Game Economist Cast
What does the new wave of open economies mean for monetization? Will negative externalities overcome cosmetics economies in the long run? What exactly does a game economist do?
Game Economist Cast is a roundtable discussion of the latest developments in mobile, HD, and crypto games through a bunch of people figuring it out using the economic tool kit.
Game Economist Cast
E29: Can a Stick of Butter Keep You Sober?
Can a stick of butter keep you sober during high-stakes diplomatic meetings? Eric is sure to tell us. Chris provides a compelling theory for the death of mods, while Phil undergoes a Supercell detox after analyzing Squad Buster's launch. The crew gets back in touch with their microeconomic selves and looks at one of the most interesting explanations for the gender pay gap.
I heard about Henry Kissinger. I don't know if this is true or not, but I heard that Henry Kissinger would, before going to diplomatic meetings, would eat a stick of butter, because at these dinners like everyone's drinking and stuff right, and then there's some like negotiation. That happened after dinner and by you don't want to not drink because then they like have that same sort of social vibe. But by eating the stick of butter it made him become less drunk and therefore have the upper hand in negotiation. By eating the stick of butter, it made him become less drunk and therefore have the upper hand in negotiation. Does butter do that? I don't even know if this is a true story. I feel like I heard this one time in college.
Speaker 2:Where do you find these stories? I feel like you've always got these insane stories. How many calories are in a stick of butter.
Speaker 1:I Googled it and I didn't find anything, so maybe it's not true. Wow.
Speaker 2:There's only 810 calories in a stick of butter. I honestly would have dude. I would have guessed 3000 calories, minimum at a zero to that it's nothing. How much butter is 2000 calories? People really want to know how much butter they can consume in one day. Now this one says three tablespoons. Don't know if I believe this Whipped butter is even lighter calories because lightweight.
Speaker 1:Let's start with utility.
Speaker 3:I don't understand what it even means.
Speaker 1:Everybody has some kind of utils in their head that they're calibrating.
Speaker 2:There's hardly anything that hasn't been used for money.
Speaker 1:In fact, there may be a fundamental problem in modeling.
Speaker 3:In fact, there may be a fundamental problem in modeling what I want to model Game Economist Cast, episode 29. We're back. We're back on a regular schedule-ish. It's summer. Everyone in Europe is on vacation. Hi, I'm Phil Black from Game Economist Consulting. I am joined by my two other wonderful co-hosts today.
Speaker 2:Chris from Star Atlas. Chris, how are you Doing? Well, it's the summer of data review, data maintenance and data analysis. We are in a holding pattern in terms of new feature releases. Just get to finally kick back and do all the stuff that I've been putting off for months.
Speaker 3:And of course, Eric from not Superlayer anymore.
Speaker 1:Yeah, I'm at a second dinner now.
Speaker 3:Still having a good time.
Speaker 1:Yeah, we had the last week off. It's a July 4th week off.
Speaker 2:Whoa.
Speaker 1:Yeah, if your employer didn't give you that week off, that means they don't love America as much as second dinner does. You should make sure they know that. Yeah, it's been pretty good. It's been hot, but otherwise no complaints.
Speaker 3:We have an exciting episode planned, as always. Not only are we covering a piece I wrote on Squadbusters because I can't help myself playing too much Squadbusters we are introducing a new segment to the Game Economist cast, which may prove divisive for our audience, but we are going to be reviewing a paper, an economics paper, that may have been published recently or may have been published a long time ago. We're talking about just doing Nobel Prize winning papers, although we won't be doing that this week. We'll be looking at a really interesting paper called One Cohort at a Time and New Perspective on the Declining Gender Pay Gap. Sometimes there will be a games tie in, sometimes there will not be a games tie in tie-in, sometimes there will not be a games tie-in. But I think one of the things that we found is that a lot of the really cool research and thinking that happens in academia doesn't really make it outside the walls of the ivy, so to speak, and we want to be connected to our roots, our home base, and a lot of that happens in very traditional economic research. So this is an opportunity to expose some of that research and some of that beautiful thinking to people who may be in games and also people who may not be in games, and so we're going to start that off today and I think it's gonna be a lot of fun.
Speaker 3:Also, we're finally in video. I fucked it up last time, you may have noticed. I said we're going to be in sparkling 4k and then it never happened, although you can have a high definition audio, but we have video. We're gonna be on YouTube, so we're gonna be sharing charts, graphs. People are going to have to look at us. It's not going to be good. We're going to find out what we look like and we're going to be immediately disappointed. We coordinated our t-shirts today, chris and I, eric get the message.
Speaker 1:Oh man, it's a plain monochrome t-shirt. Yeah.
Speaker 3:Before we talk about Squadbusters and that paper, let's talk about what we've been playing.
Speaker 1:Got a selection of good things on sale stranger.
Speaker 3:Eric, what have you been playing?
Speaker 1:I have been playing Inkbound a little bit. It's a roguelike.
Speaker 3:Oh, surprise Shit, Is there a drinking game for Game Economist cast? Yet that's going in my middle space.
Speaker 1:The special thing about this one is it's built for live service co-op, which we talked about in the past. Roguelikes are popular, but they tend to be single-player. They tend not to have live service. Don't get too excited, phil, there's no monetization on the live service, but it is built to have leaderboards and seasons and, most interestingly I think, the co-op. It's a tactic game, like a turn-based tactic game that managed to make co-op pretty good. Managed to make co-op pretty good. Uh, if you've ever tried to play co-op tactics games, they tend to be extremely slow. Played balder's gate 3 with your friends, as it takes 30 minutes to do one fight.
Speaker 2:This is just a skew. This is just like cost 15 bucks or something in steam yeah, yeah, flat price.
Speaker 1:I bought it thingsdale. I think it'd be cool to see more, like I said, live service roguelikes. Oh, I don't have to rehash that whole spiel. What?
Speaker 3:spiel. He was talking about lapso. No one, no one's really figured out monetization for this.
Speaker 1:Yeah, the roguelikes seemed like they would fit perfectly in a live service model. Play it over and over. You make minor tweaks, these progression resets them very well, but at the same time, no one's fit them into the live service monetization model, which has proven very successful for other games. Right, why is that? Why has no one cracked the nut here? One theory was always multiplayer. That live service works only for multiplayer games because you need that social component. You need that if you're going to sell cosmetics or sell power. You need someone to peacock your power to, and if you can make a social co-op like roguelike, maybe that that nail on the head does squadbusters count as a?
Speaker 3:roguelike we're going to be talking about that. I actually think there's an interesting case, for this incorporates a lot of roguelike elements and eric we're I think we're talking last episode or offline I. You could put ccgs as a roguelike in some sense. Right, it is true that you're making a deck beforehand, but you're drawing random cards, right, and you're still doing a lot of that theory crafting on the fly. Squadbusters operates in basically the same way, in that you still have to unlock characters, but the characters that you choose to bring into a match randomly select for you, so it's almost like the deck is being selected for you. That, to me, is a really interesting model. I haven't seen a lot of roguelikes implement.
Speaker 3:The reality of moving to a roguelike that has monetization is that you need some level of permanence, so you have to lock some roguelike abilities behind something that you have to monetize. You have to collect them at some level. Like you, I think you just have to sacrifice that as a roguelike game to get this thing off the ground, which which kind of it does harm a little bit of the theory crafting, since the theory crafting is only as large as your collection of abilities or things that ultimately can interact to create some sort of interesting machine design space or your ability to create those really cool. What is it? Goldberg machines. Goldberg machines like your ability to craft that would. The design space would shrink so much until you acquired so many items.
Speaker 3:I think it just radically transforms the game into something that isn't really palatable but, I still think of a designer set there started with the precept that you're going to have to collect some of these roguelike abilities. I'm sure I think they could craft a better experience to incorporate the monetization element in without killing your ability to build that machine early on.
Speaker 1:That's true Most roguelikes. You start with a subset of the available relics or power-ups and you unlock the more complex ones as you play more.
Speaker 2:Do you get out-of-round progression in Slay the Spire? I don't think so. Right, it's just in-round.
Speaker 1:Yeah, you get more cards and relics that get added to the.
Speaker 2:You get access to.
Speaker 1:Okay, yeah, and the ones you unlock are generally stronger, but it's not very dramatic.
Speaker 2:Because I found Vampire Survivors overwhelming. I enjoyed the in-round progression. The out-of-round progression was just like with all the characters, the unlocks, like all the different quests. I'm curious because you've got belatro, which is actually the game I've been playing recently. You've got belatro where there's no out of round progression, from what I have been able to glean.
Speaker 3:I think you get decks. You can get the different deck foreman, the different card backs which have different abilities, so you get that's the one thing they have. Okay, that's a deck, would say you, you don't interest, you get 10 more currency per win, but that's it. It's just that one ability.
Speaker 2:I'm just curious what the what's, the marginal? I don't know what the word I'm looking for is. What's the margin? The marginal additional hours played or the marginal retention for out of round progression versus in round progression? You could imagine a game that's zero in round progression, all out of round, and then vice versa, something that's more like us not Bellatro, but Bellatro or Slay the Spire, where a lot of it is in round, and then versus something like Vampire Spires where it's mostly out of round progression. At least that's the majority of the actual progress they're going to make. And I'm just curious, if there's a trade-off, how would you set up that RCT? I think it's pretty. You could imagine. It's pretty simple to do practically, but or sorry, it's very easy to set up the theory but it would be difficult to actually implement. It would be costly. You could basically just like Vampire Survivors, take out the out-of-round progression, which is still a fun game, vampire Survivors with the in-round progression, random cohorts.
Speaker 3:Because in Vampire Survivors the only out-of-round is that you get new characters right and the new characters have a default roguelike ability attached to them.
Speaker 2:So you get new characters, you get new weapons, you get currency that allows you to purchase perks and you level up your characters. So there's both. You can go wide and you can go deep. So there's vertical and horizontal progression Out of round. Yeah, you could imagine a more subtle one where maybe there's one additional mechanic out of round mechanic and there's versus you add in a single additional out of round mechanic. The problem is, I think this would be a really long term study and that would be a really long-term study and that would be extremely costly. Um, this isn't oh what?
Speaker 2:let's run this experiment for 48 hours and have one yellow button and one purple button and see which one gets clicked more. This is, like, probably just too large. You would have to probably do more of a meta analysis than an actual um experiment, I think yeah, I mean it's like core to the design of the game, right?
Speaker 1:not like a simple knob you tweak yes, I see what you're saying.
Speaker 2:I don't think that's necessarily true. Like for me, out of round progression as long as you can get better at the game and you can like actually accomplish beating a level like, I think, vampire survivors is a great example of this, because the out of round progression really just makes you ridiculously powerful. But you can beat a level with a completely unleveled out of complete. You could never touch the out of round progression and still beat each of the levels in vampire survivors versus something like maybe slay the spire is actually a good example. There are some combos that if you don't have those combos, you will not be able to beat some of the cities. Or I'm not able to beat the cities without certain combos, I don't know.
Speaker 1:I guess another way of thinking about it is like how many tries does a player need to take to beat a level right? And one way games traditionally help people is if you fail repeatedly, your like stats go up, you level up and you get better gear, you get better whatever, and it makes it easier to beat it. Another way roguelikes handle this is just the randomness. You might just get lucky rolls. You might just get a certain set of relics that tend to happen to be really powerful. You might have the experience in Bellatro where you beat a difficult level just by playing it over and over until you get a lucky draw.
Speaker 2:I've been playing Bellatro. That's the game I've been playing. Tell us about it so far. It's interesting. I got it on sale from Steam for $13. Did they?
Speaker 1:put it on sale too early.
Speaker 2:That's a question I have. It's $2. I would have bought it either way. I didn't know it was on sale. I didn't really care that it was on sale, so maybe you're right.
Speaker 3:Maybe they didn't need to do that and they've actually moved away from that. If you go to SteamDBcom, you can see the price history for a lot of games. You look at some of the recent paid releases. The discounts are not as deep and they're not as frequent anymore.
Speaker 3:But the research I always heard from Valve and we do know someone who works at Valve is that the discounting was always effective at Steam. The usual economist response to discounting is well, maybe, but if you continue to discount and you have predictable discounts, then people will withhold their purchasing behavior until there's a discount and then only purchase when there's a discount. But it turns out so many customers are time insensitive that the revenue maximizing strategy is to it essentially just becomes price discrimination of okay, when you release this full price, you're going to get everyone at the full price, and then when you discount, you're only targeting a very specific set of customers that wouldn't have purchased otherwise and so like the revenue maximizing strategy is to do these discounts over the long run, although it feels like there's been a shift in how you get the song, how you get the rhythm. But again, I just really love for Valve to publish anything on this, like anyone to publish anything on this would be really nice.
Speaker 2:I was going to say something else, but I can't remember what it was. Oh, it has to do with discounting. This is just a pet peeve and I'm just going to use this. This is an opportunity to bitch. I was trying to buy the Elden Ring DLC the Shadow of the Earth tree DLC the Shadow of the Earth Tree and I tried to take pictures of my experience, but I have an iPhone 7 and it doesn't take pictures anymore.
Speaker 2:So, like you go to the main page, the main Xbox store, shadow of the Earth Tree you click on it and you're like, oh, that's cool, I'm going to go buy that. And it says here are the three versions you have available to purchase. One is $50, one is $60 and one is $80 or something the different versions. And I had heard that the DLC was $40. And I was like that's weird. I don't see the $40 version, I just see $50.
Speaker 2:I had to do some Googling to figure out how to get to the $40 version of the game. You have to scroll all the way down, go to the different pieces that are in the package and you have to specifically click on and it's not obvious that this is something you can click on, but you have to specifically click on urge shadow, the earth tree, dlc. And then there's also a fucking like song or something that comes with it and that's the ten dollars extra. So I like, clicked on that. I finally found the 39.99 skew. I was so pissed off and I wonder. I really would love to know what's the delta like? What are the proportion of people who just shoot the? Just they just buy it for 50 bucks because they don't know that the $40 skew exists.
Speaker 1:It's like Phil was saying time insensitive. There's also just like UX pain in the ass.
Speaker 3:Insensitive, ignorant, yeah, I mean, I mean the big things I used to always preach with a lot of the devs is that you want to have your DLC as a single purchase item on the Steam page. Right, because you want everything at the point of sale. You want to maximize revenue at the point of sale. You don't want things split out into those drop down menus on Steam. Like sometimes you have to say show more and you can buy the bundles. You want everything as easy as possible and you want to be able to buy it with the main skew too.
Speaker 3:Because, like people, I think sometimes the heuristic is I want the best, I want the best. And because games generate so much overwhelming value for customers relative to their price that, like the difference between 40 and $70 is relatively minor, like they're not very a lot of these games aren't very inelastic for those time insensitive customers Literally just put the money. Like this price of special editions have been creeping up, like when people talk about the price of games not moving in the last 20, 25 years when you adjust for inflation and sometimes even declining, that they're forgetting that, like we have increased price, we've just done it through special editions.
Speaker 2:Yeah, that's fair. My question for you guys is this good or bad use of monetization? And I think it's bad. I think it's hiding information and I think that it's bad for the consumer. We had had the big. Is our bots good or bad for the consumer's discussion? I'm curious your thoughts on this. If you can make it ambiguous what the true price is and then basically throw, it's almost like an experiment. It's almost like an econ experiment. You have a true value of a product that the player doesn't know and then you give them a slew of options and they have to evaluate the value of the product based on what they know about it which might be a limited set of information and choose the correct price. I had those. I had that 50, 60, $80 skew option. I didn't even have the $40 option. But to me it's a bad thing and I think if you were to run an experiment, people would almost always click the cheapest available option, assuming they get the product.
Speaker 1:It's a similar question to that is free-to-play games. Good, right, it is allowing the firm to price discriminate more effectively, better for social welfare.
Speaker 3:What are they hiding, Chris? I guess I don't understand what's being hid in those different SKU options. Different SKUs, they're different value propositions. That seems fine.
Speaker 1:He knew what he wanted. He wanted the base DLC and nothing else, at the lowest price.
Speaker 2:And the only reason I didn't purchase the 50 is because someone told me it was supposed to cost 40. And that was a human being who told me that I never saw an advertisement for go buy Elden Ring DLC $40. I never saw that. I just knew that the Elden Ring DLC was coming out and it was basically like Elden Ring 2. So $50 price tag would make sense. The only reason I didn't purchase it is because I knew it was supposed to be $40.
Speaker 3:What was in the $50 SKU that wasn't in the $40 SKU?
Speaker 2:I think it was some sort of soundtrack.
Speaker 3:Okay so they're basically trying to create a special edition of the DLC to push the average transaction size higher and the one you actually wanted. So they used a bundling strategy to push average transaction price higher and they made information to acquire the $40 base SKU more expensive to acquire.
Speaker 2:Is that ethically?
Speaker 3:wrong. Is that the right way to think about it?
Speaker 2:I think that's a pretty good way to put it. I think it's wrong. Here's the cherry on top for me, when I clicked on the big Elden Ring DLC banner, only three SKUs showed up, and none of those SKUs were the $40 version. Each one was a different type of bundle. There was no. There was this. When you give me options, I assume that those are all the options. When you give me three different versions of a type of product I can buy, I kind of those are. Probably. If I go to the Apple website to buy an iPhone, I have a choice of color and a choice of the storage capacity. I don't expect that there are other options that aren't secretly behind the scenes that I need to search for. You know what I mean. So to me that's really. I definitely agree with your statement. They made it more expensive for me to discover the different SKUs, but to me, the way they did it was like not economic, was manipulative I guess what I see this feels like choice architecture for me I would go back to is the.
Speaker 3:What I would argue is like good ethical principles that every game economist should subscribe to don't lie, don't cheat and don't steal and it's not the case that those are easy to apply everywhere. They're simple rules but it gets complicated and there's murky water. And what I would argue here is I think it becomes a lie. If they're going out and they're advertising that $40 price and they're making it very expensive for you to find out how to actually acquire that $40 price, that to me starts to become a borderline lie. That to me feels like manipulation. If it's very costly to acquire it like it's, it's a tough argument to make, because is that the margin?
Speaker 3:And the philosophers would beat us over the head with a, with a bludgeon, because, hey, usually they don't think at the margin. And hey, does something become unethical or ethical just based on a marginal analysis of it's? Okay, if they had moved it above the fold or they put it in this place, does it suddenly become ethical? But I would argue it only becomes unethical if they were swinging hard at the price being $40 and they were advertising it. I don't think that was the case. I rarely see price in a really bold format for a DLC package. Certainly not.
Speaker 2:Anyway, I think it's a really interesting discussion To me. You take that classic what's the choice architecture book they like talk about cast sunstein and failure.
Speaker 2:Nudge, nudge, nudge. The nudge framework is put the product that you think that the person should have as the default up. You walk into a grocery store and the cheerios are at eye level and the off-brand is at the bottom. Now, in my opinion, xbox didn't even put the original the, the base skew, on the shelf. It was in the back and I had to go ask the manager to take it out for me. That's my problem.
Speaker 3:To me, something that would be clearly unethical is like advertising something in a heavily discounted price but having a very limited quantity of it in stock like you only have one in stock, but, of course, like it comes back to the margin okay, what if you only have 100 in stock? What if there's 200 in stock? Like you only have one in stock, but, of course, like it comes back to the margin Okay, what if you only have a hundred in stock? What if there's 200 in stock? Like it becomes like a lot of this has to go back to intent, right, which is such a fuzzy concept. Has to go back to intent. What was the intent?
Speaker 2:There's infinity in stock right. Zero marginal cost product here.
Speaker 3:True, but in the case of the X, the example I, yeah, chris, what have you been playing? Elden rings?
Speaker 2:I told you I, I was, I was playing bellatro, but I also fell victim to the multimedia transmedia, to a transmedia thesis and after watching, after finishing fallout, the tv show on I think it was on Amazon Prime I enjoyed that a lot and started playing Fallout 4 and I've been enjoying it. It's free on Xbox Game Pass, so it's a no brainer. There's a lot of old school games. I think I'm going to get into the Gears series next. I used to play that when I was a kid. Love Gears of War.
Speaker 3:Don't do that for yourself.
Speaker 2:no, that's never been good since epic passed it off to the coalition oh, like gears 5, I wasn't gonna play gears 5, I'll probably play like gears 2 yeah, there we go, now you're cooking.
Speaker 2:But yeah, no, I fell victim to the transmedia thing and I was like I'm gonna go play this and I've tried playing gears but not gears fallout before and I didn't really enjoy it. So we'll see. This is the beauty of xbox game pass I can churn which have no more than a three-day retention and go on and download Gears 3 or 2. I've chatted with some friends and they're like, yeah, it's still worth it for me. And I asked them I was like at what point is it not worth it anymore? Because Xbox is going to find that. They're going to find it. They're going to find that price point. If they have any decent data scientists or game economists on their staff, they're going to go and try and find out.
Speaker 3:It's so tempting, once you're a subscriber, once you're a subscription-based business, to basically look at your MAU, realize that your MAU is not really growing and just say, if I raise price by $1, it's MAU times price increase. Oh, that's a lot of money. Okay, what if we multiply it by $2? What about $3? Okay, what's the result in churn? And it's really hard to argue the result in churn is going to be bigger than the increase in price. The only problem is new customers facing that higher price that you struggle with. So you've got to figure out how you're going to discount that. You're going to get people into it.
Speaker 3:And the churn effect is often delayed as well. So you can, you really, when it really hits this is enough.
Speaker 2:Netflix playbook aren't they mostly monthly anyway, though? Nowadays netflix is monthly for sure.
Speaker 1:I think it's more that when do you notice, let's say, xbox increases the price by two dollars? When are you going to notice that?
Speaker 2:true, that is true, don't you think? Yeah, geez, this is just the ethics podcast now. Don't you think you should get a fucking notification like to your bank account? Hey, chris, I actually do get notifications. If a subscription increases, chase will message me and be like hey, like you used to get fantastic to see there's. I guess we call that the nudge strategy.
Speaker 3:There's now a nudge counterattack and it makes sense, right? You should not be spending resources on thinking about that. It's far more cost efficient for someone else to tell you that this is happening rather than for you to monitor it. Monitoring costs are way too high. Oh, I totally agree.
Speaker 2:Yeah, that's counter free market, though it's not the consumer's job to figure out what they're paying Worst experience I had was with fucking Coinbase.
Speaker 1:Coinbase signed me up for a $30 a month subscription Yep For a product I never paid for. I never asked for this thing, literally made zero trades with it, found out a year and a half later, fucking livid.
Speaker 2:Oh, my God yeah.
Speaker 3:Fuck Coinbase. I'm on that team too. I really fucking hate.
Speaker 2:Coinbase. I get like a notification every fucking month that I have USDC rewards and I'm like fuck, I have to like I'm going to get a tax document for this bullshit, because I have a dollar in your fucking bank account and you're giving me 5% savings. Yeah, it's annoying.
Speaker 1:Phil, what have you been playing these days?
Speaker 3:I am back. I am back on the Apex Legends train and I've also been dabbling in a little bit of Overwatch as well. My fascination with Apex goes way back the game and really think of it as an acquisition platform. Because we know, like acquisition is one of the most costly things right now in gaming. And so how can you take the mass amount of players playing a given franchise and kind of grow and expand the hours and audience? And so Warzone was one of the most successful examples of this.
Speaker 3:So they took base Call of Duty, remember. They added in zombies, which is a co-op mode which is coming back, although that one didn't really didn't seem to work to the same heights. But then they added in the Battle Royale mode. They also added in an extraction mode. So they're starting to bolt in all these modes in Call of Duty and kind of see what works. What can we point our user base to and see if it's successful. And Apex, which I think was one of the first ones to do this, they added a. This was a couple of years ago. They added a Counter-Strike ish mode where you had in-round progression, although the in-round progression was pretty weak and it never really did that well, and they've added Team Deathmatch as well.
Speaker 3:So they're trying to add these different modes, but they're just not capturing a significant share of users, and I still think it's the right strategy. I still think it's the right thing to do. The problem is that a lot of game teams are afraid to kill things, and it sounds really tempting because once it's already created, it feels like it's zero marginal cost and it's a sunk cost as well. But what happens is very similar to politics that you create this cottage industry within your firm when you launch these modes of people who are vested interests in this thing continuing to survive, because usually there's a lead designer that's appointed to a given mode, when, in fact, if it's not working, you either need to pivot pretty quickly or you need to kill it, because if you continue to let this thing run at only 5% of the AU engaging with it, it becomes a drain on resources, it becomes a drain on the engine. There's just a lot of these hidden taxes which I think sometimes large corporations have a hard time seeing, but they can really slow down development. But I still think the strategy of taking these given games and using them as platforms to launch other games is super interesting.
Speaker 3:The thing I was talking to a colleague about was that mods are pretty much dead. Valve's entire portfolio came from mods CSGO, dota, team, fortress 2. Like almost all of their games came from mods, even their failed games. The auto chess game was originally a Dota 2 mod. Amethy artifact was one which explains a lot, but all mods have essentially died. What games are being spawned by mods anymore? There's none of them Fortnite, pubg. What do you mean by?
Speaker 1:that. Yeah, it was an Arma mod, then it was an H1Z1 mod and then it was PUBG, but that's only been 10 years.
Speaker 2:What was Fortnite a mod of? Or are you just saying?
Speaker 1:Fortnite copied.
Speaker 3:PUBG oh okay.
Speaker 3:That was also using it as a platform, right, because that started as a base-building Defender game. That pivoted into Battle Royale, but I think Roblox has really killed it. That pivoted into battle royale, but I think roblox has really killed it. Roblox and, to a much lesser degree, fortnite, creative. But I'm just surprised we haven't seen more games come out of mods. It feels like there's less of it today than there was 10, 20 years ago and it feels the mood is that shouldn't be the case. It should be the exact opposite. I'm just surprised by that.
Speaker 1:I don't think we have a good explanation for that yet I've thought a lot about this because I worked on TFT, which did come out of a mod, but you're right, it's much rarer now. I think it's just game development tools are way better now, so a lot of the people who would work on a mod just make an indie game right. Oh.
Speaker 3:I really like this explanation, Eric, but you would think they would use it to acquire an audience, right? Because if it's a mod, you already have a built-in user base right that built-in user base right, that's true, but you also capture very little of the upside.
Speaker 1:Yeah, that's also true because it's all free. But on the flip side, I think a lot of people are making the same observation you are and they're like, maybe we just need to make the platform for it.
Speaker 3:That's fortnite creator and roblox and all that, and there is a company called modio which is trying to be a third-party service which attaches onto a game that you can plug in and create a modding ecosystem very quickly. Because one of the challenges with with a lot of these mods is that it's a pain in the ass to actually put them in a game because they're usually not explicitly developer supported if you go in and try to build a witch, to try to integrate in a witcher mod. So after I watched the witcher tv series, okay, I'm gonna get into the witcher game series. And what was the first thing I wanted? I wanted henry cavill's face, you know, as a Witcher, and it was like just a lot of hoopla to get this thing in and I ended up churning from that experience Like anytime you make me go into my C drive and I'm in program files I'm peacing out. I don't fuck that either.
Speaker 2:I have an alternative theory. The labor force of people who were doing mods is now just basically integrated with the actual companies themselves. There was this fringe and they're now just assimilated into the actual companies themselves, and what used to be a mod is now just like a feature update.
Speaker 3:So basically the game industry has grown so much that they basically just absorb people. There was an outstrip of supply relative to demand. Now demand has caught up to supply.
Speaker 2:Exactly. I liked your post on somebody's LinkedIn, your comments on somebody's LinkedIn posts, just like the share of employees or sorry, the count of employees in the games industry over the last 20 years. I think that's just going to be the that needs to be like the blanket response for flooding the industry. Oh, cutting our jobs, killing us? Yeah, but we've also seen a hundred% growth in the last 10 years.
Speaker 1:I'm curious about this theory, chris. So this implies that it's those skilled game developers who in the past, were just making mods. They're much easier for the companies to find and hire them, and maybe the companies have bigger budgets to hire them. So what would be the mechanism behind this? Yeah, what has changed in the market that has gotten them hired?
Speaker 2:the mechanism that's gotten them hired. Uh, I think it's just that the industry has grown. They have more money, they have more disposable income and they can hire more people. And who are the people? Who are qualified, the people who are already building mods?
Speaker 1:it's really more of a selection story than a mechanism you're saying there was like a fixed number of people interesting and developing games, but the number of companies have been growing.
Speaker 2:Assume yeah, assume that the labor demand is X in time period zero and the supply is Y, or is X plus two, in time period zero and then in time period one you don't even need, so the supply is up here. You don't even need growth in the supply, the labor supply, but you ended up with growth in the labor demand. You're just naturally going to increase the labor force and so that share of non-market participants decreases because more people are actually working for the game companies.
Speaker 3:Great anecdotal evidence exactly to that point, chris, is that Frogmind I may have that name wrong is the guy who was behind some of the dota stuff, behind dota 2, hired by valve ice frog, thank you. He got hired by valve and is now working on deadlock, by the way, which is valve's third person shooter, which I just got a playtest access to, full review and coming soon. But that's really great antidotal evidence to support your point we'll take it.
Speaker 2:We'll take m equal one. And speaking of labor markets, yeah all right, let's.
Speaker 3:Let's talk about squad buff. I spent way too much time on the deconstruction of it. I made a lot of claims that ended up being, uh, not well received when the game actually launched. One of the claims I made is that this game would have historic download volume. And I made that claim because squad busters from supercell combines the ip of all Supercell games into one. And just one side note here I haven't seen anyone make like the post late stage capitalism point that basically all franchises are moving towards this multiverse future, which could not be more like late stage capitalism, stage capitalism. So we think about like mcu, because moving towards these merged universes, like we have got deadpool and wolverine coming out pretty soon and of course you have the different universes that's always been colliding. For that you had spider-man no way home that did this.
Speaker 3:I had another great example of this happening fortnight just puts every ip in there fortnight puts every ip and it has the same portal mechanic right of all these worlds colliding. There actually is a narrative to fortnight that tries to string this together.
Speaker 2:It's not right.
Speaker 3:Magic gathering is doing this. It has. Warhammer 40k is now showing up iju cards or, excuse me, kaijus are now showing up in the game. Overwatch is now doing a transformers bundle. Like everyone is just like saying fuck it to their ip and saying let's put all the toys together and see what happens. And squadbustersusters is, in many ways, a perfect example of this, because it's not just a stunt.
Speaker 3:The whole IP is supposed to be a reflection of all of Supercell's IP, and I thought Supercell was really building brand equity, really thought they were building brand recognition. One of the things that Rob Lowe, who's a marketing executive at Supercell, mentioned in an interview is that the Barbarian from Clash of Clans is one of the few recognizable characters from mobile IP. And when you think about kind of Western mobile IP, it's not in a good place. Like it's been very hard for mobile to build IP. Like when you think about, okay, what is the mobile IP that's been really successful. We got Rovio. Angry Birds was successful like maybe 10 years ago, but the best that could do in terms of branching out beyond mobile was like a mediocre film series. In terms of branching out beyond mobile was like a mediocre film series and then the only other one I could really come to mind is a failed Mario Lopez Candy Crush TV show that'll never let King live down.
Speaker 3:You guys really missed this. You haven't seen clips from this Mario Lopez. It's a game show. You're like, how would you do a Candy Crush game show? And if you're?
Speaker 2:curious about that question. Candy Crush has no IP. I don't understand.
Speaker 3:Like there's a fucking bean like a jelly bean I I know it's hilarious, like japanese game show style, like they have the board and they're like crawling on the board trying to match gems boy. It's embarrassing. And there's been like a little bit of subway server stuff if you're familiar with that franchise. They're like I'm trying to do an animated series but relative to the amount of revenue and amount of downloads and active users that all of these IPs have generated, like this is the best we can do. It's embarrassing, but I did think Supercell is different than that and what I expected is a lot of that IP to drive so many downloads. And it turns out. So I want to do analysis and figure out hey, what's going on? Why isn't this working? I ended up developing an R package for this which should be out soon I won't be in the show notes, but maybe next week.
Speaker 3:So what we're looking at right now is we're looking at Squad Buster's download collapse and what we're doing is we're benchmarking it against the other three Supercell titles which are relevant enough in terms of download. So if you think about Hay Day and Clash of Clans, those are not included here because they had very weird launches almost a decade ago, I think in 2013,. They actually had staggered launches on iOS and Android a year apart. So I chose not to include this in the analysis because too much noise and isn't really relevant. So Squadbusters started with a huge number of downloads 8 million downloads about on its first day, which is way more than Clash Royale during its launch, way more than Brawl Stars during its launch and way more than Boom Beach during its launch. But as you can see in this graph, it collapses so, so quickly, so quickly collapses. And we look at this as a running sum. You can really see the cumulative download curve bend inward on day five, whereas if you look at Clash Royale or Brawl Stars, there is no such bend. It looks like it's not linear. There is a bend, but it doesn't bend like a twig like Squadbuster does on day five.
Speaker 3:So that's your first clue that hey, something might be weird here. This isn't what we'd normally expect, and so what you end up finding when you look at daily active users and all of this data is pulled from Sensor Tower is that there is a huge decline in daily active users in Squadbusters during the first 10 days. There's a rapid decline, and when you benchmark that against clash royale and brawl stars, they are stacking dau. In their first 10 days they're actually growing and they start to level off. Really, around day 15 they find an equilibrium, whereas squadbusters continues to decline and decline and decline and that is a huge problem. Not just that they're declining and not stacking DAU, but if you also just look at the relative values here Clash Royale and Brawl Stars by day 30, they're hitting their equilibrium at around 12.5, 15 million DAU, squadbusters around 2.5. So huge difference in magnitude and a very different slope between the two games. So the plots are in thick in here. Right, very different launch, very different numbers, dau not stacking just like troubling signs. And then what you do is you compare it on a absolute basis. So we're just looking at the first 30 days since launch.
Speaker 3:But let's just look at the Supercell portfolio and judge it relative to where the Super is right now, and this to me is the most damning graph of all. Squadbusters is the lowest engagement game out of Supercell's entire portfolio, behind even Hay Day, which is still running. By the way, hay Day has been going on for over a decade is impressive, but it's the worst performing game in their portfolio and it doesn't have any impact. It's just not doing anything, so extremely troubling so far. Another damning graph here is if you just looked at it from a portfolio level. So if we summed up all the games in Supercell's portfolio, all of the DAU together, is the portfolio increasing or decreasing since Squadbusters launched? Because the other thing that could happen is cannibalization. Right, other players from other Supercell games might turn out of those games and go to Squadbusters. That could be a potential problem here.
Speaker 3:But when you just look at the portfolio DAU since Squadbusters launch, it's actually on the decline.
Speaker 3:And you might say there could be other factors. Right, you just told me, phil, that Squadbusters is only doing 2 million DAU, whereas, let's say, brawl Stars is doing like 60 million DAU. Of course, it wouldn't have any impact on the portfolio. And, yeah, that's the point, it doesn't have an impact on the portfolio, not a good sign. And so you start to think about, okay, what could be some reasons for this game kind of having this bowed out curve, this quick decline in DAU. There are a lot that we'll get to, but the first thing I want to do is look at it on a regional basis. And when you look at it on a regional basis and you divide it between tier one, tier two and tier three and you divide it between tier one, tier two and tier three and these are just different collections of country that we usually group by monetization level. So we think about tier one country Germany, united States when you think about tier two Sweden, ireland and tier three might be India or the Philippines, just regions that don't have really high monetization values.
Speaker 3:When you break out the cumulative amount of downloads for Squadbusters and all the other Supercell games between tier one, tier two and tier three, that one of the places Squadbusters is struggling in the most is tier one and tier two, and a lot of its downloads actually come from tier three. So why would a lot of his downloads be coming from tier three? What's going on here? And so one of the things that was really interesting about Squadbusters was pre-registrations. So this is a new tactic that Google Play has developed and Apple has developed I wouldn't say new, it was round-roaring brawl stars, but they're pushing it more and more. So what you can do is you can go in the App Store, you can say pre-register and then when the game comes out on Google Play, it will automatically download to your phone. And when you're in the App Store you get a notification. You just have to go in and download it. And they were pushing these pre-registrations and so many other of Supercell's titles, and not only that, they actually created a really good public goods game, almost. So what would happen if you're playing Clash of Clans? You can go in and you can pre-register for the game in Clash of Clans and if they hit certain milestones let's say 5 million pre-registration then you actually get a reward in Clash of Clans based on the number of pre-registrations for Squadbusters.
Speaker 3:So a lot of people ended up doing these flimsy pre-registrations and so what happens is that when you break down those games by region as well, you find that Clash of Clans, clash Royale and Brawl Stars all have a very significant portion of their DAU from Tier 3 regions. So Clash of Clans, to give you an example, has about 85% of its DAU is from tier three and that's the biggest game in its portfolio. So a lot of Clash of Clans users, I would argue, and Clash Royale and Brawl Stars players ended up pre-registering and ended up going into Squadbusters and that explains a lot of their tier three DAU share. And that also would go back and I would explain why so many people churn from the game so quickly is that they were pre-registration users. They were tier three, which has lower retention rates to begin with, and they have real opportunity cost, which is the other Supercell games. So they pre-registered and they went into the game and then they churned. That would be my basic explanation for why Squadbusters had so much fat trimmed very quickly versus the other titles.
Speaker 2:And they pre-registered because that was incentivized. Yep.
Speaker 3:They would get kickbacks almost. I think there's another explanation here, which is a lot of brand marketing spent which I've ridiculed as being inefficient. So they ran this really big campaign where you saw Will Arnett and Chris Helmsworth almost cosplay as Squadbusters players, and to me there's a question about whether or not this sort of brand marketing is really effective and not to poke too much at Second Dinner. But they also pulled this stuff for Marvel Snaps launch, if you remember, way back when they had this very expensive Samuel L Jackson ad where he played Nick Fury, and I heard this was extremely expensive. Samuel L Jackson is one of the most expensive actors actually of our generation and to do commercials they are extremely expensive. I heard and this is a rumor that this was $10 million to put together this creative, which is mind boggling. So are you going to get more than $10 million back in revenue is skeptical. But the whole theory of brand marketing is something that, to me, is antithesis to a lot of economic thinking.
Speaker 3:When I speak to brand marketers, it feels like the playbook is the same. They want to create this like confluence of events. You want to have all these different things triggering at once and they want to create noise and that creates what they call awareness, which is the first part of the funnel. You create awareness, then someone gets more interested in the product and we can think about all the different funnels that someone would go through. But it starts with awareness. And if you want to create awareness, you need to create noise. And if you want to create noise you need this confluence of events, you need everything to be strung together. When you think about a movie, it's the best example of this, right, it's all about that first weekend and creating as much buzz as possible and creating this confluence of events. But I think about like an economist, right? It's not clear to me that when you sum all these parts together, that the whole is greater than the sum of their parts. I actually feel like you get a lot of diminishing returns from each of these parts.
Speaker 3:For instance, I know if I was thinking about how to maximize my marketing spend, I could do a billboard, which is a stupid idea. You could do a billboard and that costs X number of dollars and that gets X number of impressions, and then I could also do like a TV ad. But what can happen is you can, like, double target a user. So not only do they see the billboard, they also see the TV ad, and I would argue that when you double target a user, the additional revenue you get from that it actually decreases. There's there's any like special interaction effect. I think you're just double paying for a user, right? So not only do you have to pay for them to see the billboard ad, you also have to pay for them to see the TV ad.
Speaker 3:Like a lot of that thinking, I think, actually strikes at the whole theory of brand marketing that this is just ineffective to start with, that there isn't these interaction effects that make the individual parts summed greater than the individual parts, that there's actually a negative effect here, that you're actually double paying for the same user to see the same ad across multiple platforms. That would be my kind of counter, my counter thesis, to brand marketing. So I think they may have misspent on a lot of brand marketing, and one of the things that I also looked into because Supercell has done this brand marketing before is they ran this infamous 2015 Liam Neeson Super Bowl ad for Clash of Clans. Do you guys remember this?
Speaker 2:This was huge.
Speaker 3:This was huge at the time and this got so much positive buzz because they got 80 million YouTube views. And they got not only 80 million YouTube views, they also got 120 million views during the Super Bowl, right, and then you look at what the impact was on downloads it was peanuts. It was peanuts their download ranking way back when this aired in 2018 during the Super Bowl Patriots Super Bowl, the Patriots versus Seahawks Super Bowl, which is fucking great. By the way, it's probably the best Super Bowl of all time was minimal.
Speaker 3:The download ranking was very marginal. If you do a lot of back of the envelope math, you think about how many downloads this drove. It's just simply not enough relative to the cost here. So I think there's a lot of reasons to be pretty skeptical on brand. And the other thing that I think was really important thinking about the role of brand was a piece by Eric Seifert. It was a brilliant individual and one of the things he really challenges is just how brand marketers think of themselves. And he says a CMO, a chief marketing officer, should own the company's outward facing growth initiatives. A lot of CMOs don't think that way. They think they should own the brand vision or the thoughts that are conjured up when customers think of the product. Really, what marketers should drive instead is actually growth Like that, ultimately, is what marketers need to drive growth. They shouldn't be driving brand vision or brand the thoughts that you think of when you think about a product. They should be driving growth. That, ultimately, is the KPI they're responsible for, not some of this artsy, fartsy stuff.
Speaker 3:When I talk to a lot of brand marketers, the things that they're most proud of are like stunts. I was talking to one brand marketer and they were doing Jay-Z's biography and they were telling me about to do Jay-Z's biography. They did this thing where they had a projector and it would project pages from Jay-Z's biography and they would project these pages at the locations that the biography was referencing. On that page Jay-Z was talking about a particular hotel in Los Angeles. That was important. They would project the page that his biography talks about that LA hotel on the hotel.
Speaker 3:But when you think about that as an economist, what were the level of impressions for that, chris? It doesn't seem very high. It doesn't seem like there'd be a very high number of impressions. So it feels like there's just a lot of stuntiness and not a lot of thinking about what happens down funnel. Is the benefits going to outweigh the cost? So I'd say, look, I think this is a marketing story. It's a tale of pre-registrations that ultimately didn't deliver, that weren't effective, and it's a tale of brand marketing that ultimately is ineffective at being able to acquire high quality users.
Speaker 2:And that to me right now is the best story, phil. One of the charts I was really excited about was the per user spent. Like ARPDAU was actually higher for Squad Busters than any other game, and I think it was either last episode of the or two episodes ago where we talked about. They're pretty aggressive in round spending. Like you can spend every single round infinitely. You can buy an infinite amount of power up and I think that there's an interesting you can't spend almost like limitless, like every round.
Speaker 2:You could spend money if you really wanted to true, but you are capped.
Speaker 3:So there are the keys you can buy, which let you open chests, and you have to keep buying more keys, but you can't. There's a limit to how many keys that you can use in a given round, so there still is a monetization ceiling per round.
Speaker 2:But still much higher cap than a traditional Brawl Stars or something like that, where there's a limited amount of money you could spend and, honestly, they don't incentivize you to spend money that much. In Brawl Stars, they give you tons of free stuff. So that's one thing that I think is a bit bullish and I almost wonder. Supercell is creating a portfolio of games. Obviously, brawl Stars, clash Royale these are like their kind of bread and butter, where a lot of their users come from, and very retentive games Like I still play brawl stars every once in a while. It's super simple, super easy to jump into. Even if I haven't played it in a while, I can jump in, play a game in three minutes. That's for a specific demographic.
Speaker 2:Is it possible that squadbusters was supposed to lock into a different type of demo that they might not have? Right? We talked about cannibalization. Does it behoove you to create a Brawl Stars killer as Supercell? Probably not. You don't want to kill off Brawl Stars, you want to access this different group of people. Now, I'm assuming the marketing spend was non-trivial for this and I don't think that this was like a low cost product, low marginal cost, low marginal benefit. But I almost wonder is there a room for a product that could still be successful, that's capturing new players, new people. That's not what's happening here, especially when you look at this pre-registration picture, but it does make you think. A giant company like Supercell with this portfolio of games and products, and I would argue I don't play Clash Royale, I do play Brawl Stars I think there's two completely different types of games that are potentially targeting different demos. Is that, is that being too nice, or is this possible?
Speaker 3:I think you're, I think a lot of your observations is right. Like the idea that you're going to launch in a game that let's say, but that you're going to launch in a game that let's say, but that you're going to launch another game and a given player is going to play two games at once is a farce. Like the probability of that happening is very low. Like the idea that you're going to stack these LTVs for an individual player is nonsense to me. So you're looking at cannibalization, and so then you're thinking about okay, what's the LTV of a given player, of that given player in one game and what's the ltv of player in the other game? And I want to funnel them to the revenue, maximize it game, like I want. Just like we say, wang goes to its highest value use. We want players to go to their highest value game and it, first of all, it's not quite that squad busters, but the more important point is they didn't even do that, like they didn't even cannibalize brawl stars. There was a small decrease in brawl stars, ball stars is dau during this period, but it's just minuscule. It's absolutely minuscule. Like to your point, chris, like when your portfolio grows to a certain level you need to think about exactly what you mentioned, like how is this game going to grow the portfolio rather than just rearrange the deck chairs? Yeah, and I think that becomes an increasing problem for a lot of different companies, like how do I grow the portfolio versus just rearranged deckchairs? And I think this game never had a really strong thesis on the target demo Like you do.
Speaker 3:Rob Lowe's the marketing executive came out and said they were the super. It's the super smash brothers in terms of all the IP combination. There's actually a lot of parallels to Mario Kart as well in terms of the rubber banding, it being accessible and it fits into the kids category, being for kids, just like Brawl Stars is. I was in Rome recently and it's crazy how many kids I see with Brawl Stars swag. But that, to me, doesn't grow the portfolio right. If you're going after the same demo like, you need to go after different types of users. Like Hay Day is an economy management game that goes after females. It's very different than Supercell's other portfolio. So I would say it's the same thing for any company that has like so many different planks. It's like, at some point, this is how you do imperialism right. You go after different things, rather than just capitalize your own portfolio.
Speaker 2:And it didn't just have a weak marketing thesis. I think it has a weak game design thesis. I don't know what Squadbusters is. I've played a few games, I've tried to get into it, I've heard what Eric has to say about it and I'm not convinced that this is a game with a clear vision or a clear thesis. It's like when you're a company this size, what was this game for? Who was this game for? To me it felt like Brawl Stars, but I was like I said this on day one when we first started talking about Squadbusters. I was like I don't like Brawl Stars, but I prefer Brawl Stars over Squadbusters because it's simpler, cleaner. What did you ultimately attribute the poor performance to? Is it just a bad game?
Speaker 3:I think there's in a debate between when we think about Leaky Bucket is something marketers like to say I'm pumping a bunch of users into something and they're churning them, so I would attribute this to not finding us. First of all, they shouldn't have launched. They didn't go through soft launch. That's the whole point of the article. More than anything else is that you got to go through soft launch to figure out whether or not you should scale this game. Like KPIs and soft launch are predictive of games at global. They are Like there is some scale effects, but you can use some pretty reasonable models on this.
Speaker 3:To me right now, I think, more than anything, the game's just feature short. The game's feature short. They went heavy on brand marketing and pre-registrations are full of shit. They shouldn't have done as many pre-registrations as they did. They pushed that way too hard. I don't even know if you want to do pre-registration, like who wants someone from Ball Stars to go to this game Exactly what we were talking about earlier. They might have lower LTVs, like it's not clear to me that's what you want to do. So I think pre-registrations are. I'm extremely skeptical of pre-registrations, the theory of pre-registrations, and I'm extremely skeptical of how they spent their brand marketing money. Those, to me, harmed rather than helped Squadbusters. But more than anything, this game just didn't have the features that it needed to have at launch to be able to maintain a strong user base.
Speaker 2:Okay, I'd buy that, but it performed abysmally like it's bad. I'm not convinced of the like the pre or the this. Do you think they could have fixed the problems that they had done soft launch beforehand?
Speaker 3:I think you just need more time to add more features.
Speaker 3:There's no social features there's no social pvp is is just lackluster. It almost is non-existent. It's a very flimsy leaderboard. They need time to get the level four evolutions in which would have increased spend cap quite a bit. The LiveOps playbook is really basic right now. Like I play this game religiously and to me the LiveOps are like sometimes they have this card thing and sometimes, like I get out of a round and it says I can flip a card and I get a couple of resources and I can flip another card. That is their live ops right now. That's it, it's all they have. This game needed more time to really hone its identity and just deliver on the basics, like.
Speaker 3:One of the other things that was really interesting to look into is like, for the first seven days of this game's launch, the ratings on the app store were abysmal. So the average rating was around like a 3.5 in the United States, which is really poor. Usually you need four or above to get featuring on the App Store and this was really low and I was like, oh my God, what's happening? Why is this so low? Usually it's a technical problem. Usually what happens is that you didn't whitelist the right devices.
Speaker 3:So people who are like in India, who tend to have lower processing power, will download the game and then it runs really poorly and they review it a-star review. But it turns out the problem with this game is that they didn't do review prompting. Review prompting is when you get a message in game that says, hey, if you like this game, you should leave a review on the app store, and they only implemented that. I think it was 15 days after launch and you can very clearly see the change from having a really poor one-star reviews to having a ton of five-star reviews. Not only does that share change, but the N increases, so they go from having 250 reviews a day to having 2,000 to 5,000 a day when you review prompting. That, to me, is just a sign that this thing was not ready to go.
Speaker 2:Because the people who aren't being prompted are typically dissatisfied.
Speaker 3:Yes, disproportionately satisfied, they self-select into bitching.
Speaker 2:Yeah, exactly.
Speaker 3:It's review, bombing on steam almost. It's interesting. All right, let's talk about our first paper in our new section. The paper is called one cohort at a time a new perspective on the declining gender pay gap. This was a fucking great paper in my opinion, for so many different reasons. One it was only 29 pages, which is pretty unusual for econ papers these days. Usually they're was only 29 pages, which is pretty unusual for econ papers these days. Usually they're about like 80 pages long. This is pretty reasonable.
Speaker 3:Not only that, this had a real result, that was applicable not only to policymakers, but I think also to NGOs and to a wide variety of people, and I thought the analysis was rather convincing as well.
Speaker 3:And so what is this paper actually about? So the paper tries to explain the gender pay gap between men and women, the differences in wages between both of them, not necessarily as a result of any discrimination, but instead the theory that they offer is that there's actually this effect where a lot of older men are not leaving the workforce at the rates they once did, and so it's creating a bottleneck in wages. So a lot of younger men are getting to the point where they should be going to higher level careers that pay more, and instead they're not able to move to those careers because they're preoccupied by men who are older, and so what that means is that they're in lower paying jobs than they might otherwise be in, and so that is actually what's closing the gap. It's that a lot of these men are actually getting paid lower than they would previously, and that's actually converging with where women are, and we'll now start to go into some of the evidence for this.
Speaker 1:I just made a connection with the Biden situation. Old man doesn't want to leave office.
Speaker 3:What is it? The genocracy? What's the word again?
Speaker 1:Gerontocracy. Gerontocracy Rule by boomers.
Speaker 3:The paper uses data from the USA and from Italy, which is pretty interesting. The first graph that we're looking at here is just the gender gap in weekly earnings for both the United States and Italy as a time series, and it starts at 1976 and it goes to 2020. And against all good reason and judgment, they made this a dual access graph, which drives me fucking insane because there's so many tricks you can play with dual access. But by and large, the trend is pretty similar. You can see as a time series, the gender pay gap is radically declining almost linearly in both the USA and Italy, although at different levels from 1976 to 2020.
Speaker 2:Notice, the gender gap is smaller in Italy, which is a country that prioritizes leisure time over the United States, and one of the big contributors to the gender pay gap in the United States has to do with leisure and outside of work hours. We're going to see later that the pay gap starts to expand, or sorry, it becomes larger as you grow older within a cohort, and that's largely due to women leaving the workforce altogether. That's true, I think the other, and that's largely due to women leaving the workforce altogether.
Speaker 3:That's true. I think the other thing that's important to note here when you look at the dual axis, I want to strangle someone, but when you look at the log weekly earnings for USA workers on the gender pay gap, it starts at 0.8 and it goes down to 0.3 in terms of the gap. So there was a much more radical decline in the United States on an absolute basis than in Italy.
Speaker 1:Wait, chris, can you talk more about that leisure time thing? So you're saying in Italy there's more leisure time, which means that women have fewer time pressures and therefore are more capable to work the jobs?
Speaker 2:So that's part of it. So there's a labor supply story where the protections for women leaving the workforce are better in Europe largely than they are in the United States. So you get more paid time off, you can leave the workforce for I think in some countries up to two years, like Scandinavia. You can leave the workforce for two years, though, right and get and still be on and still be on your, your salary, job or whatever. So there's a lot better kind of benefits for leaving the workforce. So that is one of the big contributors. The other thing is just that's disproportionately impacting women and then you have, like, uniformly across both genders, you just have this more interest in leisure time, more time spent on leisure in European countries than you have in the United States. So that's what's also going to just shift the kind of there's a lower top end of the distribution which shrinks the gap, if that makes sense.
Speaker 3:Because a lot of the gap isn't the high income earners, which is what Peralta supports.
Speaker 2:Yeah, sorry, just an interesting little point.
Speaker 3:And I think this next figure that we're looking at right now is the gender pay gap between and within cohorts.
Speaker 3:So what this is showing is that linear decline in the gender pay gap in the United States and in Italy.
Speaker 3:But what this also layers in is looking at the gender pay gap for people by birth year, and what it'll do is it'll separate out different generations into those cohorts.
Speaker 3:So we can say, for people born between, let's say, 1952 and 1956, how is the gender pay gap for people born in that era changing over time and different effects in both the United States and in Italy? And what you'll see is that over time the gender pay gap for the United States, as each of those cohorts is aging, starts to decline. There's almost this like complete flattening, or at least this kind of flattening, where the gender pay gap stops increasing as workers get older and older around the year 2000 for almost all of the cohorts, which I thought was really interesting. There's some socioeconomic effects that appear to be happening in the United States, whereas when you look at Italy, within each of the cohorts, as the cohorts are getting older and older, the gender pay gap is increasing in each of them, although over time that relative increase is shifting downward. So it's still the case that as the cohorts are getting older, the gender pay gap is increasing, but the absolute level of that gender pay gap has been shifting down between each of the cohorts.
Speaker 2:I'm honestly surprised they didn't see more dramatic differences, though I would argue that they do see dramatic differences.
Speaker 1:Between Italy and US Between.
Speaker 2:Italy and US. What's interesting and notable about the Italy graph is, you'll notice, within each cohort, the gap continues to increase forever. And in the US it increases and then it starts to steady off and then sometimes it even goes down, like in the case of the 1947 cohort. I would assume that this is due to the fact that in Italy you're, as a woman, once you have a child, you're much less likely to ever enter back into the workforce. They're probably much more. They're also probably more what's the word I'm looking for Single home, single income households in Italy versus the US. So that's probably attributing survival bias.
Speaker 1:Yeah, that makes a lot of sense to me. They're also starting just from a way higher point in the US literally double the pay gap in the US compared to Italy.
Speaker 3:I still think this flattening in the United States is very interesting around 96 to 2000, that the gender pay gap within each cohort it's not increasing like it is in Italy. It's flattening out against all cohorts, which suggests to me that there was some sort of socioeconomic policy happening around then that kind of helped push this along. Interesting.
Speaker 2:I don't know if I would like look at these two graphs and say that there's a definitive difference between the slope in the top and the bottom. Yeah, the US is flattening out, but so too is Italy. Italy has, potentially based on all of our previous comments about the US having so much more room to stick, to go, maybe you would expect it to be reversed. The US gap is more quickly declining than Italy, or no? Actually it makes more sense because Italy is more binary, more single-family homes. There's more room to grow or not grow for the gender pay gap, I don't know. As you normalize women entering the workforce in Italy versus the United States. It's been pretty normal for the last 40 years.
Speaker 3:So, if we go to this next figure four and this to me is the paper coming together this next figure four and this to me is the paper coming together it is looking at the positions in pay distribution for 25-year-olds the mean man at age 25, and the mean woman at age 25. It's doing it as a time series. So what we're doing is we're looking at okay, for the mean man at age 25 in 1976, where were they in terms of the percentile of wage distribution? And for the mean man, let's say, at age 25 in 2020, where are they in terms of wage distribution in that given year?
Speaker 3:So, if you're starting in 1976, the mean man at age 25 was at the 50th percentile of wages. So 50% of people earned more than that mean man at age 25, and 50% people earned less than that mean man at age 25. Whereas for women, in 1976, they were at the 31st percentile. So there was 70% of people earned more than women at age 25, and 30% earned less than women age 25. And what you can see is that for the mean man at age 25, their pay percentile is decreasing over time. So, 1976, you're starting at 50th and by 2020, you hit the 38th percentile, whereas for women it is largely stable, at least in the United States, whereas when we turn to Italy, we just see linear declines. For men it starts at 47 and ends up at 33. There's a lot less noise in there, whereas for women it starts at 33 and goes to 27. So not as quite a steep slope as men.
Speaker 1:And Italy is a downward trend because they got more old people.
Speaker 3:Yes, yes, there would be more people who are refusing to leave the workforce, more older men in head-paying positions in Italy that are refusing to leave the workforce. That would explain the relatively larger decline in male earnings at age 25.
Speaker 2:Discuss the very basics of the model with respect to the old versus young and top versus bottom. Did we talk about that?
Speaker 3:Yeah, Do we want to talk about that?
Speaker 2:Yeah, I think it's so the mechanic here why we look at the top graph and we see the. Let's just assume this is a beautiful graph and I'm not as convinced of this graph actually. But let's assume that we see the mean man at age 25 in the US go from the 50th percentile down to the 25th percentile, so they drop a 25 percentile point and then you have the mean woman stays exactly the same at. I guess that's not 25, whatever that number is, it's a little blurry. What's happening here, supposedly, is that there are disproportionately more men in high paying positions than there are women, and there's also a higher cost of employing a woman for the firm than there is of employing a man. So what this ends up doing is it creates a dynamic where, as men are sticking around in higher-paying positions, that doesn't impact women, because women were never in those higher-paying positions anyway. This they're not being impacted by this old man sticking around, because they were never impacted to begin with. However, the men are because they were in those positions. So in early times men were rising up into these high paying positions. So it's actually very similar to the black white wage gap literature, which is basically just is this, and this is my area of expertise, that's why I'm referring to it. But it's basically like a lot of the black-white wage gap is statistical. The fact that white people hold more prestigious, more high-paying positions than black people it's not necessarily that within a single firm a black person is being paid less than a white person. There's not necessarily this within firm discrimination that's happening. So it's similar with this paper.
Speaker 2:And what I like about this paper is it most gender pay gap literature is our statistical approaches, because it's obvious that women and men and blacks and whites are completely equivalent, right In terms of their skill distributions. But there's a statistical component where, for one reason or another, people end up getting different jobs. So this takes a statistical approach and it says because there are fewer, there were already fewer women in high paying positions. To begin with they weren't impacted by this aging top tier male cohort, but the top tier young males are being impacted, so their wages are decreasing. Women are staying the same. So I like it because it's a statistical approach to a question.
Speaker 2:You also have the classic approach. The classic model is also statistical and it says as an employer, I know that this woman is probably going to end up having to take maternity leave. She might never come back, maybe she has a child, and that's why we saw the gender pay gap increasing throughout the lifetime of a cohort. So I statistically discriminate against women and I hire a male for this prestigious position because I know he's not going to go anywhere, or at least I think he has a lower probability of going somewhere than somebody else. So that's like the very basics of the model. The aging male population at high paying jobs is disproportionately impacting young males compared to young females. That's the model.
Speaker 3:And to me the graph that really drove this home was looking at the workforce aging in the United States. So if you look at the mean age in the US population, so we're not looking at the workforce on the first graph, but if you look at just the US population, the mean age of the US population. So we're not looking at the workforce on the first graph, but if you look at just the US population, the mean age of the US population has been increasing. So it used to be about 32 and a half and right now the mean age is actually 38.5, which is justa function of the baby boom. You just have this older cohort which is just moving through the population increasing. So this is going to tip and go down after the baby boomers are gone.
Speaker 3:And then we look at panel B, which is the mean age of US private employees. You basically see the same effect that the mean age of US employees has been increasing. So it started in 1976 at around 37.5 years of age and now it's actually 41.5 years of age, which is fucking crazy. By the way, when you think about the mean age of a US employee, it's 41 and a half years old is wild to me. But this, to me, is just like the base level of evidence.
Speaker 3:I think we also need to get this thing off the ground, and I think they did a good job of presenting it, and to me it's pretty clear. Now they go on, they build a model, of course, and they go on to argue that this kind of bottlenecking of male wages, it explains all of the gender pay gap, and I don't know if I'm on that boat yet, but to me this is a pretty novel approach and the thing that shocked me is just how flat women's wages were over this period. I really expect to see more growth in women's wages and I'm surprised we haven't seen it to the degree we did in this paper.
Speaker 2:No, we're wrong.
Speaker 3:The gender pay gap is increasing, but I expected, like the percentile of women's wages, to be much higher than they are.
Speaker 2:This is percentile of the mean woman at age 25. Yeah, that is true. The mean female wage has increased during this time period. So don't, if you're looking at this and you're not like an econ, phd or something like that, like this isn't saying that people are getting paid less, it's just saying like percentile, you know.
Speaker 3:but everyone's getting paid more. Yeah, that that is important.
Speaker 2:I want to piggyback off of what you said, phil, with respect to this does not explain the entire diminishing of the gender pay gap, because this is a an interesting mechanism that probably explains some of it. But I don't want to throw shade because I haven't dug into the paper super, super deeply. But oftentimes you'll see a trend and it could be backed up by multiple theories. So they make a lot of really cool plots that back up what they say. But it's. This is why I'm curious where this paper will end up. Right now this is a working paper, so this is just like available for free on the internet. They wrote it, they haven't published it yet, so this is a pre-publication paper and I'm curious where it's going to end up, because they don't actually have any quasi-experimental structure. So they talk about counterfactuals at one point of the paper. But beyond just doing like large cohort counterfactuals at one point of the paper, but beyond just doing like large cohort counterfactuals, in order to really be successful in this literature and this is probably the biggest problem with the entire labor economic literature, especially personnel economics and personnel economics talks about pay gaps, black, white wage gap, female or sorry, gender pay gap, these types of things. The problem with this entire field is, in order to do a true counterfactual, you need to compare two people within firm and you can't do that in this literature. What you can do, especially with these surveys, they use the CPS, cpsacs American Community Service, current Population Survey and then I don't think they use the NLSY, but they use, like some of these big giant surveys with lots of tens of thousands of individuals and observation, which is great, and there's lots of industry and occupation details. How much do you make Even opinion questions? But the problem with them is you cannot compare two people in the same organization. So if you want to study like statistical versus taste-based discrimination in like the black white wage gap, you would literally need to follow joe and bob joe is white, bob is black when they're in the same berm and see if there's any differential. So it's not too dissimilar with the.
Speaker 2:This paper is really cool. It has some excellent visuals that really compel the theory. But you can write a theory to fit any visual visualization that you find, and so that doesn't necessarily mean that you're like the only. It doesn't mean that you're right. The only way that this paper is going to become a big hit is if it can convince the typical academic and economics that the theory is very sound and also general, because the statement that much of the diminishing in the pay gap can be attributed to this aging of the male top labor force. It's a cool theory and there's some graphs that can back it up, but there's no counterfactual that you can actually run to study this. It's very difficult to set up an experimental structure. That's not to say this isn't going to be an impactful paper.
Speaker 2:I think 20 years ago I talked about Bjork 2007 earlier. I think it's David Bjork 2007. He wrote a very similar type of paper. He had a theoretical model and then he had a bunch of really awesome visualizations that backed the theoretical model up. Now that was 20 years ago, almost Totally different environment. I don't want to say it was much easier, but it was much easier to publish those types of papers where you have a theory, you show some cross tabulations and some cohort analysis that backs up your theory. Much easier to publish that back in the early 2000s, late 19. These days I'm not so sure the types of papers that are coming out are so heavily structural and so heavily unless you were running an experiment in a small village in the Congo. You're not going to publish something like this. I don't think this is going to get super great publication.
Speaker 3:Then I'm on an identification strategy, because there's no quasi-experimental method, there's no synthetic control.
Speaker 3:There's no diff and diff. There's nothing here. We should talk more about this because there was another really interesting paper that kind of looked about how science is actually done and we always say RCTs, randomized controlled trials, are the gold standard and I bought into that. But this paper tried to argue that a lot of scientific knowledge actually doesn't come through RCTs. It comes in a variety of ways and there's still something to causal inference like this. There's still something to that that I think can advance the cause. But I would agree you're totally right. Right, like in modern academia, like no identification strategy, no rct, you don't have us-based data, these things. It makes it really hard to get your argument across the lines, sometimes for the better and sometimes for the worse no, I, I think that's where I was.
Speaker 2:I was headed with this. I think, like it is a valuable paper, it's an extremely valuable. Just the visualizations that they put together with the various different you you don't typically see I don't typically see Italian data in labor economics papers. Scandinavian data is very common. I think they did Luxembourg as well. So two pretty unique labor economies Scandinavia.
Speaker 3:This is the one thing people don't realize is like you can just in Sweden, you can go on. You can request anyone's tax returns and you can get them. Sweden, you can go on. You can request anyone's tax returns and you can get them. Like you can go on and google right now, you can get my tax returns.
Speaker 2:You see there's no gap. Uh, wage gap scandinavia? Oh, I highly doubt that's the case.
Speaker 3:I thought that's what you said no, that that actually isn't the case. There's actually a paper that identifies that, like there actually is a pretty significant wage gap in scandinavia, but again it doesn't. It's not controlling for profession choice, but you can get really good wage data, which is why you see so many swedish papers come out, as it's very easy, and finland, oldest canadian countries that have high levels of trust. There's almost no privacy here, which drive americans crazy like you can go, but you can do it, wait.
Speaker 1:But didn't you say you do a bunch of funky tax stuff with your income, so that like doesn't that? Mean like the publicly available tax return is going to be like fudged because everyone's doing funky tax stuff.
Speaker 3:No, not in that sense. It depends on how income is counted. So, like, when you request someone's tax return, you don't get like these crazy If you work two jobs, that's not going to be reflected in the public tax return that you request. You're just going to see a wage rate and it's going to merge both of those things together. So there's stuff like that that you're not going to get breakdowns on. It's what you fill in. The tax form, which is a lot of stuff, ends up getting merged. It isn't broken out like your dividends, like things like dividends get broken out, but like this we get. We could talk about scandinavian tax strategy. There's a lot we could go into in that, but by and large, it's what's reported to skataberkit, which is the tax agency here. This is why there's a great lottery paper for swedish lottery winners recently that we should talk about at some point.
Speaker 1:I guess one big takeaway I had from this paper was just like looking at the cohorts, and looking at the start and the end of each cohort is really important, right, if you just lump the gender wage gap and aggregate one trend. But if you look at within a group, a different trend and a lot of the pattern is explained by these population humps moving through the different age brackets.
Speaker 3:That's the thing with games Like look at your card. We're just looking at ArpDAO and Squadbusters. Arpdao looked really high, but it's only because they assassinated half their users.
Speaker 2:Exactly. Yeah, that's a lie. You said something earlier, actually at the beginning, before we started recording. Eric, oh, I'm surprised that even today, there are still papers that are coming out that are using ACS and the CPS and doing basically like crosstabs, cohort analysis, like this type of stuff is. It just goes to show that there's still so much out there that we haven't even looked at. With respect to these surveys and when I was finishing up my PhD four years ago now, I had this discussion it was like oh, there's not much you can do with the ACS, I wouldn't spend too much time with that. It's actually maybe, probably not true. So I think that's an interesting point. There's still so much growth to be to be had and so much to discover from these data sets that have been absolutely researched. The hell out of that's a wrap.
Speaker 3:Send us a text message. If you liked our new format. It's going. We're going to do game papers too. We're gonna do Steve Levitt's Candy Crush paper. There's another one from a friend of the cast, julian Rouge published.
Speaker 3:We're going to be going through that one, and if you have recommendations, please send us our way. We're probably going to do a little bit more Nobel Prize winning papers let's talk about that and we'll be tying it more into games too. I think there's a lot you can get out of this manager as well. All right, cool game economist cats in the can nice, bye, guys, see y'all we should teach this to our children. Economics is major.
