irrelevant to the question of how Network Rationalization will affect mailers. Only Phase 2 addresses that question specifically.
The Postal Service’s Reply Brief now takes a strange tack in its argument. It says the following: “Astute observers of market research might take it one step further by asking and answering the question: what will the impact on mail volume be if all of the matters described in the Phase 1 market research go forward except for MPNR? The two phases of research provide a ready response: mail volume will be approximately the Phase 1 research minus the Phase 2 research results, or approximately an 8.5 percent decrease.”
The Reply Brief doesn’t explain where the 8.5 percent comes from. The Phase 1 total volume (7.7%) minus the Phase 2 (1.7%) equals 6%, and the Phase 1 Single Piece volume (10.3%) minus Phase 2 Single Piece (2.8%) equals 7.5%. Maybe someone got it mixed up and subtracted Phase 2 total from Phase 1 Single Piece, which equals 8.6%. Anyway, the Postal Service suggests that the difference between the two surveys — 8.5 percent — must be attributed to causes other than Network Rationalization.
But how valid, asks the Postal Service, is this projection? “Two ready avenues for assessing the validity of this hypothetical future mail volume fall readily to hand,” answers the Postal Service.
The first avenue is looking at the “primary driver” of losses — electronic diversion. Those projections are substantially less than 8.5 percent, and even if one includes the recession, “one does still not approach 8.5 percent.”
The prose or the logic (or both) are so convoluted in this section of the Brief, it's hard to understand what the Postal Service is driving at. Apparently it's taking the difference between the two surveys, and then asking us to “assess the validity” of projected losses on the magnitude of 8.5 percent a year. Since even the Internet and recession aren’t being blamed for losses anywhere near that big — a typical projection for First-Class mail declines is 3.5 percent a year, as in this oft-cited BCG study — it’s unlikely that Saturday delivery and closing post offices could be responsible for losses of that magnitude. Therefore, the Postal Service concludes, the Phase 1 survey is not reliable.
But if that’s a correct reading of this passage in the Reply Brief, the Postal Service’s argument begins by assuming the very thing it's trying to prove — that the losses projected by Phase 2 are reliable. If, however, the Phase 2 results seriously underestimate losses due to Network Rationalization, the difference between the two surveys would be much smaller and more reasonably attributed to eliminating Saturday delivery and closing post offices.
The Postal Service’s Brief then suggests a second way to assess the validity of the 8.5 percent projection: “Which if any of all the various initiatives and challenges embodied in the Phase 1 concept statement are either likely or unlikely to be implemented?” The Postal Service observes that 5-Day Delivery continues to be prohibited, as it has been for nearly four decades, and closing post offices isn’t happening either, especially as the new plan, POStPlan, “goes forward largely unopposed.”
The Postal Service then says this: “So we also know that the various possibilities identified in the Phase 1 market research concept statement are unlikely to reach fruition. Hence, any volume projection from that research (assuming its final results resemble the initial, tentative estimates available) would be a poor foundation for projecting future mail volumes.”
In other words, because they may not eliminate Saturday delivery and since they’re not going to close post offices, the Phase 1 research isn’t relevant. (Besides, the results were just "tentative.")
This argument is also flawed. The Phase 1 research may have shown the effect of multiple causes, but it doesn’t matter whether these other initiatives actually happen. The point is that these other causes — using the Postal Service’s own claims — don’t come anywhere near to adding up to an 8.5 percent loss. That doesn’t mean the Phase 1 survey is necessarily unreliable. It could also mean that the service standards will be responsible for the lion’s share of the 10.3 percent loss, so that it’s the Phase 2 survey that’s flawed, not Phase 1.
The Postal Service concludes with the following remark: “Examination of the two phases of market research thus illustrates that the widespread exhortation that the Phase 1 market research results constitute a better estimate of future mail volumes if MPNR is implemented than the Phase 2 results is mere rhetoric, or the result of party voting, while lacking any reasoned, logical or empirical foundation.”
The Postal Service thus calls those who accept the Phase 1 research unreasonable and biased, even though the Postal Service itself commissioned and helped design the research and the Vice President of the market research firm described it as “very good research.” Wouldn't the Postal Service be more reasonable if it simply acknowledged that the Phase 1 research has much to recommend it and then developed volume and revenue projections based on both studies?
The Phase 1 research needs to be adjusted to account for the unintended influence on customers of the references to Saturday delivery and closing post offices, and the Phase 2 needs to be adjusted because of the way it modified the concept statement and introduced a “solely attributable” factor. A reasonable projection would probably be somewhere in between what the two studies found — perhaps around a 5 percent decline in total volumes, with contribution losses around $1.4 billion. But rather working its way toward some middle ground, the Postal Service sticks to its original testimony and accuses those who question the numbers of “mere rhetoric,” “party voting,” and “wishful thinking.”
The Postal Service would never do that either
Having attacked its opponents for engaging in rhetoric rather than reasoning, the Postal Service’s Reply Brief then addresses one of the more serious accusations that have been made against it — the charge that it kept the Phase 1 research secret. “Claims that the Postal Service hid its preliminary market research results from Phase 1 lack any foundation in fact or law,” says the Postal Service.
Did the Postal Service hide the survey? The answer to that question probably depends on the meaning of the word “hid.” The Postal Service certainly chose not to mention the survey when it submitted a Request for an Advisory Opinion. It provided a mountain of testimony, but none of it mentioned the existence of a phase 1 survey.
More important, when the Postal Service witnesses were initially presented with interrogatories that provided an opportunity to acknowledge its existence, they chose not to. For example, in January, the NALC posed the following interrogatory to the Postal Service manager who helped design the survey:
“In its quantitative market research, did USPS seek to make any estimate of how much volume, revenue and contribution it would lose if USPS implemented both (1) the end of Saturday delivery and (2) the change in service standards proposed in this case? If so, please provide these estimates and explain how they were calculated.” (NALC/USPS-T12-14)
The response: “The market research upon which the Postal Service relies in this docket examines mail volume changes arising from the service standard changes alone. The Postal Service has not undertaken similar research that isolates changes arising from the sum of service standard changes and the end of Saturday delivery.”
That answer may have been technically true, but as we would learn weeks later, the Postal Service did do “similar research.” But because the Phase 1 research included factors like closing post offices and postal reform legislation and not just service standards and Saturday delivery, the Postal Service apparently did not think it was necessary to mention this research.
There were several other interrogatories about market research on “cumulative impacts,” and in each case, the Postal Service produced a nuanced response that did not mention the Phase 1 survey. Only when the interrogatories became much more explicit did the Postal Service decide to reveal the existence of Phase 1.
If the Postal Service didn’t “hide” the Phase 1 research, it certainly can’t be said to have been very forthcoming about it.
One September afternoon
One can only surmise about the chain of events over the past year, but it’s not unreasonable to imagine the following:
The executives of the Postal Service decided they wanted to consolidate 250 processing plants. They were convinced that it would save money. How could eliminating tens of thousands of jobs not?
They then realized they’d need to produce some market research for the Advisory Opinion. They had done market research for the Saturday delivery case, and there was no way the PRC would think it unnecessary in this case, which promises to have an even bigger impact on the mail. So last summer the Postal Service contracted ORC to do some research.
When the results came back, postal management realized they had a problem, and someone quickly called a meeting. So picture a conference room at L’Enfant Plaza, one afternoon in September 2011. Let's listen in.
Someone is giving a presentation about the survey and explaining that they’re looking at $2 billion in contribution losses.
“Yikes,” someone says, “that wipes out almost the entire $2.6 billion in savings we're hoping to realize. This plan will never fly with those kinds of numbers.”
“But wait, look," someone joins in, "the concept statement mentions eliminating Saturday delivery and closing post offices. Maybe the survey is actually showing the cumulative effects of all three factors.”
“Good thought,” someone else says, “but we already know that Five-Day will lose $200 million and closing post offices won’t lose anything, so that leaves us with a $1.8 billion loss for service standards. How are we going to explain that to the BOG?”
That’s when the smartest guy in the room says, “Let’s have them do the survey again. We’ll get it right the second time around. We'll just bury this one.”
So the market research firm goes back to work, and the Postal Service revises the concept statement and makes it sound as if the changes would be good for the big mailers. They introduce a “solely attributable” factor and tweak the survey in some other ways to help ensure that the second time around produces the desired results.
Now the Postal Service says there’s no basis for saying it hid the survey, the Phase 1 results are irrelevant, the Phase 2 results are definitive, and anyone who says otherwise is biased, unscientific, and engaged in wishful thinking. It wants to ignore all the evidence that the consolidations may do more harm than good, and rather than waiting for the Advisory Opinion to be issued, it has already implemented the changes in service standards.
Once the Postal Service made up its mind that Network Rationalization would be a good thing, nothing could convince it that it might be wrong. The Postal Service still refuses to believe that the consolidations will have a very deleterious effect on mail volumes and revenues.
Talk about wishful thinking.