WEBVTT 00:00:04.938 --> 00:00:06.806 align:middle line:79% position:50% size:53% Alright, so it's my pleasure to introduce 00:00:06.873 --> 00:00:11.177 align:middle line:79% position:50% size:60% Prof. Feng Zhu, who is a professor at Harvard 00:00:11.244 --> 00:00:15.014 align:middle line:79% position:50% size:63% Business School he is an expert in digital markets 00:00:15.081 --> 00:00:18.518 align:middle line:79% position:50% size:63% especially those in which platforms bring together 00:00:18.585 --> 00:00:22.822 align:middle line:79% position:50% size:63% different types of agents so for example Uber which 00:00:22.889 --> 00:00:26.059 align:middle line:79% position:50% size:63% brings together consumers on the one hand and 00:00:26.126 --> 00:00:33.099 align:middle line:79% position:50% size:68% drivers on the other. As we all know I think markets 00:00:33.166 --> 00:00:37.704 align:middle line:79% position:50% size:55% like these are subject to substantial network 00:00:37.771 --> 00:00:40.974 align:middle line:79% position:50% size:60% effects and these create economies of scale and 00:00:41.040 --> 00:00:43.243 align:middle line:79% position:50% size:70% therefore they're prone to tipping. So, in markets like 00:00:43.309 --> 00:00:46.646 align:middle line:79% position:50% size:60% these we oftentimes will see one or two dominant 00:00:46.713 --> 00:00:50.517 align:middle line:79% position:50% size:68% firms. But professors Zhu's work doesn't stop there 00:00:50.583 --> 00:00:54.387 align:middle line:79% position:50% size:65% right he looks under the hood and he identifies the 00:00:54.454 --> 00:00:58.425 align:middle line:79% position:50% size:63% factors that explain whether platforms succeed 00:00:58.491 --> 00:01:02.295 align:middle line:79% position:50% size:60% or fail and an important contribution of his is 00:01:02.362 --> 00:01:05.498 align:middle line:79% position:50% size:73% he doesn't just assume that a platform is plopped down 00:01:05.565 --> 00:01:10.603 align:middle line:79% position:50% size:50% in a market and that it inherits all the 00:01:10.670 --> 00:01:14.140 align:middle line:79% position:50% size:65% characteristics of that rather, he envisions it he 00:01:14.207 --> 00:01:17.043 align:middle line:79% position:50% size:65% asks sort of what are the strategic decisions that a 00:01:17.110 --> 00:01:20.547 align:middle line:79% position:50% size:55% platform can make that affect a lot of the 00:01:20.613 --> 00:01:23.249 align:middle line:79% position:50% size:63% attributes that faces for example the strengths of 00:01:23.316 --> 00:01:27.454 align:middle line:79% position:50% size:63% its network effects whether consumers find it 00:01:27.520 --> 00:01:32.358 align:middle line:79% position:50% size:58% easy or difficult to multi-home so on and so 00:01:32.425 --> 00:01:33.026 align:middle line:85% position:50% size:15% forth. 00:01:33.092 --> 00:01:37.130 align:middle line:79% position:50% size:65% Extrapolating a bit without attribution to him 00:01:37.197 --> 00:01:41.267 align:middle line:79% position:50% size:55% a lot of those choices I think have clear 00:01:41.334 --> 00:01:46.105 align:middle line:79% position:50% size:63% implication for antitrust and what side of the line 00:01:46.172 --> 00:01:51.744 align:middle line:79% position:50% size:65% firms are on, as they make choices. And that makes 00:01:51.811 --> 00:01:54.981 align:middle line:79% position:50% size:63% today's talk especially relevant for us so please 00:01:55.048 --> 00:01:58.151 align:middle line:79% position:50% size:50% join me in welcoming Prof. Feng Zhu. 00:01:58.218 --> 00:02:05.358 align:middle line:85% position:50% size:48% [Audience Applauds] 00:02:05.425 --> 00:02:07.360 align:middle line:79% position:50% size:63% Thanks so much Jeff for the introduction and also 00:02:07.427 --> 00:02:09.596 align:middle line:79% position:50% size:70% for the invitation. So, it's a really a great pleasure 00:02:09.662 --> 00:02:13.099 align:middle line:79% position:50% size:63% to be here and so as Jeff mentioned I've been doing 00:02:13.166 --> 00:02:15.235 align:middle line:79% position:50% size:63% the research around the platforms for a number of 00:02:15.301 --> 00:02:18.004 align:middle line:79% position:50% size:65% years so it's really great to have the chance to 00:02:18.071 --> 00:02:21.608 align:middle line:79% position:50% size:65% share some of my latest research with you guys and 00:02:21.674 --> 00:02:25.278 align:middle line:79% position:50% size:63% also get us some feedback because much of the work 00:02:25.345 --> 00:02:28.414 align:middle line:79% position:50% size:60% now some of the work at least is still very much 00:02:28.481 --> 00:02:32.018 align:middle line:79% position:50% size:63% working progress. So feel free to interrupt between 00:02:32.085 --> 00:02:34.454 align:middle line:79% position:50% size:60% the talk by raising your hand and I'd be happy to 00:02:34.521 --> 00:02:37.924 align:middle line:79% position:50% size:65% hear your thoughts on some of the work I have done. 00:02:37.991 --> 00:02:40.660 align:middle line:79% position:50% size:65% And also for this talk I'd like to kind of take the 00:02:40.727 --> 00:02:44.998 align:middle line:79% position:50% size:65% opportunity to present multiple research projects 00:02:45.064 --> 00:02:48.768 align:middle line:79% position:50% size:50% related to this idea around the platform 00:02:48.835 --> 00:02:53.106 align:middle line:79% position:50% size:60% performance. And many of these projects that have 00:02:53.172 --> 00:02:55.942 align:middle line:79% position:50% size:63% working papers on my own website so if you want to 00:02:56.009 --> 00:02:59.312 align:middle line:79% position:50% size:65% know more about individual papers feel free to email 00:02:59.379 --> 00:03:01.881 align:middle line:79% position:50% size:63% me or go to my website to download these papers. 00:03:01.948 --> 00:03:05.685 align:middle line:79% position:50% size:65% So many people will say that today we're living in 00:03:05.752 --> 00:03:08.354 align:middle line:79% position:50% size:50% the age of platforms because as Jeff also 00:03:08.421 --> 00:03:09.155 align:middle line:85% position:50% size:33% mentioned it. 00:03:09.222 --> 00:03:13.159 align:middle line:79% position:50% size:65% So like examples like Uber, Upwork, Airbnb, iOS, 00:03:13.226 --> 00:03:16.796 align:middle line:79% position:50% size:58% Android, EBay, Care.com these are all kind of 00:03:16.863 --> 00:03:19.232 align:middle line:79% position:50% size:63% platforms we interact for many people even on the 00:03:19.299 --> 00:03:20.066 align:middle line:85% position:50% size:30% daily basis. 00:03:20.133 --> 00:03:24.270 align:middle line:79% position:50% size:63% So a network clustering can also put more kind of 00:03:24.337 --> 00:03:26.339 align:middle line:79% position:50% size:65% structure world describing platforms so when we're 00:03:26.406 --> 00:03:28.074 align:middle line:79% position:50% size:60% thinking about platforms we're really thinking 00:03:28.141 --> 00:03:30.176 align:middle line:79% position:50% size:53% about like market intermediaries right. 00:03:30.243 --> 00:03:33.479 align:middle line:79% position:50% size:65% So market intermediaries a middle man here connecting 00:03:33.546 --> 00:03:35.248 align:middle line:79% position:50% size:53% different groups of customers right there 00:03:35.315 --> 00:03:37.951 align:middle line:79% position:50% size:65% should be different groups of customers and as a 00:03:38.017 --> 00:03:41.020 align:middle line:79% position:50% size:58% result of existence on this intermediary right 00:03:41.087 --> 00:03:43.923 align:middle line:79% position:50% size:65% these groups of customers can interact directly with 00:03:43.990 --> 00:03:46.759 align:middle line:79% position:50% size:65% each other right so in the case of Uber because of 00:03:46.826 --> 00:03:50.463 align:middle line:79% position:50% size:63% existent Uber riders can interact with drivers and 00:03:50.530 --> 00:03:53.833 align:middle line:79% position:50% size:55% they interact directly with each other and 00:03:53.900 --> 00:03:56.769 align:middle line:79% position:50% size:73% it turns out that if you look across all different kinds 00:03:56.836 --> 00:03:58.938 align:middle line:79% position:50% size:58% of platform markets and you suddenly start to 00:03:59.005 --> 00:04:01.174 align:middle line:79% position:50% size:50% realize actually the performance of these 00:04:01.240 --> 00:04:04.277 align:middle line:79% position:50% size:45% platforms can vary significantly. 00:04:04.344 --> 00:04:07.280 align:middle line:79% position:50% size:63% So for instance Uber even though it's like a public 00:04:07.347 --> 00:04:09.549 align:middle line:79% position:50% size:60% company you would think it's a dominant platform 00:04:09.616 --> 00:04:12.952 align:middle line:79% position:50% size:60% with very high valuation like $68 billion dollars 00:04:13.019 --> 00:04:16.155 align:middle line:79% position:50% size:48% but the Uber still losing money today. 00:04:16.222 --> 00:04:18.157 align:middle line:79% position:50% size:50% So I will often tell students like one 00:04:18.224 --> 00:04:20.994 align:middle line:79% position:50% size:63% interesting example is that okay if you go on to 00:04:21.060 --> 00:04:23.963 align:middle line:79% position:50% size:53% the street, give away hamburgers for free 00:04:24.030 --> 00:04:26.833 align:middle line:79% position:50% size:60% you can practically kill McDonald's but then 00:04:26.899 --> 00:04:29.469 align:middle line:79% position:50% size:63% questions is whether this business model is a good 00:04:29.535 --> 00:04:32.839 align:middle line:79% position:50% size:53% business model or not right or is Uber like 00:04:32.905 --> 00:04:36.809 align:middle line:79% position:50% size:63% doing that right yeah but if you think about other 00:04:36.876 --> 00:04:38.711 align:middle line:79% position:50% size:65% sharing economy companies when we talk about sharing 00:04:38.778 --> 00:04:41.247 align:middle line:79% position:50% size:60% economy companies two examples would typically 00:04:41.314 --> 00:04:45.084 align:middle line:79% position:50% size:65% mentioned would be Uber or an Airbnb right but Airbnb 00:04:45.151 --> 00:04:48.154 align:middle line:79% position:50% size:65% right is a private company whose value around $40 00:04:48.221 --> 00:04:50.490 align:middle line:79% position:50% size:65% billion dollars but if you talk Airbnb it turns out 00:04:50.556 --> 00:04:53.259 align:middle line:79% position:50% size:53% that Airbnb is extremely profitable. 00:04:53.326 --> 00:04:55.695 align:middle line:79% position:50% size:63% Why one sharing economy company is losing so much 00:04:55.762 --> 00:04:58.998 align:middle line:79% position:50% size:55% money and is actually expecting to lose more 00:04:59.065 --> 00:05:01.467 align:middle line:79% position:50% size:63% money going forward but the other one is actually 00:05:01.534 --> 00:05:03.870 align:middle line:79% position:50% size:63% very profitable, and the TaskRabbit that you would 00:05:03.936 --> 00:05:06.572 align:middle line:79% position:50% size:65% argue is also kind of sharing economy company to 00:05:06.639 --> 00:05:10.109 align:middle line:79% position:50% size:60% some extent right and was value at around $120 00:05:10.176 --> 00:05:13.179 align:middle line:79% position:50% size:63% million dollars and later on got acquired by IKEA. 00:05:13.246 --> 00:05:17.417 align:middle line:79% position:50% size:58% And, Groupon used to be valued at $13 billion 00:05:17.483 --> 00:05:20.887 align:middle line:79% position:50% size:63% dollars at IPO, today if you look at its valuation 00:05:20.953 --> 00:05:24.290 align:middle line:79% position:50% size:63% it's only like 10% of it $1.4 billion today right. 00:05:24.357 --> 00:05:28.494 align:middle line:79% position:50% size:63% And we all know for our own kind of conversations 00:05:28.561 --> 00:05:31.864 align:middle line:79% position:50% size:63% with friends or reading from newspapers like many 00:05:31.931 --> 00:05:34.233 align:middle line:79% position:50% size:48% platforms startups failed to scale up. 00:05:34.300 --> 00:05:36.769 align:middle line:79% position:50% size:65% Lots of platform companies failed even though you 00:05:36.836 --> 00:05:39.138 align:middle line:79% position:50% size:63% think there would be network effects right for 00:05:39.205 --> 00:05:42.075 align:middle line:79% position:50% size:65% these businesses. And also many incumbent platforms 00:05:42.141 --> 00:05:43.876 align:middle line:79% position:50% size:55% such as like Uber they actually struggling 00:05:43.943 --> 00:05:46.713 align:middle line:79% position:50% size:63% essentially to maintain their market positions to 00:05:46.779 --> 00:05:49.549 align:middle line:79% position:50% size:63% kind of generate positive returns for their 00:05:49.615 --> 00:05:50.817 align:middle line:85% position:50% size:28% businesses. 00:05:50.883 --> 00:05:52.952 align:middle line:79% position:50% size:55% So much of my research right in the past few 00:05:53.019 --> 00:05:55.588 align:middle line:79% position:50% size:55% years were trying to understand like why do 00:05:55.655 --> 00:05:57.890 align:middle line:79% position:50% size:55% some platforms perform so much better, so 00:05:57.957 --> 00:06:02.028 align:middle line:79% position:50% size:70% profitable and thrive, while other platforms couldn't 00:06:02.095 --> 00:06:05.465 align:middle line:79% position:50% size:60% do that and I think this work in the end helps us 00:06:05.531 --> 00:06:07.667 align:middle line:79% position:50% size:58% understand like whether some of the platform 00:06:07.734 --> 00:06:09.802 align:middle line:79% position:50% size:60% markets are evolving towards winner takes all 00:06:09.869 --> 00:06:13.072 align:middle line:79% position:50% size:55% but others are not and whether some platform 00:06:13.139 --> 00:06:15.141 align:middle line:79% position:50% size:58% markets are more contestable than others 00:06:15.208 --> 00:06:18.478 align:middle line:79% position:50% size:68% so this has implications on barrier to entry whether 00:06:18.544 --> 00:06:20.747 align:middle line:79% position:50% size:60% some of the markets have high barrier to entry or 00:06:20.813 --> 00:06:25.618 align:middle line:79% position:50% size:68% not. And most of my work in the end focuses very much 00:06:25.685 --> 00:06:29.522 align:middle line:79% position:50% size:58% on the essentially the structure of the market 00:06:29.589 --> 00:06:30.757 align:middle line:79% position:50% size:60% because when we think about the platform 00:06:30.823 --> 00:06:32.692 align:middle line:79% position:50% size:65% markets, we're really thinking about essentially 00:06:32.759 --> 00:06:34.427 align:middle line:79% position:50% size:70% a network right, platform being in the middle 00:06:34.494 --> 00:06:37.663 align:middle line:79% position:50% size:50% connecting consumers to kind of a service 00:06:37.730 --> 00:06:39.866 align:middle line:79% position:50% size:58% providers or app developers on the other 00:06:39.932 --> 00:06:43.402 align:middle line:79% position:50% size:65% side, essentially building a tools idea network. 00:06:43.469 --> 00:06:46.472 align:middle line:79% position:50% size:63% It turns out that many of the network properties, 00:06:46.539 --> 00:06:49.942 align:middle line:79% position:50% size:60% characteristics actually would be able to affect 00:06:50.009 --> 00:06:53.379 align:middle line:79% position:50% size:60% the performance of these platform significantly 00:06:53.446 --> 00:06:55.748 align:middle line:79% position:50% size:60% So, in this talk I'd like to kind of work 00:06:55.815 --> 00:06:58.751 align:middle line:79% position:50% size:65% through just a few network properties we have 00:06:58.818 --> 00:07:04.056 align:middle line:79% position:50% size:63% identify in our research, me and my collaborators. 00:07:04.123 --> 00:07:06.626 align:middle line:79% position:50% size:53% And it turns out that these five kind of 00:07:06.692 --> 00:07:08.961 align:middle line:79% position:50% size:60% factors right, they have actually very strong 00:07:09.028 --> 00:07:13.699 align:middle line:79% position:50% size:65% influences on platform performances. So the first 00:07:13.766 --> 00:07:19.138 align:middle line:79% position:50% size:58% one is actually kind of common to economists is 00:07:19.205 --> 00:07:21.407 align:middle line:79% position:50% size:63% around the network effect but here I'd like to kind 00:07:21.474 --> 00:07:24.977 align:middle line:79% position:50% size:60% of also emphasize is not just the presence of 00:07:25.044 --> 00:07:27.079 align:middle line:79% position:50% size:60% network effects it's actually the strength of 00:07:27.146 --> 00:07:29.816 align:middle line:79% position:50% size:58% network effects that matter significantly in 00:07:29.882 --> 00:07:32.318 align:middle line:79% position:50% size:58% driving the competitive dynamics on network 00:07:32.385 --> 00:07:34.520 align:middle line:79% position:50% size:63% structures right. So when we think about network 00:07:34.587 --> 00:07:37.123 align:middle line:79% position:50% size:63% effects right typically in platform setting there 00:07:37.190 --> 00:07:40.259 align:middle line:79% position:50% size:65% could be like three types of network effects, one is 00:07:40.326 --> 00:07:41.661 align:middle line:85% position:50% size:58% Direct Network Effects. 00:07:41.727 --> 00:07:45.431 align:middle line:79% position:50% size:58% Direct Network Effects arise when the platform 00:07:45.498 --> 00:07:48.768 align:middle line:79% position:50% size:53% enable like basically communication across 00:07:48.835 --> 00:07:51.604 align:middle line:79% position:50% size:65% different groups of agents for instance in the case 00:07:51.671 --> 00:07:54.607 align:middle line:79% position:50% size:63% of Facebook, WeChat, because you can chat with 00:07:54.674 --> 00:07:57.109 align:middle line:79% position:50% size:65% others right so there is a direct network effects in 00:07:57.176 --> 00:07:59.912 align:middle line:79% position:50% size:63% sense that you are a utility for adopting that 00:07:59.979 --> 00:08:03.049 align:middle line:79% position:50% size:60% particular platform will increase as a result of 00:08:03.115 --> 00:08:06.219 align:middle line:79% position:50% size:65% other participants on platform right and another 00:08:06.285 --> 00:08:11.557 align:middle line:79% position:50% size:63% common one would be this Indirect Network Effects. 00:08:11.624 --> 00:08:13.893 align:middle line:79% position:50% size:60% This takes place between like essentially 00:08:13.960 --> 00:08:17.230 align:middle line:79% position:50% size:65% consumers and service providers so the reason we 00:08:17.296 --> 00:08:20.566 align:middle line:79% position:50% size:63% like iOS or Android it's because of number of apps 00:08:20.633 --> 00:08:23.936 align:middle line:79% position:50% size:65% on top of this platforms and the reason some of the 00:08:24.003 --> 00:08:27.273 align:middle line:79% position:50% size:65% app developers are interested developing apps 00:08:27.340 --> 00:08:30.610 align:middle line:79% position:50% size:63% for a particular platform is because there is a 00:08:30.676 --> 00:08:33.946 align:middle line:79% position:50% size:63% large installed base of these platforms right and 00:08:34.013 --> 00:08:38.284 align:middle line:79% position:50% size:60% the third network effect is Data Network Effects 00:08:38.351 --> 00:08:42.622 align:middle line:79% position:50% size:60% become increasingly important because of the 00:08:42.688 --> 00:08:44.590 align:middle line:79% position:50% size:63% race of big data and also machine learning right so 00:08:44.657 --> 00:08:48.294 align:middle line:79% position:50% size:55% part of the reason why Google has such a big 00:08:48.361 --> 00:08:50.529 align:middle line:79% position:50% size:50% market share not only in the U.S. 00:08:50.596 --> 00:08:53.499 align:middle line:79% position:50% size:60% but in many countries is because Google actually 00:08:53.566 --> 00:08:56.068 align:middle line:79% position:50% size:58% leverages these Data Network Effects or many 00:08:56.135 --> 00:08:58.404 align:middle line:79% position:50% size:53% effects is not really because of Indirect 00:08:58.471 --> 00:09:00.072 align:middle line:79% position:50% size:38% Network Effects in a sense. 00:09:00.139 --> 00:09:02.508 align:middle line:79% position:50% size:60% Oh, the reason I want to kind of browse Google is 00:09:02.575 --> 00:09:05.611 align:middle line:79% position:50% size:63% because I love ads served on Google, that's not the 00:09:05.678 --> 00:09:09.715 align:middle line:79% position:50% size:65% reason right and so the idea here is that the more 00:09:09.782 --> 00:09:12.852 align:middle line:79% position:50% size:58% people using Google to search for a particular 00:09:12.919 --> 00:09:15.988 align:middle line:79% position:50% size:68% item right and Google could capture that consumers 00:09:16.055 --> 00:09:18.791 align:middle line:79% position:50% size:60% behavior right, linking your search query with a 00:09:18.858 --> 00:09:23.029 align:middle line:79% position:50% size:65% link you clicked on and you that kind of result to 00:09:23.095 --> 00:09:25.498 align:middle line:79% position:50% size:65% refine its algorithm to kind of make the algorithm 00:09:25.564 --> 00:09:29.335 align:middle line:79% position:50% size:68% even better. So as a result when the next person is 00:09:29.402 --> 00:09:33.706 align:middle line:79% position:50% size:65% using Google to search for a particular query, Google 00:09:33.773 --> 00:09:36.642 align:middle line:79% position:50% size:60% can do a better job in displaying in the result 00:09:36.709 --> 00:09:39.278 align:middle line:79% position:50% size:63% right so the Data Network Effect actually allows 00:09:39.345 --> 00:09:42.081 align:middle line:79% position:50% size:65% Google to continuously to improve the quality of its 00:09:42.148 --> 00:09:45.618 align:middle line:79% position:50% size:60% search engine right so this also generates this 00:09:45.685 --> 00:09:47.954 align:middle line:85% position:50% size:58% positive feedback loop. 00:09:48.020 --> 00:09:50.556 align:middle line:79% position:50% size:65% So, when I was a doctorate student right at that time 00:09:50.623 --> 00:09:53.659 align:middle line:79% position:50% size:65% most of the papers around platforms actually studied 00:09:53.726 --> 00:09:56.762 align:middle line:79% position:50% size:65% models for economics basically the idea is that 00:09:56.829 --> 00:09:59.498 align:middle line:79% position:50% size:58% two platforms would compete and then derive 00:09:59.565 --> 00:10:03.169 align:middle line:79% position:50% size:63% some statistics right but it turns out that much of 00:10:03.235 --> 00:10:07.807 align:middle line:79% position:50% size:65% the work right essentially around that time right was 00:10:07.873 --> 00:10:09.875 align:middle line:79% position:50% size:65% the focus on the existence of networks effects 00:10:09.942 --> 00:10:13.079 align:middle line:79% position:50% size:58% regardless whether it's theoretical work or 00:10:13.145 --> 00:10:16.082 align:middle line:79% position:50% size:58% empirical work right so people have documented 00:10:16.148 --> 00:10:19.585 align:middle line:79% position:50% size:65% direct across a variety of these markets right VCR, 00:10:19.652 --> 00:10:22.989 align:middle line:79% position:50% size:60% DVD, video game consoles there exists significant 00:10:23.055 --> 00:10:26.792 align:middle line:79% position:50% size:58% network effect right but the problem is that 00:10:26.859 --> 00:10:30.663 align:middle line:79% position:50% size:63% knowing like the presence of network effect itself 00:10:30.730 --> 00:10:32.832 align:middle line:79% position:50% size:50% it's insufficient in helping us really 00:10:32.898 --> 00:10:35.568 align:middle line:79% position:50% size:63% understand like different competitive dynamics 00:10:35.634 --> 00:10:37.937 align:middle line:79% position:50% size:60% across different markets or different market 00:10:38.004 --> 00:10:39.772 align:middle line:79% position:50% size:60% structures across different markets right. 00:10:39.839 --> 00:10:42.908 align:middle line:79% position:50% size:68% So take like for instance a video game console as one 00:10:42.975 --> 00:10:46.278 align:middle line:79% position:50% size:58% example, you actually observe that the market 00:10:46.345 --> 00:10:49.749 align:middle line:79% position:50% size:63% leadership switches every few years right versus 00:10:49.815 --> 00:10:51.984 align:middle line:79% position:50% size:65% like a Google for instance right it has been a 00:10:52.051 --> 00:10:55.588 align:middle line:79% position:50% size:68% dominant player for a long period of time right, so in 00:10:55.654 --> 00:10:59.091 align:middle line:79% position:50% size:63% other markets like in the DVD or Hi-Def DVD market 00:10:59.158 --> 00:11:01.427 align:middle line:79% position:50% size:65% it's really winner takes all right Blu-ray suddenly 00:11:01.494 --> 00:11:03.396 align:middle line:79% position:50% size:45% became the dominant standard. 00:11:03.462 --> 00:11:09.068 align:middle line:79% position:50% size:63% So how would we explain like these factors across 00:11:09.135 --> 00:11:11.637 align:middle line:79% position:50% size:58% the market dynamics why some markets evolve 00:11:11.704 --> 00:11:15.007 align:middle line:79% position:50% size:63% towards a situation where multiple kind of players 00:11:15.074 --> 00:11:18.978 align:middle line:79% position:50% size:63% can co-exists right and the market leadership can 00:11:19.045 --> 00:11:21.847 align:middle line:79% position:50% size:58% change while others are evolving towards like 00:11:21.914 --> 00:11:24.650 align:middle line:79% position:50% size:43% monopoly like market structure. 00:11:24.717 --> 00:11:26.919 align:middle line:79% position:50% size:60% So at that time I basically worked with my 00:11:26.986 --> 00:11:29.355 align:middle line:79% position:50% size:55% advisor Marco Iansiti and we published paper 00:11:29.422 --> 00:11:32.591 align:middle line:79% position:50% size:53% basically looking at suppose a theoretical 00:11:32.658 --> 00:11:36.429 align:middle line:79% position:50% size:63% model with some empirical evidence supporting the 00:11:36.495 --> 00:11:38.164 align:middle line:79% position:50% size:53% theoretical model, basically we tried to 00:11:38.230 --> 00:11:41.167 align:middle line:79% position:50% size:63% model the competition between an entrant with a 00:11:41.233 --> 00:11:44.270 align:middle line:79% position:50% size:60% high quality essentially platform in a sense the 00:11:44.336 --> 00:11:47.006 align:middle line:79% position:50% size:65% entrant platform is perceived to offer greater 00:11:47.073 --> 00:11:49.708 align:middle line:79% position:50% size:60% utility to customers. And, also incumbent with 00:11:49.775 --> 00:11:52.411 align:middle line:79% position:50% size:65% greater installed base so that's incumbent advantage 00:11:52.478 --> 00:11:55.915 align:middle line:79% position:50% size:65% and I will allow for indirect network effect in 00:11:55.981 --> 00:11:58.350 align:middle line:79% position:50% size:55% this market and tried to develop essentially 00:11:58.417 --> 00:12:01.020 align:middle line:79% position:50% size:60% a dynamic model tried to predict the evolution of 00:12:01.087 --> 00:12:04.824 align:middle line:79% position:50% size:60% this market. And what we discovered is that if we 00:12:04.890 --> 00:12:07.693 align:middle line:79% position:50% size:63% allow significant network effect the strength of 00:12:07.760 --> 00:12:11.363 align:middle line:79% position:50% size:63% indirect network effect here we use a parameter E 00:12:11.430 --> 00:12:14.200 align:middle line:79% position:50% size:53% to capture that can actually help predict 00:12:14.266 --> 00:12:16.268 align:middle line:79% position:50% size:60% essentially the entrance essentially equilibrium 00:12:16.335 --> 00:12:18.704 align:middle line:79% position:50% size:38% market share in the long term. 00:12:18.771 --> 00:12:22.374 align:middle line:79% position:50% size:58% So when the strength is extremely high it turns 00:12:22.441 --> 00:12:25.578 align:middle line:79% position:50% size:55% out like as intuition would suggest that the 00:12:25.644 --> 00:12:28.280 align:middle line:79% position:50% size:65% market will move towards a monopoly situation, where 00:12:28.347 --> 00:12:30.783 align:middle line:79% position:50% size:60% the incumbent eventually become a monopoly. 00:12:30.850 --> 00:12:34.286 align:middle line:79% position:50% size:63% And the entrant will have zero percent market share 00:12:34.353 --> 00:12:37.790 align:middle line:79% position:50% size:65% over the years but when the strength is below that 00:12:37.857 --> 00:12:40.326 align:middle line:79% position:50% size:50% thresh hold it turns out that the entrant 00:12:40.392 --> 00:12:44.196 align:middle line:79% position:50% size:65% eventually will still become the dominant player 00:12:44.263 --> 00:12:47.133 align:middle line:79% position:50% size:65% in the market if you just allow two players and when 00:12:47.199 --> 00:12:52.071 align:middle line:79% position:50% size:63% E is a moderate so it's not moderately strong and 00:12:52.138 --> 00:12:56.142 align:middle line:79% position:50% size:60% it turns out the entrant will eventually become a 00:12:56.208 --> 00:12:59.345 align:middle line:79% position:50% size:60% monopoly platform in the end but when E is below 00:12:59.411 --> 00:13:03.482 align:middle line:79% position:50% size:65% like one right, in between zero to one, we have a way 00:13:03.549 --> 00:13:07.253 align:middle line:79% position:50% size:65% to measure this E right it turns out that the market 00:13:07.319 --> 00:13:09.555 align:middle line:79% position:50% size:58% will evolve towards a oligopolistic structure 00:13:09.622 --> 00:13:12.558 align:middle line:79% position:50% size:65% where the entrant actually would have a higher market 00:13:12.625 --> 00:13:13.225 align:middle line:85% position:50% size:15% share. 00:13:13.292 --> 00:13:16.061 align:middle line:79% position:50% size:58% So in other words even with network effect the 00:13:16.128 --> 00:13:21.333 align:middle line:79% position:50% size:63% entrants may not be deter from entering the market 00:13:21.400 --> 00:13:23.602 align:middle line:79% position:50% size:65% depending on its strengths and the entrants may still 00:13:23.669 --> 00:13:25.404 align:middle line:79% position:50% size:45% take over the market leadership. 00:13:25.471 --> 00:13:29.041 align:middle line:79% position:50% size:63% So, empirical if you look at if you compare these 00:13:29.108 --> 00:13:31.210 align:middle line:79% position:50% size:38% two markets you will see why? 00:13:31.277 --> 00:13:33.445 align:middle line:79% position:50% size:60% So, we're actually empirically look at this 00:13:33.512 --> 00:13:35.614 align:middle line:79% position:50% size:65% video game console market and estimated its networks 00:13:35.681 --> 00:13:38.817 align:middle line:79% position:50% size:65% strength of indirect network effect we actually 00:13:38.884 --> 00:13:41.220 align:middle line:79% position:50% size:58% found that actually network effects in that 00:13:41.287 --> 00:13:43.522 align:middle line:79% position:50% size:63% market is not that strong it's significantly but is 00:13:43.589 --> 00:13:44.857 align:middle line:85% position:50% size:40% not that strong. 00:13:44.924 --> 00:13:47.493 align:middle line:79% position:50% size:65% Part of the reason is that this market is really hit 00:13:47.560 --> 00:13:52.231 align:middle line:79% position:50% size:65% driven in a sense that average video game console 00:13:52.298 --> 00:13:56.101 align:middle line:79% position:50% size:65% owner only buys like maybe six to seven video game 00:13:56.168 --> 00:13:57.536 align:middle line:85% position:50% size:48% titles per console. 00:13:57.603 --> 00:14:01.707 align:middle line:79% position:50% size:58% So as a result if a new comer can come up with 00:14:01.774 --> 00:14:05.744 align:middle line:79% position:50% size:60% like six to seven hit titles the new comer can 00:14:05.811 --> 00:14:08.881 align:middle line:79% position:50% size:58% actually have a chance to put up a good battle 00:14:08.948 --> 00:14:10.449 align:middle line:85% position:50% size:45% against incumbent. 00:14:10.516 --> 00:14:14.186 align:middle line:79% position:50% size:65% But if you at like the VHS and the Betamax market so 00:14:14.253 --> 00:14:16.989 align:middle line:79% position:50% size:48% some other academic studies have also 00:14:17.056 --> 00:14:19.258 align:middle line:79% position:50% size:65% estimated that the network effects there, the network 00:14:19.325 --> 00:14:21.760 align:middle line:79% position:50% size:65% effects actually significantly stronger and 00:14:21.827 --> 00:14:25.598 align:middle line:79% position:50% size:65% part of the reason is that average household per 00:14:25.664 --> 00:14:29.702 align:middle line:79% position:50% size:65% month will rent like four to five movie titles. 00:14:29.768 --> 00:14:33.138 align:middle line:79% position:50% size:65% So as a result the dependence on a variety of 00:14:33.205 --> 00:14:36.141 align:middle line:79% position:50% size:63% movie titles is much more higher compared with the 00:14:36.208 --> 00:14:38.944 align:middle line:79% position:50% size:55% video game case, video game console case. 00:14:39.011 --> 00:14:43.682 align:middle line:79% position:50% size:63% So as result the market could quickly tip towards 00:14:43.749 --> 00:14:46.318 align:middle line:79% position:50% size:63% this incumbent with installed base advantage, 00:14:46.385 --> 00:14:50.623 align:middle line:79% position:50% size:63% this is also the case for High-Def DVD players like 00:14:50.689 --> 00:14:52.891 align:middle line:85% position:50% size:58% Blu-ray and the HD-DVD. 00:14:55.027 --> 00:14:58.731 align:middle line:79% position:50% size:58% So I also later on learned a case the case 00:14:58.797 --> 00:15:01.767 align:middle line:79% position:50% size:58% actually also helps to illustrate one very key 00:15:01.834 --> 00:15:03.269 align:middle line:79% position:50% size:63% point is that network effects right even though 00:15:03.335 --> 00:15:05.671 align:middle line:79% position:50% size:60% we tried to provide empirical estimation and 00:15:05.738 --> 00:15:09.074 align:middle line:79% position:50% size:65% show a particular number for a particular market it 00:15:09.141 --> 00:15:11.143 align:middle line:79% position:50% size:60% turns out that the firms strategies can also 00:15:11.210 --> 00:15:13.912 align:middle line:79% position:50% size:63% influence the strength of network effect in that 00:15:13.979 --> 00:15:14.847 align:middle line:85% position:50% size:45% particular market. 00:15:14.913 --> 00:15:17.349 align:middle line:79% position:50% size:63% So, let me give you just one example from a case I 00:15:17.416 --> 00:15:19.018 align:middle line:85% position:50% size:45% learned on Fasten. 00:15:19.084 --> 00:15:23.822 align:middle line:79% position:50% size:60% So Fasten is actually a startup in Boston trying 00:15:23.889 --> 00:15:26.558 align:middle line:79% position:50% size:63% to challenge Uber or Lyft in the Boston market it's 00:15:26.625 --> 00:15:28.694 align:middle line:85% position:50% size:58% a ride sharing company. 00:15:28.761 --> 00:15:32.131 align:middle line:79% position:50% size:55% The business model for Fasten is somewhat 00:15:32.197 --> 00:15:34.967 align:middle line:79% position:50% size:60% different from Uber in that Fasten only takes a 00:15:35.034 --> 00:15:38.837 align:middle line:79% position:50% size:63% $1 instead of 20-25 percent charge by Uber or 00:15:38.904 --> 00:15:42.408 align:middle line:79% position:50% size:63% Lyft, $1 per trip because Fasten takes less, 00:15:42.474 --> 00:15:46.879 align:middle line:79% position:50% size:60% so in the end Fasten could charge less on the 00:15:46.945 --> 00:15:50.449 align:middle line:79% position:50% size:58% consumer rider side and still give driver more 00:15:50.516 --> 00:15:51.483 align:middle line:85% position:50% size:25% money, no? 00:15:51.550 --> 00:15:53.786 align:middle line:79% position:50% size:60% Initially because Fasten has a smaller installed 00:15:53.852 --> 00:15:57.323 align:middle line:79% position:50% size:63% base as a result in order to get a Fasten car the 00:15:57.389 --> 00:16:00.659 align:middle line:79% position:50% size:63% rider may have to wait a little bit longer but you 00:16:00.726 --> 00:16:01.527 align:middle line:85% position:50% size:35% will pay less. 00:16:01.593 --> 00:16:03.962 align:middle line:79% position:50% size:53% So, it turns out it's extremely popular in 00:16:04.029 --> 00:16:06.765 align:middle line:79% position:50% size:63% Boston for a while and lots of drivers of course 00:16:06.832 --> 00:16:09.168 align:middle line:79% position:50% size:60% would like to multi-home and sign up on Fasten as 00:16:09.234 --> 00:16:12.671 align:middle line:79% position:50% size:65% well because unless Uber can lower the fee of their 00:16:12.738 --> 00:16:13.605 align:middle line:85% position:50% size:40% entire schedule. 00:16:13.672 --> 00:16:18.844 align:middle line:79% position:50% size:58% Fasten expanded in to Texas, Austin and claim 00:16:18.911 --> 00:16:23.282 align:middle line:79% position:50% size:60% that it's actually the first profitable company 00:16:23.349 --> 00:16:28.754 align:middle line:79% position:50% size:38% in the ride sharing market. 00:16:28.821 --> 00:16:30.923 align:middle line:79% position:50% size:38% So but one year later it was... 00:16:30.989 --> 00:16:36.061 align:middle line:79% position:50% size:65% Fasten started in 2015 and they became profitable in 00:16:36.128 --> 00:16:38.397 align:middle line:85% position:50% size:63% 2017 according to Fasten. 00:16:38.464 --> 00:16:43.769 align:middle line:79% position:50% size:58% But suddenly Fasten quit the market in 2018 00:16:43.836 --> 00:16:48.907 align:middle line:79% position:50% size:58% and part of the reason was actually do to this 00:16:48.974 --> 00:16:51.310 align:middle line:79% position:50% size:48% pooling service introduced by Uber. 00:16:51.377 --> 00:16:53.112 align:middle line:79% position:50% size:55% So if you think about the pooling service it 00:16:53.178 --> 00:16:55.948 align:middle line:79% position:50% size:60% actually adds another layer of network effects 00:16:56.014 --> 00:16:57.149 align:middle line:85% position:50% size:40% into the market. 00:16:57.216 --> 00:17:00.419 align:middle line:79% position:50% size:65% So essentially the idea here is that for a pooling 00:17:00.486 --> 00:17:03.021 align:middle line:79% position:50% size:65% rider I need to essentially match you with 00:17:03.088 --> 00:17:06.158 align:middle line:79% position:50% size:63% another rider going the same direction around the 00:17:06.225 --> 00:17:07.259 align:middle line:85% position:50% size:25% same time. 00:17:07.326 --> 00:17:09.428 align:middle line:79% position:50% size:53% So the installed base becomes extremely 00:17:09.495 --> 00:17:13.031 align:middle line:79% position:50% size:60% important, so there is essentially some kind of 00:17:13.098 --> 00:17:15.634 align:middle line:79% position:50% size:63% direct network effect being introduced by Uber. 00:17:15.701 --> 00:17:20.372 align:middle line:79% position:50% size:55% So with the stronger network effects Fasten 00:17:20.439 --> 00:17:22.708 align:middle line:79% position:50% size:50% couldn't match their service because the 00:17:22.775 --> 00:17:24.209 align:middle line:79% position:50% size:55% installed base is much smaller so in the end 00:17:24.276 --> 00:17:27.279 align:middle line:79% position:50% size:65% Fasten actually decided to kind of essentially get 00:17:27.346 --> 00:17:30.149 align:middle line:79% position:50% size:65% acquire by another company in Russia and they 00:17:30.215 --> 00:17:33.519 align:middle line:79% position:50% size:60% eventually decided to leave the Boston market. 00:17:33.585 --> 00:17:35.020 align:middle line:79% position:50% size:58% So now the strength of network effects may not 00:17:35.087 --> 00:17:38.323 align:middle line:79% position:50% size:65% stay stable in a market so as a result it's actually 00:17:38.390 --> 00:17:41.059 align:middle line:79% position:50% size:65% very useful to be diligent for companies to be 00:17:41.126 --> 00:17:44.530 align:middle line:79% position:50% size:60% diligent like pay attention to the how the 00:17:44.596 --> 00:17:46.932 align:middle line:79% position:50% size:65% network effects may change or may evolve over time. 00:17:46.999 --> 00:17:51.804 align:middle line:79% position:50% size:63% So then later on I sort of started to think after 00:17:51.870 --> 00:17:55.374 align:middle line:79% position:50% size:60% teaching the Fasten case I started think about 00:17:55.441 --> 00:17:57.576 align:middle line:79% position:50% size:60% essentially the differences between Uber 00:17:57.643 --> 00:18:01.013 align:middle line:79% position:50% size:68% and Airbnb because why Uber still losing so much money 00:18:01.079 --> 00:18:03.449 align:middle line:79% position:50% size:43% and why Airbnb is so profitable? 00:18:03.515 --> 00:18:10.722 align:middle line:79% position:50% size:63% So one of the factors I identified in my research 00:18:10.789 --> 00:18:13.859 align:middle line:79% position:50% size:70% is this network structure issue the clustering factor. 00:18:13.926 --> 00:18:20.532 align:middle line:79% position:50% size:63% So if you think about the Uber network, it's really 00:18:20.599 --> 00:18:23.735 align:middle line:79% position:50% size:63% a network with lots of local clusters in a sense 00:18:23.802 --> 00:18:27.606 align:middle line:79% position:50% size:63% that if you are in DC as a rider you probably only 00:18:27.673 --> 00:18:29.908 align:middle line:79% position:50% size:60% care about the number of drivers in DC. You don't 00:18:29.975 --> 00:18:32.177 align:middle line:79% position:50% size:53% really care about the number of drivers in 00:18:32.244 --> 00:18:34.780 align:middle line:79% position:50% size:60% San Francisco unless you travel so frequently to 00:18:34.847 --> 00:18:36.515 align:middle line:85% position:50% size:35% San Francisco. 00:18:36.582 --> 00:18:38.784 align:middle line:79% position:50% size:65% And if you're a driver in DC of course you only care 00:18:38.851 --> 00:18:42.254 align:middle line:79% position:50% size:65% about the number of users in DC mostly, unless there 00:18:42.321 --> 00:18:46.158 align:middle line:79% position:50% size:60% are lots of travelers coming into the DC area. 00:18:46.225 --> 00:18:49.328 align:middle line:79% position:50% size:55% And as a result so the Uber network has this 00:18:49.394 --> 00:18:51.430 align:middle line:79% position:50% size:45% local cluster feature where some 00:18:51.497 --> 00:18:54.199 align:middle line:79% position:50% size:60% interconnections because of this mobile kind of 00:18:54.266 --> 00:18:57.102 align:middle line:79% position:50% size:65% users essentially some people may travel from one 00:18:57.169 --> 00:18:57.903 align:middle line:85% position:50% size:40% city to another. 00:18:57.970 --> 00:19:02.908 align:middle line:79% position:50% size:60% Then, if you look at the Airbnb network it's 00:19:02.975 --> 00:19:05.911 align:middle line:79% position:50% size:65% completely the opposite in a sense that if you are 00:19:05.978 --> 00:19:09.882 align:middle line:79% position:50% size:60% the host, Airbnb host in DC you're not interested 00:19:09.948 --> 00:19:13.519 align:middle line:79% position:50% size:65% attracting local residents nobody is going to stay at 00:19:13.585 --> 00:19:16.321 align:middle line:79% position:50% size:40% the Airbnb host in a local city. 00:19:16.388 --> 00:19:18.724 align:middle line:79% position:50% size:65% They're interested in basically people traveling 00:19:18.790 --> 00:19:21.527 align:middle line:79% position:50% size:45% outside of DC from anywhere else and 00:19:21.593 --> 00:19:27.633 align:middle line:79% position:50% size:63% similarly if you are like a local people in DC why 00:19:27.699 --> 00:19:31.169 align:middle line:79% position:50% size:65% visit the Airbnb you probably care about Airbnb 00:19:31.236 --> 00:19:33.906 align:middle line:79% position:50% size:60% properties outside of DC you're not interested in 00:19:33.972 --> 00:19:35.407 align:middle line:79% position:50% size:35% staying at the same location. 00:19:35.474 --> 00:19:40.012 align:middle line:79% position:50% size:60% So as a result Airbnb network is more globally 00:19:40.078 --> 00:19:43.682 align:middle line:79% position:50% size:53% connected than Uber network, these things 00:19:43.749 --> 00:19:48.453 align:middle line:79% position:50% size:65% would have implications to their defensibility, think 00:19:48.520 --> 00:19:51.757 align:middle line:79% position:50% size:55% about the Uber network structure if companies 00:19:51.823 --> 00:19:54.693 align:middle line:79% position:50% size:60% like Fasten wanted to enter and challenge Uber 00:19:54.760 --> 00:19:57.796 align:middle line:79% position:50% size:63% all it needs to do is to concentrate on one market 00:19:57.863 --> 00:19:59.965 align:middle line:85% position:50% size:33% and go there. 00:20:00.032 --> 00:20:03.535 align:middle line:79% position:50% size:60% But if a company wants to challenge Airbnb it's 00:20:03.602 --> 00:20:07.940 align:middle line:79% position:50% size:60% actually a lot harder so for any host in DC you 00:20:08.006 --> 00:20:11.343 align:middle line:79% position:50% size:65% really have to announce to the rest of the world yes, 00:20:11.410 --> 00:20:16.281 align:middle line:79% position:50% size:63% yes I'm host platform and if you travel to DC you 00:20:16.348 --> 00:20:17.916 align:middle line:85% position:50% size:35% could stay at. 00:20:17.983 --> 00:20:22.788 align:middle line:79% position:50% size:63% So as a result it's a lot costly for you to start a 00:20:22.854 --> 00:20:27.626 align:middle line:79% position:50% size:55% company to essentially match the service by 00:20:27.693 --> 00:20:30.395 align:middle line:79% position:50% size:55% Airbnb once Airbnb has already scaled up. 00:20:30.462 --> 00:20:34.800 align:middle line:79% position:50% size:65% Later it turns out that if you look at across many 00:20:34.866 --> 00:20:37.569 align:middle line:79% position:50% size:60% platform markets you can actually classified them 00:20:37.636 --> 00:20:39.938 align:middle line:79% position:50% size:43% into one of these two groups. 00:20:40.005 --> 00:20:42.708 align:middle line:79% position:50% size:58% So Uber like Uber, Groupon, TaskRabbit any 00:20:42.774 --> 00:20:45.877 align:middle line:79% position:50% size:58% platform based on local matching will belong to 00:20:45.944 --> 00:20:47.846 align:middle line:79% position:50% size:45% this kind of network structure. 00:20:47.913 --> 00:20:53.819 align:middle line:79% position:50% size:63% And Upwork, EBay, Expedia these platforms are going 00:20:53.885 --> 00:20:57.389 align:middle line:79% position:50% size:63% to belong to this kind of global connecting network 00:20:57.456 --> 00:20:59.758 align:middle line:79% position:50% size:60% structure, that actually affect the defensibility 00:20:59.825 --> 00:21:02.427 align:middle line:85% position:50% size:43% of the platforms. 00:21:02.494 --> 00:21:08.200 align:middle line:79% position:50% size:60% So based on this insight my co-author and I also 00:21:08.266 --> 00:21:11.770 align:middle line:79% position:50% size:65% did some literature review and found that it turns 00:21:11.837 --> 00:21:14.940 align:middle line:79% position:50% size:65% out that the most existing work always implicit 00:21:15.007 --> 00:21:17.709 align:middle line:79% position:50% size:60% assumes this kind of network structure around 00:21:17.776 --> 00:21:19.044 align:middle line:85% position:50% size:35% the platforms. 00:21:19.111 --> 00:21:22.247 align:middle line:79% position:50% size:58% So we decided to extend the leadership and 00:21:22.314 --> 00:21:26.218 align:middle line:79% position:50% size:58% basically try to take a closer look at how this 00:21:26.284 --> 00:21:28.587 align:middle line:79% position:50% size:55% network clustering may affect the incumbent 00:21:28.654 --> 00:21:29.921 align:middle line:85% position:50% size:30% performance. 00:21:29.988 --> 00:21:33.492 align:middle line:79% position:50% size:63% So we build a theoretical model, it's a game 00:21:33.558 --> 00:21:35.861 align:middle line:79% position:50% size:60% theoretical model essentially we model the 00:21:35.927 --> 00:21:37.829 align:middle line:79% position:50% size:65% competitive interactions between incumbent platform 00:21:37.896 --> 00:21:40.999 align:middle line:79% position:50% size:63% which outputs across in multiple markets think of 00:21:41.066 --> 00:21:44.569 align:middle line:79% position:50% size:60% an Uber it operates in multiple markets and the 00:21:44.636 --> 00:21:48.540 align:middle line:79% position:50% size:70% one entrant think of Fasten seeking to enter one market. 00:21:48.607 --> 00:21:50.442 align:middle line:79% position:50% size:65% And the market allows some connectivity in the sense 00:21:50.509 --> 00:21:53.345 align:middle line:79% position:50% size:58% that factual buyers are mobile in a sense that 00:21:53.412 --> 00:21:56.648 align:middle line:79% position:50% size:65% okay I live in Boston but I travel to DC I could use 00:21:56.715 --> 00:21:59.651 align:middle line:79% position:50% size:60% Uber here. And so that's the idea and if the 00:21:59.718 --> 00:22:03.989 align:middle line:79% position:50% size:80% entrant would invest to build awareness of its offering. 00:22:04.056 --> 00:22:07.259 align:middle line:79% position:50% size:58% So the model is very simple but it turns out 00:22:07.325 --> 00:22:09.194 align:middle line:79% position:50% size:48% that it has some interesting kind of 00:22:09.261 --> 00:22:10.929 align:middle line:85% position:50% size:63% predictions, why is that? 00:22:10.996 --> 00:22:15.133 align:middle line:79% position:50% size:60% We found that even if there is no cost for the 00:22:15.200 --> 00:22:17.936 align:middle line:79% position:50% size:65% entrant to build awareness basically even if the 00:22:18.003 --> 00:22:21.940 align:middle line:79% position:50% size:60% advertising is free it turns out, we found that 00:22:22.007 --> 00:22:24.943 align:middle line:79% position:50% size:60% the entrant doesn't want to build awareness among 00:22:25.010 --> 00:22:26.344 align:middle line:85% position:50% size:60% all people in that city. 00:22:26.411 --> 00:22:29.114 align:middle line:79% position:50% size:63% The reason is that if you do that it will trigger 00:22:29.181 --> 00:22:33.218 align:middle line:79% position:50% size:63% competitive reaction from the incumbent and this is 00:22:33.285 --> 00:22:36.455 align:middle line:79% position:50% size:63% actually is not good it's actually less optimal 00:22:36.521 --> 00:22:39.291 align:middle line:79% position:50% size:63% compared to the situation where you just reach a 00:22:39.357 --> 00:22:43.061 align:middle line:79% position:50% size:50% certain, you build awareness on a small 00:22:43.128 --> 00:22:45.964 align:middle line:79% position:50% size:50% fraction of the market participants. 00:22:46.031 --> 00:22:48.967 align:middle line:79% position:50% size:63% So this matches perfectly with the Fasten case 00:22:49.034 --> 00:22:50.969 align:middle line:79% position:50% size:63% because when inquire with the founders they said 00:22:51.036 --> 00:22:54.606 align:middle line:79% position:50% size:53% well we never thought about taking over the 00:22:54.673 --> 00:22:57.809 align:middle line:79% position:50% size:63% market share from Uber we only just want to kind of 00:22:57.876 --> 00:23:01.580 align:middle line:79% position:50% size:65% target a small segment of the Boston market and make 00:23:01.646 --> 00:23:04.683 align:middle line:79% position:50% size:60% some money that actually matches with the 00:23:04.750 --> 00:23:05.917 align:middle line:85% position:50% size:53% strategies very well. 00:23:05.984 --> 00:23:09.588 align:middle line:79% position:50% size:58% And we also discovered that having more mobile 00:23:09.654 --> 00:23:13.759 align:middle line:79% position:50% size:63% buyers should make it easier for the entrant to 00:23:13.825 --> 00:23:14.926 align:middle line:85% position:50% size:38% enter actually. 00:23:14.993 --> 00:23:17.462 align:middle line:79% position:50% size:60% So the reason is that to say there are lots of 00:23:17.529 --> 00:23:20.932 align:middle line:79% position:50% size:63% people traveling from San Francisco or DC to Boston 00:23:20.999 --> 00:23:24.102 align:middle line:79% position:50% size:63% and these people they have no idea about Fasten 00:23:24.169 --> 00:23:27.139 align:middle line:79% position:50% size:68% because Fasten is only local to the Boston market. 00:23:27.205 --> 00:23:29.608 align:middle line:79% position:50% size:65% So these people are essentially captive to the 00:23:29.674 --> 00:23:32.077 align:middle line:79% position:50% size:55% Uber network, you're arriving in Boston you 00:23:32.144 --> 00:23:34.946 align:middle line:85% position:50% size:65% only use Uber application. 00:23:35.013 --> 00:23:39.284 align:middle line:79% position:50% size:58% So as a result if Uber started to launch a war 00:23:39.351 --> 00:23:43.121 align:middle line:79% position:50% size:65% against the Fasten in the Boston market, is going to 00:23:43.188 --> 00:23:45.957 align:middle line:79% position:50% size:65% kind of reduce the price equally also for all these 00:23:46.024 --> 00:23:50.962 align:middle line:79% position:50% size:63% travelers who actually to who Uber actually has a 00:23:51.029 --> 00:23:52.964 align:middle line:79% position:50% size:38% monopoly power to some extent. 00:23:53.031 --> 00:23:56.868 align:middle line:79% position:50% size:63% So is actually suboptimal as a result for Uber to 00:23:56.935 --> 00:23:59.137 align:middle line:79% position:50% size:65% fight aggressively against a Fasten if there are a 00:23:59.204 --> 00:24:03.408 align:middle line:79% position:50% size:80% lots of travelers from outside Boston, drive in Boston. 00:24:03.475 --> 00:24:06.745 align:middle line:79% position:50% size:65% So as a result this interconnectivity actually 00:24:06.812 --> 00:24:09.915 align:middle line:79% position:50% size:60% would make it easier for the entrant to enter the 00:24:09.981 --> 00:24:13.485 align:middle line:79% position:50% size:63% market because it reduces the incentive for the 00:24:13.552 --> 00:24:14.920 align:middle line:85% position:50% size:48% incumbent to fight. 00:24:14.986 --> 00:24:19.391 align:middle line:79% position:50% size:63% So in the case of Fasten actually even though Uber 00:24:19.457 --> 00:24:21.993 align:middle line:79% position:50% size:58% from the very beginning were aware of the 00:24:22.060 --> 00:24:23.328 align:middle line:85% position:50% size:50% existence of Fasten. 00:24:23.395 --> 00:24:26.932 align:middle line:79% position:50% size:63% Uber actually didn't reduce its price to fight 00:24:26.998 --> 00:24:30.268 align:middle line:79% position:50% size:60% against a Fasten, it did broke Uber accounts for 00:24:30.335 --> 00:24:33.939 align:middle line:79% position:50% size:65% all Fasten founders basically but it didn't do 00:24:34.005 --> 00:24:35.106 align:middle line:85% position:50% size:43% much beyond that. 00:24:35.173 --> 00:24:38.944 align:middle line:79% position:50% size:63% And we also discover that our model showed stronger 00:24:39.010 --> 00:24:41.947 align:middle line:79% position:50% size:65% network effects in general help deter entry so is in 00:24:42.013 --> 00:24:44.950 align:middle line:79% position:50% size:60% a favor of the incumbent and benefit Uber. 00:24:45.016 --> 00:24:47.619 align:middle line:79% position:50% size:60% So these is indeed the case after the Uber pool 00:24:47.686 --> 00:24:50.288 align:middle line:79% position:50% size:53% service became very popular Fasten had to 00:24:50.355 --> 00:24:51.790 align:middle line:85% position:50% size:43% leave the market. 00:24:51.857 --> 00:24:53.959 align:middle line:79% position:50% size:60% So after this exercise I was actually amaze in a 00:24:54.025 --> 00:24:57.128 align:middle line:79% position:50% size:55% sense that Fasten was so strategic, I really 00:24:57.195 --> 00:25:01.333 align:middle line:79% position:50% size:63% understood these dynamics including like they knew 00:25:01.399 --> 00:25:04.903 align:middle line:79% position:50% size:63% it's time to basically exit, including that even 00:25:04.970 --> 00:25:07.439 align:middle line:79% position:50% size:63% its decision to exit was actually somewhat optimal 00:25:07.505 --> 00:25:11.643 align:middle line:79% position:50% size:40% according to our model result. 00:25:11.710 --> 00:25:16.648 align:middle line:79% position:50% size:63% So the certain factor, if there is a question feel 00:25:16.715 --> 00:25:17.749 align:middle line:85% position:50% size:60% free to raise your hand. 00:25:17.816 --> 00:25:22.087 align:middle line:79% position:50% size:60% So the certain factor we looked very close at is 00:25:22.153 --> 00:25:25.557 align:middle line:79% position:50% size:65% this idea of multi-homing, in a sense that most of 00:25:25.624 --> 00:25:29.127 align:middle line:79% position:50% size:63% drivers they use kind of multiple apps they're are 00:25:29.194 --> 00:25:31.096 align:middle line:79% position:50% size:63% on Uber, they're on Lyft, they're on like in New 00:25:31.162 --> 00:25:33.198 align:middle line:79% position:50% size:63% York they have other platforms like a Juno and 00:25:33.265 --> 00:25:34.666 align:middle line:85% position:50% size:40% other platforms. 00:25:34.733 --> 00:25:39.104 align:middle line:79% position:50% size:58% And actually many times consumers also have 00:25:39.170 --> 00:25:41.940 align:middle line:79% position:50% size:65% multiple apps installed so I mean, I guess many of us 00:25:42.007 --> 00:25:43.608 align:middle line:79% position:50% size:38% have both Uber, Lyft installed. 00:25:43.675 --> 00:25:47.946 align:middle line:79% position:50% size:50% So this actually has implications for 00:25:48.013 --> 00:25:50.815 align:middle line:79% position:50% size:63% competitive dynamics in a sense that it basically 00:25:50.882 --> 00:25:54.953 align:middle line:79% position:50% size:63% makes it more difficult essentially for basically 00:25:55.020 --> 00:25:57.956 align:middle line:79% position:50% size:65% one platform to dominate the market in a sense that 00:25:58.023 --> 00:26:01.559 align:middle line:79% position:50% size:50% if switching cost or adaption cost for 00:26:01.626 --> 00:26:05.730 align:middle line:79% position:50% size:58% additional platform is so trivial that the new 00:26:05.797 --> 00:26:09.067 align:middle line:79% position:50% size:65% platform essentially can enter and pay a very small 00:26:09.134 --> 00:26:11.903 align:middle line:79% position:50% size:50% price to incentivize customers or service 00:26:11.970 --> 00:26:15.240 align:middle line:79% position:50% size:65% providers to overcome the switching or adaption cost 00:26:15.307 --> 00:26:16.241 align:middle line:85% position:50% size:65% to adopt the new platform. 00:26:16.308 --> 00:26:20.745 align:middle line:79% position:50% size:60% So Fasten was leveraging that see this is one ad 00:26:20.812 --> 00:26:25.917 align:middle line:79% position:50% size:60% made by Fasten in Boston actually on Facebook 00:26:25.984 --> 00:26:26.885 align:middle line:85% position:50% size:30% essentially. 00:26:26.952 --> 00:26:29.421 align:middle line:79% position:50% size:63% So Fasten basically shows that a clear message 00:26:29.487 --> 00:26:33.224 align:middle line:79% position:50% size:65% saying that different apps same driver so regardless 00:26:33.291 --> 00:26:37.262 align:middle line:79% position:50% size:65% you use Uber app or Fasten app so the left color is 00:26:37.329 --> 00:26:39.597 align:middle line:79% position:50% size:55% Uber color the right color is Fasten color. 00:26:39.664 --> 00:26:42.968 align:middle line:79% position:50% size:65% So but in the end you get same custom driver but you 00:26:43.034 --> 00:26:45.103 align:middle line:85% position:50% size:63% pay less that was the ad. 00:26:45.170 --> 00:26:48.773 align:middle line:79% position:50% size:65% So the idea is to leverage multi-homing to enter 00:26:48.840 --> 00:26:52.277 align:middle line:79% position:50% size:63% here, so if you look at other platforms like this 00:26:52.344 --> 00:26:55.613 align:middle line:79% position:50% size:65% food delivery networks, so lots of restaurants are 00:26:55.680 --> 00:26:59.451 align:middle line:79% position:50% size:63% working with multiple platforms to deliver food 00:26:59.517 --> 00:27:01.886 align:middle line:79% position:50% size:58% and lots of customers are interested in using 00:27:01.953 --> 00:27:05.790 align:middle line:79% position:50% size:53% multiple platforms to order food as well. 00:27:05.857 --> 00:27:08.893 align:middle line:79% position:50% size:60% So I worked on multiple research projects around 00:27:08.960 --> 00:27:11.730 align:middle line:79% position:50% size:65% the multi-homing so in the interest of time I would 00:27:11.796 --> 00:27:15.500 align:middle line:79% position:50% size:58% just focus on the first one it's also a paper 00:27:15.567 --> 00:27:17.402 align:middle line:79% position:50% size:50% we're still revising for our journal. 00:27:17.469 --> 00:27:22.574 align:middle line:79% position:50% size:63% So this paper was motivated by this kind of 00:27:22.640 --> 00:27:26.911 align:middle line:79% position:50% size:65% intuition that even though it turns out that in many 00:27:26.978 --> 00:27:28.913 align:middle line:79% position:50% size:53% of the markets we saw like this two sided 00:27:28.980 --> 00:27:31.750 align:middle line:79% position:50% size:65% multi-homing in a sense both consumers and service 00:27:31.816 --> 00:27:33.952 align:middle line:79% position:50% size:65% providers are they're multi-homing they adopting 00:27:34.019 --> 00:27:36.488 align:middle line:79% position:50% size:40% kind of multiple platforms. 00:27:36.554 --> 00:27:40.258 align:middle line:79% position:50% size:63% But an issue it turns out that most papers would 00:27:40.325 --> 00:27:43.061 align:middle line:79% position:50% size:60% tend to assume that only one side is multi-homing 00:27:43.128 --> 00:27:45.430 align:middle line:79% position:50% size:63% because in the leadership there is this implicit 00:27:45.497 --> 00:27:48.767 align:middle line:79% position:50% size:63% assumption that if one side is multi-homing then 00:27:48.833 --> 00:27:50.769 align:middle line:79% position:50% size:60% there is no need for the other side to multi-home 00:27:50.835 --> 00:27:53.938 align:middle line:79% position:50% size:65% because without multi-home you can access all service 00:27:54.005 --> 00:27:55.573 align:middle line:79% position:50% size:43% providers in one platform already. 00:27:55.640 --> 00:27:58.743 align:middle line:79% position:50% size:60% So but it doesn't really match with the empirical 00:27:58.810 --> 00:28:02.213 align:middle line:79% position:50% size:63% phenomena and also in the literature most of the 00:28:02.280 --> 00:28:05.550 align:middle line:79% position:50% size:60% papers were to focus on how platforms use prices 00:28:05.617 --> 00:28:09.054 align:middle line:79% position:50% size:63% to influence multi-homing but in reality platforms 00:28:09.120 --> 00:28:11.890 align:middle line:79% position:50% size:58% use a variety of strategies to influence 00:28:11.956 --> 00:28:13.892 align:middle line:79% position:50% size:43% this multi-homing tendencies. 00:28:13.958 --> 00:28:17.629 align:middle line:79% position:50% size:58% So here are just some examples when Microsoft 00:28:17.695 --> 00:28:20.899 align:middle line:79% position:50% size:63% was trying to grow its Windows phone system many 00:28:20.965 --> 00:28:23.902 align:middle line:79% position:50% size:58% years ago Microsoft was reported to offer like 00:28:23.968 --> 00:28:28.173 align:middle line:79% position:50% size:55% $100k or more to many popular app developers 00:28:28.239 --> 00:28:30.742 align:middle line:79% position:50% size:60% asking them oh why don't you kind port your 00:28:30.809 --> 00:28:33.912 align:middle line:79% position:50% size:60% application from iOS or Android to Windows phone 00:28:33.978 --> 00:28:34.646 align:middle line:85% position:50% size:18% system? 00:28:34.712 --> 00:28:39.250 align:middle line:79% position:50% size:63% So there would be like a same app in Windows phone 00:28:39.317 --> 00:28:43.588 align:middle line:79% position:50% size:60% system and very recently you hear from the news 00:28:43.655 --> 00:28:46.424 align:middle line:79% position:50% size:65% that many of the merchants on the Alibaba platform 00:28:46.491 --> 00:28:49.427 align:middle line:79% position:50% size:58% started to complaint that Alibaba is kind of 00:28:49.494 --> 00:28:53.932 align:middle line:79% position:50% size:63% implicitly trying to push them to adopt to work 00:28:53.998 --> 00:28:57.202 align:middle line:79% position:50% size:60% exclusively with Alibaba not with its competitors 00:28:57.268 --> 00:28:59.104 align:middle line:85% position:50% size:38% like JD or PDD. 00:28:59.170 --> 00:29:03.541 align:middle line:79% position:50% size:63% So the thought is that if you started to work with 00:29:03.608 --> 00:29:06.611 align:middle line:79% position:50% size:55% Alibaba's competitors, Alibaba can adjust its 00:29:06.678 --> 00:29:09.214 align:middle line:79% position:50% size:55% algorithm so that when people search for a 00:29:09.280 --> 00:29:11.483 align:middle line:79% position:50% size:60% particular product your product won't show up on 00:29:11.549 --> 00:29:16.387 align:middle line:79% position:50% size:63% the first few pages which could significantly 00:29:16.454 --> 00:29:19.557 align:middle line:79% position:50% size:48% influence the sales of these merchants. 00:29:19.624 --> 00:29:24.662 align:middle line:79% position:50% size:65% We also heard stories that EBay actually sued Amazon 00:29:24.729 --> 00:29:28.500 align:middle line:79% position:50% size:65% saying that Amazon kind of had attempted to lure the 00:29:28.566 --> 00:29:34.806 align:middle line:79% position:50% size:60% top sellers on EBay's market place asking them 00:29:34.873 --> 00:29:38.510 align:middle line:79% position:50% size:58% sell on Amazon's market places by exploiting 00:29:38.576 --> 00:29:41.479 align:middle line:79% position:50% size:68% essentially EBay's internal messaging systems. 00:29:41.546 --> 00:29:45.250 align:middle line:79% position:50% size:63% So firms strategies turns out can have significant 00:29:45.316 --> 00:29:48.920 align:middle line:79% position:50% size:60% influence on the multi-homing tendencies, 00:29:48.987 --> 00:29:51.422 align:middle line:79% position:50% size:63% they change essentially the competitive dynamics. 00:29:51.489 --> 00:29:55.760 align:middle line:79% position:50% size:65% So the research project we worked specifically on was 00:29:55.827 --> 00:29:58.429 align:middle line:79% position:50% size:65% looking at the competition between Groupon and The 00:29:58.496 --> 00:30:01.432 align:middle line:79% position:50% size:50% Living Social before they are merged. 00:30:01.499 --> 00:30:06.271 align:middle line:79% position:50% size:63% So it turns out that in 2010, so Groupon's action 00:30:06.337 --> 00:30:10.208 align:middle line:79% position:50% size:65% made some changes to its counter so previously when 00:30:10.275 --> 00:30:13.044 align:middle line:79% position:50% size:65% you go to Groupon, Groupon is going to give you like 00:30:13.111 --> 00:30:17.282 align:middle line:79% position:50% size:55% a clear message on how many deals have been 00:30:17.348 --> 00:30:22.787 align:middle line:79% position:50% size:60% bought for a particular offer so here 471 people 00:30:22.854 --> 00:30:26.291 align:middle line:79% position:50% size:60% have purchased this particular offer and the 00:30:26.357 --> 00:30:28.293 align:middle line:85% position:50% size:28% deal is on. 00:30:28.359 --> 00:30:30.962 align:middle line:79% position:50% size:65% But this was before Groupon's IPO and it turns 00:30:31.029 --> 00:30:33.965 align:middle line:79% position:50% size:60% out that many analysts started to use this deal 00:30:34.032 --> 00:30:38.236 align:middle line:79% position:50% size:65% counter to infer Groupon's profitability because you 00:30:38.303 --> 00:30:40.805 align:middle line:79% position:50% size:63% know like how many people have purchased, you know 00:30:40.872 --> 00:30:44.375 align:middle line:79% position:50% size:65% the price, you know roughly the cost, etc. And 00:30:44.442 --> 00:30:48.346 align:middle line:79% position:50% size:65% so as a result Groupon had to actually revise its 00:30:48.413 --> 00:30:52.917 align:middle line:79% position:50% size:65% statement for IPO several times so Groupon was upset 00:30:52.984 --> 00:30:56.921 align:middle line:79% position:50% size:65% about this so Groupon said okay now I'm going to 00:30:56.988 --> 00:31:00.892 align:middle line:79% position:50% size:65% change my counter so that instead of telling you the 00:31:00.959 --> 00:31:03.094 align:middle line:79% position:50% size:60% exact number I'm going to give you some kind of 00:31:03.161 --> 00:31:07.465 align:middle line:79% position:50% size:60% vague threshold and also going to play with the 00:31:07.532 --> 00:31:10.101 align:middle line:79% position:50% size:63% number so instead of showing the actual number 00:31:10.168 --> 00:31:12.203 align:middle line:79% position:50% size:55% I'm going to display a number that has been 00:31:12.270 --> 00:31:14.405 align:middle line:79% position:50% size:45% reduced by a random percentage. 00:31:14.472 --> 00:31:17.609 align:middle line:79% position:50% size:63% So as a result like even if it says over 50 bought 00:31:17.675 --> 00:31:22.547 align:middle line:79% position:50% size:60% you have no idea whether it means 471 or like 51. 00:31:22.614 --> 00:31:27.719 align:middle line:79% position:50% size:55% So then in the end you have no clue like the 00:31:27.785 --> 00:31:30.455 align:middle line:79% position:50% size:55% profitability of the company but in the end 00:31:30.521 --> 00:31:34.292 align:middle line:79% position:50% size:63% what happens is that this has spillover effects for 00:31:34.359 --> 00:31:38.129 align:middle line:79% position:50% size:60% its rivals Living Social because before this 00:31:38.196 --> 00:31:40.465 align:middle line:79% position:50% size:60% counter change many people were saying other 00:31:40.531 --> 00:31:43.968 align:middle line:79% position:50% size:60% competitor can easy look at this information and 00:31:44.035 --> 00:31:47.305 align:middle line:79% position:50% size:63% know what are the popular merchants and they could 00:31:47.372 --> 00:31:49.474 align:middle line:79% position:50% size:55% just approach the same merchants so that they 00:31:49.540 --> 00:31:52.477 align:middle line:79% position:50% size:65% could offer deal on Living Social later on or offer 00:31:52.543 --> 00:31:55.146 align:middle line:79% position:50% size:58% deal on some other platforms but after the 00:31:55.213 --> 00:31:57.982 align:middle line:79% position:50% size:63% counter change so it will be very difficult for you 00:31:58.049 --> 00:32:01.252 align:middle line:79% position:50% size:60% to know oh which deal is the very popular which 00:32:01.319 --> 00:32:05.590 align:middle line:79% position:50% size:58% deal is the worst like as our effort to poach. 00:32:05.657 --> 00:32:10.428 align:middle line:79% position:50% size:60% So we did some empirical analysis we did discover 00:32:10.495 --> 00:32:12.430 align:middle line:79% position:50% size:60% after this policy change with the deal counter 00:32:12.497 --> 00:32:15.266 align:middle line:79% position:50% size:58% Living Social actually copied a few deals from 00:32:15.333 --> 00:32:18.603 align:middle line:79% position:50% size:63% Groupon but there is also interesting that after 00:32:18.670 --> 00:32:21.439 align:middle line:79% position:50% size:60% Living Social started to copy few deals Living 00:32:21.506 --> 00:32:24.609 align:middle line:79% position:50% size:65% Social also expanded its effort to source new deals 00:32:24.676 --> 00:32:29.280 align:middle line:79% position:50% size:53% and as a result it's actually was worth it 00:32:29.347 --> 00:32:31.115 align:middle line:79% position:50% size:63% enhancing in a sense that the variety of deals in 00:32:31.182 --> 00:32:32.950 align:middle line:79% position:50% size:53% the market increased because Living Social 00:32:33.017 --> 00:32:36.954 align:middle line:79% position:50% size:45% started to look for its own deals. 00:32:37.021 --> 00:32:40.291 align:middle line:79% position:50% size:60% And of course this operation is very costly 00:32:40.358 --> 00:32:42.560 align:middle line:79% position:50% size:63% like previously you could copy now you have to do 00:32:42.627 --> 00:32:45.129 align:middle line:79% position:50% size:65% discovery, discovery is usually more costly but at 00:32:45.196 --> 00:32:49.133 align:middle line:79% position:50% size:63% the same time we also observe there is a seesaw 00:32:49.200 --> 00:32:52.637 align:middle line:79% position:50% size:55% effect in a sense that when there is less 00:32:52.704 --> 00:32:56.307 align:middle line:79% position:50% size:63% multi-homing the merchant decide in a sense that 00:32:56.374 --> 00:32:59.143 align:middle line:79% position:50% size:65% okay the few merchants are offering deals on both 00:32:59.210 --> 00:33:01.746 align:middle line:79% position:50% size:65% platforms now specially these popular ones and the 00:33:01.813 --> 00:33:03.915 align:middle line:79% position:50% size:50% two platforms become multi-differentiated 00:33:03.981 --> 00:33:06.851 align:middle line:79% position:50% size:65% actually then we saw actually more multi-homing 00:33:06.918 --> 00:33:08.920 align:middle line:79% position:50% size:58% on the consumer side it turns out as a consumer 00:33:08.986 --> 00:33:11.589 align:middle line:79% position:50% size:55% because the user so differentiated now and 00:33:11.656 --> 00:33:14.926 align:middle line:79% position:50% size:58% more willing to sign up for deal offers on both 00:33:14.992 --> 00:33:15.927 align:middle line:85% position:50% size:25% platforms. 00:33:15.993 --> 00:33:18.863 align:middle line:79% position:50% size:58% So as a result Living Social even though like 00:33:18.930 --> 00:33:24.936 align:middle line:79% position:50% size:60% financially it has incur more cost but it's the 00:33:25.002 --> 00:33:27.271 align:middle line:79% position:50% size:60% position on the consumer side it's actually 00:33:27.338 --> 00:33:30.475 align:middle line:79% position:50% size:58% becoming stronger after this so in a sense the 00:33:30.541 --> 00:33:34.712 align:middle line:79% position:50% size:68% more consumers are willing to visit Living Social now. 00:33:34.779 --> 00:33:39.717 align:middle line:79% position:50% size:65% So as a result there is multi-homing on both sides 00:33:39.784 --> 00:33:43.454 align:middle line:79% position:50% size:63% it's actually more difficult essentially for 00:33:43.521 --> 00:33:45.957 align:middle line:79% position:50% size:58% the incumbent to drive a competitor out of the 00:33:46.023 --> 00:33:49.293 align:middle line:79% position:50% size:60% market because of this kind of seesaw dynamics. 00:33:49.360 --> 00:33:52.130 align:middle line:79% position:50% size:65% So in order to actually drive a competitor outside 00:33:52.196 --> 00:33:54.632 align:middle line:79% position:50% size:65% the market you need to really fight on both sides 00:33:54.699 --> 00:33:58.636 align:middle line:79% position:50% size:60% making sure this multi-homing tendency is 00:33:58.703 --> 00:34:02.073 align:middle line:79% position:50% size:65% reduced on both sides of the market so that will be 00:34:02.140 --> 00:34:05.910 align:middle line:79% position:50% size:53% a managerial takeaway from this research. 00:34:05.977 --> 00:34:12.917 align:middle line:79% position:50% size:73% So there is a force factor we also worked a lot of times 00:34:12.984 --> 00:34:15.453 align:middle line:79% position:50% size:53% and thinking a lot on this idea around the 00:34:15.520 --> 00:34:20.091 align:middle line:79% position:50% size:63% disintermediation so the idea of disintermediation 00:34:20.158 --> 00:34:23.261 align:middle line:79% position:50% size:63% is that platforms they are essentially middlemen 00:34:23.327 --> 00:34:27.432 align:middle line:79% position:50% size:63% intermediaries in air balling matching but once 00:34:27.498 --> 00:34:29.934 align:middle line:79% position:50% size:50% you're match with a service provider the 00:34:30.001 --> 00:34:32.103 align:middle line:79% position:50% size:60% question is do we really have to stay on the 00:34:32.170 --> 00:34:35.106 align:middle line:79% position:50% size:45% platform afterwards or not? 00:34:35.173 --> 00:34:37.275 align:middle line:79% position:50% size:38% Could you leave the platform? 00:34:37.341 --> 00:34:39.777 align:middle line:79% position:50% size:60% So there is a, you could also describe this 00:34:39.844 --> 00:34:43.114 align:middle line:79% position:50% size:63% disintermediation problem as a leakage issue 00:34:43.181 --> 00:34:44.015 align:middle line:85% position:50% size:30% essentially. 00:34:45.516 --> 00:34:47.785 align:middle line:79% position:50% size:55% So they might be able to essentially connect 00:34:47.852 --> 00:34:52.290 align:middle line:79% position:50% size:60% directly with each other afterwards without the 00:34:52.356 --> 00:34:54.992 align:middle line:79% position:50% size:58% help from the platforms but I started thinking 00:34:55.059 --> 00:34:58.129 align:middle line:79% position:50% size:60% about this idea actually when I first started at 00:34:58.196 --> 00:35:02.233 align:middle line:79% position:50% size:40% the Harvard MBA program in 2013. 00:35:02.300 --> 00:35:06.904 align:middle line:79% position:50% size:63% So we had this case about this hundred year old 00:35:06.971 --> 00:35:08.906 align:middle line:85% position:50% size:63% company called Li & Fung. 00:35:08.973 --> 00:35:11.242 align:middle line:79% position:50% size:63% And Li & Fung some of you may have heard of this 00:35:11.309 --> 00:35:14.745 align:middle line:79% position:50% size:63% company is based in Hong Kong and is company known 00:35:14.812 --> 00:35:18.249 align:middle line:79% position:50% size:53% for connecting retailers in the U.S. 00:35:18.316 --> 00:35:22.653 align:middle line:79% position:50% size:50% like H&M, Gap, with manufactures in Asia 00:35:22.720 --> 00:35:26.724 align:middle line:79% position:50% size:60% mostly in China for instance if H&M needs to 00:35:26.791 --> 00:35:30.094 align:middle line:79% position:50% size:63% make some cloths they could, Li & Fung can help 00:35:30.161 --> 00:35:32.396 align:middle line:79% position:50% size:55% you identify the right manufacturer with the 00:35:32.463 --> 00:35:35.600 align:middle line:79% position:50% size:65% right capacity and they could also do all kinds of 00:35:35.666 --> 00:35:38.603 align:middle line:79% position:50% size:63% monitoring to make sure that the practice follows 00:35:38.669 --> 00:35:42.440 align:middle line:79% position:50% size:60% the standards and Li & Fung also make sure that 00:35:42.507 --> 00:35:45.610 align:middle line:79% position:50% size:63% the cloth are going to be shipped on time to the 00:35:45.676 --> 00:35:46.611 align:middle line:85% position:50% size:35% U.S. retailer. 00:35:46.677 --> 00:35:48.946 align:middle line:79% position:50% size:65% So this company has been doing extremely well so at 00:35:49.013 --> 00:35:54.619 align:middle line:79% position:50% size:58% the time I was teaching this case and Li & Fung 00:35:54.685 --> 00:35:58.289 align:middle line:79% position:50% size:60% probably was performing probably at its peak but 00:35:58.356 --> 00:36:00.291 align:middle line:79% position:50% size:65% afterwards it turns out that Li & Fung performance 00:36:00.358 --> 00:36:04.061 align:middle line:79% position:50% size:55% started to deteriorate and the company today 00:36:04.128 --> 00:36:06.397 align:middle line:79% position:50% size:55% actually lost probably more than 80% of its 00:36:06.464 --> 00:36:08.232 align:middle line:85% position:50% size:60% market value afterwards. 00:36:08.299 --> 00:36:10.902 align:middle line:79% position:50% size:60% So I started to investigate what's going 00:36:10.968 --> 00:36:14.071 align:middle line:79% position:50% size:65% with Li & Fung why hundred year old company couldn't 00:36:14.138 --> 00:36:17.742 align:middle line:85% position:50% size:53% kind of thrive today? 00:36:17.808 --> 00:36:21.746 align:middle line:79% position:50% size:63% So then I discover all these examples in a sense 00:36:21.812 --> 00:36:25.249 align:middle line:79% position:50% size:65% that, oh Walmart initially signed like a multiyear 00:36:25.316 --> 00:36:27.919 align:middle line:79% position:50% size:65% contract to work with Li & Fung but after a few years 00:36:27.985 --> 00:36:30.421 align:middle line:79% position:50% size:63% Walmart discover actually I could just work with 00:36:30.488 --> 00:36:33.257 align:middle line:79% position:50% size:63% these manufactures directly why do I have to 00:36:33.324 --> 00:36:35.593 align:middle line:79% position:50% size:60% pay a huge commission to Li & Fung to coordinate 00:36:35.660 --> 00:36:36.761 align:middle line:85% position:50% size:33% all the work. 00:36:36.827 --> 00:36:40.665 align:middle line:79% position:50% size:60% So this is partly due to the fact Walmart started 00:36:40.731 --> 00:36:44.268 align:middle line:79% position:50% size:65% to know these manufactures extremely well also partly 00:36:44.335 --> 00:36:47.605 align:middle line:79% position:50% size:58% due to the fact that during because we're in 00:36:47.672 --> 00:36:50.942 align:middle line:79% position:50% size:65% the digital age as a result these is actually a 00:36:51.008 --> 00:36:53.778 align:middle line:79% position:50% size:65% lot more information about Chinese manufacturers 00:36:53.844 --> 00:36:56.948 align:middle line:79% position:50% size:60% today you can Google and look for its reputation, 00:36:57.014 --> 00:36:58.616 align:middle line:79% position:50% size:45% look for all the complaints about a 00:36:58.683 --> 00:37:00.051 align:middle line:85% position:50% size:60% particular manufacturer. 00:37:00.117 --> 00:37:04.722 align:middle line:79% position:50% size:73% So as a result you could, you could without the help of 00:37:04.789 --> 00:37:08.392 align:middle line:79% position:50% size:65% Li & Fung you could start to have some trust in some 00:37:08.459 --> 00:37:08.893 align:middle line:85% position:50% size:35% manufacturers. 00:37:08.960 --> 00:37:13.230 align:middle line:79% position:50% size:53% Then I thought as I started to study more 00:37:13.297 --> 00:37:16.934 align:middle line:79% position:50% size:60% platforms and I discover that this issue of 00:37:17.001 --> 00:37:20.905 align:middle line:79% position:50% size:55% disintermediation is a big problem for many 00:37:20.972 --> 00:37:21.906 align:middle line:85% position:50% size:25% platforms. 00:37:21.973 --> 00:37:26.210 align:middle line:79% position:50% size:55% So for instance for Airbnb, so if you have 00:37:26.277 --> 00:37:29.580 align:middle line:79% position:50% size:63% traveled to a city before and stayed with an Airbnb 00:37:29.647 --> 00:37:31.916 align:middle line:79% position:50% size:58% host the next time you travel to that city you 00:37:31.983 --> 00:37:35.252 align:middle line:79% position:50% size:63% probably can just reach out to that host directly 00:37:35.319 --> 00:37:38.422 align:middle line:79% position:50% size:63% and ask him whether you can book the room without 00:37:38.489 --> 00:37:41.592 align:middle line:79% position:50% size:53% going to Airbnb and pay the 15-20 percent 00:37:41.659 --> 00:37:43.928 align:middle line:85% position:50% size:65% commission for the nights. 00:37:43.995 --> 00:37:46.764 align:middle line:79% position:50% size:50% And EBay is also struggling with this 00:37:46.831 --> 00:37:51.268 align:middle line:79% position:50% size:65% issue, TaskRabbit also has this issue because if you 00:37:51.335 --> 00:37:56.107 align:middle line:79% position:50% size:65% use TaskRabbit to hire for instance some cleaner or 00:37:56.173 --> 00:37:59.610 align:middle line:79% position:50% size:65% plumber or people who can assemble furniture in your 00:37:59.677 --> 00:38:03.547 align:middle line:79% position:50% size:75% house and then after the first interaction you probably 00:38:03.614 --> 00:38:05.549 align:middle line:79% position:50% size:65% would have all the contact information about that 00:38:05.616 --> 00:38:09.387 align:middle line:79% position:50% size:60% person and then the next time you want to kind of 00:38:09.453 --> 00:38:12.556 align:middle line:79% position:50% size:60% have someone do the same service, conditioning on 00:38:12.623 --> 00:38:15.226 align:middle line:79% position:50% size:65% the first time interaction was positive you probably 00:38:15.292 --> 00:38:17.895 align:middle line:79% position:50% size:75% won't go to TaskRabbit again and paid the commission. 00:38:17.962 --> 00:38:22.433 align:middle line:79% position:50% size:65% And the reality that the problem with this platform 00:38:22.500 --> 00:38:25.569 align:middle line:79% position:50% size:65% is that if you have positive experience hiring 00:38:25.636 --> 00:38:29.573 align:middle line:79% position:50% size:65% a person the next time you will just interact with 00:38:29.640 --> 00:38:30.908 align:middle line:85% position:50% size:53% that person directly. 00:38:30.975 --> 00:38:33.911 align:middle line:79% position:50% size:65% But if you have a negative experience the first time 00:38:33.978 --> 00:38:37.081 align:middle line:79% position:50% size:63% the next time you will go to another platform to 00:38:37.148 --> 00:38:37.915 align:middle line:85% position:50% size:35% hire a person. 00:38:37.982 --> 00:38:40.584 align:middle line:79% position:50% size:63% So regardless whether the experience is positive or 00:38:40.651 --> 00:38:44.522 align:middle line:79% position:50% size:65% negative the platform will not have that customer 00:38:44.588 --> 00:38:48.926 align:middle line:79% position:50% size:65% again that's a problem and the same if you hire a 00:38:48.993 --> 00:38:52.263 align:middle line:79% position:50% size:65% nanny on Care.com so do we really go back to Care.com 00:38:52.329 --> 00:38:54.098 align:middle line:79% position:50% size:63% every time you want to hire the same person like 00:38:54.165 --> 00:38:55.433 align:middle line:85% position:50% size:33% probably not. 00:38:55.499 --> 00:38:59.603 align:middle line:79% position:50% size:65% And HomeTry used to be a very popular company doing 00:38:59.670 --> 00:39:02.206 align:middle line:79% position:50% size:55% like this kind offline kind of matching 00:39:02.273 --> 00:39:04.675 align:middle line:79% position:50% size:63% essentially for you to hire someone to clean the 00:39:04.742 --> 00:39:07.445 align:middle line:79% position:50% size:50% home and actually died because of this 00:39:07.511 --> 00:39:11.549 align:middle line:79% position:50% size:65% intermediation because they found that they spent 00:39:11.615 --> 00:39:14.885 align:middle line:79% position:50% size:60% a lot of money trying to scale up but the people 00:39:14.952 --> 00:39:17.054 align:middle line:79% position:50% size:43% come and they leave afterwards. 00:39:17.121 --> 00:39:19.890 align:middle line:79% position:50% size:55% So essentially you're paying money for first 00:39:19.957 --> 00:39:24.228 align:middle line:79% position:50% size:60% interaction so which was very stressful for the 00:39:24.295 --> 00:39:28.065 align:middle line:79% position:50% size:55% company and later on I learned the case about 00:39:28.132 --> 00:39:34.572 align:middle line:79% position:50% size:73% this company Upwork it's an online labor market place. 00:39:34.638 --> 00:39:38.375 align:middle line:79% position:50% size:60% So the idea here is that if you want someone to 00:39:38.442 --> 00:39:42.313 align:middle line:79% position:50% size:60% kind of write a piece of software program you can 00:39:42.379 --> 00:39:46.584 align:middle line:79% position:50% size:78% go on Upwork and hire someone mostly outside the U.S. 00:39:46.650 --> 00:39:49.587 align:middle line:79% position:50% size:63% at a lower rate if you want someone to copy edit 00:39:49.653 --> 00:39:52.957 align:middle line:79% position:50% size:75% a paper you can find a copy editor there and my Ph.D. 00:39:53.023 --> 00:39:56.260 align:middle line:79% position:50% size:65% students even told me that you can hire someone to 00:39:56.327 --> 00:39:59.096 align:middle line:79% position:50% size:58% write the presentation, I did not do that. 00:39:59.163 --> 00:40:01.999 align:middle line:79% position:50% size:65% So but you can find a variety of services there. 00:40:02.066 --> 00:40:06.937 align:middle line:79% position:50% size:65% So another case about this company and also discover 00:40:07.004 --> 00:40:09.607 align:middle line:79% position:50% size:60% that the company has been kind of essentially 00:40:09.673 --> 00:40:11.942 align:middle line:79% position:50% size:58% fighting with disintermediation issue 00:40:12.009 --> 00:40:17.348 align:middle line:79% position:50% size:65% and have tried essentially lots of weapons to rid of 00:40:17.414 --> 00:40:18.949 align:middle line:85% position:50% size:45% disintermediation. 00:40:19.016 --> 00:40:23.120 align:middle line:85% position:50% size:48% So later on a Ph.D. 00:40:23.187 --> 00:40:28.626 align:middle line:79% position:50% size:65% student on my grade school, she basically made 00:40:28.692 --> 00:40:30.961 align:middle line:79% position:50% size:58% some arrangement so she actually worked with 00:40:31.028 --> 00:40:35.466 align:middle line:79% position:50% size:65% online kind of freelancer platform so she went there 00:40:35.533 --> 00:40:39.303 align:middle line:79% position:50% size:58% and over the summer she basically investigated 00:40:39.370 --> 00:40:41.972 align:middle line:79% position:50% size:58% disintermediation issue for that particular 00:40:42.039 --> 00:40:45.142 align:middle line:79% position:50% size:58% platform and she also elaborated some further 00:40:45.209 --> 00:40:48.646 align:middle line:79% position:50% size:60% experiments to show to actually kind of support 00:40:48.712 --> 00:40:50.981 align:middle line:85% position:50% size:65% some of her policies here. 00:40:51.048 --> 00:40:54.652 align:middle line:79% position:50% size:43% So one of the key hypothesis in her 00:40:54.718 --> 00:40:59.490 align:middle line:79% position:50% size:63% presentation research was that, it turns out that 00:40:59.557 --> 00:41:02.993 align:middle line:79% position:50% size:63% for a present business to viable and the typical 00:41:03.060 --> 00:41:06.263 align:middle line:79% position:50% size:58% kind of a textbook recommendation would be 00:41:06.330 --> 00:41:09.099 align:middle line:79% position:50% size:55% the platform needs to build lots of trust so 00:41:09.166 --> 00:41:11.936 align:middle line:79% position:50% size:63% that the two sides can interact with each other. 00:41:12.002 --> 00:41:15.439 align:middle line:79% position:50% size:60% So without trust nobody is going to hire an Uber 00:41:15.506 --> 00:41:18.108 align:middle line:79% position:50% size:60% driver essentially you need to have some degree 00:41:18.175 --> 00:41:19.109 align:middle line:85% position:50% size:23% of trust. 00:41:19.176 --> 00:41:21.278 align:middle line:79% position:50% size:53% So they way platforms builds trust is like 00:41:21.345 --> 00:41:24.615 align:middle line:79% position:50% size:58% reputation systems, background checks, I.D. 00:41:24.682 --> 00:41:27.284 align:middle line:79% position:50% size:65% verification, sometimes in the online labor market 00:41:27.351 --> 00:41:32.289 align:middle line:79% position:50% size:65% space you do skill test in a sense if you claim you 00:41:32.356 --> 00:41:36.627 align:middle line:79% position:50% size:65% can kind of translate English into French then I 00:41:36.694 --> 00:41:39.463 align:middle line:79% position:50% size:60% give you a test to make sure you do know French. 00:41:39.530 --> 00:41:43.934 align:middle line:79% position:50% size:65% So essentially the platforms in order to grow 00:41:44.001 --> 00:41:46.804 align:middle line:79% position:50% size:65% their business to enable transaction to make people 00:41:46.871 --> 00:41:48.639 align:middle line:79% position:50% size:55% feel comfortable transacting each other 00:41:48.706 --> 00:41:52.309 align:middle line:79% position:50% size:65% they kind of tried to do a lot to build the trust but 00:41:52.376 --> 00:41:56.981 align:middle line:79% position:50% size:60% if you started to build so much trust then value 00:41:57.047 --> 00:41:59.984 align:middle line:79% position:50% size:60% creation of the platform in terms of facilitating 00:42:00.050 --> 00:42:03.254 align:middle line:79% position:50% size:63% the transactions decrease significantly in a sense 00:42:03.320 --> 00:42:08.259 align:middle line:79% position:50% size:63% if I know that this programmer has been doing 00:42:08.325 --> 00:42:12.930 align:middle line:79% position:50% size:65% perfect job for the past hundred transactions for a 00:42:12.997 --> 00:42:17.601 align:middle line:79% position:50% size:63% different client then why do I need to actually do 00:42:17.668 --> 00:42:19.770 align:middle line:79% position:50% size:63% all the work on the platform and then pay the 00:42:19.837 --> 00:42:23.440 align:middle line:79% position:50% size:63% commission and in order to get some like services 00:42:23.507 --> 00:42:26.110 align:middle line:79% position:50% size:58% such as insurance potential refund if the 00:42:26.176 --> 00:42:30.614 align:middle line:79% position:50% size:60% job doesn't work out, if there is a dispute the 00:42:30.681 --> 00:42:31.949 align:middle line:85% position:50% size:58% platform can intervene. 00:42:32.016 --> 00:42:35.519 align:middle line:79% position:50% size:68% So these services are become less valuable to me. 00:42:35.586 --> 00:42:39.323 align:middle line:79% position:50% size:65% So as a result I have lots of incentives to take the 00:42:39.390 --> 00:42:42.626 align:middle line:79% position:50% size:65% transactions off the platform to deal with this 00:42:42.693 --> 00:42:44.795 align:middle line:79% position:50% size:48% service provider directly, directly. 00:42:44.862 --> 00:42:49.300 align:middle line:79% position:50% size:58% So she did find that in the platform she worked 00:42:49.366 --> 00:42:52.803 align:middle line:79% position:50% size:63% with as platforms started to increase this value 00:42:52.870 --> 00:42:57.241 align:middle line:79% position:50% size:65% creation in matching the two sides and you actually 00:42:57.308 --> 00:42:59.510 align:middle line:79% position:50% size:55% observe more disintermediation as a 00:42:59.576 --> 00:42:59.977 align:middle line:85% position:50% size:18% result. 00:43:00.044 --> 00:43:04.315 align:middle line:79% position:50% size:63% People after being assure that the other side is 00:43:04.381 --> 00:43:07.918 align:middle line:79% position:50% size:63% very reliable very high quality yes and they need 00:43:07.985 --> 00:43:11.155 align:middle line:79% position:50% size:58% to hire that person but I'm more willing to 00:43:11.221 --> 00:43:13.257 align:middle line:79% position:50% size:58% actually work with this person off the platform 00:43:13.324 --> 00:43:15.926 align:middle line:79% position:50% size:60% rather than on the platform to avoid paying 00:43:15.993 --> 00:43:17.261 align:middle line:85% position:50% size:40% the service fee. 00:43:17.328 --> 00:43:21.732 align:middle line:79% position:50% size:65% So then it becomes a dilemma for many platforms 00:43:21.799 --> 00:43:24.435 align:middle line:79% position:50% size:65% in terms of how they actually thrive in a sense 00:43:24.501 --> 00:43:28.439 align:middle line:79% position:50% size:60% that ok you can pay lots of money to build like 00:43:28.505 --> 00:43:31.608 align:middle line:79% position:50% size:58% some installed base to make your platform very 00:43:31.675 --> 00:43:34.578 align:middle line:79% position:50% size:63% popular but the question is how do you retain this 00:43:34.645 --> 00:43:37.948 align:middle line:79% position:50% size:63% people specially the good ones, the good ones are 00:43:38.015 --> 00:43:42.286 align:middle line:79% position:50% size:58% more likely to be hired to go off the platform 00:43:42.353 --> 00:43:44.688 align:middle line:79% position:50% size:55% so that becomes a very difficult issue. 00:43:44.755 --> 00:43:48.625 align:middle line:79% position:50% size:60% So after writing this paper we also started to 00:43:48.692 --> 00:43:50.627 align:middle line:79% position:50% size:48% kind of look around for solutions. 00:43:50.694 --> 00:43:53.964 align:middle line:79% position:50% size:63% So it turns out that yeah actually some solutions 00:43:54.031 --> 00:43:56.133 align:middle line:79% position:50% size:43% platforms can use to mitigate the 00:43:56.200 --> 00:43:57.634 align:middle line:85% position:50% size:60% disintermediation issue. 00:43:57.701 --> 00:44:01.739 align:middle line:79% position:50% size:65% So one example is this platform called Thumbtack, 00:44:01.805 --> 00:44:05.242 align:middle line:79% position:50% size:65% Thumbtack is also a platform basically for you 00:44:05.309 --> 00:44:09.246 align:middle line:79% position:50% size:60% to look for some helpers like plumbers for 00:44:09.313 --> 00:44:12.016 align:middle line:79% position:50% size:50% instance, someone to clean your house. 00:44:12.082 --> 00:44:15.586 align:middle line:79% position:50% size:60% So their business model is more about charging a 00:44:15.652 --> 00:44:18.422 align:middle line:79% position:50% size:65% commission on a particular task, their business model 00:44:18.489 --> 00:44:21.492 align:middle line:79% position:50% size:63% is based on ok charging a fee while the two sides 00:44:21.558 --> 00:44:23.427 align:middle line:79% position:50% size:58% are matched essentially before you actually 00:44:23.494 --> 00:44:25.662 align:middle line:79% position:50% size:60% interact with each other the first time before I 00:44:25.729 --> 00:44:27.765 align:middle line:79% position:50% size:60% tell you the contact information of the other 00:44:27.831 --> 00:44:30.267 align:middle line:79% position:50% size:65% party the service provider will have to pay. 00:44:30.334 --> 00:44:32.603 align:middle line:79% position:50% size:65% Essentially Thumbtack will reach out to the service 00:44:32.669 --> 00:44:34.671 align:middle line:79% position:50% size:63% provider, here is a like a description of the task 00:44:34.738 --> 00:44:38.275 align:middle line:79% position:50% size:53% then you pay a fee in order to get contact 00:44:38.342 --> 00:44:39.777 align:middle line:79% position:50% size:38% information of the other side. 00:44:39.843 --> 00:44:43.947 align:middle line:79% position:50% size:60% So that was the business model but again this 00:44:44.014 --> 00:44:46.316 align:middle line:79% position:50% size:65% business model is actually works for the first 00:44:46.383 --> 00:44:48.419 align:middle line:79% position:50% size:63% transaction but it doesn't work for repeated 00:44:48.485 --> 00:44:52.289 align:middle line:79% position:50% size:60% transactions when the plumber actually did his 00:44:52.356 --> 00:44:56.126 align:middle line:79% position:50% size:58% or her first project in a home then afterwards 00:44:56.193 --> 00:45:00.898 align:middle line:79% position:50% size:65% Thumbtack won't be able to prevent disintermediation. 00:45:00.964 --> 00:45:04.568 align:middle line:79% position:50% size:65% So Upwork later on started to kind of have this 00:45:04.635 --> 00:45:11.241 align:middle line:79% position:50% size:58% basically a new pricing structure to encourage 00:45:11.308 --> 00:45:12.576 align:middle line:79% position:50% size:40% people to stay on the platform. 00:45:12.643 --> 00:45:15.079 align:middle line:79% position:50% size:53% So for the first interaction Upwork is 00:45:15.145 --> 00:45:18.082 align:middle line:79% position:50% size:60% going to charge like ten percent but if you work 00:45:18.148 --> 00:45:21.585 align:middle line:79% position:50% size:58% repeatedly then the accumulative amount for 00:45:21.652 --> 00:45:24.922 align:middle line:79% position:50% size:55% the same payer exceeds like say $500 then the 00:45:24.988 --> 00:45:27.257 align:middle line:79% position:50% size:63% commission is going to be lower, exceeds some other 00:45:27.324 --> 00:45:29.927 align:middle line:79% position:50% size:75% amounts the commission is going to drop even further. 00:45:29.993 --> 00:45:33.597 align:middle line:79% position:50% size:65% But the problem of course is that once you build the 00:45:33.664 --> 00:45:36.100 align:middle line:79% position:50% size:65% sufficient trust specially from this repeated 00:45:36.166 --> 00:45:39.103 align:middle line:79% position:50% size:63% transaction do you really want to pay even if it's 00:45:39.169 --> 00:45:40.070 align:middle line:85% position:50% size:43% just two percent? 00:45:40.137 --> 00:45:41.939 align:middle line:79% position:50% size:43% Do we really want to pay that? 00:45:42.005 --> 00:45:45.509 align:middle line:79% position:50% size:50% So that's another basically a dilemma. 00:45:45.576 --> 00:45:48.445 align:middle line:79% position:50% size:55% So we searched also further for additional 00:45:48.512 --> 00:45:50.781 align:middle line:79% position:50% size:63% solutions because we feel like these solutions are 00:45:50.848 --> 00:45:53.117 align:middle line:85% position:50% size:60% probably not ideal ones. 00:45:53.183 --> 00:45:56.954 align:middle line:79% position:50% size:53% So I came across this Chinese equivalent of 00:45:57.020 --> 00:46:01.391 align:middle line:79% position:50% size:65% Upwork so this company has also been struggling with 00:46:01.458 --> 00:46:03.260 align:middle line:79% position:50% size:43% disintermediation for ten years. 00:46:03.327 --> 00:46:06.763 align:middle line:79% position:50% size:63% So it turns out that disintermediation is even 00:46:06.830 --> 00:46:08.732 align:middle line:79% position:50% size:55% more prominent in the Chinese market because 00:46:08.799 --> 00:46:12.236 align:middle line:79% position:50% size:65% people are very kind of sensitive to the price and 00:46:12.302 --> 00:46:18.075 align:middle line:79% position:50% size:58% ZBJ's platform was charging twenty percent 00:46:18.142 --> 00:46:21.044 align:middle line:79% position:50% size:60% the commission and then the company actually did 00:46:21.111 --> 00:46:25.582 align:middle line:79% position:50% size:63% some kind of a survey of its service providers and 00:46:25.649 --> 00:46:28.919 align:middle line:79% position:50% size:65% also users and it discover actually more than 90 00:46:28.986 --> 00:46:30.420 align:middle line:79% position:50% size:43% percent of the transactions were 00:46:30.487 --> 00:46:31.755 align:middle line:85% position:50% size:43% disintermediated. 00:46:31.822 --> 00:46:34.458 align:middle line:79% position:50% size:65% So the company capture the ten percent essentially of 00:46:34.525 --> 00:46:36.026 align:middle line:85% position:50% size:53% the value it created. 00:46:36.093 --> 00:46:39.263 align:middle line:79% position:50% size:60% So it was a big headache because the company has 00:46:39.329 --> 00:46:41.598 align:middle line:79% position:50% size:65% been looking for solutions for ten years and in the 00:46:41.665 --> 00:46:45.335 align:middle line:79% position:50% size:65% end it actually discovered one action that worked 00:46:45.402 --> 00:46:46.603 align:middle line:79% position:50% size:38% really well for the company. 00:46:46.670 --> 00:46:52.276 align:middle line:79% position:50% size:63% So ZBJ did a lots of data analytics of all the 00:46:52.342 --> 00:46:56.613 align:middle line:79% position:50% size:65% transactions on ZBJ and it discovered that many kind 00:46:56.680 --> 00:46:59.950 align:middle line:79% position:50% size:63% of clients, many users are using the platform to 00:47:00.017 --> 00:47:03.220 align:middle line:79% position:50% size:63% design logos because when it comes to design logos 00:47:03.287 --> 00:47:05.889 align:middle line:79% position:50% size:55% you really need some creativity so why they 00:47:05.956 --> 00:47:07.891 align:middle line:79% position:50% size:58% don't have that kind of creativity you want to 00:47:07.958 --> 00:47:09.893 align:middle line:79% position:50% size:63% outsource it to professionals they design 00:47:09.960 --> 00:47:10.894 align:middle line:85% position:50% size:35% logos for you. 00:47:10.961 --> 00:47:14.064 align:middle line:79% position:50% size:60% And this platform is actually very well known 00:47:14.131 --> 00:47:16.900 align:middle line:79% position:50% size:58% for local design and it turns out that after 00:47:16.967 --> 00:47:20.404 align:middle line:79% position:50% size:63% people design logo on the platform the next thing 00:47:20.470 --> 00:47:25.642 align:middle line:79% position:50% size:65% many clients do is to do business registration they 00:47:25.709 --> 00:47:27.077 align:middle line:85% position:50% size:50% start a new company. 00:47:27.144 --> 00:47:31.682 align:middle line:79% position:50% size:55% Now turns out that in China because business 00:47:31.748 --> 00:47:34.251 align:middle line:79% position:50% size:65% registrations have to deal with the government the 00:47:34.318 --> 00:47:36.753 align:middle line:79% position:50% size:63% process is extremely complicated and very slow 00:47:36.820 --> 00:47:42.426 align:middle line:79% position:50% size:60% and offline, you usually have to hire a law firm 00:47:42.492 --> 00:47:46.930 align:middle line:79% position:50% size:65% from the offline market and that law firm tends to 00:47:46.997 --> 00:47:47.931 align:middle line:85% position:50% size:53% charge lots of money. 00:47:47.998 --> 00:47:51.602 align:middle line:79% position:50% size:58% So ZBJ said ok we could probably provide this 00:47:51.668 --> 00:47:55.505 align:middle line:79% position:50% size:63% service ourselves so the company started to design 00:47:55.572 --> 00:48:00.877 align:middle line:79% position:50% size:63% its IT system to connect with government websites. 00:48:00.944 --> 00:48:04.114 align:middle line:79% position:50% size:58% So that the company can actually automatically 00:48:04.181 --> 00:48:07.050 align:middle line:79% position:50% size:63% file for registration and also company develop lots 00:48:07.117 --> 00:48:11.555 align:middle line:79% position:50% size:60% of tools to kind of help the clients to estimate 00:48:11.622 --> 00:48:14.558 align:middle line:79% position:50% size:63% the likelihood of success because some of the logos 00:48:14.625 --> 00:48:18.161 align:middle line:79% position:50% size:63% if it's a duplicate of some other logos and that 00:48:18.228 --> 00:48:20.430 align:middle line:79% position:50% size:38% is going to get it rejected. 00:48:20.497 --> 00:48:24.568 align:middle line:79% position:50% size:60% So because this is an IT company and it uses a 00:48:24.635 --> 00:48:27.904 align:middle line:79% position:50% size:65% digital solution to enable this process the company 00:48:27.971 --> 00:48:31.742 align:middle line:79% position:50% size:65% was able to charge a lower rate than this offline law 00:48:31.808 --> 00:48:32.743 align:middle line:85% position:50% size:15% firms. 00:48:32.809 --> 00:48:36.413 align:middle line:79% position:50% size:53% And within one or two years the company 00:48:36.480 --> 00:48:39.516 align:middle line:79% position:50% size:53% essentially became the largest trademark 00:48:39.583 --> 00:48:42.753 align:middle line:79% position:50% size:58% registration company in China, the whole China 00:48:42.819 --> 00:48:46.590 align:middle line:79% position:50% size:65% it's the largest one and for that service alone the 00:48:46.657 --> 00:48:51.928 align:middle line:79% position:50% size:60% company generated enough profit to be thriving. 00:48:51.995 --> 00:48:56.600 align:middle line:79% position:50% size:60% So then the company said ok so now giving that I 00:48:56.667 --> 00:48:59.636 align:middle line:79% position:50% size:65% have this service and this service is generating lots 00:48:59.703 --> 00:49:01.538 align:middle line:79% position:50% size:53% of revenue I actually don't have to charge 00:49:01.605 --> 00:49:02.873 align:middle line:85% position:50% size:38% commission now. 00:49:02.939 --> 00:49:05.042 align:middle line:79% position:50% size:60% So the company basically reduced the commission 00:49:05.108 --> 00:49:09.546 align:middle line:79% position:50% size:48% rate to zero and so there won't be any 00:49:09.613 --> 00:49:12.382 align:middle line:79% position:50% size:53% disintermediation now because if it's zero 00:49:12.449 --> 00:49:14.718 align:middle line:79% position:50% size:60% percent commission, everybody wants to bring 00:49:14.785 --> 00:49:16.420 align:middle line:79% position:50% size:55% their terminal transactions online so 00:49:16.486 --> 00:49:19.489 align:middle line:79% position:50% size:55% that they can showcase essentially how many 00:49:19.556 --> 00:49:23.060 align:middle line:79% position:50% size:63% projects I have completed essentially as a way to 00:49:23.126 --> 00:49:25.729 align:middle line:79% position:50% size:55% attract new clients or new service providers. 00:49:25.796 --> 00:49:30.901 align:middle line:79% position:50% size:55% So lots of people basically brought even 00:49:30.967 --> 00:49:32.903 align:middle line:79% position:50% size:65% their offline projects through the platform which 00:49:32.969 --> 00:49:36.573 align:middle line:79% position:50% size:63% gave the platform even more sales leads for this 00:49:36.640 --> 00:49:39.576 align:middle line:79% position:50% size:53% trademark service the company is running. 00:49:39.643 --> 00:49:45.549 align:middle line:79% position:50% size:63% So in the end the company became a unicorn is about 00:49:45.615 --> 00:49:50.087 align:middle line:79% position:50% size:65% to go IPO this year and so after the company had this 00:49:50.153 --> 00:49:52.756 align:middle line:79% position:50% size:65% success the company also replicated this strategies 00:49:52.823 --> 00:49:56.460 align:middle line:79% position:50% size:60% in kind of in different verticals so the company 00:49:56.526 --> 00:49:59.763 align:middle line:79% position:50% size:58% also started to provide accounting services for 00:49:59.830 --> 00:50:02.933 align:middle line:79% position:50% size:58% small enterprises, etc. so for these service 00:50:02.999 --> 00:50:05.435 align:middle line:79% position:50% size:63% providers on the platform if you need like some 00:50:05.502 --> 00:50:07.938 align:middle line:79% position:50% size:60% accounting services the platform can provide it. 00:50:08.004 --> 00:50:16.146 align:middle line:79% position:50% size:63% So from this example once we could see that one 00:50:16.213 --> 00:50:20.450 align:middle line:79% position:50% size:58% solution to deal with disintermediation is to 00:50:20.517 --> 00:50:25.455 align:middle line:79% position:50% size:65% essentially avoid this commission fee essentially 00:50:25.522 --> 00:50:27.958 align:middle line:79% position:50% size:45% and generate lots of revenue selling 00:50:28.024 --> 00:50:29.760 align:middle line:85% position:50% size:58% complementary services. 00:50:29.826 --> 00:50:33.196 align:middle line:79% position:50% size:63% So later on I started to realize this was also one 00:50:33.263 --> 00:50:36.133 align:middle line:79% position:50% size:65% of the reasons why Alibaba won the battle against 00:50:36.199 --> 00:50:37.467 align:middle line:85% position:50% size:35% EBay in China. 00:50:37.534 --> 00:50:40.637 align:middle line:79% position:50% size:65% So EBay actually went into Chinese market before 00:50:40.704 --> 00:50:44.141 align:middle line:79% position:50% size:55% Alibaba with its acquisition of a local 00:50:44.207 --> 00:50:47.644 align:middle line:79% position:50% size:63% market leader essentially there but EBay's business 00:50:47.711 --> 00:50:50.981 align:middle line:79% position:50% size:65% model is about charging a commission fee essentially 00:50:51.047 --> 00:50:55.152 align:middle line:79% position:50% size:63% so in the US market it charges 10-15 percent, in 00:50:55.218 --> 00:50:57.220 align:middle line:79% position:50% size:48% Chinese market EBay decided to charge 00:50:57.287 --> 00:50:59.322 align:middle line:79% position:50% size:58% essentially more or less the same amount of 00:50:59.389 --> 00:51:01.091 align:middle line:85% position:50% size:28% commission. 00:51:01.158 --> 00:51:04.094 align:middle line:79% position:50% size:65% And when Alibaba came into the market and Alibaba 00:51:04.161 --> 00:51:07.931 align:middle line:79% position:50% size:60% started to say ok given we're the follower we're 00:51:07.998 --> 00:51:09.766 align:middle line:79% position:50% size:65% not going to charge you anything we just offer the 00:51:09.833 --> 00:51:12.602 align:middle line:79% position:50% size:63% service Taobao, initially we offer completely for 00:51:12.669 --> 00:51:14.438 align:middle line:79% position:50% size:35% free to buyers and sellers. 00:51:14.504 --> 00:51:19.609 align:middle line:79% position:50% size:60% And so when you charge a commission fee one thing 00:51:19.676 --> 00:51:23.580 align:middle line:79% position:50% size:53% you need to do is you don't want buyers and 00:51:23.647 --> 00:51:26.950 align:middle line:79% position:50% size:70% seller to communicate before the transaction is over. 00:51:27.017 --> 00:51:28.952 align:middle line:79% position:50% size:65% So on EBay it turns out it's very difficult to buy 00:51:29.019 --> 00:51:30.954 align:middle line:79% position:50% size:63% a product because EBay won't allow the two sides 00:51:31.021 --> 00:51:35.959 align:middle line:79% position:50% size:70% to talk but Alibaba came in with no transaction fees. 00:51:36.026 --> 00:51:38.128 align:middle line:79% position:50% size:60% So Alibaba actually encourage people to talk 00:51:38.195 --> 00:51:43.300 align:middle line:79% position:50% size:65% so many buyers and sellers found that it's just far 00:51:43.366 --> 00:51:46.470 align:middle line:79% position:50% size:60% more convenient essentially to just kind 00:51:46.536 --> 00:51:49.139 align:middle line:79% position:50% size:58% of be able to talk to each other, discuss the 00:51:49.206 --> 00:51:52.142 align:middle line:79% position:50% size:60% product, and discuss the negotiated price before 00:51:52.209 --> 00:51:53.477 align:middle line:85% position:50% size:60% the transaction is over. 00:51:53.543 --> 00:51:56.313 align:middle line:79% position:50% size:63% So many people decided to switch and within four 00:51:56.379 --> 00:51:59.282 align:middle line:79% position:50% size:50% years Alibaba had basically the market 00:51:59.349 --> 00:52:01.585 align:middle line:79% position:50% size:63% leadership with an 80 percent market share EBay 00:52:01.651 --> 00:52:03.086 align:middle line:79% position:50% size:38% had to quit the Chinese market. 00:52:03.153 --> 00:52:05.922 align:middle line:79% position:50% size:63% But then people started to question this business 00:52:05.989 --> 00:52:08.425 align:middle line:79% position:50% size:63% model it's like giving away free hamburger right 00:52:08.492 --> 00:52:12.095 align:middle line:79% position:50% size:73% you kill the EBay so then how would you survive given 00:52:12.162 --> 00:52:15.098 align:middle line:79% position:50% size:65% that you're not making any money but it turns out 00:52:15.165 --> 00:52:19.102 align:middle line:79% position:50% size:65% that once you build a huge market place you then find 00:52:19.169 --> 00:52:20.937 align:middle line:79% position:50% size:53% other ways to generate the profits. 00:52:21.004 --> 00:52:24.274 align:middle line:79% position:50% size:63% So today actually Alibaba's Taobao still is 00:52:24.341 --> 00:52:27.277 align:middle line:79% position:50% size:48% free to buyers and sellers, the Taobao 00:52:27.344 --> 00:52:30.113 align:middle line:79% position:50% size:65% platform generates most of revenues from advertising 00:52:30.180 --> 00:52:33.483 align:middle line:79% position:50% size:60% so because there are so many merchants there and 00:52:33.550 --> 00:52:36.786 align:middle line:79% position:50% size:63% when consumers search for a particular product is 00:52:36.853 --> 00:52:39.789 align:middle line:79% position:50% size:63% not clear which merchants to list the first but if 00:52:39.856 --> 00:52:41.958 align:middle line:79% position:50% size:53% you want to be listed first you need to pay 00:52:42.025 --> 00:52:43.960 align:middle line:79% position:50% size:58% essentially advertising fee essentially. 00:52:44.027 --> 00:52:46.563 align:middle line:79% position:50% size:65% So this is essentially the Google model the search 00:52:46.630 --> 00:52:49.132 align:middle line:79% position:50% size:65% engine is completely free but most of the money will 00:52:49.199 --> 00:52:50.800 align:middle line:85% position:50% size:55% come from advertisers. 00:52:50.867 --> 00:52:54.804 align:middle line:79% position:50% size:45% So as a result one way to deal with 00:52:54.871 --> 00:52:59.242 align:middle line:79% position:50% size:63% disintermediation from this research is to start 00:52:59.309 --> 00:53:01.912 align:middle line:79% position:50% size:63% thinking about getting rid of the commission fee 00:53:01.978 --> 00:53:04.147 align:middle line:79% position:50% size:50% maybe generate money from other sources. 00:53:06.983 --> 00:53:09.920 align:middle line:79% position:50% size:63% So another network strategy we discovered is 00:53:09.986 --> 00:53:13.924 align:middle line:79% position:50% size:58% about the bridging strategy so the idea is 00:53:13.990 --> 00:53:17.227 align:middle line:79% position:50% size:65% that if you have built one successful network if you 00:53:17.294 --> 00:53:20.430 align:middle line:79% position:50% size:63% have built one successful network one way to grow 00:53:20.497 --> 00:53:24.334 align:middle line:79% position:50% size:60% further is by bridging a current network with new 00:53:24.401 --> 00:53:27.938 align:middle line:79% position:50% size:65% networks to each other so as if in the end like they 00:53:28.004 --> 00:53:30.040 align:middle line:79% position:50% size:53% behave just like just a single big network. 00:53:30.106 --> 00:53:33.610 align:middle line:79% position:50% size:65% And it turns out that this strategy is extremely 00:53:33.677 --> 00:53:36.680 align:middle line:79% position:50% size:63% effective in the platform setting because much of 00:53:36.746 --> 00:53:40.951 align:middle line:79% position:50% size:63% the kind of assets or resources of a particular 00:53:41.017 --> 00:53:44.454 align:middle line:79% position:50% size:58% platform is around the installed base or data. 00:53:44.521 --> 00:53:48.224 align:middle line:79% position:50% size:65% And installed bases and the data are arguably very 00:53:48.291 --> 00:53:51.294 align:middle line:79% position:50% size:63% valuable in multiple scenarios across multiple 00:53:51.361 --> 00:53:51.962 align:middle line:85% position:50% size:20% markets. 00:53:52.028 --> 00:53:58.201 align:middle line:79% position:50% size:68% So I learned a case about this fintech startup so now 00:53:58.268 --> 00:54:00.337 align:middle line:79% position:50% size:65% it's actually the largest fintech company in 00:54:00.403 --> 00:54:04.074 align:middle line:79% position:50% size:60% the world now called Ant Financial which is a 00:54:04.140 --> 00:54:05.909 align:middle line:85% position:50% size:53% spinoff from Alibaba. 00:54:05.976 --> 00:54:12.916 align:middle line:79% position:50% size:68% So this company started like a basically -- started 00:54:12.983 --> 00:54:15.585 align:middle line:79% position:50% size:65% as a result of this Alibaba's market place one 00:54:15.652 --> 00:54:19.255 align:middle line:79% position:50% size:63% is Taobao which is C-to-C market and TMall which is 00:54:19.322 --> 00:54:19.923 align:middle line:85% position:50% size:35% B-to-C market. 00:54:19.990 --> 00:54:23.093 align:middle line:79% position:50% size:53% So Alibaba at that time was already very 00:54:23.159 --> 00:54:26.496 align:middle line:79% position:50% size:65% successful like attracting like millions of sellers 00:54:26.563 --> 00:54:29.766 align:middle line:79% position:50% size:60% connecting with millions of buyers but in China 00:54:29.833 --> 00:54:32.002 align:middle line:79% position:50% size:63% because of the low penetration of the credit 00:54:32.068 --> 00:54:35.438 align:middle line:79% position:50% size:65% cards Alibaba had to develop the payment method 00:54:35.505 --> 00:54:37.774 align:middle line:85% position:50% size:60% by itself called Alipay. 00:54:37.841 --> 00:54:42.112 align:middle line:79% position:50% size:63% And later on Alibaba also opened up Alipay not just 00:54:42.178 --> 00:54:45.048 align:middle line:79% position:50% size:60% to merchant or buyers on Alibaba platform also to 00:54:45.115 --> 00:54:47.450 align:middle line:79% position:50% size:63% other kind of online platforms or even offline 00:54:47.517 --> 00:54:47.951 align:middle line:85% position:50% size:25% merchants. 00:54:48.018 --> 00:54:52.288 align:middle line:79% position:50% size:63% So Alibaba also decided to create a separate unit 00:54:52.355 --> 00:54:55.458 align:middle line:79% position:50% size:55% called Ant Financial to manage this payment 00:54:55.525 --> 00:55:00.230 align:middle line:79% position:50% size:68% service. From there because Ant Financial collected so 00:55:00.296 --> 00:55:03.900 align:middle line:79% position:50% size:63% much data as a result of this payment tool because 00:55:03.967 --> 00:55:06.903 align:middle line:79% position:50% size:65% it knows like for instance how many times people go 00:55:06.970 --> 00:55:09.506 align:middle line:79% position:50% size:65% to theaters, it knows when you travel, whether you 00:55:09.572 --> 00:55:12.409 align:middle line:79% position:50% size:58% fly business class or economy class, it knows 00:55:12.475 --> 00:55:17.080 align:middle line:79% position:50% size:65% like what kind of products you buy, how expensive 00:55:17.147 --> 00:55:18.081 align:middle line:85% position:50% size:48% these products are? 00:55:18.148 --> 00:55:21.751 align:middle line:79% position:50% size:60% So as a result from this data, Ant Financial was 00:55:21.818 --> 00:55:25.922 align:middle line:79% position:50% size:63% able to create credit scores to individuals and 00:55:25.989 --> 00:55:29.592 align:middle line:79% position:50% size:63% also to all the merchants on Alibaba network in a 00:55:29.659 --> 00:55:31.928 align:middle line:79% position:50% size:60% sense that Ant Financial knows like for instance 00:55:31.995 --> 00:55:35.432 align:middle line:79% position:50% size:55% how many units are you selling and how much 00:55:35.498 --> 00:55:37.767 align:middle line:79% position:50% size:55% inventory you have etc., on the platform. 00:55:37.834 --> 00:55:41.438 align:middle line:79% position:50% size:63% So once you have the credit score, around that 00:55:41.504 --> 00:55:44.607 align:middle line:79% position:50% size:60% time like most of the Chinese citizens they do 00:55:44.674 --> 00:55:47.944 align:middle line:79% position:50% size:63% not have a credit score the infrastructure is not 00:55:48.011 --> 00:55:51.614 align:middle line:79% position:50% size:58% as develop as in the US but once you have the 00:55:51.681 --> 00:55:53.450 align:middle line:79% position:50% size:55% credit score you could actually issue loans 00:55:53.516 --> 00:55:56.286 align:middle line:79% position:50% size:60% because you can evaluate risk very accurate. 00:55:56.352 --> 00:55:59.956 align:middle line:79% position:50% size:55% So the company started this online banking 00:56:00.023 --> 00:56:03.727 align:middle line:79% position:50% size:65% service called MYBank and MYBank was able to kind of 00:56:03.793 --> 00:56:06.396 align:middle line:79% position:50% size:58% essentially issue loans to all the merchants on 00:56:06.463 --> 00:56:12.569 align:middle line:79% position:50% size:65% Alibaba because these merchants contributed lots 00:56:12.635 --> 00:56:14.571 align:middle line:79% position:50% size:60% of data essentially to this company to evaluate 00:56:14.637 --> 00:56:18.408 align:middle line:79% position:50% size:58% their performance and Alibaba's Ant Financial 00:56:18.475 --> 00:56:22.412 align:middle line:79% position:50% size:63% also could issue loans to consumers based on you 00:56:22.479 --> 00:56:25.582 align:middle line:79% position:50% size:63% past purchasing behaviors and these loans actually 00:56:25.648 --> 00:56:30.053 align:middle line:79% position:50% size:65% most of them are very kind of small loans like two 00:56:30.120 --> 00:56:32.922 align:middle line:79% position:50% size:58% thousand dollars, three thousand dollars and 00:56:32.989 --> 00:56:35.592 align:middle line:79% position:50% size:63% traditionally it's almost impossible for you to get 00:56:35.658 --> 00:56:38.261 align:middle line:79% position:50% size:60% these kind of loans from the banks because banks 00:56:38.328 --> 00:56:41.931 align:middle line:79% position:50% size:63% are not interested in customers like asking for 00:56:41.998 --> 00:56:45.335 align:middle line:79% position:50% size:63% small loans because in order for bank to kind of 00:56:45.401 --> 00:56:47.937 align:middle line:79% position:50% size:63% issue a loan they have to do so much paperwork to 00:56:48.004 --> 00:56:49.939 align:middle line:79% position:50% size:63% figure out how many house properties you have, 00:56:50.006 --> 00:56:51.307 align:middle line:79% position:50% size:38% whether you own a car or not. 00:56:51.374 --> 00:56:54.611 align:middle line:79% position:50% size:65% So if you request for a small loan if they justify 00:56:54.677 --> 00:56:58.615 align:middle line:79% position:50% size:65% all these transactions cost there is just no data 00:56:58.681 --> 00:57:02.218 align:middle line:79% position:50% size:65% analytics, Ant Financial was able to basically kind 00:57:02.285 --> 00:57:06.222 align:middle line:79% position:50% size:75% of help these merchants or individual consumers in China. 00:57:06.289 --> 00:57:12.395 align:middle line:79% position:50% size:65% And then because in Alipay many users are moving 00:57:12.462 --> 00:57:14.898 align:middle line:79% position:50% size:58% their money from their bank accounts to Alipay 00:57:14.964 --> 00:57:18.067 align:middle line:79% position:50% size:63% account so that you can make purchases so lots of 00:57:18.134 --> 00:57:21.404 align:middle line:79% position:50% size:75% money essentially would be seating in the Alipay account. 00:57:21.471 --> 00:57:24.541 align:middle line:79% position:50% size:58% So Alibaba also started this money market fund 00:57:24.607 --> 00:57:28.578 align:middle line:79% position:50% size:53% called Yu'e Bao which allows customers, 00:57:28.645 --> 00:57:31.915 align:middle line:79% position:50% size:65% consumers to actually kind of invest in this fund and 00:57:31.981 --> 00:57:36.186 align:middle line:79% position:50% size:63% within actually four years this is already the 00:57:36.252 --> 00:57:38.755 align:middle line:79% position:50% size:50% largest money market fund in the world. 00:57:38.822 --> 00:57:44.928 align:middle line:79% position:50% size:63% So as you can see that in the end Ant Financial is 00:57:44.994 --> 00:57:47.096 align:middle line:79% position:50% size:50% almost like another platform essentially 00:57:47.163 --> 00:57:50.934 align:middle line:79% position:50% size:63% linking kind of consumers with not just the 00:57:51.000 --> 00:57:53.169 align:middle line:79% position:50% size:58% merchants here but also all kinds of merchants 00:57:53.236 --> 00:57:56.739 align:middle line:79% position:50% size:43% outside Alibaba's echo system. 00:57:56.806 --> 00:57:59.943 align:middle line:79% position:50% size:63% So this two networks are being bridge together and 00:58:00.009 --> 00:58:03.546 align:middle line:79% position:50% size:50% essentially co-grow, co-evolve together. 00:58:03.613 --> 00:58:07.550 align:middle line:79% position:50% size:63% It sounds like as much as you are talking about the 00:58:07.617 --> 00:58:10.386 align:middle line:79% position:50% size:60% network structures being indeterminate of market 00:58:10.453 --> 00:58:12.722 align:middle line:79% position:50% size:60% structure business success, you keep coming 00:58:12.789 --> 00:58:14.557 align:middle line:79% position:50% size:58% back to kind of the differentiated business 00:58:14.624 --> 00:58:17.560 align:middle line:79% position:50% size:58% models being an equally significant variable in 00:58:17.627 --> 00:58:19.729 align:middle line:79% position:50% size:60% terms of the way the market structure evolves 00:58:19.796 --> 00:58:21.898 align:middle line:79% position:50% size:55% and in terms whether business will succeed? 00:58:21.965 --> 00:58:24.901 align:middle line:79% position:50% size:45% Yeah, I completely agree, yes. 00:58:24.968 --> 00:58:30.573 align:middle line:79% position:50% size:58% Totally, so network structures in this case 00:58:30.640 --> 00:58:34.377 align:middle line:79% position:50% size:55% I would say are very indulgenes, indulgenes 00:58:34.444 --> 00:58:37.113 align:middle line:79% position:50% size:63% is not like you see oh this is a platform market 00:58:37.180 --> 00:58:40.183 align:middle line:79% position:50% size:65% that's a network structure and there is another 00:58:40.250 --> 00:58:42.585 align:middle line:79% position:50% size:53% platform market there it is like a network 00:58:42.652 --> 00:58:44.053 align:middle line:79% position:50% size:40% structure has to be different. 00:58:44.120 --> 00:58:49.425 align:middle line:79% position:50% size:65% But it turns out that firm strategies, also external 00:58:49.492 --> 00:58:51.127 align:middle line:79% position:50% size:48% factors can trigger network structures 00:58:51.194 --> 00:58:54.430 align:middle line:79% position:50% size:65% significantly and that has important implications for 00:58:54.497 --> 00:58:57.433 align:middle line:79% position:50% size:48% the competitiveness in that market. 00:58:57.500 --> 00:59:03.206 align:middle line:79% position:50% size:60% So again so in this case it's a delivery strategy 00:59:03.273 --> 00:59:07.043 align:middle line:79% position:50% size:65% in bridging the networks and making sure that these 00:59:07.110 --> 00:59:09.212 align:middle line:79% position:50% size:60% two networks can support each other essentially 00:59:09.279 --> 00:59:12.715 align:middle line:79% position:50% size:65% Ant Financial becomes very successful because so much 00:59:12.782 --> 00:59:15.218 align:middle line:79% position:50% size:50% data are coming from Alibaba's platform. 00:59:15.285 --> 00:59:19.222 align:middle line:79% position:50% size:65% And Taobao or Zhima become more successful because 00:59:19.289 --> 00:59:22.225 align:middle line:79% position:50% size:53% Ant Financial started to kind of enable its 00:59:22.292 --> 00:59:24.394 align:middle line:79% position:50% size:60% transactions, issuing loans so that the people 00:59:24.460 --> 00:59:26.696 align:middle line:79% position:50% size:53% can purchase more and startups can actually 00:59:26.763 --> 00:59:28.564 align:middle line:79% position:50% size:45% build up inventory accordingly. 00:59:28.631 --> 00:59:32.902 align:middle line:79% position:50% size:63% So these two networks can co-evolve but then 00:59:32.969 --> 00:59:36.506 align:middle line:79% position:50% size:60% we started to think okay maybe if that's the case 00:59:36.572 --> 00:59:40.576 align:middle line:79% position:50% size:65% lots of positive feedbacks you can identify in this 00:59:40.643 --> 00:59:41.244 align:middle line:85% position:50% size:33% two networks. 00:59:41.311 --> 00:59:43.746 align:middle line:79% position:50% size:65% So is this like a winner takes all, should we worry 00:59:43.813 --> 00:59:47.917 align:middle line:79% position:50% size:58% that ok then in the end a monopoly is going to 00:59:47.984 --> 00:59:49.252 align:middle line:79% position:50% size:35% emerge in this market place? 00:59:49.319 --> 00:59:53.589 align:middle line:79% position:50% size:63% It turns out not, it turns out that because of 00:59:53.656 --> 00:59:57.260 align:middle line:79% position:50% size:65% network bridging other giants can also leverage a 00:59:57.327 --> 01:00:00.096 align:middle line:79% position:50% size:60% network bridging getting into like for example 01:00:00.163 --> 01:00:01.264 align:middle line:85% position:50% size:43% payment services. 01:00:01.331 --> 01:00:05.435 align:middle line:79% position:50% size:63% So Tencent which has this very popular kind of 01:00:05.501 --> 01:00:08.938 align:middle line:79% position:50% size:63% application called WeChat so it's like kind of 01:00:09.005 --> 01:00:14.277 align:middle line:79% position:50% size:60% Facebook of China, quite similar but because 01:00:14.344 --> 01:00:17.013 align:middle line:79% position:50% size:60% Facebook is also blocked in China so WeChat has 01:00:17.080 --> 01:00:22.452 align:middle line:79% position:50% size:55% almost like a monopoly where is close to like 01:00:22.518 --> 01:00:24.454 align:middle line:79% position:50% size:63% probably 80-90 percent of penetration pretty much 01:00:24.520 --> 01:00:26.222 align:middle line:85% position:50% size:65% all Chinese are on WeChat. 01:00:26.289 --> 01:00:30.193 align:middle line:79% position:50% size:63% So because people are chatting so much and also 01:00:30.259 --> 01:00:33.463 align:middle line:79% position:50% size:63% there are opportunities for people to essentially 01:00:33.529 --> 01:00:36.799 align:middle line:79% position:50% size:63% kind of transfer money to each other so if we kind 01:00:36.866 --> 01:00:39.302 align:middle line:79% position:50% size:60% of go to eat together I want to kind of transfer 01:00:39.369 --> 01:00:41.437 align:middle line:79% position:50% size:50% some money to you or during the Chinese 01:00:41.504 --> 01:00:46.309 align:middle line:79% position:50% size:58% New Year it's kind of a cultural that people 01:00:46.376 --> 01:00:48.978 align:middle line:79% position:50% size:53% actually send the red pockets to each other 01:00:49.045 --> 01:00:50.146 align:middle line:79% position:50% size:40% essentially with some money. 01:00:50.213 --> 01:00:52.648 align:middle line:79% position:50% size:58% So WeChat also has this feature built in to it, 01:00:52.715 --> 01:00:54.650 align:middle line:79% position:50% size:38% based on the social network. 01:00:54.717 --> 01:00:59.989 align:middle line:79% position:50% size:68% So as a result WeChat also successful essentially 01:01:00.056 --> 01:01:02.091 align:middle line:79% position:50% size:60% another network bridging started to grow another 01:01:02.158 --> 01:01:02.925 align:middle line:85% position:50% size:43% payment services. 01:01:02.992 --> 01:01:07.196 align:middle line:79% position:50% size:63% So today like WeChat is also very big having like 01:01:07.263 --> 01:01:12.268 align:middle line:79% position:50% size:65% basically in terms of market share it's probably 01:01:12.335 --> 01:01:14.937 align:middle line:79% position:50% size:65% almost at the same level as Alipay now depending on 01:01:15.004 --> 01:01:16.939 align:middle line:79% position:50% size:48% the data source essentially equally 01:01:17.006 --> 01:01:20.943 align:middle line:79% position:50% size:65% popular because the WeChat product itself it's 01:01:21.010 --> 01:01:23.946 align:middle line:79% position:50% size:60% extremely popular so as a result even though you 01:01:24.013 --> 01:01:26.616 align:middle line:79% position:50% size:63% might think oh lots network effects, positive 01:01:26.682 --> 01:01:29.952 align:middle line:79% position:50% size:63% feedback groups, the market is not going to be 01:01:30.019 --> 01:01:33.289 align:middle line:79% position:50% size:60% competitive but it turns out the other giants can 01:01:33.356 --> 01:01:36.459 align:middle line:79% position:50% size:65% also leverage this network bridging in entering this 01:01:36.526 --> 01:01:38.661 align:middle line:79% position:50% size:45% market becomes and even competitor. 01:01:38.728 --> 01:01:42.498 align:middle line:79% position:50% size:60% So today actually many Chinese would be on both 01:01:42.565 --> 01:01:46.636 align:middle line:79% position:50% size:63% WeChat and Alipay because they use these two 01:01:46.702 --> 01:01:48.504 align:middle line:79% position:50% size:50% products in different scenarios. 01:01:48.571 --> 01:01:51.307 align:middle line:79% position:50% size:60% If you want to transfer money in WeChat then you 01:01:51.374 --> 01:01:53.476 align:middle line:79% position:50% size:65% probably need to use WeChat Pay but if you want 01:01:53.543 --> 01:01:56.646 align:middle line:79% position:50% size:63% to buy on Alibaba's platform, you have to use 01:01:56.712 --> 01:02:00.083 align:middle line:79% position:50% size:50% Alipay because Taobao/TMall doesn't 01:02:00.149 --> 01:02:00.983 align:middle line:85% position:50% size:35% except WeChat. 01:02:03.986 --> 01:02:08.424 align:middle line:79% position:50% size:58% So as you can see that there are like a lot of 01:02:08.491 --> 01:02:11.761 align:middle line:79% position:50% size:63% factors, lots of factors here I mentioned the five 01:02:11.828 --> 01:02:15.431 align:middle line:79% position:50% size:58% but presumably as I continue to do research 01:02:15.498 --> 01:02:17.467 align:middle line:79% position:50% size:65% and other scholars continue to do research we 01:02:17.533 --> 01:02:20.269 align:middle line:79% position:50% size:48% could identify more factors influencing 01:02:20.336 --> 01:02:21.938 align:middle line:79% position:50% size:45% essentially the competitiveness of 01:02:22.004 --> 01:02:25.441 align:middle line:79% position:50% size:63% platform markets and also the performance of these 01:02:25.508 --> 01:02:26.943 align:middle line:85% position:50% size:48% platform companies. 01:02:27.009 --> 01:02:30.613 align:middle line:79% position:50% size:65% Could you speak a little bit to the degree in which 01:02:30.680 --> 01:02:34.684 align:middle line:79% position:50% size:55% the actual industry may limit the way that 01:02:34.750 --> 01:02:36.452 align:middle line:85% position:50% size:55% platforms can compete? 01:02:36.519 --> 01:02:40.556 align:middle line:79% position:50% size:65% So for example with Uber you have this local market 01:02:40.623 --> 01:02:43.793 align:middle line:79% position:50% size:65% structure because it would be inefficient for cars to 01:02:43.860 --> 01:02:49.799 align:middle line:79% position:50% size:58% drive across different states, where as in the 01:02:49.866 --> 01:02:54.237 align:middle line:79% position:50% size:65% sale of products on Amazon it's easier to transport 01:02:54.303 --> 01:02:56.973 align:middle line:79% position:50% size:45% materials across different regions? 01:02:57.039 --> 01:02:57.907 align:middle line:85% position:50% size:33% Totally, yes. 01:02:57.974 --> 01:03:02.912 align:middle line:79% position:50% size:63% So, this is to do with the clustering story in a 01:03:02.979 --> 01:03:05.114 align:middle line:79% position:50% size:65% sense whether they matching or eventually the 01:03:05.181 --> 01:03:08.718 align:middle line:79% position:50% size:65% service or transaction has to take place locally so 01:03:08.784 --> 01:03:11.254 align:middle line:79% position:50% size:65% within a geographical region or within some kind 01:03:11.320 --> 01:03:12.588 align:middle line:85% position:50% size:30% of a region. 01:03:12.655 --> 01:03:16.259 align:middle line:79% position:50% size:55% Some kind of network cluster or where their 01:03:16.325 --> 01:03:20.062 align:middle line:79% position:50% size:63% products can move freely from one area to another. 01:03:20.129 --> 01:03:23.266 align:middle line:79% position:50% size:65% Certainly markets factors would change this and then 01:03:23.332 --> 01:03:26.068 align:middle line:79% position:50% size:48% you see in addition to market factors 01:03:26.135 --> 01:03:28.070 align:middle line:79% position:50% size:43% technologies can also change this. 01:03:28.137 --> 01:03:29.939 align:middle line:79% position:50% size:48% So think about like driverless car 01:03:30.006 --> 01:03:32.942 align:middle line:79% position:50% size:60% technologies you think that okay right now like 01:03:33.009 --> 01:03:35.611 align:middle line:79% position:50% size:60% an Uber, like Lyft, these are like two sided 01:03:35.678 --> 01:03:38.614 align:middle line:79% position:50% size:55% networks you can think must be some kind of 01:03:38.681 --> 01:03:41.450 align:middle line:79% position:50% size:65% network effect even though they may not necessarily 01:03:41.517 --> 01:03:43.452 align:middle line:79% position:50% size:55% be very strong between riders and drivers. 01:03:43.519 --> 01:03:46.489 align:middle line:79% position:50% size:58% But think about the war when we have driverless 01:03:46.556 --> 01:03:50.126 align:middle line:79% position:50% size:58% car technologies then essentially one side is 01:03:50.193 --> 01:03:52.695 align:middle line:79% position:50% size:43% gone, like driver side is gone. 01:03:52.762 --> 01:03:56.966 align:middle line:79% position:50% size:65% So there is no network effects in the end because 01:03:57.033 --> 01:04:01.404 align:middle line:79% position:50% size:65% any new entrant even like a BMW or Toyota as long as 01:04:01.470 --> 01:04:03.906 align:middle line:79% position:50% size:65% you can manufacture cars, you can just or driverless 01:04:03.973 --> 01:04:06.409 align:middle line:79% position:50% size:63% cars, you can just deploy like maybe thousands of 01:04:06.475 --> 01:04:09.245 align:middle line:79% position:50% size:58% cars, they just drive alone you can enter the 01:04:09.312 --> 01:04:09.912 align:middle line:85% position:50% size:40% market that way. 01:04:09.979 --> 01:04:13.416 align:middle line:79% position:50% size:55% So because there is no network effect anymore 01:04:13.482 --> 01:04:17.086 align:middle line:79% position:50% size:65% so technology can actually kill network effects but 01:04:17.153 --> 01:04:20.256 align:middle line:79% position:50% size:63% the story didn't end here but then companies can 01:04:20.323 --> 01:04:22.325 align:middle line:79% position:50% size:58% actually regenerate the network effects again 01:04:22.391 --> 01:04:26.262 align:middle line:79% position:50% size:63% because once you are in the driverless car you're 01:04:26.329 --> 01:04:28.297 align:middle line:79% position:50% size:65% not going to pay attention to the road situation 01:04:28.364 --> 01:04:32.602 align:middle line:79% position:50% size:55% anymore you started to kind of care about for 01:04:32.668 --> 01:04:35.104 align:middle line:79% position:50% size:63% instance on the content provided in the car maybe 01:04:35.171 --> 01:04:37.673 align:middle line:79% position:50% size:55% a very big screen with lots of YouTube videos 01:04:37.740 --> 01:04:41.611 align:middle line:79% position:50% size:65% etc. Then content delivery can generate the different 01:04:41.677 --> 01:04:44.447 align:middle line:79% position:50% size:63% kinds of network effects because you maybe want to 01:04:44.513 --> 01:04:46.782 align:middle line:79% position:50% size:55% kind of go to like for instance Google car 01:04:46.849 --> 01:04:49.952 align:middle line:79% position:50% size:65% because Google has YouTube and YouTube is exclusive 01:04:50.019 --> 01:04:51.287 align:middle line:85% position:50% size:63% available for Google car. 01:04:51.354 --> 01:04:54.223 align:middle line:79% position:50% size:65% And then you're going to be like I want to finish a 01:04:54.290 --> 01:04:58.461 align:middle line:79% position:50% size:60% movie so I use Google to find one and then maybe 01:04:58.527 --> 01:05:02.565 align:middle line:79% position:50% size:63% like Apple has like kind of a car services so like 01:05:02.632 --> 01:05:06.402 align:middle line:79% position:50% size:55% oh iTune services exclusively offered in 01:05:06.469 --> 01:05:10.740 align:middle line:79% position:50% size:65% Apple driverless car then in that case actually that 01:05:10.806 --> 01:05:13.409 align:middle line:79% position:50% size:60% allows these competitors to reintroduce network 01:05:13.476 --> 01:05:18.080 align:middle line:79% position:50% size:63% effects in the market to drive kind of essentially 01:05:18.147 --> 01:05:20.416 align:middle line:79% position:50% size:58% platform competition essential they become a 01:05:20.483 --> 01:05:22.752 align:middle line:79% position:50% size:58% platform again but it's with a different set of 01:05:22.818 --> 01:05:28.424 align:middle line:79% position:50% size:80% players is not with Uber it's all this content providers. 01:05:28.491 --> 01:05:30.926 align:middle line:79% position:50% size:60% So both these market structures technology as 01:05:30.993 --> 01:05:35.765 align:middle line:79% position:50% size:65% they evolve they could kind of make the situation 01:05:35.831 --> 01:05:37.266 align:middle line:85% position:50% size:48% even more exciting. 01:05:37.333 --> 01:05:40.102 align:middle line:79% position:50% size:63% So that's why I predict I could still have my job 01:05:40.169 --> 01:05:44.440 align:middle line:79% position:50% size:55% for many years to come yeah, and just to 01:05:44.507 --> 01:05:46.108 align:middle line:79% position:50% size:58% summarize what are some of the implications? 01:05:46.175 --> 01:05:49.445 align:middle line:79% position:50% size:60% So not all platform businesses are viable or 01:05:49.512 --> 01:05:51.947 align:middle line:79% position:50% size:65% sustainable because of all of these network structure 01:05:52.014 --> 01:05:55.951 align:middle line:79% position:50% size:63% issues and sometimes you just cannot overcome some 01:05:56.018 --> 01:05:58.954 align:middle line:79% position:50% size:65% of the challenges and even with significant network 01:05:59.021 --> 01:06:01.891 align:middle line:79% position:50% size:65% effects it's not always the platform being used as 01:06:01.957 --> 01:06:04.060 align:middle line:79% position:50% size:60% necessarily winner takes all, winner takes most, 01:06:04.126 --> 01:06:07.396 align:middle line:79% position:50% size:65% you have to take a holistic view of all these 01:06:07.463 --> 01:06:10.900 align:middle line:79% position:50% size:58% factors to determine whether the platform is 01:06:10.966 --> 01:06:14.236 align:middle line:79% position:50% size:63% actually going to become a monopoly like platform. 01:06:14.303 --> 01:06:17.406 align:middle line:79% position:50% size:63% And the entry barrier can be low when the incumbent 01:06:17.473 --> 01:06:20.576 align:middle line:79% position:50% size:48% or the entrant can leverage multi-home 01:06:20.643 --> 01:06:24.914 align:middle line:79% position:50% size:50% network clustering or network bridging. 01:06:24.980 --> 01:06:26.782 align:middle line:79% position:50% size:53% So you really have to think about all these 01:06:26.849 --> 01:06:29.251 align:middle line:79% position:50% size:60% factors to think about whether some markets are 01:06:29.318 --> 01:06:34.090 align:middle line:79% position:50% size:58% contestable or not and I made have seem to pay 01:06:34.156 --> 01:06:36.592 align:middle line:79% position:50% size:53% close attention to it turns out that these 01:06:36.659 --> 01:06:40.262 align:middle line:79% position:50% size:60% factors I mentioned none of them are specific or 01:06:40.329 --> 01:06:42.765 align:middle line:79% position:50% size:58% fixed for a particular platform market, firm's 01:06:42.832 --> 01:06:45.434 align:middle line:79% position:50% size:55% strategy, government regulations, all these 01:06:45.501 --> 01:06:48.938 align:middle line:79% position:50% size:60% technology kind of advancements, they could 01:06:49.004 --> 01:06:51.073 align:middle line:79% position:50% size:48% all change these network properties. 01:06:51.140 --> 01:06:53.776 align:middle line:79% position:50% size:55% So as a result we have to kind of pay close 01:06:53.843 --> 01:06:57.113 align:middle line:79% position:50% size:63% attention to these external factors and then 01:06:57.179 --> 01:06:59.448 align:middle line:79% position:50% size:63% re-evaluate the situation from time to time. 01:06:59.515 --> 01:07:04.420 align:middle line:79% position:50% size:65% So that concludes my talk and thank you so much feel 01:07:04.487 --> 01:07:05.388 align:middle line:85% position:50% size:43% free to email me. 01:07:05.454 --> 01:07:05.888 align:middle line:85% position:50% size:25% Thank you. 01:07:05.955 --> 01:07:16.899 align:middle line:85% position:50% size:25% [Applause] 01:07:16.966 --> 01:07:17.900 align:middle line:85% position:50% size:55% Any questions for him? 01:07:17.967 --> 01:07:24.740 align:middle line:79% position:50% size:60% So, if we can go back to the first example, first 01:07:24.807 --> 01:07:27.243 align:middle line:79% position:50% size:55% issue with strength of network effects I'm 01:07:27.309 --> 01:07:29.779 align:middle line:79% position:50% size:58% interested can you talk a little bit more about 01:07:29.845 --> 01:07:34.984 align:middle line:79% position:50% size:65% specifically the examples of the video game console, 01:07:35.050 --> 01:07:40.322 align:middle line:79% position:50% size:65% and DVD or VHS, Betamax, DVD, Blu-ray and all that? 01:07:44.994 --> 01:07:49.932 align:middle line:79% position:50% size:60% I'm interested if you've done any thinking about 01:07:49.999 --> 01:07:53.769 align:middle line:79% position:50% size:63% whether the important factor in the strength of 01:07:53.836 --> 01:07:58.941 align:middle line:79% position:50% size:55% network effects is the price of acquiring the 01:07:59.008 --> 01:08:02.878 align:middle line:79% position:50% size:63% platform versus the later acquisitions like the 01:08:02.945 --> 01:08:05.614 align:middle line:79% position:50% size:80% price of the video game consoles versus the video games. 01:08:05.681 --> 01:08:08.350 align:middle line:79% position:50% size:53% If there is something important about the 01:08:08.417 --> 01:08:11.887 align:middle line:79% position:50% size:60% differentiation in those prices or if has more to 01:08:11.954 --> 01:08:15.891 align:middle line:79% position:50% size:60% do with the number of options available versus 01:08:15.958 --> 01:08:17.760 align:middle line:79% position:50% size:53% the number of options you're actually going 01:08:17.827 --> 01:08:20.896 align:middle line:79% position:50% size:63% take, can you talk a little bit more about the 01:08:20.963 --> 01:08:23.399 align:middle line:79% position:50% size:50% strength of network effects and what the 01:08:23.466 --> 01:08:24.400 align:middle line:79% position:50% size:43% important factors are in that? 01:08:24.467 --> 01:08:28.737 align:middle line:79% position:50% size:65% Surprise is usually a very important in terms of 01:08:28.804 --> 01:08:30.573 align:middle line:79% position:50% size:60% influencing multi-homing tendencies. 01:08:30.639 --> 01:08:35.077 align:middle line:79% position:50% size:63% So if the prices are low consumers are more likely 01:08:35.144 --> 01:08:38.347 align:middle line:79% position:50% size:63% to multi-home so for instance like Disney also 01:08:38.414 --> 01:08:41.083 align:middle line:79% position:50% size:63% introduces its streaming services to rival against 01:08:41.150 --> 01:08:43.919 align:middle line:85% position:50% size:20% Netflix. 01:08:43.986 --> 01:08:46.755 align:middle line:79% position:50% size:60% It turns out that if you look at the Disney price 01:08:46.822 --> 01:08:50.426 align:middle line:79% position:50% size:65% less than $10/month and Netflix subscription price 01:08:50.493 --> 01:08:53.028 align:middle line:79% position:50% size:63% also less than $10/month, they are relative low so 01:08:53.095 --> 01:08:56.265 align:middle line:79% position:50% size:53% you would expect that in that case lots of 01:08:56.332 --> 01:08:58.434 align:middle line:79% position:50% size:58% consumers may subscribe to both services. 01:08:58.501 --> 01:09:03.038 align:middle line:79% position:50% size:60% So the result it may be quite easy for Disney to 01:09:03.105 --> 01:09:04.874 align:middle line:79% position:50% size:58% enter this market place even though you would 01:09:04.940 --> 01:09:07.543 align:middle line:79% position:50% size:63% think Netflix has lots of customers installed base 01:09:07.610 --> 01:09:09.879 align:middle line:79% position:50% size:65% already, lots of content already because of the low 01:09:09.945 --> 01:09:11.881 align:middle line:85% position:50% size:63% price for both platforms. 01:09:11.947 --> 01:09:16.151 align:middle line:79% position:50% size:65% In terms of network effects, so I'm also doing 01:09:16.218 --> 01:09:19.889 align:middle line:79% position:50% size:65% some research some more thinking about essentially 01:09:19.955 --> 01:09:23.626 align:middle line:79% position:50% size:65% what really kind of determines essentially the 01:09:23.692 --> 01:09:25.561 align:middle line:79% position:50% size:40% strength of the network effects. 01:09:25.628 --> 01:09:27.563 align:middle line:79% position:50% size:65% What are some of the factors we can look at and 01:09:27.630 --> 01:09:29.231 align:middle line:79% position:50% size:53% the result through empirical analysis we 01:09:29.298 --> 01:09:31.567 align:middle line:79% position:50% size:65% could know that oh network effects is strong on 01:09:31.634 --> 01:09:32.902 align:middle line:85% position:50% size:53% network effects wake. 01:09:32.968 --> 01:09:37.273 align:middle line:79% position:50% size:60% So one factor as you mentioned is essentially 01:09:37.339 --> 01:09:40.342 align:middle line:79% position:50% size:53% the need for variety, that's actually very 01:09:40.409 --> 01:09:40.910 align:middle line:85% position:50% size:25% important. 01:09:40.976 --> 01:09:45.247 align:middle line:79% position:50% size:65% So if you don't care about variety say like an on 01:09:45.314 --> 01:09:48.217 align:middle line:79% position:50% size:65% Uber network I don't really care who is driving 01:09:48.284 --> 01:09:51.420 align:middle line:79% position:50% size:63% the car or what kind of car you have, then I only 01:09:51.487 --> 01:09:55.090 align:middle line:79% position:50% size:58% care about just my wait time then the network 01:09:55.157 --> 01:09:59.261 align:middle line:79% position:50% size:65% effect cannot be very strong because a new comer 01:09:59.328 --> 01:10:03.933 align:middle line:79% position:50% size:53% can just deploy, dump enough money to hire 01:10:03.999 --> 01:10:07.770 align:middle line:79% position:50% size:65% enough drivers to offer service in a local city to 01:10:07.836 --> 01:10:08.604 align:middle line:85% position:50% size:30% get started. 01:10:08.671 --> 01:10:13.776 align:middle line:79% position:50% size:65% So as a result the barrier entry cannot be very high, 01:10:13.842 --> 01:10:16.946 align:middle line:79% position:50% size:70% well another thing is also whether this market is a hit 01:10:17.012 --> 01:10:19.281 align:middle line:79% position:50% size:43% driven or is a long term market. 01:10:19.348 --> 01:10:23.452 align:middle line:79% position:50% size:63% If everybody's preference concentrate on just the 01:10:23.519 --> 01:10:28.457 align:middle line:79% position:50% size:63% few hits in that market's place so that Airbnb 01:10:28.524 --> 01:10:33.095 align:middle line:79% position:50% size:65% everyone just want to stay in like a small number of 01:10:33.162 --> 01:10:35.230 align:middle line:79% position:50% size:65% properties we don't really care about the variety 01:10:35.297 --> 01:10:38.300 align:middle line:79% position:50% size:58% then that's also the network effects in that 01:10:38.367 --> 01:10:40.369 align:middle line:79% position:50% size:53% case is also going to be relatively low. 01:10:40.436 --> 01:10:44.974 align:middle line:79% position:50% size:63% Because a new comer could just get like a few hits 01:10:45.040 --> 01:10:47.309 align:middle line:79% position:50% size:50% and then you could start your business. 01:10:47.376 --> 01:10:50.813 align:middle line:79% position:50% size:58% The barrier to entry is not going to be very 01:10:50.879 --> 01:10:53.816 align:middle line:79% position:50% size:63% strong but the good thing for Airbnb is that it 01:10:53.882 --> 01:10:55.818 align:middle line:79% position:50% size:58% turns out that people's preferences are very 01:10:55.884 --> 01:10:58.988 align:middle line:79% position:50% size:65% diverse because when you travel to a place yeah you 01:10:59.054 --> 01:11:01.256 align:middle line:79% position:50% size:65% probably want to stay in the downtown area, someone 01:11:01.323 --> 01:11:02.758 align:middle line:79% position:50% size:38% will want to stay in suburb. 01:11:02.825 --> 01:11:05.928 align:middle line:79% position:50% size:65% So they have like different constraints as a 01:11:05.995 --> 01:11:07.596 align:middle line:79% position:50% size:58% result that they prefer different kinds of 01:11:07.663 --> 01:11:10.099 align:middle line:79% position:50% size:63% properties so in that case the variety actually 01:11:10.165 --> 01:11:12.935 align:middle line:79% position:50% size:45% matters a lot in that case as well. 01:11:13.002 --> 01:11:15.104 align:middle line:79% position:50% size:60% So these are the factors we could use to think 01:11:15.170 --> 01:11:17.940 align:middle line:79% position:50% size:53% about the strength of network effects here. 01:11:18.007 --> 01:11:20.776 align:middle line:79% position:50% size:60% So in example I showed it's really because like 01:11:20.843 --> 01:11:27.116 align:middle line:79% position:50% size:65% in the DVD or the VHS, Betamax case consumers are 01:11:27.182 --> 01:11:29.618 align:middle line:79% position:50% size:53% actually renting lots of titles per months. 01:11:29.685 --> 01:11:33.122 align:middle line:79% position:50% size:60% So the variety becomes very important so around 01:11:33.188 --> 01:11:36.558 align:middle line:79% position:50% size:60% that time like the close of the rental stores, 01:11:36.625 --> 01:11:39.962 align:middle line:79% position:50% size:63% rental businesses also enhanced it in a sense if 01:11:40.029 --> 01:11:43.298 align:middle line:79% position:50% size:65% you operate a rental store you also need to decide 01:11:43.365 --> 01:11:46.468 align:middle line:79% position:50% size:63% like how to allocate your shelf spaces in a sense 01:11:46.535 --> 01:11:52.975 align:middle line:79% position:50% size:65% maybe 50% will be VHS tapes, 50% will be Betamax 01:11:53.042 --> 01:11:55.477 align:middle line:79% position:50% size:63% tapes, like it turns out that as the market starts 01:11:55.544 --> 01:11:59.314 align:middle line:79% position:50% size:60% to evolve the choice of this rental store owners 01:11:59.381 --> 01:12:03.118 align:middle line:79% position:50% size:63% is a function essentially of the market share of 01:12:03.185 --> 01:12:06.088 align:middle line:79% position:50% size:58% these two standards and actually amplifies this 01:12:06.155 --> 01:12:08.924 align:middle line:79% position:50% size:63% network effects in the end because rental stores 01:12:08.991 --> 01:12:13.262 align:middle line:79% position:50% size:65% owners after realizing, oh VHS is actually about to 01:12:13.328 --> 01:12:16.098 align:middle line:79% position:50% size:55% take off then I should just forget about the 01:12:16.165 --> 01:12:18.834 align:middle line:79% position:50% size:60% Betamax standard and desire to also influence 01:12:18.901 --> 01:12:22.438 align:middle line:79% position:50% size:63% consumers choice of which kind of machine to get in 01:12:22.504 --> 01:12:26.442 align:middle line:79% position:50% size:65% the end in a local market, so all these things should 01:12:26.508 --> 01:12:28.010 align:middle line:79% position:50% size:35% be taken into consideration. 01:12:31.013 --> 01:12:34.116 align:middle line:79% position:50% size:55% I was wondering if you comment at all about 01:12:34.183 --> 01:12:39.121 align:middle line:79% position:50% size:63% alternative strategies to prevent multi-homing, 01:12:39.188 --> 01:12:44.626 align:middle line:79% position:50% size:70% I was wondering if you could give us your thoughts on 01:12:44.693 --> 01:12:46.962 align:middle line:79% position:50% size:63% alternative strategies to prevent multi-homing and 01:12:47.029 --> 01:12:50.299 align:middle line:79% position:50% size:55% you brought up the one example in China where 01:12:50.365 --> 01:12:53.969 align:middle line:79% position:50% size:63% basically they enter this space that was super 01:12:54.036 --> 01:12:59.641 align:middle line:79% position:50% size:58% complicated as a way to sort of enabling the 01:12:59.708 --> 01:13:01.744 align:middle line:79% position:50% size:45% network, are there other strategies? 01:13:01.810 --> 01:13:07.916 align:middle line:79% position:50% size:63% Yes, certainly so part of the reason why lots of 01:13:07.983 --> 01:13:10.519 align:middle line:79% position:50% size:63% Amazon merchants actually I mean some of them must 01:13:10.586 --> 01:13:14.089 align:middle line:79% position:50% size:63% be multi-homing but it turns out that Amazon can 01:13:14.156 --> 01:13:16.058 align:middle line:79% position:50% size:50% reduce multi-homing, actually slow this 01:13:16.125 --> 01:13:18.827 align:middle line:79% position:50% size:58% fulfilled by the Amazon Services because you 01:13:18.894 --> 01:13:22.598 align:middle line:79% position:50% size:55% provide this logistics support for all the 01:13:22.664 --> 01:13:22.931 align:middle line:85% position:50% size:25% merchants. 01:13:22.998 --> 01:13:25.300 align:middle line:79% position:50% size:45% Merchants send all inventory to your 01:13:25.367 --> 01:13:28.203 align:middle line:79% position:50% size:58% warehouse and then you would ship them and you 01:13:28.270 --> 01:13:28.937 align:middle line:85% position:50% size:58% handle all the returns. 01:13:29.004 --> 01:13:33.008 align:middle line:79% position:50% size:65% So if you look at Amazon's pricing structure any 01:13:33.075 --> 01:13:35.444 align:middle line:79% position:50% size:60% merchants even if you're not selling products on 01:13:35.511 --> 01:13:38.614 align:middle line:79% position:50% size:55% Amazon you can use its warehouse because you 01:13:38.680 --> 01:13:41.016 align:middle line:79% position:50% size:58% create economy of scale in the shipping part. 01:13:41.083 --> 01:13:46.622 align:middle line:79% position:50% size:60% And so they welcome you but on the other hand if 01:13:46.688 --> 01:13:49.958 align:middle line:79% position:50% size:65% you're not selling on Amazon actually Amazon has 01:13:50.025 --> 01:13:53.662 align:middle line:79% position:50% size:65% increase the fee structure for you basically you have 01:13:53.729 --> 01:13:54.630 align:middle line:85% position:50% size:30% to pay more. 01:13:54.696 --> 01:13:57.800 align:middle line:79% position:50% size:65% So that actually so as a result if you're operating 01:13:57.866 --> 01:14:00.402 align:middle line:79% position:50% size:50% on EBay and you try to leverage Amazon's 01:14:00.469 --> 01:14:04.072 align:middle line:79% position:50% size:60% fulfillment services the cost structure is a lot 01:14:04.139 --> 01:14:07.576 align:middle line:79% position:50% size:63% higher compared to if you operate on Amazon and use 01:14:07.643 --> 01:14:09.578 align:middle line:79% position:50% size:55% Amazon's fulfillment services, fulfilled by 01:14:09.645 --> 01:14:09.912 align:middle line:85% position:50% size:18% Amazon. 01:14:09.978 --> 01:14:12.247 align:middle line:79% position:50% size:58% So that actually reduce incentive for the 01:14:12.314 --> 01:14:14.917 align:middle line:79% position:50% size:65% merchants once you started the two years FBA, 01:14:14.983 --> 01:14:17.319 align:middle line:79% position:50% size:63% fulfilled by Amazon Services to multi-home on 01:14:17.386 --> 01:14:19.087 align:middle line:79% position:50% size:33% EBay that was one strategy. 01:14:19.154 --> 01:14:21.924 align:middle line:79% position:50% size:63% And also this Prime thing also that deals with 01:14:21.990 --> 01:14:25.427 align:middle line:79% position:50% size:65% consumer multi-homing so today when I buy, look for 01:14:25.494 --> 01:14:29.097 align:middle line:79% position:50% size:63% some products I no longer go through Google I go 01:14:29.164 --> 01:14:31.366 align:middle line:79% position:50% size:55% through all the products availability. 01:14:31.433 --> 01:14:33.936 align:middle line:79% position:50% size:65% Oh this on Amazon, this is the price on Walmart, this 01:14:34.002 --> 01:14:36.705 align:middle line:79% position:50% size:60% is the price if you're a Prime member a natural 01:14:36.772 --> 01:14:39.107 align:middle line:79% position:50% size:58% starting point to go to would be actually the 01:14:39.174 --> 01:14:39.942 align:middle line:85% position:50% size:38% Amazon website. 01:14:40.008 --> 01:14:42.978 align:middle line:79% position:50% size:60% So this also reduces multi-homing tendencies, 01:14:43.045 --> 01:14:46.281 align:middle line:79% position:50% size:63% so Amazon actually manage to reduce multi-homing 01:14:46.348 --> 01:14:49.117 align:middle line:79% position:50% size:63% essentially on both sides of the market both on the 01:14:49.184 --> 01:14:51.453 align:middle line:79% position:50% size:55% merchant side and also the consumer side. 01:14:51.520 --> 01:14:53.789 align:middle line:79% position:50% size:55% So as we saw from this Groupon versus Living 01:14:53.856 --> 01:14:56.959 align:middle line:79% position:50% size:65% Social example this strategy actually is going 01:14:57.025 --> 01:14:58.794 align:middle line:79% position:50% size:58% to be effective whether it'll be actually very 01:14:58.861 --> 01:15:01.897 align:middle line:79% position:50% size:58% effective that probably contributes to the fact 01:15:01.964 --> 01:15:03.632 align:middle line:79% position:50% size:53% that Amazon is valued so much today. 01:15:09.972 --> 01:15:12.140 align:middle line:79% position:50% size:65% So early you were talking about network bridging and 01:15:12.207 --> 01:15:16.745 align:middle line:79% position:50% size:65% based on these large companies that essentially 01:15:16.812 --> 01:15:20.582 align:middle line:79% position:50% size:60% amassed a certain amount of data and that certain 01:15:20.649 --> 01:15:24.653 align:middle line:79% position:50% size:55% small markets are able to itemize for money 01:15:24.720 --> 01:15:27.589 align:middle line:79% position:50% size:63% transfers, credit scores, things like that. 01:15:27.656 --> 01:15:31.760 align:middle line:79% position:50% size:60% Is there anything unique about how those markets 01:15:31.827 --> 01:15:35.197 align:middle line:79% position:50% size:55% operate in and as far as entry is concern so 01:15:35.264 --> 01:15:37.766 align:middle line:79% position:50% size:63% someone who is looking to enter into those markets 01:15:37.833 --> 01:15:42.271 align:middle line:79% position:50% size:55% that itemize from the business behemoths you 01:15:42.337 --> 01:15:44.606 align:middle line:79% position:50% size:55% know is there anything unique about entry or 01:15:44.673 --> 01:15:47.109 align:middle line:79% position:50% size:65% should we think about those markets as operating 01:15:47.175 --> 01:15:50.612 align:middle line:79% position:50% size:63% in the same way as frankly any other market? 01:15:50.679 --> 01:15:57.286 align:middle line:79% position:50% size:65% What kinds of factors are you thinking about like in 01:15:57.352 --> 01:15:58.453 align:middle line:85% position:50% size:50% terms of uniqueness? 01:15:58.520 --> 01:16:01.890 align:middle line:79% position:50% size:65% For example would you have to enter into like in some 01:16:01.957 --> 01:16:04.159 align:middle line:79% position:50% size:58% other market that the behemoth is playing in, 01:16:04.226 --> 01:16:07.029 align:middle line:79% position:50% size:53% are your strategies effective against the 01:16:07.095 --> 01:16:12.567 align:middle line:79% position:50% size:58% behemoth essentially has you research showed 01:16:12.634 --> 01:16:15.437 align:middle line:79% position:50% size:63% anything interesting or unique about this market? 01:16:15.504 --> 01:16:20.409 align:middle line:79% position:50% size:63% Totally, so when there is a possibility for bigger 01:16:20.475 --> 01:16:23.578 align:middle line:79% position:50% size:60% players to leverage a network bridging to grow 01:16:23.645 --> 01:16:26.748 align:middle line:79% position:50% size:63% the opportunity for small player is actually very 01:16:26.815 --> 01:16:30.919 align:middle line:79% position:50% size:60% limited because in terms of network bridging the 01:16:30.986 --> 01:16:33.088 align:middle line:79% position:50% size:65% idea is that across multiple networks there is 01:16:33.155 --> 01:16:35.457 align:middle line:79% position:50% size:58% actually synergy across this multiple networks. 01:16:35.524 --> 01:16:39.828 align:middle line:79% position:50% size:65% So for a new comer, for an new startup is actually 01:16:39.895 --> 01:16:42.998 align:middle line:79% position:50% size:60% very very difficult to survive in a competitive 01:16:43.065 --> 01:16:45.367 align:middle line:79% position:50% size:53% environment like this because you're only 01:16:45.434 --> 01:16:48.937 align:middle line:79% position:50% size:65% operating in one product line but the incumbent has 01:16:49.004 --> 01:16:49.938 align:middle line:79% position:50% size:58% a portfolio of products that are interconnected 01:16:50.005 --> 01:16:51.039 align:middle line:85% position:50% size:40% with each other. 01:16:51.106 --> 01:16:55.944 align:middle line:79% position:50% size:65% So as a result most of the time when we observe this 01:16:56.011 --> 01:16:58.280 align:middle line:79% position:50% size:65% kind of market competition we're are really looking 01:16:58.347 --> 01:17:00.716 align:middle line:79% position:50% size:65% at the competition between the giants among the 01:17:00.782 --> 01:17:01.883 align:middle line:85% position:50% size:18% giants. 01:17:01.950 --> 01:17:03.685 align:middle line:79% position:50% size:60% So in terms of payment services another example 01:17:03.752 --> 01:17:06.888 align:middle line:79% position:50% size:65% is like for instance Grab, Grab is basically Uber 01:17:06.955 --> 01:17:08.390 align:middle line:79% position:50% size:40% equivalent in South East Asia. 01:17:08.457 --> 01:17:10.726 align:middle line:79% position:50% size:65% They are also thinking about leveraging this ride 01:17:10.792 --> 01:17:14.229 align:middle line:79% position:50% size:58% sharing network to add this payment feature as 01:17:14.296 --> 01:17:17.399 align:middle line:79% position:50% size:63% well as a way to enter the payment services area 01:17:17.466 --> 01:17:21.403 align:middle line:79% position:50% size:55% because I mean somehow the payment service I 01:17:21.470 --> 01:17:23.739 align:middle line:79% position:50% size:63% predicted that in the future like it's going to 01:17:23.805 --> 01:17:27.075 align:middle line:79% position:50% size:63% get more and more crowded because this is actually 01:17:27.142 --> 01:17:30.579 align:middle line:79% position:50% size:55% one of the fundamental need of everybody in a 01:17:30.645 --> 01:17:33.415 align:middle line:79% position:50% size:55% sense that you have to pay basically everyday 01:17:33.482 --> 01:17:34.583 align:middle line:85% position:50% size:38% multiple times. 01:17:34.649 --> 01:17:38.420 align:middle line:79% position:50% size:65% So this is actually very attractable market both in 01:17:38.487 --> 01:17:41.923 align:middle line:79% position:50% size:58% terms of enlarging the installed base and also 01:17:41.990 --> 01:17:44.926 align:middle line:79% position:50% size:65% collecting lots of data to enable any other kind of 01:17:44.993 --> 01:17:45.927 align:middle line:85% position:50% size:23% services. 01:17:45.994 --> 01:17:49.264 align:middle line:79% position:50% size:63% So as a result in that space is going to be very 01:17:49.331 --> 01:17:52.601 align:middle line:79% position:50% size:60% competitive but there is actually not so easy for 01:17:52.667 --> 01:17:55.937 align:middle line:79% position:50% size:65% startup to enter very easy once you are facing the 01:17:56.004 --> 01:18:00.042 align:middle line:79% position:50% size:55% competitors based on this network bridging. 01:18:00.108 --> 01:18:06.381 align:middle line:79% position:50% size:65% But clearly even like in other settings like in the 01:18:06.448 --> 01:18:09.384 align:middle line:79% position:50% size:63% case of Uber you see that the startup can still 01:18:09.451 --> 01:18:11.386 align:middle line:79% position:50% size:63% leverage some other network properties like a 01:18:11.453 --> 01:18:14.389 align:middle line:79% position:50% size:53% multi-homing, network clustering to enter 01:18:14.456 --> 01:18:15.223 align:middle line:85% position:50% size:33% successfully. 01:18:15.290 --> 01:18:18.226 align:middle line:79% position:50% size:60% In New York for instance in addition to Uber and 01:18:18.293 --> 01:18:21.396 align:middle line:79% position:50% size:63% Lyft I heard there are actually lots of startups 01:18:21.463 --> 01:18:23.632 align:middle line:79% position:50% size:48% of the ride sharing services there. 01:18:23.698 --> 01:18:27.969 align:middle line:79% position:50% size:63% Ok, we probably have time for one more question. 01:18:32.274 --> 01:18:34.910 align:middle line:79% position:50% size:60% Alright, thank you everyone for joining us. 01:18:34.976 --> 01:18:36.978 align:middle line:85% position:50% size:25% [Applause] 01:18:37.045 --> 01:18:39.381 align:middle line:85% position:50% size:25% Thank you.