Compounding Incumbent Advantage
"Getting your foot in the door" is permanent permission to enter the fast lane
It is a truth universally acknowledged that incumbent politicians tend to win elections. The magnitude of this effect is quite large and has been documented for decades, so in the context of politics it isn’t that interesting. Once you’re in office, you receive strong name recognition, the presumption of competence in your role, and ease of fundraising that doesn’t come to challenger politicians. This is true in national elections down to local ones (where the effect tends to be more pronounced, not less).
Is there any truth to the idea that incumbent politicians do better than challengers once in office? Put another way: do politicians perform worse during their first term?
To say nothing of how one would quantify a politician’s job performance as better or worse, I don’t think this is a workable question. My irresponsible speculation is that there may be some truth to this; perhaps they would underperform because it takes a while for politicians to adjust to their role, understand the exercises of power available to them, make the personal connections necessary to get work done, etc. If this were true, then incumbent advantage would have some justification.
But there are two problems. The first problem is that politicians tend to propose or enact the most sweeping changes quickly upon receiving power. There is less to do on the margin upon incumbent re-election—they’ve already spent years in the same office. Also, often approval for an incumbent is lower in subsequent terms, even if they win. For example, for the US presidency, incumbent presidents tend to lose ground in Congress by the start of their second term, so it’s impossible to get as much done in the first place. If we tried to compare first terms to subsequent terms, we would inevitably run into this effect.
The second problem is that I can’t find an objective way to measure performance. Although many studies do it anyway, I refuse to use approval ratings as a proxy because approval is pretty incoherent. Objective measures come mostly in the form of economic benchmarks like GDP, which is probably decent. Voters reward politicians who preside over good economic performance: Aytaç (2018) found that the economic performance of a region relative to surrounding regions directly correlates (p < .01) with re-election rates. Sadly this tells us nothing about whether or not first- or subsequent-term politicians presided over these periods, and very little about competence—it’s possible to preside over a period of high economic growth without having much effect on it.
So we must proceed without an answer to my first question, but we can still make some progress. Hirano & Snyder (2009) found that the incumbent advantage still applies between candidates in the same party with similar policy positions. In sum, we can postulate that the electorate generally believes incumbents are more competent in the job, even compared to a substantially similar challenger.1
I’m pretty sure this applies to everything.
I graduated from university with an economics degree and a B+ average. I interned for Apple for two and a half years by then and had a pretty good degree of exposure to front-end web development, even if much of my time was non-technical. I probably submitted around 1400 applications for entry-level jobs across several industries, seldom getting a response. I ultimately landed a data entry & software testing job in Dallas. I made less than I would’ve needed to simultaneously pay rent, eat decent food and service my student loans—not a very glamorous starting point.
Later, I worked as a cost accountant with a technical focus, using SQL to obsolete what had previously been stored in Excel sheets. Then I got a job as a software engineer, and after around two months of that, I started to get relatively constant recruitment emails and messages which continue to this day.
Why the stark contrast between the starting point and the end? Less than two years separate the bookends to this story. Did I really have so little employability at the beginning and so much by the end? I don’t think so—by a month into each job I think I had grokked the technical skills I needed, so if a company were to have just borne with me for three months out of college I would’ve been fine.
Why didn’t this happen? Well, mostly signaling: there’s no direct way to differentiate an entry-level worker who will do well from one who will fall flat, so employers look for proxies like a related college degree and an excellent GPA. I had middling viability according to both those metrics, so employers mostly ignored me for a couple of years until I had better signals like a history of technical work and good references.
But wait, that sounds a lot like the incumbency presumption of competence! There’s no signal as good as prior experience doing the same thing, and that’s exactly what already having “software engineer” on the resume did for me. Because someone else already went through the basic vetting of “is this person remotely viable as a developer?” it’s easier for the next company to presume I can do the job.
But it’s deeper than that. As a professional software engineer, I’m able to spend 40+ hours a week developing software, learning about software, reading books on software, and talking with my technical and non-technical colleagues about software and working on a software team. That’s a lot more time than someone outside the industry gets!
Knowledge-worker incumbency advantage compounds on itself. Over time, my basic signals get stronger—I have more years on the resume and slowly gain a reputation in my areas of expertise—but also my actual skills tend to get stronger at a pace that would be unsustainable for someone who isn’t already in the industry and has to spend 40+ hours a week doing something else.
The calculus might look like this:
Applicant A learned to code on the side during an unrelated full-time job and has done an average of one hour of open source work per day over a span of three years. Therefore, Applicant A has around 1100 hours of technical experience.
Applicant B codes daily in the course of a full-time job and has been employed in such a role for just over six months. Therefore, Applicant B has around 1100 hours of technical experience.
Because of previous full-time experience in the job, Applicant B has some greater degree of implicit knowledge about working full-time as a software developer.
Therefore, Applicant B is most likely a better candidate than Applicant A.
Consider three more factors: first, Applicant A had to work very hard—almost unrealistically hard—to get 1100 hours of experience; I don’t know any newcomer to coding who averages an hour a day for three years. Applicant B didn’t have to do anything except regular full-time work and got a comparable amount of experience six times faster. It seems that X years coding on the side is at least quadratically less valuable than X years coding full-time. Notice also that statement #3 is a nod to an incumbent's presumed competence relative to a challenger.
I wouldn’t be surprised if this is why “you need experience to get experience” is such a common phenomenon. Just a few months of full-time experience in a field is likely to net as many hours of experience as a very studious schedule on the side. Plus, we are all constrained by the same amount of time in a day—it’s effectively impossible for Applicant A to catch up.
This isn’t to say it’s futile to try to get into software development. I’m using it as an example because it’s what I know the most about, and it’s perfectly reasonable to get an entry-level job by doing what Applicant A did (and it’ll take much less than three years if done well). My point is that there are two lanes in any field of expertise—those already in and those trying to get in—and already being in is such a tremendous advantage that the sole goal of anyone outside should be to get in. It doesn’t matter if you start in a subpar job with subpar pay. “Getting your foot in the door” equates to “permanent permission to enter the fast lane.”
This is also why switching industries is so hard.
If I’m spending forty hours a week writing software, but I want to be a violinist, I’ll be outcompeted by entry-level folks despite my best efforts. This is because children, teenagers, and college students have more free time to practice than I do. If a college student is on an 8-week summer break and practices full-time on the violin with a half-day on Saturdays, they’ll accomplish as much as I could by spending ten hours a week on it for six months.2 Putting in that amount of effort outside work doesn’t come naturally for people with competing family obligations, either. It’s a tremendous strain.
Going to night school is likely more effort than this, and I honestly don’t know how people do it. I admire it greatly, both as a sign of extreme dedication to self-improvement and as a means to force oneself to follow through (else a great deal of time and money would be wasted quitting partway). It’s also (in some industries, probably not tech) a time-efficient3 way to improve one’s signaling to potential employers.
In this way, if I have any other interests whatsoever (such as performing open source work, keeping up my reading and writing schedule, etc.), the switching costs to another industry would be too high for me, and I know plenty of people with the same sentiment.4 As far as this newsletter goes, I love how it’s going and appreciate the rather larger amount of attention I’ve received so far than I expected—thank you everyone!—but I remain envious of professional writers who have much more time to hone their craft than I do. Life looking as it is right now, my writing frequency is the highest it could comfortably be, and I certainly don’t have time to learn more about music production, etc. without sacrificing something that matters more to me.
That’s okay. Adult life involves constant trade-offs about how you spend your time. What’s most important is to make thoughtful, deliberate choices about how that time should be spent. There’s much more to say here, from the importance of habit-forming to the importance of correcting for hyperbolic discounting—all to be examined in the coming weeks.
It is commonly believed, and I would agree, that some of the effect also lies in name recognition, but decomposing incumbent advantage further is dicey. We can confidently say for our purposes that presumed confidence is a component of full-size incumbency advantage.
To say nothing of the time it takes to enter a flow state when starting an activity, which in my experience makes the first hour much less productive than the subsequent!
But not necessarily cost-efficient? I don’t know; I suppose there is no better option for many industries than to continue education. I can’t imagine I’d ever be taken seriously if I wanted to go work in management consulting, for example, unless I went back to school. This is unfortunate considering the price tag.
Although I love my industry, so I’m lucky in this regard!