Every year my employer schedules two weeks for Open Enrollment in the benefits plan. This is the time period where you can freely change health plans, life insurance, etc. In the weeks before Open Enrollment we receive numerous emails reminding us that it is coming up. During the Open Enrollment period we receive numerous emails reminding us of the deadline. The day of the deadline we receive a final email saying that today is the deadline.
Then, every year after the deadline passes the deadline is extended an additional week, and the many people who procrastinated and missed the first deadline are given a reprieve. This happens so consistently that many people know it is coming and understand that the “real” deadline is the second one. So many that it is reasonable to assume that a sizeable number of these people procrastinate up to the second deadline and miss that one too. But there is never a third deadline.
You notice this kind of thing happening a lot. Artificial deadlines that you can be forgiven for missing but only once. It’s a puzzle because whatever causes the first deadline to be extended should eventually cause the second deadline to be extended once everyone figures out that the second deadline is the real one. For example if the reason for the first deadline extension is that year after year many people miss the first deadline and flood the Benefits office with requests, then you would expect that this will eventually happen with the second deadline.
The deadline-setters must feel on firmer ground denying a second request by saying “We warned you about the first deadline, then we were so nice and gave you an extension and you still missed that one.” But if a huge number of people missed the second deadline, no doubt they could still mount enough pressure for a second extension. Indeed the original speech of “We sent you emails every day warning you about the deadline and you still missed it” didn’t stem that tide.
Is it just that every nth deadline is a new coordination game among all the employees and the equilibrium number of deadline extensions is simply the first one in which the “no extension request” equilibrium is selected? I think there is more structure than that because you rarely see more than just one, and you almost never see zero.
I think there is scope for some original theory here and it could be very interesting. What’s your theory?
Cheap Talk 
6 comments
Comments feed for this article
November 22, 2010 at 11:45 pm
mich
Have you actually confirmed that requests for a later deadline are denied? I’ve experienced that after a second deadline individual requests will often be honored, but the overall deadline remains for those who don’t ask.
November 23, 2010 at 9:07 am
Anonymous
This is part of the puzzle. It appears that after the final deadline (whatever n that is) there is another soft deadline. You can get an extension but only if you have the courage to ask and only before some hidden final deadline. Presumably this soft deadline is endogenous. If every day before you someone asked for an extension then that’s enough activity to keep the option open. If a week goes by and nobody has asked for an extension, then the next to ask is probably not going to get it because the office can confidently say that you are an outlier.
This endgame itself has multiple equilibria.
November 29, 2010 at 10:10 am
twicker
Mich, I agree w/you — and I suspect what happens is the following in coordinating:
1) Initial deadline set — all is good.
2) Large # of people request extensions — an N(asked) (meaning N of people who asked for extensions) large enough that it “seems” unfair to not simply let everyone have a later deadline.
3) Also, the deadline-setter sees a large enough N(asked) to make an assumption that there’s a large pool of people who:
(a) want to respond,
(b) have not responded, for whatever reason, and
(c) haven’t formally asked for an extension (N(quiet)).
Meaning, there’s acually N(asked) + N(quiet) non-responders
4) The deadline-setter sees more to gain from extending the deadline than not extending it (more on this in a moment).
5) Thus, deadline is extended.
If N(asked) is sufficiently small, then, as Mich mentions, the extensions will be one-offs. Part of the assumption *here* would be that N(quiet) is trending rapidly towards zero, as people’s senses of urgency/need for action overtakes any embarrassment at asking.
Now — back to #4, it being in the interest of the deadline-setter to extend the deadline:
If I’m running the open benefits stuff, then I like keeping my fellow employees happy and being known as someone they can work with. Remember, I *work* with these people; they’re my friends, and the ones who aren’t are still people I see every day. Even beyond my personnel evaluation, I have strong social reasons to be helpful, not hurtful.
Now say I’m an apartment owner: yeah, you get that first “extension” (in my complex, rent is “due” on the 1st, but we have until the 5th without penalty), but that’s it. If you have a *really* good payment history and you’re on good terms with them, they’ll forgive you if you don’t get it in until the 6th or 7th — but they file court papers on the 11th.
Period.
Why? Well, they have a stong incentive to make sure they don’t have deadbeats. Deadbeats raise costs all over the place — and the apartment managers want to make sure they’re ferreted out and removed ASAP. So, the “deadline extension” (the 5th) does become the de facto deadline (with a certain N of tenants paying early to avoid the stress, and almost everyone paying by the 5th). However, there really isn’t any expectation of a deadline extension beyond the 5th.
(and now I need to get back to my day job as a PhD student. Shout out to Mich, Jeff, et al. from Durham, NC — yep, that university, and biz skool to boot [Management & Organizations]. Credit them w/anything intelligent I might say; all stupidity is my own and predates my affiliation w/Buck Duke’s university.)
yrs,
twicker
November 23, 2010 at 8:25 am
Lee
This is similar (to me) to the problem of dealing with NetID password expirations. At what point is there too much early notice of a deadline? The cycle is something like 90 or 120 days, but when I last compared that to the first notice that your password needs to be changed, that notice was coming almost half-way into the cycle!
Of course you can change your password at the time of first notice, which maybe some people do, but that increases the number of passwords you have to come up with and remember. I like to keep my password as long as allowed, so I’ve resorted finally to scheduling a meeting with myself in our calendar to change the password on the day before — so that I can safely ignore all the subsequent, egregiously early warnings. It’s too much for me to parse each e-mail and figure out when I’ll need to execute the change to be “on time”.
I think the worst of it is that for regular staff like myself who come in daily, it would probably catch a vast majority of cases if the most important warning were a business-day-before e-mail, but there is no such thing.
November 26, 2010 at 6:07 pm
TomGrey
Isn’t this much like a rule of thumb for speeding: up to 10 mph over the limit and you’re probably safe?
Everybody knows that people, especially Other People, need limits and borders. But for me, this time, this time it’s special, plus I’m special, so I need an exemption. (Pres. Obama hasn’t gone thru a TSA groping).
People support the limits, in theory, but want the slack for themselves, at least a little. Especially if there was slack last time, and the cost of slack is so low…
Meetings should always start on time, to avoid that slope. With some pointed joke/ shame at those coming late.
April 21, 2011 at 6:23 am
Game theory and the budget — Marginal Revolution
[…] and option value. If you’re not sure of the true model, wait rather than commit. Here is Jeff on deadlines. Another way to put this point is that we can’t, from current Republican inaction, infer […]