1 00:00:01,355 --> 00:00:02,488 Welcome everyone. 2 00:00:02,506 --> 00:00:04,995 Today, we're going to be talking about O-ring theory. 3 00:00:05,273 --> 00:00:08,334 Turns out O-ring theory has implications for development, 4 00:00:08,334 --> 00:00:11,690 but also has implications for the industrial organization 5 00:00:11,690 --> 00:00:13,233 of developed economies, 6 00:00:13,264 --> 00:00:17,086 that is, for how firms and workers are organized -- 7 00:00:17,224 --> 00:00:21,852 when maximizing the value of production requires that we work as a team. 8 00:00:22,276 --> 00:00:24,519 O-ring theory is also going to have implications 9 00:00:24,519 --> 00:00:26,920 for our understanding of inequality. 10 00:00:27,339 --> 00:00:28,511 Let's take a look. 11 00:00:30,154 --> 00:00:32,885 The O-ring theory of development is primarily due to 12 00:00:32,885 --> 00:00:35,654 a very creative economist called Michael Kremer. 13 00:00:36,342 --> 00:00:40,103 Now, we're going to call O-ring production or an O-ring production function -- 14 00:00:40,413 --> 00:00:42,930 if the production task has the following attributes: 15 00:00:43,288 --> 00:00:46,936 It's going to depend upon completing a series of tasks. 16 00:00:47,455 --> 00:00:50,393 Moreover, failure at any one of these tasks 17 00:00:50,486 --> 00:00:53,505 is going to reduce the value of the entire product, 18 00:00:53,505 --> 00:00:54,584 perhaps to zero. 19 00:00:54,896 --> 00:00:57,200 So, this is the weakest link problem. 20 00:00:57,450 --> 00:01:01,941 One weak link in a chain destroys the entire value of the chain. 21 00:01:02,413 --> 00:01:05,403 It's also going to be the case, with an O-ring production function, 22 00:01:05,403 --> 00:01:08,665 that we can't substitute quantity for quality. 23 00:01:08,855 --> 00:01:10,662 I'll come back to that in a minute. 24 00:01:11,006 --> 00:01:13,355 Let me give you some examples: Take microchips. 25 00:01:13,525 --> 00:01:19,245 A single spec of dust on a microchip run ruins all of those microchips. 26 00:01:19,451 --> 00:01:20,919 Or think about a souffle. 27 00:01:21,526 --> 00:01:24,702 To make a souffle, you've got to get the ingredients right, 28 00:01:24,978 --> 00:01:26,887 you've got to get the temperature right, 29 00:01:26,887 --> 00:01:30,114 you've got to take the souffle out of the oven at just the right time. 30 00:01:30,313 --> 00:01:32,871 Mess up on any one of these tasks 31 00:01:32,871 --> 00:01:35,117 and the souffle will collapse. 32 00:01:35,561 --> 00:01:36,930 What about a musical? 33 00:01:37,004 --> 00:01:41,102 To make a great musical, you're going to require an excellent lyricist, 34 00:01:41,179 --> 00:01:42,812 a wonderful composer, 35 00:01:42,812 --> 00:01:44,074 a great director, 36 00:01:44,074 --> 00:01:45,530 superb performers. 37 00:01:45,642 --> 00:01:47,541 You need an entire team -- 38 00:01:47,746 --> 00:01:50,489 and the talents of that team must all come together 39 00:01:50,489 --> 00:01:53,051 at the right time, at the right place. 40 00:01:53,467 --> 00:01:57,013 Notice what we mean here by: You can't substitute quantity for quality. 41 00:01:57,344 --> 00:01:59,984 Two mediocre chefs are not going to be better 42 00:01:59,984 --> 00:02:02,055 at making a wonderful souffle -- 43 00:02:02,252 --> 00:02:04,648 than one great chef. 44 00:02:04,648 --> 00:02:05,784 For a musical, -- 45 00:02:06,102 --> 00:02:07,948 if you don't have Stephen Sondheim, -- 46 00:02:08,595 --> 00:02:10,860 then you can't replace him by saying, 47 00:02:10,860 --> 00:02:16,567 "We'll add three, or four, or five, or more mediocre composers, 48 00:02:16,567 --> 00:02:19,767 to make up for the fact that we don't have one great composer." 49 00:02:19,767 --> 00:02:21,404 It doesn't work that way. 50 00:02:21,804 --> 00:02:24,126 Now, why do we call this O-ring production? 51 00:02:24,126 --> 00:02:28,068 Well, as you probably know, this is due to the destruction, -- 52 00:02:28,671 --> 00:02:32,777 the explosion of the Challenger space shuttle. 53 00:02:33,301 --> 00:02:37,198 As famously shown by the physicist Richard Feynman, -- 54 00:02:37,406 --> 00:02:42,212 the mayor cause of that explosion was the failure of a very simple product, 55 00:02:42,212 --> 00:02:46,027 a failure of the O-ring to expand because it was too cold. 56 00:02:46,154 --> 00:02:49,082 So, this cheap really inconsequential product; 57 00:02:49,082 --> 00:02:52,246 it was only worth, you know, ten dollars or something like that; 58 00:02:52,246 --> 00:02:54,655 because it didn't expand at the right time, 59 00:02:54,655 --> 00:02:57,829 because this one piece of the puzzle didn't work right, 60 00:02:57,829 --> 00:03:02,267 the entire spacecraft along with seven people was lost. 61 00:03:02,433 --> 00:03:04,130 So, that's O-ring production. 62 00:03:05,024 --> 00:03:07,125 Let's formalize this model a little bit. 63 00:03:07,137 --> 00:03:10,660 Assume that there are N tasks, one worker per task. 64 00:03:10,714 --> 00:03:13,253 Doesn't have to be that way, but simplifies things. 65 00:03:13,396 --> 00:03:18,183 We're going to let qi be the quality level of worker i or task i. 66 00:03:18,644 --> 00:03:20,563 So, qi equal to .9; 67 00:03:20,563 --> 00:03:22,544 this can be interpreted in different ways 68 00:03:22,544 --> 00:03:24,966 but an easy interpretation is to think that -- 69 00:03:25,496 --> 00:03:30,026 .9 means there's 90% chance of completing the task perfectly 70 00:03:30,228 --> 00:03:32,945 and a 10% chance of complete failure. 71 00:03:33,580 --> 00:03:34,769 It could also mean, -- 72 00:03:34,879 --> 00:03:37,995 there's a 50% chance of completing the task perfectly -- 73 00:03:38,383 --> 00:03:42,352 and a 50% chance of reducing the value by 20%. 74 00:03:42,671 --> 00:03:47,552 That is, 1/2 times 1 plus 1/2 times .8, 75 00:03:47,552 --> 00:03:52,358 reducing the value of the entire product by 20% equals .9. 76 00:03:52,548 --> 00:03:55,703 So, there are different ways of interpreting these quality levels. 77 00:03:56,693 --> 00:04:00,014 Output is going to equal the number of tasks -- 78 00:04:00,392 --> 00:04:04,337 times the quality level in each task, -- 79 00:04:04,337 --> 00:04:06,049 all multiplied together. 80 00:04:06,252 --> 00:04:09,977 So, this is really the key to the model, to multiply these quality levels 81 00:04:10,179 --> 00:04:11,815 in each task altogether. 82 00:04:12,291 --> 00:04:13,647 So, for example, -- 83 00:04:14,882 --> 00:04:16,795 if there are ten tasks -- 84 00:04:17,258 --> 00:04:20,481 and the quality level of every worker is point .99, -- 85 00:04:20,565 --> 00:04:24,974 then output will be equal to 10 times .99 to the power of 10 -- 86 00:04:25,102 --> 00:04:27,026 or 9.04. 87 00:04:27,199 --> 00:04:31,636 So, notice that if each worker were perfect, at a equality level of 1, 88 00:04:31,738 --> 00:04:33,525 then the output would be 10, 89 00:04:33,525 --> 00:04:36,927 because each worker has a 1% chance of messing up, -- 90 00:04:37,209 --> 00:04:40,702 the output of the expected output is 9. 91 00:04:41,945 --> 00:04:46,077 If the quality level of the workers went down to .95, 92 00:04:46,077 --> 00:04:48,570 notice that the output would fall to 6. 93 00:04:48,778 --> 00:04:51,958 So, just a small decrease in the quality level -- 94 00:04:52,220 --> 00:04:54,433 decreases the output by a lot. 95 00:04:54,999 --> 00:04:57,424 If the quality level went down to .9; -- 96 00:04:57,705 --> 00:04:59,453 again, not that big a drop; 97 00:04:59,453 --> 00:05:02,349 the output level goes down to 3.5, -- 98 00:05:02,541 --> 00:05:05,413 an awfully big drop in output 99 00:05:05,413 --> 00:05:08,582 for a relatively small drop in the quality level. 100 00:05:10,633 --> 00:05:13,497 One of the most important implications of the O-ring model 101 00:05:13,497 --> 00:05:15,358 is that we'll have quality matching. 102 00:05:15,667 --> 00:05:19,900 That is, output will be higher if we put all the high-quality workers together 103 00:05:19,900 --> 00:05:22,052 and all the low quality workers together 104 00:05:22,052 --> 00:05:24,640 compared with if we mixed the workers up. 105 00:05:24,730 --> 00:05:26,313 Let's do a simple example. 106 00:05:26,609 --> 00:05:30,565 Suppose we have two high-quality workers and two low quality workers. 107 00:05:30,991 --> 00:05:34,147 If we put the high-quality workers together in one firm, 108 00:05:34,147 --> 00:05:35,450 then the output we get is 109 00:05:35,450 --> 00:05:40,423 2 the number of tasks times qh times qh or 2qh quared. 110 00:05:40,540 --> 00:05:44,751 The low quality workers then are 2ql squared for the same reasons. 111 00:05:45,200 --> 00:05:47,588 If we mix, we then have again two firms. 112 00:05:47,588 --> 00:05:52,276 In one firm we get 2 the number of tasks times qh times ql 113 00:05:52,276 --> 00:05:53,999 and same thing for the second firm. 114 00:05:54,555 --> 00:05:56,294 Okay. Which one of these is bigger? 115 00:05:56,676 --> 00:05:58,472 Well, we can get rid of the 2s. 116 00:05:58,667 --> 00:06:01,116 That gives us qh squared plus ql squared 117 00:06:01,116 --> 00:06:03,673 compared with getting rid of the 2s here; 118 00:06:04,085 --> 00:06:06,403 we still have 2qhql. 119 00:06:07,046 --> 00:06:08,567 Which one of those is bigger? 120 00:06:08,735 --> 00:06:10,932 Let's just put into numbers. 121 00:06:10,932 --> 00:06:14,341 So, suppose that qh is 1 and ql is 1/2. 122 00:06:15,115 --> 00:06:17,600 Therefore, for the match group, 123 00:06:17,887 --> 00:06:20,903 we get 1 squared plus 1/2 squared, 124 00:06:20,903 --> 00:06:25,095 and for the mix group we get 2 times 1 times 1/2. 125 00:06:25,868 --> 00:06:26,659 Let's see. 126 00:06:26,820 --> 00:06:30,219 Well, that's a quarter, one and a quarter for the match group 127 00:06:30,219 --> 00:06:32,391 and just one the mix group. 128 00:06:32,531 --> 00:06:34,261 Therefore, you won a match. 129 00:06:34,567 --> 00:06:37,212 What about other examples? Okay, let's do a general proof. 130 00:06:37,981 --> 00:06:39,948 If qh is bigger than ql, 131 00:06:39,995 --> 00:06:45,111 then notice that qh minus ql squared is certainly bigger than 0. 132 00:06:45,477 --> 00:06:48,725 Well, just using FOIL, multiplying these out, 133 00:06:48,725 --> 00:06:53,492 we get qh squared plus ql squared minus 2qhql. 134 00:06:53,640 --> 00:06:56,455 Let's put the 2qhql on the other side, 135 00:06:56,455 --> 00:07:01,893 we get qh squared plus ql squared is bigger than 2qhql. 136 00:07:02,035 --> 00:07:06,249 But notice that this is just exactly our statement here 137 00:07:06,298 --> 00:07:11,310 for comparing the match output with the mix output. 138 00:07:11,310 --> 00:07:14,831 So, this tells us that, for any qh and any ql, 139 00:07:14,831 --> 00:07:18,429 the match output is bigger than the mix output. 140 00:07:20,918 --> 00:07:24,218 Now, it's not too hard to show that in a competitive economy, 141 00:07:24,218 --> 00:07:28,788 with quality matching, higher output is going to mean higher wages. 142 00:07:29,235 --> 00:07:33,325 So, remember now that the effect of quality on output, which we now know, 143 00:07:33,325 --> 00:07:36,160 is going to be the same as the effect of quality on wages 144 00:07:36,160 --> 00:07:37,755 this is highly non-linear. 145 00:07:38,037 --> 00:07:42,483 So, if all the out-workers have a quality level of 1, 146 00:07:42,662 --> 00:07:45,017 then output is going to be 10, up here. 147 00:07:45,165 --> 00:07:49,264 Notice that if quality falls just a little bit to .9, -- 148 00:07:49,616 --> 00:07:52,415 output falls a huge amount to less than 4. 149 00:07:52,846 --> 00:07:56,086 So, you get a big drop in quality, 150 00:07:56,086 --> 00:07:59,616 big drop in output with a fairly small drop in quality. 151 00:07:59,909 --> 00:08:02,847 Indeed, notice that if quality falls in half, -- 152 00:08:03,206 --> 00:08:05,384 output is basically going to zero. 153 00:08:05,580 --> 00:08:08,994 You can't even see in this graph how small output is 154 00:08:09,198 --> 00:08:11,249 with a drop in quality of a half. 155 00:08:11,517 --> 00:08:13,542 So, what this says is that, -- 156 00:08:14,403 --> 00:08:19,272 if there are differences in quality levels across countries, 157 00:08:19,566 --> 00:08:25,536 then one country may have much, much smaller GDP per capita 158 00:08:25,536 --> 00:08:26,894 than the other country 159 00:08:26,894 --> 00:08:29,445 even though the quality levels are not that different. 160 00:08:29,445 --> 00:08:32,929 A fairly small decrease in the quality levels of the workers 161 00:08:32,977 --> 00:08:35,627 creates a big decrease in wages. 162 00:08:37,675 --> 00:08:41,258 We can also this at a national level. There's slightly different way. 163 00:08:41,842 --> 00:08:45,509 Suppose the talent distribution is something like this -- 164 00:08:45,792 --> 00:08:47,200 on the left hand side, 165 00:08:47,292 --> 00:08:51,476 that is, most of the workers have a talent level of 1, 166 00:08:52,856 --> 00:08:54,928 this is an arbitrary number here. 167 00:08:55,233 --> 00:08:59,336 I don't think of it as all being perfect, just think of it as an arbitrary scale. 168 00:08:59,432 --> 00:09:02,355 But suppose most of the workers have this talent distribution 169 00:09:02,355 --> 00:09:04,545 somewhere around here. Okay? 170 00:09:05,406 --> 00:09:09,257 When you map that into the wage distribution, 171 00:09:09,625 --> 00:09:13,059 taking into account the fact we have O-ring production, -- 172 00:09:13,368 --> 00:09:17,505 what you get is wages, you get a big right hand tail. 173 00:09:17,637 --> 00:09:21,192 You get wages are much more unequal than talent. 174 00:09:21,192 --> 00:09:26,759 So, a fairly equal distribution of talent when you map that into the O-ring model, 175 00:09:26,871 --> 00:09:31,657 turns into an unequal distribution of wages. 176 00:09:32,236 --> 00:09:35,163 So, in particular, notice that over here, -- 177 00:09:36,054 --> 00:09:41,100 there's hardly anybody who has a talent level of 2 or greater. 178 00:09:41,277 --> 00:09:45,092 Very very few people in this economy have a talent level of 2 or greater 179 00:09:45,305 --> 00:09:47,077 and yet, wages, -- 180 00:09:47,194 --> 00:09:51,535 a large fraction of the wages are going to go to people 181 00:09:51,535 --> 00:09:54,644 who have a talent level of 2 or more. 182 00:09:54,762 --> 00:10:01,125 So, this shows you how an O-ring model magnifies the distribution of talent -- 183 00:10:01,274 --> 00:10:06,044 turning it into a much more unequal distribution of wages. 184 00:10:07,512 --> 00:10:09,471 Here's another implication of the model. 185 00:10:09,736 --> 00:10:13,358 In an O-ring model, workers performing the same tasks -- 186 00:10:13,558 --> 00:10:16,461 will earn higher wages in a high-skill firm 187 00:10:16,752 --> 00:10:18,341 than in a low-skill firm. 188 00:10:18,541 --> 00:10:19,669 So, for example, 189 00:10:19,669 --> 00:10:24,034 highest quality secretaries will work with the highest quality CEO's 190 00:10:24,049 --> 00:10:27,272 simply because a mistake by one of those secretaries 191 00:10:27,272 --> 00:10:32,035 is going to be so much more damaging when she works for a high-quality CEO 192 00:10:32,035 --> 00:10:33,796 than for a low quality CEO. 193 00:10:34,615 --> 00:10:35,320 Apple. 194 00:10:35,590 --> 00:10:38,278 They'll hire the best programmers and the best designers. 195 00:10:38,278 --> 00:10:42,258 They'll also want to hire the best janitors and they'll pay them the most, 196 00:10:42,258 --> 00:10:45,861 at least to the extent that the output of those janitors contributes 197 00:10:45,861 --> 00:10:47,840 to the output of the entire product. 198 00:10:48,490 --> 00:10:51,111 The same idea applies with the economy as a whole. 199 00:10:51,392 --> 00:10:53,861 High-quality workers will be paid more 200 00:10:53,861 --> 00:10:56,514 when there are more high quality workers to work with. 201 00:10:56,974 --> 00:10:59,256 Talent likes to work with other talent. 202 00:10:59,256 --> 00:11:01,090 There's a multiplier effect here. 203 00:11:01,263 --> 00:11:04,791 The more high-talented, high-quality workers you're surrounded with, -- 204 00:11:04,925 --> 00:11:06,814 the more your earnings are going to be. 205 00:11:07,123 --> 00:11:09,201 That's one reason I like to work with Tyler. 206 00:11:10,585 --> 00:11:11,845 It's not just; -- 207 00:11:12,148 --> 00:11:14,797 we have to take a longer picture view of this as well; -- 208 00:11:15,092 --> 00:11:17,891 is when there's a lot of high-quality workers around, 209 00:11:17,891 --> 00:11:22,426 it pays you to invest in being a high-quality worker. 210 00:11:22,589 --> 00:11:25,399 Similarly, if there's just a lot of low-skill workers around, 211 00:11:25,399 --> 00:11:28,437 it doesn't pay you to be a high-quality worker. 212 00:11:28,720 --> 00:11:32,022 So, if we think about a very smart person in a poor country 213 00:11:32,022 --> 00:11:34,059 surrounded by low quality workers, 214 00:11:34,059 --> 00:11:35,888 that person isn't going to earn a lot. 215 00:11:36,284 --> 00:11:39,688 They, in fact, may not want to invest in an education, 216 00:11:39,688 --> 00:11:41,796 or in building up their skills 217 00:11:41,796 --> 00:11:44,129 because the skills won't pay very much 218 00:11:44,129 --> 00:11:47,358 when they don't have those people to work on their team, -- 219 00:11:47,648 --> 00:11:50,459 when they can't combine with other high-quality people, 220 00:11:50,589 --> 00:11:52,650 when they can't get that high pay-off 221 00:11:52,650 --> 00:11:56,228 which comes with multiplying high quality by high quality. 222 00:11:56,798 --> 00:12:00,655 This indicates that, in this model, there's a potential for multiple equilibrium. 223 00:12:00,923 --> 00:12:03,114 You can have a high-quality -- 224 00:12:04,279 --> 00:12:06,883 equilibrium where everyone wants to be high skilled, 225 00:12:06,883 --> 00:12:09,745 but you could also have for the exact the same people, 226 00:12:09,961 --> 00:12:12,848 for exactly the same people, you might also end up 227 00:12:12,848 --> 00:12:14,911 in a low quality equilibria 228 00:12:15,086 --> 00:12:16,993 where no one is getting high skilled 229 00:12:16,993 --> 00:12:20,812 and no one thinks it's worthwhile to become a high-skilled worker. 230 00:12:22,850 --> 00:12:24,255 In an O-ring model, -- 231 00:12:24,322 --> 00:12:27,075 capital wants to work with high-quality workers 232 00:12:27,211 --> 00:12:28,789 for exactly the same reasons that 233 00:12:28,789 --> 00:12:30,996 high-quality workers want to work with each other. 234 00:12:31,428 --> 00:12:33,184 In particular, note that -- 235 00:12:33,226 --> 00:12:36,912 more capital in these models doesn't substitute -- 236 00:12:37,154 --> 00:12:38,907 for lower-skilled workers. 237 00:12:39,246 --> 00:12:40,925 So, if you give me a Stradivarius, 238 00:12:40,925 --> 00:12:43,117 I'm not going to be a better violin player. 239 00:12:44,833 --> 00:12:46,658 You don't want to do something like this, therefore, 240 00:12:46,658 --> 00:12:49,689 you don't want to have Homer Simpson running the nuclear power plant. 241 00:12:49,689 --> 00:12:52,456 This a failure of quality matching. 242 00:12:52,566 --> 00:12:53,733 Don't do that. 243 00:12:53,858 --> 00:12:56,092 Instead, what you want to do is you want to match 244 00:12:56,092 --> 00:12:59,203 the best workers with the most expensive machines. 245 00:12:59,301 --> 00:13:02,220 So, what you want Itzhak Perlman with a Stradivarius. 246 00:13:03,378 --> 00:13:04,758 Implication of this -- 247 00:13:04,964 --> 00:13:07,612 is that poor countries will have more workers 248 00:13:07,823 --> 00:13:10,759 in less capital intensive primary productions, 249 00:13:10,759 --> 00:13:12,429 so, for example, agriculture. 250 00:13:12,678 --> 00:13:16,338 This force magnifies all the other forces we're talking about earlier. 251 00:13:16,630 --> 00:13:17,844 So, in particular, -- 252 00:13:17,911 --> 00:13:20,866 capital is going to flow away from countries 253 00:13:20,866 --> 00:13:22,864 which have low-skilled workers 254 00:13:22,864 --> 00:13:25,834 could want to go to countries which have high-skilled workers. 255 00:13:25,976 --> 00:13:27,107 This means that -- 256 00:13:27,624 --> 00:13:29,943 you're going to have lower income and wages 257 00:13:29,943 --> 00:13:31,542 in countries with low-skilled workers. 258 00:13:31,542 --> 00:13:34,487 This magnifies all the effects we're talking about earlier. 259 00:13:35,371 --> 00:13:39,605 Similarly, poor countries will have more workers in production tasks 260 00:13:39,605 --> 00:13:41,802 or production jobs that are simpler, -- 261 00:13:42,118 --> 00:13:44,063 that require fewer tasks. 262 00:13:44,414 --> 00:13:45,955 Let's take a closer look at this. 263 00:13:47,445 --> 00:13:50,155 So, what we're showing here is three jobs scaled 264 00:13:50,155 --> 00:13:54,783 so that you get the 100% of output when the workers are perfectly high quality. 265 00:13:55,186 --> 00:13:58,616 Job 1 requires the workers to get 5 tasks right. 266 00:13:58,785 --> 00:14:01,002 Job 2, that they get 10 tasks 267 00:14:01,002 --> 00:14:03,896 and Job 3 that they get 40 tasks right. 268 00:14:03,958 --> 00:14:05,033 Now, here's point. 269 00:14:05,271 --> 00:14:08,777 If you take workers of reasonably high skilled level, .9, 270 00:14:08,938 --> 00:14:13,518 but you assign them to a job which requires that they get 40 things right, 271 00:14:13,715 --> 00:14:16,963 where getting 1 thing wrong, one of those tasks wrong 272 00:14:16,963 --> 00:14:19,208 can reduce the value of the entire product, -- 273 00:14:19,367 --> 00:14:23,566 then your chances of getting full output are virtually nil. 274 00:14:24,515 --> 00:14:25,579 On the other hand, -- 275 00:14:25,670 --> 00:14:29,161 if you take those same workers and you assign them to a job 276 00:14:29,161 --> 00:14:31,887 which requires that they just get 5 things right, 277 00:14:31,887 --> 00:14:35,375 then your chances of getting full output are much higher. 278 00:14:35,706 --> 00:14:39,521 So, think about this as being bicycle production, car production 279 00:14:39,618 --> 00:14:41,672 and space shuttle production. 280 00:14:41,941 --> 00:14:45,036 What it says is that countries with lower-skilled work forces; 281 00:14:45,214 --> 00:14:48,534 they're going to specialize in things like bicycle production -- 282 00:14:48,777 --> 00:14:50,752 which tend to be lower paid. 283 00:14:51,277 --> 00:14:53,555 The value of the entire product tends to be lower 284 00:14:53,555 --> 00:14:56,473 when the number of tasks required is lower. 285 00:14:57,250 --> 00:14:59,132 What this also says is that, -- 286 00:14:59,337 --> 00:15:02,275 if you have a job requiring a lot of tasks, -- 287 00:15:02,413 --> 00:15:03,885 like a space shuttle, 288 00:15:03,885 --> 00:15:06,823 where you've got to get even hundreds of tasks right 289 00:15:06,823 --> 00:15:09,028 in order to get the full value of the product. 290 00:15:09,105 --> 00:15:13,437 Then you really want to be working with workers of the very highest skill level 291 00:15:13,658 --> 00:15:16,745 to have any chance of producing that high-quality product. 292 00:15:19,988 --> 00:15:24,252 We can apply many of these ideas at the level of the economy as whole, 293 00:15:24,321 --> 00:15:27,624 in which case it becomes a theory of bottlenecks, linkages 294 00:15:27,624 --> 00:15:29,237 and complementarities. 295 00:15:29,511 --> 00:15:33,937 So, let's think about N industries each performing a single task. 296 00:15:34,025 --> 00:15:35,417 In following Kremer, -- 297 00:15:35,594 --> 00:15:40,682 let's suppose the quality falls by half in just of these tasks or industries. 298 00:15:40,854 --> 00:15:41,990 So, for example, 299 00:15:41,990 --> 00:15:46,231 the electricity production becomes more spotty and subject to blackouts. 300 00:15:46,326 --> 00:15:48,554 Corruption increases at the licence bureau 301 00:15:48,554 --> 00:15:51,831 requiring us to spend a lot more to get a licence. 302 00:15:52,425 --> 00:15:57,043 Well, even though, quality has fallen in just 2 of the many tasks 303 00:15:57,043 --> 00:15:58,748 which we need to complete, 304 00:15:58,748 --> 00:16:02,276 output is going to fall immediately by 75%. 305 00:16:03,527 --> 00:16:06,165 Moreover, there are going to be knock-on effects. 306 00:16:06,348 --> 00:16:10,308 Wages in every other sector of the economy are also going to fall 307 00:16:10,308 --> 00:16:13,199 because the total value of the product has fallen. 308 00:16:13,457 --> 00:16:15,972 And this fall in wages is going to greatly reduce 309 00:16:15,972 --> 00:16:20,066 the incentive to invest in quality in all those other sectors 310 00:16:20,066 --> 00:16:23,849 and that reduces output further in the long run. 311 00:16:24,207 --> 00:16:26,073 So, you could have a bottleneck, 312 00:16:26,141 --> 00:16:30,427 and a bottleneck affects not just that sector of the economy, 313 00:16:30,427 --> 00:16:33,235 but every other sector of the economy. 314 00:16:33,698 --> 00:16:35,702 Because you have complementarities, 315 00:16:35,771 --> 00:16:40,294 when one sector goes down, the other sectors also go down as well. 316 00:16:40,701 --> 00:16:43,023 This shows, by the way, the importance of trade 317 00:16:43,023 --> 00:16:47,328 as a method of avoiding those bottlenecks, of rooting around bottlenecks. 318 00:16:47,620 --> 00:16:50,464 If you can import a fairly high-quality good, 319 00:16:50,556 --> 00:16:53,236 perform just a few tasks in country, 320 00:16:53,236 --> 00:16:55,059 and then export that good, 321 00:16:55,088 --> 00:16:58,570 then you can root around the bottleneck and get some development 322 00:16:58,570 --> 00:17:01,884 even when not every sector in your economy is working well. 323 00:17:02,299 --> 00:17:05,016 If, instead, you've got to produce everything in house, 324 00:17:05,016 --> 00:17:06,447 everything in the country, 325 00:17:06,447 --> 00:17:09,601 then you need every sector in that economy to be working well 326 00:17:09,709 --> 00:17:11,706 when you're working with O-ring production, 327 00:17:11,706 --> 00:17:15,506 because you have all these complementarities across industries. 328 00:17:18,196 --> 00:17:21,403 Okay, let's give some final thoughts about O-ring production. 329 00:17:21,496 --> 00:17:22,233 First, -- 330 00:17:22,360 --> 00:17:24,611 not every industry is an O-ring industry, 331 00:17:24,700 --> 00:17:29,034 but, perhaps in our modern world, more industries are moving in this direction. 332 00:17:29,419 --> 00:17:32,728 That tells us something about the sources of growing inequality. 333 00:17:33,535 --> 00:17:37,316 O-ring production also reminds us that production is complex, 334 00:17:37,518 --> 00:17:41,532 that production often requires that every member of a team be working 335 00:17:41,532 --> 00:17:43,650 in the same direction like a sports team. 336 00:17:43,721 --> 00:17:47,081 Every member has got to be performing at their highest level of ability 337 00:17:47,081 --> 00:17:48,732 in order to get those wins. 338 00:17:49,097 --> 00:17:52,032 This tells us that organizational capital, -- 339 00:17:52,330 --> 00:17:55,597 the ability to bring together high-skilled workers -- 340 00:17:55,757 --> 00:17:57,896 with expensive capital, -- 341 00:17:57,990 --> 00:18:01,818 and to get all those workers into that capital working together 342 00:18:01,818 --> 00:18:04,940 in a team at the highest level of skill. 343 00:18:05,108 --> 00:18:08,374 The ability to do that is incredibly important, -- 344 00:18:08,622 --> 00:18:11,827 and the more complex production grows, 345 00:18:11,827 --> 00:18:16,546 the more tasks that are required to achieve maximum production, 346 00:18:16,667 --> 00:18:19,517 the more valuable organizational capital, 347 00:18:19,517 --> 00:18:22,080 the ability to bring these workers and capital together, 348 00:18:22,080 --> 00:18:24,543 the more valuable those skills are going to become. 349 00:18:25,644 --> 00:18:27,018 In an O-ring model, 350 00:18:27,018 --> 00:18:29,576 you can get virtuous and vicious cycles. 351 00:18:29,897 --> 00:18:33,120 When one industry in a O-ring model increases its ability, 352 00:18:33,120 --> 00:18:36,699 that increases the incentive of every other industry 353 00:18:36,699 --> 00:18:39,344 to perform at its highest level of ability. 354 00:18:39,637 --> 00:18:41,905 But the same thing is also true in reverse. 355 00:18:41,999 --> 00:18:43,469 When one team member 356 00:18:43,469 --> 00:18:46,521 or when one industry is performing at a low level, 357 00:18:46,616 --> 00:18:49,480 that reduces the incentive of all the other industries 358 00:18:49,480 --> 00:18:51,234 to perform at a high level. 359 00:18:51,448 --> 00:18:55,140 Hence you can get growth miracles and growth disasters. 360 00:18:55,695 --> 00:18:58,901 Now, what does it take to coordinate a team? 361 00:18:59,051 --> 00:19:02,762 To coordinate an economy on a high-skilled equilibrium 362 00:19:02,762 --> 00:19:05,157 where everyone is working at their maximum level? 363 00:19:05,295 --> 00:19:07,540 Well this is an incredibly hard problem. 364 00:19:07,707 --> 00:19:10,203 This is all about what culture is about. 365 00:19:10,267 --> 00:19:12,132 It's an incredibly hard problem, 366 00:19:12,132 --> 00:19:16,833 but also an incredibly important problem, something important to think about. 367 00:19:17,244 --> 00:19:18,431 Thanks very much.