1 00:00:00,000 --> 00:00:11,040 Welcome to the Cannabis Data Science Meetup Group. 2 00:00:11,040 --> 00:00:13,040 You're in for a special treat today. 3 00:00:13,040 --> 00:00:15,880 Got incredibly good material. 4 00:00:15,880 --> 00:00:21,680 Some of my thoughts have really come together about the industry in general and the economic 5 00:00:21,680 --> 00:00:25,720 effects of the permitted cannabis industry. 6 00:00:25,720 --> 00:00:31,840 So without further ado, we'll go ahead and dive right into it and I'll start sharing 7 00:00:31,840 --> 00:00:36,960 my screen. 8 00:00:36,960 --> 00:00:43,200 Thanks. 9 00:00:43,200 --> 00:00:55,320 So last week we began to just see how Massachusetts is publishing a good amount of public data. 10 00:00:55,320 --> 00:01:08,880 So you have everything from just the number of licensees, you have sales by product type, 11 00:01:08,880 --> 00:01:21,960 and at the end of last week we were beginning to look at the production side. 12 00:01:21,960 --> 00:01:28,000 So we saw, wow, they're doing a good job at publishing data. 13 00:01:28,000 --> 00:01:34,480 They're publishing daily totals on an almost real time basis. 14 00:01:34,480 --> 00:01:35,480 Look at this. 15 00:01:35,480 --> 00:01:45,400 This was just updated on September 27th, so just on Monday. 16 00:01:45,400 --> 00:01:53,480 So we have incredibly recent data, which is hard to come by data that's so fresh. 17 00:01:53,480 --> 00:02:01,480 So we've got fresh data and on top of that we can access it through an API just to make 18 00:02:01,480 --> 00:02:04,640 our lives simple and easy. 19 00:02:04,640 --> 00:02:12,560 And we have a rich data set here with a number of interesting data points. 20 00:02:12,560 --> 00:02:19,520 We've got the number of plants in their various stages, so we can see what plants are mature, 21 00:02:19,520 --> 00:02:22,120 which are flowering. 22 00:02:22,120 --> 00:02:31,640 We can see what packages are on hand, so we can get a gauge of inventory at any given 23 00:02:31,640 --> 00:02:34,280 time. 24 00:02:34,280 --> 00:02:42,480 Furthermore, we can get a couple more interesting data points, such as the number of strains, 25 00:02:42,480 --> 00:02:50,760 products, and last but not least, we saw, wow, they're tracking the number of employees 26 00:02:50,760 --> 00:02:53,760 in the cannabis industry. 27 00:02:53,760 --> 00:03:02,360 And earlier in this series, in particular, we were looking at Colorado and we were trying 28 00:03:02,360 --> 00:03:09,120 to look at production functions and we were talking about capital inputs, labor inputs, 29 00:03:09,120 --> 00:03:16,680 and that was where we estimated the competitive rate of return on capital, as well as the 30 00:03:16,680 --> 00:03:19,600 competitive wage rate in Colorado. 31 00:03:19,600 --> 00:03:24,520 And so what's real interesting with this data is we could repeat this exact same analysis 32 00:03:24,520 --> 00:03:27,600 for Massachusetts. 33 00:03:27,600 --> 00:03:33,800 And we'll probably do that, but we can do something new and fresh today. 34 00:03:33,800 --> 00:03:39,840 So first things first, let's get the data. 35 00:03:39,840 --> 00:03:48,280 So just using a handful of standard Python packages, we'll be using Python, but the data 36 00:03:48,280 --> 00:03:57,320 is accessible through an API, so you can use your favorite programming languages. 37 00:03:57,320 --> 00:04:03,480 Just writing a handful of helper functions here, I can talk a bit more about these when 38 00:04:03,480 --> 00:04:06,640 we use them here. 39 00:04:06,640 --> 00:04:16,040 So first things first, let's get this data and look at it ourselves. 40 00:04:16,040 --> 00:04:23,600 So you can register for Socratic to get an app token. 41 00:04:23,600 --> 00:04:26,120 You're not necessarily required. 42 00:04:26,120 --> 00:04:34,400 You just get throttled if you make too many requests without an app token. 43 00:04:34,400 --> 00:04:39,040 So we can use one at the end of the row. 44 00:04:39,040 --> 00:04:41,800 So we'll just initialize the API. 45 00:04:41,800 --> 00:04:45,120 So we're just saying, okay, this is the URL. 46 00:04:45,120 --> 00:04:47,320 So what is an API? 47 00:04:47,320 --> 00:04:53,960 It's really just a simple, simple website that you can interact with programmatically 48 00:04:53,960 --> 00:04:56,000 to get data. 49 00:04:56,000 --> 00:05:01,480 So it's not going to be marked up in a nice, beautiful website format. 50 00:05:01,480 --> 00:05:05,440 It's just going to be raw data. 51 00:05:05,440 --> 00:05:09,600 That's what we're after, the bare bones. 52 00:05:09,600 --> 00:05:24,120 So we'll just ping this website, backslash our data set ID, which you can get right here, 53 00:05:24,120 --> 00:05:28,480 the data set identifier. 54 00:05:28,480 --> 00:05:30,720 We'll pass a handful of parameters. 55 00:05:30,720 --> 00:05:36,720 So we'll basically say, okay, we don't want more than 2000 observations. 56 00:05:36,720 --> 00:05:39,200 And we actually want to order this. 57 00:05:39,200 --> 00:05:43,600 So here I've done a little bit of scouting. 58 00:05:43,600 --> 00:05:50,040 And then I saw, okay, look, they give us a timestamp. 59 00:05:50,040 --> 00:05:54,960 So we can actually order our data by time. 60 00:05:54,960 --> 00:06:01,920 Here they give us an example of how you can do a query for a specific date. 61 00:06:01,920 --> 00:06:19,160 Luckily, we can use a search query language command descending. 62 00:06:19,160 --> 00:06:28,280 To specify in our query that we want everything sorted by the date. 63 00:06:28,280 --> 00:06:30,280 So enough of that. 64 00:06:30,280 --> 00:06:42,440 And so long story short, we're just going to ping the website and get our data. 65 00:06:42,440 --> 00:06:44,160 So we should have our data here. 66 00:06:44,160 --> 00:06:53,320 So for example, in the programming language, 200 is success. 67 00:06:53,320 --> 00:06:56,040 And so what did we just get? 68 00:06:56,040 --> 00:07:02,880 Well, instead of getting a website, we just got a giant data dump. 69 00:07:02,880 --> 00:07:09,920 So this is instead of delivering us a website, they just deliver us a bunch of what's called 70 00:07:09,920 --> 00:07:12,120 JSON objects. 71 00:07:12,120 --> 00:07:18,640 So JavaScript object notation data. 72 00:07:18,640 --> 00:07:27,080 And so this is how you can think about a data point where it's essentially an observation 73 00:07:27,080 --> 00:07:28,080 here. 74 00:07:28,080 --> 00:07:30,480 So this is one observation. 75 00:07:30,480 --> 00:07:37,080 So this is the date, September 27th. 76 00:07:37,080 --> 00:07:47,920 On that date, there were 129,000 immature plants and so on and so forth. 77 00:07:47,920 --> 00:07:53,000 So we can work with this. 78 00:07:53,000 --> 00:07:58,400 What's a glorified simplified Excel spreadsheet in programming language? 79 00:07:58,400 --> 00:08:00,000 So a data frame. 80 00:08:00,000 --> 00:08:02,760 So we'll just put this in the data frame. 81 00:08:02,760 --> 00:08:09,520 And here I use one of these helper functions to basically put everything into chronological 82 00:08:09,520 --> 00:08:13,160 order. 83 00:08:13,160 --> 00:08:18,240 Because we wanted to get the most recent data, but now we actually need to put it in chronological 84 00:08:18,240 --> 00:08:26,880 order just because that's how we do time series analysis from the past to the present or even 85 00:08:26,880 --> 00:08:28,840 the future. 86 00:08:28,840 --> 00:08:37,880 So let's put this into a data frame just so that we can work with it easily. 87 00:08:37,880 --> 00:08:41,680 And what does this look like? 88 00:08:41,680 --> 00:08:47,400 Well this is why I was saying this looks similar to an Excel spreadsheet where you've got a 89 00:08:47,400 --> 00:08:49,120 table here. 90 00:08:49,120 --> 00:08:55,780 We've got a thousand plus rows, 1069 rows. 91 00:08:55,780 --> 00:09:01,400 So that's 1069 days. 92 00:09:01,400 --> 00:09:15,160 So just short of three years of data here. 93 00:09:15,160 --> 00:09:18,920 And like we mentioned earlier, we have all these observations. 94 00:09:18,920 --> 00:09:21,880 Here's the observations on a daily basis. 95 00:09:21,880 --> 00:09:23,120 Awesome. 96 00:09:23,120 --> 00:09:25,880 What variables do we have here? 97 00:09:25,880 --> 00:09:33,120 Well I just did a little pre-scouting and just listed, okay. 98 00:09:33,120 --> 00:09:37,280 Here we are. 99 00:09:37,280 --> 00:09:40,640 So those are the variables we're working with. 100 00:09:40,640 --> 00:09:48,120 So without further ado, let's start looking at a handful of these. 101 00:09:48,120 --> 00:09:52,440 So great because that's the first step is look at the data. 102 00:09:52,440 --> 00:10:00,680 So you can plot these. 103 00:10:00,680 --> 00:10:05,440 Awesome. 104 00:10:05,440 --> 00:10:15,960 So we have our three years of data and we see that wow, at the very beginning, the very 105 00:10:15,960 --> 00:10:19,440 first observation, there were three employees. 106 00:10:19,440 --> 00:10:27,000 So we're basically tracking the Massachusetts industry from the inception, the first company, 107 00:10:27,000 --> 00:10:40,200 the first three employees to present day where we have just shy of 10,000 employees in presumably 108 00:10:40,200 --> 00:10:45,780 plant touching cannabis businesses in Massachusetts. 109 00:10:45,780 --> 00:10:53,000 So it's now time to put this into perspective because that I think is what this is all about. 110 00:10:53,000 --> 00:10:58,440 Because there's been a lot of noise in the cannabis industry. 111 00:10:58,440 --> 00:11:06,160 They'll let you believe that everybody works in the cannabis industry and it's just booming 112 00:11:06,160 --> 00:11:10,640 and well that may be, but it's time to quantify that. 113 00:11:10,640 --> 00:11:16,200 So it's time to measure, okay, what exactly is the size, what exactly is the impact here, 114 00:11:16,200 --> 00:11:18,280 at least the economic impact. 115 00:11:18,280 --> 00:11:21,320 We can measure that. 116 00:11:21,320 --> 00:11:31,360 So without further ado, the reason why is, well, perhaps other states can look at the 117 00:11:31,360 --> 00:11:34,720 Massachusetts and see, okay, what happens? 118 00:11:34,720 --> 00:11:38,580 What is the economic effect of permitting cannabis? 119 00:11:38,580 --> 00:11:45,680 You can look at the social effects as you should, but perhaps in another study right 120 00:11:45,680 --> 00:11:50,600 now they're really focused on the economic side of things. 121 00:11:50,600 --> 00:11:55,280 So time to put this into perspective. 122 00:11:55,280 --> 00:12:01,640 So for starters, this is going to need some sort of time scale, right? 123 00:12:01,640 --> 00:12:04,400 We're looking at a time series here. 124 00:12:04,400 --> 00:12:16,280 So what we can do is actually add a date to our data and set the index on the date. 125 00:12:16,280 --> 00:12:29,680 Nothing fancy here, but what that does is we can now plot our data in the exact same 126 00:12:29,680 --> 00:12:35,720 way, and we actually now can see the time scale. 127 00:12:35,720 --> 00:12:38,680 So now we can see, oh, look at this. 128 00:12:38,680 --> 00:12:49,280 Our data is starting just about this time in 2018, around September or October of 2018, 129 00:12:49,280 --> 00:13:01,840 and we can actually quantify that as October 15th is when our data set begins. 130 00:13:01,840 --> 00:13:09,520 And as we noted, we have quite recent data here up until September 27th, so all the way 131 00:13:09,520 --> 00:13:11,400 until Monday's totals. 132 00:13:11,400 --> 00:13:14,960 So they haven't quite published Tuesday's totals yet. 133 00:13:14,960 --> 00:13:27,920 So let's say, in my opinion, awesome that we have almost real-time data here. 134 00:13:27,920 --> 00:13:30,880 So let's start putting things into context. 135 00:13:30,880 --> 00:13:36,400 So we know how many employees are here in the campus industry. 136 00:13:36,400 --> 00:13:41,320 Well, how many employees are in Massachusetts in general? 137 00:13:41,320 --> 00:13:49,040 So this is a technique that I suggest to anyone doing data sciences. 138 00:13:49,040 --> 00:13:52,760 OK, you get your data set, right? 139 00:13:52,760 --> 00:13:55,000 Nice interesting data. 140 00:13:55,000 --> 00:13:58,840 What do we have here? 141 00:13:58,840 --> 00:14:00,840 What do we have here? 142 00:14:00,840 --> 00:14:07,640 How do you define a Massachusetts employee for the total? 143 00:14:07,640 --> 00:14:08,640 Not trying to be nasty. 144 00:14:08,640 --> 00:14:11,720 I'm just having a little difficulty with that. 145 00:14:11,720 --> 00:14:14,600 Wait, would you comment, Heather? 146 00:14:14,600 --> 00:14:22,720 Oh, meaning to say, it may not be so simple to define and say, OK, total Massachusetts-bound 147 00:14:22,720 --> 00:14:24,760 employees, that's your total. 148 00:14:24,760 --> 00:14:32,880 And then you're going to divide that by the number of cannabis-related or cannabis industry 149 00:14:32,880 --> 00:14:33,880 employees. 150 00:14:33,880 --> 00:14:34,880 Am I right? 151 00:14:34,880 --> 00:14:42,040 Because essentially, the rough idea is to just try to figure out the percentage of employees 152 00:14:42,040 --> 00:14:44,280 that are working in the cannabis industry. 153 00:14:44,280 --> 00:14:45,280 Right. 154 00:14:45,280 --> 00:14:49,400 So for cannabis, there aren't a whole lot of remote jobs. 155 00:14:49,400 --> 00:14:55,680 And I guess for me, just using Maryland as a comparison, I have difficulty even saying 156 00:14:55,680 --> 00:15:00,480 who's a Maryland employee other than those who are physically on site. 157 00:15:00,480 --> 00:15:05,600 Just because the headquarters is there, all of their managers may be on the West Coast 158 00:15:05,600 --> 00:15:07,720 somewhere as I'm finding. 159 00:15:07,720 --> 00:15:10,880 So I'm not saying that the number is going to be wrong. 160 00:15:10,880 --> 00:15:16,120 The definition of a Massachusetts employee may be more complicated than expected. 161 00:15:16,120 --> 00:15:21,680 Critical, critical, critical, critical point, Heather. 162 00:15:21,680 --> 00:15:24,200 And this is what we need more of, right? 163 00:15:24,200 --> 00:15:28,120 Because you can't just take this data at face value. 164 00:15:28,120 --> 00:15:32,800 You need to, for each, every data point, do exactly what you did. 165 00:15:32,800 --> 00:15:36,220 You just say, okay, we've got this data point here. 166 00:15:36,220 --> 00:15:37,640 What could be wrong with it? 167 00:15:37,640 --> 00:15:41,240 Well, for starters, it could easily undercount people. 168 00:15:41,240 --> 00:15:49,000 Like you said, like executives, people who are kind of gray on the books, right? 169 00:15:49,000 --> 00:15:56,320 Like are the janitors, are the janitors, are all the consultants or what have you, or all 170 00:15:56,320 --> 00:15:59,000 those people on the books. 171 00:15:59,000 --> 00:16:05,760 So I would say, like you said, and then you can start to think, okay, how does that bias 172 00:16:05,760 --> 00:16:06,760 the data? 173 00:16:06,760 --> 00:16:11,400 Well, I think it would probably bias the data downwards, right? 174 00:16:11,400 --> 00:16:19,120 Not necessarily, but I think you may have an undercount of employees here. 175 00:16:19,120 --> 00:16:26,240 So I think there could be more people than this that are associated with these businesses. 176 00:16:26,240 --> 00:16:33,600 And another interesting aspect is there's probably many people that are tangentially 177 00:16:33,600 --> 00:16:36,520 related to these businesses, like consultants, right? 178 00:16:36,520 --> 00:16:43,280 Like so you may have law firms or banks that predominantly serve these licensees, but they're 179 00:16:43,280 --> 00:16:48,120 not going to be counted as employees. 180 00:16:48,120 --> 00:16:55,720 So I think I like your note, Heather, and I think we need more of that. 181 00:16:55,720 --> 00:17:05,000 So be incredibly critical of these data points because, you know, we'll do our analysis here, 182 00:17:05,000 --> 00:17:15,880 but if our data going in has measurement errors, then that's going to have consequences on 183 00:17:15,880 --> 00:17:17,960 our analysis. 184 00:17:17,960 --> 00:17:25,480 So critical point. 185 00:17:25,480 --> 00:17:28,440 I'd like to say that's not going to stop us. 186 00:17:28,440 --> 00:17:30,600 We'll do our analysis anyways. 187 00:17:30,600 --> 00:17:34,520 You just have to take that into consideration when you're looking at the analysis. 188 00:17:34,520 --> 00:17:42,440 So when we get to our final result, we'll want to take into consideration, okay, let's 189 00:17:42,440 --> 00:17:49,720 keep in mind that total employees may be biased downwards or some of these other figures, 190 00:17:49,720 --> 00:17:53,740 you know, they may have their own measurements and their measurement errors of their own. 191 00:17:53,740 --> 00:18:01,400 So these are things that factor into our conclusion, conclusions and results. 192 00:18:01,400 --> 00:18:05,160 So every bit is worth pointing out. 193 00:18:05,160 --> 00:18:16,700 So just to keep steamrolling here, we have 15 variables. 194 00:18:16,700 --> 00:18:22,580 So that's 15 sources of potential measurement error. 195 00:18:22,580 --> 00:18:28,280 So every single one of these probably has their own flaws. 196 00:18:28,280 --> 00:18:36,680 So but nonetheless, we've got these data points here and to introduce even more uncertainty 197 00:18:36,680 --> 00:18:44,000 and measurement error, but also to add some more insight as well, we can supplement this 198 00:18:44,000 --> 00:18:45,320 data. 199 00:18:45,320 --> 00:18:55,040 And so we'll be supplementing it with data from the the Petal Roll Reserve here. 200 00:18:55,040 --> 00:19:00,720 So yes, that's right, we were looking at the number of employees. 201 00:19:00,720 --> 00:19:07,920 So once again, these numbers admittedly, and you know, people have done studies, there 202 00:19:07,920 --> 00:19:11,840 are undoubtedly measurement errors here. 203 00:19:11,840 --> 00:19:21,600 So take them into consideration, but we can still just start getting a rough measure on 204 00:19:21,600 --> 00:19:26,520 on the market, the size of the market and its impact. 205 00:19:26,520 --> 00:19:31,620 Because a rough measure is better than no measure. 206 00:19:31,620 --> 00:19:38,960 So we'll grab the number of employees here in Massachusetts. 207 00:19:38,960 --> 00:19:46,560 So note that this is going to be in thousands of people. 208 00:19:46,560 --> 00:20:00,600 So without further ado, we can get the number of Massachusetts employees here with the thread. 209 00:20:00,600 --> 00:20:05,320 Note I'm just doing your I'll show you specifically what I'm doing here. 210 00:20:05,320 --> 00:20:13,080 So we'll grab the first thing first, we need thread. 211 00:20:13,080 --> 00:20:19,080 So we've imported the packages. 212 00:20:19,080 --> 00:20:23,600 And then we need to initialize the API. 213 00:20:23,600 --> 00:20:29,040 So essentially just create a little client here that remembers our API key and makes 214 00:20:29,040 --> 00:20:36,520 requests on our behalf to the website, because once again, just like before, right, we're 215 00:20:36,520 --> 00:20:45,920 using one API, the Socratic API, we'll also use the dead Fred API, which is once again, 216 00:20:45,920 --> 00:20:52,000 just a URL that we ping and get some data back. 217 00:20:52,000 --> 00:21:01,400 Specifically, we're going to ping Federal Reserve. 218 00:21:01,400 --> 00:21:09,480 We want to real quick, I'm defining the start of our data here, right, because here's the 219 00:21:09,480 --> 00:21:16,800 Federal Reserve data set, they're going all the way back to prior 1992. 220 00:21:16,800 --> 00:21:17,800 That's awesome. 221 00:21:17,800 --> 00:21:24,160 We don't necessarily need that data at this moment. 222 00:21:24,160 --> 00:21:28,160 So we'll define a start date here. 223 00:21:28,160 --> 00:21:36,600 And like we noted earlier, our start date is October 15, 2018. 224 00:21:36,600 --> 00:21:40,720 Next, we need a data set ID. 225 00:21:40,720 --> 00:21:44,000 Notice here, they're all over the place, but there's one right there. 226 00:21:44,000 --> 00:21:46,440 You can also get it from the URL. 227 00:21:46,440 --> 00:21:51,040 So we'll get this data set ID. 228 00:21:51,040 --> 00:22:00,600 Then we'll read the employees here. 229 00:22:00,600 --> 00:22:03,880 And let's see, but yes, that's right. 230 00:22:03,880 --> 00:22:10,000 And so just, and this is what you'll notice with data sciences. 231 00:22:10,000 --> 00:22:17,160 There's these small little things that are important and you can get hung up on them. 232 00:22:17,160 --> 00:22:19,720 But here I've just done a little scouting. 233 00:22:19,720 --> 00:22:24,680 And so I'll show you a couple things that may seem simple, but they'll make the analysis 234 00:22:24,680 --> 00:22:26,440 go a lot better. 235 00:22:26,440 --> 00:22:29,960 So here we have the total Massachusetts employees. 236 00:22:29,960 --> 00:22:33,040 We have it by the month. 237 00:22:33,040 --> 00:22:35,920 So that is awesome. 238 00:22:35,920 --> 00:22:38,500 Right. 239 00:22:38,500 --> 00:22:40,800 And wow, look at that. 240 00:22:40,800 --> 00:22:49,280 So keep in mind, this is thousands of employees. 241 00:22:49,280 --> 00:22:55,080 So here we're going to multiply that by a thousand in just one second. 242 00:22:55,080 --> 00:23:02,120 But you'll notice, wow, like we noticed, we noted this last meetup. 243 00:23:02,120 --> 00:23:07,680 Yeah, something happened in April of 2020. 244 00:23:07,680 --> 00:23:13,760 So that's going to be showing in a lot of data sets. 245 00:23:13,760 --> 00:23:26,760 So there's a regime or they go by so many names. 246 00:23:26,760 --> 00:23:36,320 It depends on what field you're really studying, maybe macroeconomics, like a structural break, 247 00:23:36,320 --> 00:23:38,440 a shock. 248 00:23:38,440 --> 00:23:45,440 So anyways, anyways, we're focusing on the data here. 249 00:23:45,440 --> 00:24:05,920 So this little trick is basically if you look at our production data here, right, we've 250 00:24:05,920 --> 00:24:08,760 got daily data. 251 00:24:08,760 --> 00:24:16,800 And so basically we want to compare daily to monthly. 252 00:24:16,800 --> 00:24:30,480 So we're going to need to aggregate our production data into monthly data, which is not the end 253 00:24:30,480 --> 00:24:31,700 of the world here. 254 00:24:31,700 --> 00:24:35,440 So here we've already set our index. 255 00:24:35,440 --> 00:24:40,560 So we can just create a monthly average, right? 256 00:24:40,560 --> 00:24:44,160 Because how do you want to aggregate? 257 00:24:44,160 --> 00:24:51,640 So there's two ways predominantly that we'll be using in this demonstration, we'll use 258 00:24:51,640 --> 00:24:53,500 two ways to aggregate. 259 00:24:53,500 --> 00:24:55,760 You can aggregate by averaging. 260 00:24:55,760 --> 00:24:58,600 So what's the average during the month? 261 00:24:58,600 --> 00:25:04,320 So here we'll do that with total employees, which is the average number of employees during 262 00:25:04,320 --> 00:25:06,280 a given month. 263 00:25:06,280 --> 00:25:10,840 You could also do some, which we'll be doing later with sales. 264 00:25:10,840 --> 00:25:15,340 And then that would just be what's the total during a given month. 265 00:25:15,340 --> 00:25:19,360 So it's the total sales during a given month. 266 00:25:19,360 --> 00:25:25,440 Here we're looking at averages. 267 00:25:25,440 --> 00:25:35,200 So for example, we can now look at, okay, what's the monthly average number of employees 268 00:25:35,200 --> 00:25:39,580 in the Canvas industry in Massachusetts? 269 00:25:39,580 --> 00:25:46,680 And we can see our monthly averages here. 270 00:25:46,680 --> 00:25:55,040 Real quick, this is a cool little, oops, the cool little note here is if you actually look 271 00:25:55,040 --> 00:26:05,960 at these two plots together, you'll notice that our monthly average is essentially, you 272 00:26:05,960 --> 00:26:12,160 know, it's not quite because it changes month, right? 273 00:26:12,160 --> 00:26:17,140 The different months have different number of days, but it's roughly a 30 day moving 274 00:26:17,140 --> 00:26:18,700 average. 275 00:26:18,700 --> 00:26:29,280 So that's what you see with the moving averages, the moving average tends to smooth out the 276 00:26:29,280 --> 00:26:31,280 daily series. 277 00:26:31,280 --> 00:26:39,660 So anyways, quick little aside that I thought was interesting. 278 00:26:39,660 --> 00:26:45,240 So anyways, we have the monthly average here. 279 00:26:45,240 --> 00:26:49,160 So now we can actually compare that to the total employees. 280 00:26:49,160 --> 00:26:51,160 Oh yes, and that's right. 281 00:26:51,160 --> 00:26:52,520 I was explaining that. 282 00:26:52,520 --> 00:27:05,040 Oh, if we look at our monthly data here, the way I tend to conceptualize things is to timestamp 283 00:27:05,040 --> 00:27:08,080 them at the end of the month, right? 284 00:27:08,080 --> 00:27:17,080 Because if you're talking about the average number of employees in October of 2018, for 285 00:27:17,080 --> 00:27:25,600 me it conceptually makes a bit more sense to timestamp that as October 31st versus October 286 00:27:25,600 --> 00:27:30,320 1st, but that's entirely a preference. 287 00:27:30,320 --> 00:27:42,680 Well in this, for this analysis, it's entirely a preference, but it matters for the programming. 288 00:27:42,680 --> 00:27:53,540 So long story short, I'm using this function here to basically just take our, so what we 289 00:27:53,540 --> 00:28:06,360 have is, all I'm doing is taking our total Massachusetts employees and changing the timestamp 290 00:28:06,360 --> 00:28:14,240 from the beginning of the month to the end of the month. 291 00:28:14,240 --> 00:28:23,400 That way, you know, that way we have the same data, but it just timestamped at the end of 292 00:28:23,400 --> 00:28:24,400 the month. 293 00:28:24,400 --> 00:28:27,200 So that way we're comparing apples to apples. 294 00:28:27,200 --> 00:28:29,800 We're saying, okay, we're comparing this month to this month. 295 00:28:29,800 --> 00:28:32,440 It's real clear what we're doing here. 296 00:28:32,440 --> 00:28:36,560 Furthermore, we want to compare the same units to units. 297 00:28:36,560 --> 00:28:44,220 So changing thousands of people to number of people. 298 00:28:44,220 --> 00:28:48,840 So can do that simple enough. 299 00:28:48,840 --> 00:28:53,200 And then we can actually look at our employees here. 300 00:28:53,200 --> 00:29:04,880 So we can say, here's our monthly number of employees. 301 00:29:04,880 --> 00:29:08,760 I think we've already done this. 302 00:29:08,760 --> 00:29:12,400 Exactly. 303 00:29:12,400 --> 00:29:17,720 Then we can plot that with the total number of Massachusetts employees. 304 00:29:17,720 --> 00:29:26,360 And okay, right off the bat, what's this? 305 00:29:26,360 --> 00:29:34,280 Like is this like 3 million some employees in Massachusetts? 306 00:29:34,280 --> 00:29:37,800 Exactly. 307 00:29:37,800 --> 00:29:45,840 So you've got about 3.5 million total employees in Massachusetts. 308 00:29:45,840 --> 00:29:52,840 So quite a large number. 309 00:29:52,840 --> 00:30:00,600 Seems like too many, doesn't it? 310 00:30:00,600 --> 00:30:05,680 Real quick, let's just grab the population of Massachusetts real quick. 311 00:30:05,680 --> 00:30:12,520 So here I'm just grabbing the population of Massachusetts. 312 00:30:12,520 --> 00:30:15,560 So what was the population here? 313 00:30:15,560 --> 00:30:19,600 Total says 6.893 million as of 2019. 314 00:30:19,600 --> 00:30:23,520 Okay, that's more reasonable then. 315 00:30:23,520 --> 00:30:39,840 So you've got, well, we could do the math of the percentage of people actually working. 316 00:30:39,840 --> 00:30:43,480 We may do that here in one second. 317 00:30:43,480 --> 00:30:46,120 So let's just do it right now. 318 00:30:46,120 --> 00:30:47,120 That's what we're all about. 319 00:30:47,120 --> 00:31:06,080 The reason I was hesitant is we're going to have to 320 00:31:06,080 --> 00:31:30,480 create the annual number of Massachusetts employees, which we just want to do an average. 321 00:31:30,480 --> 00:31:36,800 If this works out well, we can do it. 322 00:31:36,800 --> 00:31:41,040 Awesome. 323 00:31:41,040 --> 00:31:44,920 So now we have the annual number of Massachusetts employees. 324 00:31:44,920 --> 00:31:51,880 And then we can just divide that by the population. 325 00:31:51,880 --> 00:31:56,400 Not quite. 326 00:31:56,400 --> 00:32:05,000 Wait one second. 327 00:32:05,000 --> 00:32:06,560 Let's look at these two data sets here. 328 00:32:06,560 --> 00:32:10,520 We've got the number of employees. 329 00:32:10,520 --> 00:32:13,520 We've got the population. 330 00:32:13,520 --> 00:32:22,800 Oh yes, that's right. 331 00:32:22,800 --> 00:32:37,640 It's just this sort of weird pandas thing going on here. 332 00:32:37,640 --> 00:32:41,920 So there we are. 333 00:32:41,920 --> 00:32:50,520 So roughly 50% of the population is working at any given time. 334 00:32:50,520 --> 00:32:58,760 And keep in mind, this is different than the unemployment rate, because the population 335 00:32:58,760 --> 00:33:01,400 includes children. 336 00:33:01,400 --> 00:33:04,120 It includes retired people. 337 00:33:04,120 --> 00:33:11,280 And children and retired people and people that aren't in the labor force, they're not 338 00:33:11,280 --> 00:33:14,960 included when you're calculating unemployment. 339 00:33:14,960 --> 00:33:21,160 So this is a different statistic here. 340 00:33:21,160 --> 00:33:24,800 So I just wanted to do it more for a sanity check, right? 341 00:33:24,800 --> 00:33:32,300 Because you can't have more people working than there are living in the state. 342 00:33:32,300 --> 00:33:38,240 So I was more doing it for a sanity check. 343 00:33:38,240 --> 00:33:43,440 So anyways, we've got the total number of employees as we saw. 344 00:33:43,440 --> 00:33:47,040 Oh yes, I wanted to comment on this real quick. 345 00:33:47,040 --> 00:33:54,520 So we've got the total number of employees here in Massachusetts. 346 00:33:54,520 --> 00:34:04,120 And if you look at the cannabis industry, you'll see how they really demonstrate how 347 00:34:04,120 --> 00:34:12,700 the cannabis industry proved to be, well, they were defined as an essential industry 348 00:34:12,700 --> 00:34:14,480 during the pandemic. 349 00:34:14,480 --> 00:34:20,020 And it definitely has an effect here on the employment here. 350 00:34:20,020 --> 00:34:30,760 So you'll notice in August of 2020, or around that time frame, that you do see a slight 351 00:34:30,760 --> 00:34:34,240 dip in employment. 352 00:34:34,240 --> 00:34:37,720 However, it remains almost constant. 353 00:34:37,720 --> 00:34:44,360 So it's almost like in the cannabis industry, there is essentially a hiring freeze, not 354 00:34:44,360 --> 00:34:52,360 a lot of new hires, or at least any new hires were essentially offset by anyone leaving. 355 00:34:52,360 --> 00:35:06,680 It looks like there were a handful of people that, for a better word, they quit being employees 356 00:35:06,680 --> 00:35:07,680 during that time. 357 00:35:07,680 --> 00:35:18,200 Maybe they lost their jobs, maybe they resigned, who knows, maybe they got fired or laid off. 358 00:35:18,200 --> 00:35:20,280 What about homegrown? 359 00:35:20,280 --> 00:35:22,560 Maybe they took up homegrown. 360 00:35:22,560 --> 00:35:24,520 Seriously, it's happening. 361 00:35:24,520 --> 00:35:26,240 People are moving for that reason. 362 00:35:26,240 --> 00:35:27,240 Thank you. 363 00:35:27,240 --> 00:35:33,360 Oh yes, and that may actually be what this crowd is. 364 00:35:33,360 --> 00:35:36,760 So these people are already in the cannabis industry here. 365 00:35:36,760 --> 00:35:44,160 Yes, people may be transitioning from banking or what have you. 366 00:35:44,160 --> 00:35:55,480 Their everyday retail jobs do exactly that, to home grow perhaps. 367 00:35:55,480 --> 00:36:03,040 We can only conjecture as to what the reasons are or what is underlying this data. 368 00:36:03,040 --> 00:36:12,160 However, I think it's interesting to remark and to note that the cannabis industry was 369 00:36:12,160 --> 00:36:21,760 quite resilient to job loss during the pandemic, which remember last week we were talking about 370 00:36:21,760 --> 00:36:31,880 how regulations often affect the structure and performance and outcomes of various industries. 371 00:36:31,880 --> 00:36:39,400 And so simply just saying, okay, this industry is essential. 372 00:36:39,400 --> 00:36:43,240 Well that likely has an impact here. 373 00:36:43,240 --> 00:36:46,840 And so, you know, does that explain everything? 374 00:36:46,840 --> 00:36:54,400 No, like there's always a multitude of factors here. 375 00:36:54,400 --> 00:37:01,300 So I don't know, I just, so this is something that's been observed, not just by me, a lot 376 00:37:01,300 --> 00:37:07,360 of people have observed this, that, you know, the cannabis industry was resilient during 377 00:37:07,360 --> 00:37:09,960 the pandemic. 378 00:37:09,960 --> 00:37:18,560 And the data shows that here in Massachusetts where the employment levels pretty much, they 379 00:37:18,560 --> 00:37:24,840 stayed constant during that time, but there wasn't a dramatic drop. 380 00:37:24,840 --> 00:37:34,760 And then, you know, basically what we see here is the trend looks like the trend resumes 381 00:37:34,760 --> 00:37:39,680 to about what the trend was, but perhaps maybe a little more volatility here. 382 00:37:39,680 --> 00:37:47,840 However, the overall rate, there is, it looks like there was an entire shift in the curve. 383 00:37:47,840 --> 00:37:54,960 So it looks like, okay, there was the shock, the adjustment, steady state growth is now 384 00:37:54,960 --> 00:38:02,720 back on track, but there was a whole shift down in employment. 385 00:38:02,720 --> 00:38:14,800 Whereas the cannabis industry, there was the shock, not much of a shock, at least to employment, 386 00:38:14,800 --> 00:38:24,120 and then resumes to steady state growth with, you know, just a slight, slight shift in the 387 00:38:24,120 --> 00:38:27,240 labor curve. 388 00:38:27,240 --> 00:38:36,000 So I think there's more insights that can be taken away here, but we're just going to 389 00:38:36,000 --> 00:38:40,960 kind of keep moving along here because there's a lot of ground to cover. 390 00:38:40,960 --> 00:38:47,480 So just to look at another statistic here. 391 00:38:47,480 --> 00:38:57,400 Oh yes, we were just going to start to gauge, okay, cannabis industry is resilient. 392 00:38:57,400 --> 00:38:58,920 Is it meaningful, right? 393 00:38:58,920 --> 00:39:02,040 What percent of the economy is this? 394 00:39:02,040 --> 00:39:06,160 Is it a noteworthy segment of the economy, right? 395 00:39:06,160 --> 00:39:11,000 Because like I said, if you talk to people in the cannabis industry, they'll make you 396 00:39:11,000 --> 00:39:16,000 believe that, you know, the cannabis industry is propping up the entire economy. 397 00:39:16,000 --> 00:39:24,800 Let's see how true that is, or at least try to quantify. 398 00:39:24,800 --> 00:39:30,060 So I think this is interesting data here. 399 00:39:30,060 --> 00:39:40,000 So here you have just the cannabis employees as a percent of all employees. 400 00:39:40,000 --> 00:39:46,120 And so obviously they were 0% to begin with. 401 00:39:46,120 --> 00:39:52,400 And keep in mind, and so this is where we're going to start talking about the economic 402 00:39:52,400 --> 00:39:55,320 impact of permitting cannabis, right? 403 00:39:55,320 --> 00:39:58,040 Because was it actually zero? 404 00:39:58,040 --> 00:40:04,440 So what we've talked about this in the past, whenever you run up to a zero bound, well, 405 00:40:04,440 --> 00:40:08,440 that may mean you're not capturing things on the other side of the bound. 406 00:40:08,440 --> 00:40:13,340 And what's on the other side of the bound here is you may have people that are working 407 00:40:13,340 --> 00:40:17,240 in illegal cannabis markets, right? 408 00:40:17,240 --> 00:40:24,840 So people doing illegal grows, people doing illegal processing, people doing illegal retail. 409 00:40:24,840 --> 00:40:28,680 And that's not going to be captured. 410 00:40:28,680 --> 00:40:31,960 That's not measured economic activity. 411 00:40:31,960 --> 00:40:37,960 That's typically not tax economic activity. 412 00:40:37,960 --> 00:40:46,520 And it has own debatable social costs, right? 413 00:40:46,520 --> 00:40:48,920 So this is debatable, right? 414 00:40:48,920 --> 00:40:57,520 Because you don't want to be tossing good people in jail for doing some activity that's 415 00:40:57,520 --> 00:40:59,080 now legal today, right? 416 00:40:59,080 --> 00:41:03,080 And so that's why there's these no... 417 00:41:03,080 --> 00:41:06,760 I want to give a shout out to them, but I don't know their specific names, but there's... 418 00:41:06,760 --> 00:41:14,480 I'll have to put some links in the chat or something, but there's movements out there 419 00:41:14,480 --> 00:41:20,400 to try to help people that were in prison for cannabis crimes that were nonviolent, 420 00:41:20,400 --> 00:41:25,400 but today would be perfectly permissible. 421 00:41:25,400 --> 00:41:31,520 But that's not to under shadow that whenever things are illegal, it does attract kind of 422 00:41:31,520 --> 00:41:33,280 shady people sometimes. 423 00:41:33,280 --> 00:41:40,720 So not saying that everybody's bad, but unfortunately when things are illegal, right? 424 00:41:40,720 --> 00:41:45,840 That's when you kind of run into the organized crime and some bad actors. 425 00:41:45,840 --> 00:41:52,640 So maybe some decent people get lumped in with some bad characters. 426 00:41:52,640 --> 00:41:57,080 So long story short, if things can be permitted, then... 427 00:41:57,080 --> 00:42:02,880 And this is where you kind of get into the judgment calls, but in my opinion, if things 428 00:42:02,880 --> 00:42:09,280 are permitted, then you can at least put them out under the light. 429 00:42:09,280 --> 00:42:20,320 You can let things operate, at least they can get measured and regulated and operate 430 00:42:20,320 --> 00:42:22,320 under this regulation series. 431 00:42:22,320 --> 00:42:26,160 But long story short, that's a little bit of a spiel there. 432 00:42:26,160 --> 00:42:33,400 Didn't mean to make any opinionated statements because we're focused here on the data. 433 00:42:33,400 --> 00:42:38,800 So anyways, back to the data. 434 00:42:38,800 --> 00:42:45,920 We've got about a quarter of a percent of people today working in the cannabis industry 435 00:42:45,920 --> 00:42:48,200 in Massachusetts. 436 00:42:48,200 --> 00:42:51,860 So I think this is a non-negligible portion. 437 00:42:51,860 --> 00:42:55,960 It seems small, but I think this is larger than you would think. 438 00:42:55,960 --> 00:43:01,200 So I think it would be interesting to compare to other professions. 439 00:43:01,200 --> 00:43:09,960 So if we had more time, I would get the data on how many teachers are there in Massachusetts, 440 00:43:09,960 --> 00:43:12,200 how many... 441 00:43:12,200 --> 00:43:15,200 I think people have done this, how many fast food workers are there. 442 00:43:15,200 --> 00:43:19,360 And so you can kind of compare how many fast food workers to how many people working in 443 00:43:19,360 --> 00:43:24,320 cannabis to kind of get some perspective, tell a story. 444 00:43:24,320 --> 00:43:25,960 So that would be beneficial. 445 00:43:25,960 --> 00:43:31,840 We're time strapped today, limited on time, so we're just going to keep moving. 446 00:43:31,840 --> 00:43:37,000 But I think it's worth pointing out that a couple of things are quick. 447 00:43:37,000 --> 00:43:38,940 It's increasing. 448 00:43:38,940 --> 00:43:46,760 It increased faster during this recession here in 2020. 449 00:43:46,760 --> 00:43:54,880 And the reason I say recession is because the Federal Reserve has flagged this period 450 00:43:54,880 --> 00:43:58,600 as a recession, these gray bars are here. 451 00:43:58,600 --> 00:44:09,400 So the Federal Reserve has designated this quarter in 2020 as a recession, I believe. 452 00:44:09,400 --> 00:44:12,840 If I'm wrong, please let me know. 453 00:44:12,840 --> 00:44:18,640 But anyways, we've got the total number of employees here. 454 00:44:18,640 --> 00:44:22,440 Here, one second. 455 00:44:22,440 --> 00:44:34,120 We may have somebody joining us real quick. 456 00:44:34,120 --> 00:44:41,600 So welcome, Hentoko. 457 00:44:41,600 --> 00:44:45,840 We've got about 15 minutes here to finish up some interesting analysis. 458 00:44:45,840 --> 00:44:48,920 So you're just in time for some good takeaways. 459 00:44:48,920 --> 00:44:53,720 And then I'd be happy to talk with you some more. 460 00:44:53,720 --> 00:45:03,040 So long story short, we're looking here at the GDP in Massachusetts, and we're comparing 461 00:45:03,040 --> 00:45:10,320 that to, okay, what's the economic output here in the cannabis industry? 462 00:45:10,320 --> 00:45:18,880 We just saw that about a quarter of a percent of people here work in the cannabis industry. 463 00:45:18,880 --> 00:45:23,480 And that may be understated because as Heather pointed out, the total number of employees 464 00:45:23,480 --> 00:45:30,080 here may be understated. 465 00:45:30,080 --> 00:45:38,200 When that's thing, well, you can kind of correlate these two, may as well just create this statistic. 466 00:45:38,200 --> 00:45:40,360 They're negatively correlated. 467 00:45:40,360 --> 00:45:57,520 So it's hard to interpret this, but basically, as the total employment in Massachusetts decreases, 468 00:45:57,520 --> 00:46:03,480 you actually have employment increasing in the cannabis industry, which is an interesting 469 00:46:03,480 --> 00:46:04,480 observation. 470 00:46:04,480 --> 00:46:08,240 So anyways, to GDP real quick. 471 00:46:08,240 --> 00:46:13,600 So in this last 15 minutes, we can do some pretty incredible stuff here. 472 00:46:13,600 --> 00:46:18,280 So we can measure GDP. 473 00:46:18,280 --> 00:46:20,200 So what is GDP? 474 00:46:20,200 --> 00:46:27,120 Well, that's the whole economic output in an economy. 475 00:46:27,120 --> 00:46:33,680 In this case, we're talking about the cannabis sector, and we can simply proxy the whole 476 00:46:33,680 --> 00:46:38,120 economic output as consumption. 477 00:46:38,120 --> 00:46:42,560 So we can actually measure cannabis consumption. 478 00:46:42,560 --> 00:46:46,400 Keep in mind, I'm simplifying this here. 479 00:46:46,400 --> 00:46:56,920 There are some good resources that discuss, okay, what is GDP? 480 00:46:56,920 --> 00:47:00,280 Let's see. 481 00:47:00,280 --> 00:47:03,120 Right. 482 00:47:03,120 --> 00:47:11,680 So in general, right, it's consumption, investment, government expenditures, and then exports. 483 00:47:11,680 --> 00:47:15,320 So but right, we're just going to proxy things as consumption. 484 00:47:15,320 --> 00:47:19,440 And this is the way you see things done in macroeconomics a lot. 485 00:47:19,440 --> 00:47:24,240 We can abstract away the government spending and the investment, and there's not going 486 00:47:24,240 --> 00:47:25,960 to be any net exports. 487 00:47:25,960 --> 00:47:31,240 So we can abstract those away for the time being and just say, okay, GDP is going to 488 00:47:31,240 --> 00:47:37,800 be consumption, which is going to be sales. 489 00:47:37,800 --> 00:47:39,640 Here I am calculating sales. 490 00:47:39,640 --> 00:47:53,800 I noticed that sales total is actually a cumulative total. 491 00:47:53,800 --> 00:47:57,200 That's not what daily data should look like. 492 00:47:57,200 --> 00:48:05,320 So we can actually take the difference here to calculate our daily data. 493 00:48:05,320 --> 00:48:16,440 And so now if you just plot sales, there is this one outlier, which I'm quite worried 494 00:48:16,440 --> 00:48:17,440 about. 495 00:48:17,440 --> 00:48:23,080 And so I think we should come and revisit what happened on this day. 496 00:48:23,080 --> 00:48:26,920 This could be a miscoding measurement error. 497 00:48:26,920 --> 00:48:30,800 So this is a problem. 498 00:48:30,800 --> 00:48:38,880 In fact, I'd probably exclude this one day, but to the time being, it doesn't overly affect 499 00:48:38,880 --> 00:48:40,600 our analysis here. 500 00:48:40,600 --> 00:48:46,600 But I think it should be pointed out and it needs to be investigated further. 501 00:48:46,600 --> 00:48:51,320 And our analysis probably needs to be redone with this one observation excluded because 502 00:48:51,320 --> 00:48:57,320 it's atypical. 503 00:48:57,320 --> 00:49:07,880 So for example, if we just look at the last 100 observations here, you'll see, okay, this 504 00:49:07,880 --> 00:49:14,800 is more typical for what daily data looks like, where we have fluctuations, right? 505 00:49:14,800 --> 00:49:21,040 Fridays may be busier than Mondays or Tuesdays or what have you. 506 00:49:21,040 --> 00:49:25,120 Which is a whole other interesting analysis of its own, right? 507 00:49:25,120 --> 00:49:28,480 I'm a big fan of looking at the days of the week. 508 00:49:28,480 --> 00:49:36,160 So exactly like I just said, how do sales on Friday compare to Monday or Tuesday? 509 00:49:36,160 --> 00:49:38,160 In fact, we could probably do that next week. 510 00:49:38,160 --> 00:49:41,840 We'll probably do that in not that much time. 511 00:49:41,840 --> 00:49:44,160 But like we said, we're strapped on time here. 512 00:49:44,160 --> 00:49:50,760 So we're going to try to stay focused and save some of these cool analyses for the future. 513 00:49:50,760 --> 00:49:58,560 We've got sales here, daily data. 514 00:49:58,560 --> 00:50:11,480 So we can aggregate this into monthly and quarterly data, which we'll do. 515 00:50:11,480 --> 00:50:13,880 We've already gotten our total employees. 516 00:50:13,880 --> 00:50:16,800 We've already gotten the population. 517 00:50:16,800 --> 00:50:21,640 It's now time to define GDP here. 518 00:50:21,640 --> 00:50:24,760 So we're going to define GDP. 519 00:50:24,760 --> 00:50:31,200 And here I'm going to put everything into millions of dollars because we're actually 520 00:50:31,200 --> 00:50:39,080 going to be comparing this to the GDP of Massachusetts to put things into perspective. 521 00:50:39,080 --> 00:50:45,600 And the GDP of Massachusetts measured by the Federal Reserve is measured in millions of 522 00:50:45,600 --> 00:50:49,640 dollars. 523 00:50:49,640 --> 00:50:57,760 So we're going to put our units in millions of dollars just for convenience sake. 524 00:50:57,760 --> 00:51:08,680 Then we are going to get the GDP from Federal Reserve and make sure everything's time stamped 525 00:51:08,680 --> 00:51:14,000 at the end of the quarter and not the beginning of the quarter. 526 00:51:14,000 --> 00:51:19,200 And so without further ado, what do these series look like? 527 00:51:19,200 --> 00:51:33,840 Well, here is quarterly cannabis GDP or quarterly cannabis sales, interchangeable in our case. 528 00:51:33,840 --> 00:51:41,280 So sales may be a bit more familiar to those people. 529 00:51:41,280 --> 00:51:50,200 And here is GDP in Massachusetts during a slightly shorter timeframe here. 530 00:51:50,200 --> 00:52:00,320 And in fact, to compare apples to apples, I'm actually going to plot just quarterly 531 00:52:00,320 --> 00:52:09,520 cannabis through quarter one of 2021 because that's all we have, data for Massachusetts 532 00:52:09,520 --> 00:52:11,960 GDP. 533 00:52:11,960 --> 00:52:19,240 Okay, now we're comparing apples to apples here. 534 00:52:19,240 --> 00:52:25,480 And what we notice is, I think this is interesting. 535 00:52:25,480 --> 00:52:40,000 So remember earlier we saw employment was almost impervious to the shock of April of 536 00:52:40,000 --> 00:52:42,960 2020. 537 00:52:42,960 --> 00:52:48,280 Sales were actually hit here. 538 00:52:48,280 --> 00:52:56,440 And so this is contrary to I think perhaps what we saw in other places or maybe just 539 00:52:56,440 --> 00:53:04,600 the general perspective that sales increased during this time. 540 00:53:04,600 --> 00:53:07,560 And so there's a couple things going on here. 541 00:53:07,560 --> 00:53:11,160 One, we may have just entirely messed up our measurement. 542 00:53:11,160 --> 00:53:12,160 That's one. 543 00:53:12,160 --> 00:53:15,360 It's not impossible. 544 00:53:15,360 --> 00:53:24,760 And you have an interesting effect here in economics where people will shift their purchases. 545 00:53:24,760 --> 00:53:30,720 So just because you see a spike in purchases, just say everybody rushes to the store and 546 00:53:30,720 --> 00:53:36,480 they buy a bunch of cannabis at this one specific time and you see a large spike, that doesn't 547 00:53:36,480 --> 00:53:41,180 necessarily mean their overall consumption is going to increase. 548 00:53:41,180 --> 00:53:50,280 So they may just change the time at which they make their purchase and then consume 549 00:53:50,280 --> 00:53:51,280 the same amount. 550 00:53:51,280 --> 00:53:55,560 So they may have purchased a lot at a given time. 551 00:53:55,560 --> 00:54:01,640 So there may have been a spike in sales during this time. 552 00:54:01,640 --> 00:54:14,360 So instead of just conjecturing, we can actually look. 553 00:54:14,360 --> 00:54:18,000 This is going to cover the right timeframe here. 554 00:54:18,000 --> 00:54:32,680 So let's see if we can't actually look at sales during this time period. 555 00:54:32,680 --> 00:54:40,000 Sorry I could probably pinpoint this a little better, but just sort of ad hoc trying to 556 00:54:40,000 --> 00:54:47,880 find this. 557 00:54:47,880 --> 00:54:50,520 Okay this is not good. 558 00:54:50,520 --> 00:54:59,760 We may even have measurement error here. 559 00:54:59,760 --> 00:55:02,460 Okay so this is not good. 560 00:55:02,460 --> 00:55:13,580 So we just zoomed in on this pandemic period here and would you look at this. 561 00:55:13,580 --> 00:55:19,520 We don't have any sales data for this period here. 562 00:55:19,520 --> 00:55:28,160 So that would actually explain our big decrease here. 563 00:55:28,160 --> 00:55:32,080 So I think we may need to do a deeper dive here. 564 00:55:32,080 --> 00:55:42,080 So I need you to look and see, okay did Massachusetts maybe suspend reporting during this period? 565 00:55:42,080 --> 00:55:51,080 Maybe they said okay you know we don't have to report sales during this period. 566 00:55:51,080 --> 00:55:53,580 There could be a lot of things going on. 567 00:55:53,580 --> 00:56:00,480 So this is sort of what data science is all about here right. 568 00:56:00,480 --> 00:56:04,080 We just had a monkey wrench thrown in. 569 00:56:04,080 --> 00:56:16,160 Welcome back Kandoko. 570 00:56:16,160 --> 00:56:22,240 I'm going to be wrapping up my spiel here then momentarily. 571 00:56:22,240 --> 00:56:26,560 That way we can get through a little discussion here because this is a pretty glaring monkey 572 00:56:26,560 --> 00:56:30,480 wrench that we're going to have to address for next week. 573 00:56:30,480 --> 00:56:35,480 I'll still show you how the analysis can be done but this is what kind of Heather was 574 00:56:35,480 --> 00:56:41,480 talking about where unfortunately if you put the data under your microscope you're going 575 00:56:41,480 --> 00:56:48,780 to find some glaring oddities sometimes. 576 00:56:48,780 --> 00:56:54,520 So that need to be explained and this is one of them. 577 00:56:54,520 --> 00:56:56,960 So we just put the data under your microscope. 578 00:56:56,960 --> 00:57:07,400 We're basically just missing data here for the most important period. 579 00:57:07,400 --> 00:57:29,200 We're one of the more important periods here. 580 00:57:29,200 --> 00:57:39,160 So let's see if we just I just want to see real quick if okay yep so it's the data set 581 00:57:39,160 --> 00:57:45,120 themselves so maybe I'll do a deeper dive and try to figure out and explain for next 582 00:57:45,120 --> 00:57:47,120 week why this is missing. 583 00:57:47,120 --> 00:57:52,360 But basically we just put the data under a microscope and we found out that we're missing 584 00:57:52,360 --> 00:57:57,760 data for this critical time period here and that's probably going to have an effect on 585 00:57:57,760 --> 00:58:06,000 our analysis right because here we are trying to make these statements about what's going 586 00:58:06,000 --> 00:58:12,420 on here in quarter two and we've got measurement errors. 587 00:58:12,420 --> 00:58:18,000 So that's a big problem and so that makes us wonder like are there measurement errors 588 00:58:18,000 --> 00:58:23,740 going on here that explain this dip. 589 00:58:23,740 --> 00:58:37,960 So we've got problems but that's life as a data scientist but just to show you the last 590 00:58:37,960 --> 00:58:42,760 little bit of analysis here because despite having measurement errors we can still calculate 591 00:58:42,760 --> 00:58:48,360 the statistics because I think they're still worthwhile to look at right. 592 00:58:48,360 --> 00:58:52,760 We'd rather calculate the statistics than not. 593 00:58:52,760 --> 00:58:55,160 And it rated the stage we've not. 594 00:58:55,160 --> 00:58:57,360 So let's calculate them real quick. 595 00:58:57,360 --> 00:59:09,200 So let's grab Massachusetts GDP maybe we've already done that but let's grab it again. 596 00:59:09,200 --> 00:59:16,880 We've plotted them we can calculate the correlation between the two. 597 00:59:16,880 --> 00:59:21,640 They were positively correlated but not perfectly correlated. 598 00:59:21,640 --> 00:59:28,060 So a lot of times you'll see much stronger correlation with sectors and GDP. 599 00:59:28,060 --> 00:59:34,280 So I'd say this is a weak correlation not I mean an actual weak correlation you know 600 00:59:34,280 --> 00:59:35,920 point one point two. 601 00:59:35,920 --> 00:59:45,160 So this is still statistically a strong correlation but not like a I would not say this is a strong 602 00:59:45,160 --> 00:59:53,280 like economic correlation which is an interesting real real real interesting observation. 603 00:59:53,280 --> 00:59:58,680 But we're going to be cursory with it. 604 00:59:58,680 --> 01:00:03,640 We're going to say okay we're quantifying it here and sorry that I'm rushing through 605 01:00:03,640 --> 01:00:07,240 this we can expand on everything next week. 606 01:00:07,240 --> 01:00:14,000 But we'll say okay what's cannabis is a percent of GDP. 607 01:00:14,000 --> 01:00:23,200 Well my takeaway is Massachusetts has a quite booming robust economy here. 608 01:00:23,200 --> 01:00:35,280 So yes cannabis is this amazing interesting sector but if you look at the data they may 609 01:00:35,280 --> 01:00:41,360 it may just be I want to say almost like a drop in the bucket of the whole economy of 610 01:00:41,360 --> 01:00:42,360 Massachusetts here. 611 01:00:42,360 --> 01:00:45,680 But like I said it's not negligible. 612 01:00:45,680 --> 01:00:53,160 You've got 0.04 percent of the entire economy. 613 01:00:53,160 --> 01:00:57,320 So it's not nothing. 614 01:00:57,320 --> 01:01:06,400 It's not everything by all means. 615 01:01:06,400 --> 01:01:20,120 So you know it is what it is but it's not it's not the it's not the end all be all. 616 01:01:20,120 --> 01:01:34,080 But let's put this into dollars so we can say okay what's the annual GDP here. 617 01:01:34,080 --> 01:01:39,960 And we can visualize that. 618 01:01:39,960 --> 01:01:44,600 Okay so like we said it's not nothing. 619 01:01:44,600 --> 01:01:51,120 So here's just the annual cannabis sales or cannabis GDP in Massachusetts. 620 01:01:51,120 --> 01:01:53,800 So like we said it's almost nothing. 621 01:01:53,800 --> 01:02:01,720 But look at this it's almost one billion dollars in 2021. 622 01:02:01,720 --> 01:02:03,160 And the year is not even over yet. 623 01:02:03,160 --> 01:02:06,400 We've still got like a whole quarter left. 624 01:02:06,400 --> 01:02:10,800 So Massachusetts is doing well for themselves. 625 01:02:10,800 --> 01:02:13,960 Like we said Massachusetts is a big economy here. 626 01:02:13,960 --> 01:02:17,840 So this is great. 627 01:02:17,840 --> 01:02:21,080 Like look at look at the size of their economy here. 628 01:02:21,080 --> 01:02:31,480 This is quarterly six hundred thousand million dollars. 629 01:02:31,480 --> 01:02:34,560 So they've got a big economy here. 630 01:02:34,560 --> 01:02:40,520 But you know a billion dollars is not nothing. 631 01:02:40,520 --> 01:02:47,640 So like you said you need to weigh that against the social costs. 632 01:02:47,640 --> 01:02:54,840 But I mean if you're going to increase GDP of a state by one billion dollars that's not 633 01:02:54,840 --> 01:02:56,840 bad in my book. 634 01:02:56,840 --> 01:02:58,840 But that's an opinion. 635 01:02:58,840 --> 01:03:07,680 And so then okay what's the annual percent of GDP here? 636 01:03:07,680 --> 01:03:15,760 Well as we noted you know it's not the biggest segment of the economy. 637 01:03:15,760 --> 01:03:20,800 But look at this you know. 638 01:03:20,800 --> 01:03:37,280 The cannabis GDP is about 0.16 percent of all of GDP. 639 01:03:37,280 --> 01:03:43,120 But I mean when you're talking about like a boost to GDP right. 640 01:03:43,120 --> 01:03:50,880 People are out there like the nation shooting like for like you know like a 2 percent or 641 01:03:50,880 --> 01:03:51,880 so. 642 01:03:51,880 --> 01:03:52,880 More than that's awesome right. 643 01:03:52,880 --> 01:03:59,240 So you know when you're talking about just trying to squeak up GDP I mean you're squeaking 644 01:03:59,240 --> 01:04:06,240 up GDP you know 0.16 of a percent that's not nothing. 645 01:04:06,240 --> 01:04:14,200 And like to put that into dollar terms well I thought this was an interesting statistic 646 01:04:14,200 --> 01:04:15,200 here. 647 01:04:15,200 --> 01:04:19,240 Well we've got the population of Massachusetts. 648 01:04:19,240 --> 01:04:25,520 We've calculated GDP which is used to say okay that's the it's used to kind of proxy 649 01:04:25,520 --> 01:04:28,960 you know the standard of living sometimes. 650 01:04:28,960 --> 01:04:37,880 And so we can say okay well how much better off is a person in Massachusetts for having 651 01:04:37,880 --> 01:04:42,020 the cannabis industry in economic terms. 652 01:04:42,020 --> 01:04:50,200 Social terms that's a whole nother can of worms that will maybe be for somebody. 653 01:04:50,200 --> 01:04:53,200 So but we can look at the economic side. 654 01:04:53,200 --> 01:04:58,680 Okay so what's the GDP per capita? 655 01:04:58,680 --> 01:05:09,400 Well in 2018 that was just about two dollars so negligible. 656 01:05:09,400 --> 01:05:15,880 Already by 2019 per capita that's around sixty five dollars. 657 01:05:15,880 --> 01:05:27,040 So per person that's man woman and child and the retired elderly everybody sixty five dollars. 658 01:05:27,040 --> 01:05:34,180 And then by 2021 almost one hundred and forty dollars. 659 01:05:34,180 --> 01:05:42,280 So that's every person in Massachusetts is about one hundred and forty dollars better 660 01:05:42,280 --> 01:05:49,000 off per year because of the cannabis industry just just passively. 661 01:05:49,000 --> 01:05:55,680 Right just you know they're you know just the economy is better right. 662 01:05:55,680 --> 01:06:03,120 It's just a bigger economy right there's more tax revenue there's just more economic activity 663 01:06:03,120 --> 01:06:05,880 and they're just they're just better off. 664 01:06:05,880 --> 01:06:13,680 So just having the cannabis industry permitted right you don't even have to partake in the 665 01:06:13,680 --> 01:06:14,680 industry. 666 01:06:14,680 --> 01:06:21,200 So great so just for some people that actually partake that's you know beneficial for them. 667 01:06:21,200 --> 01:06:23,480 But even for the people that don't partake. 668 01:06:23,480 --> 01:06:30,800 I mean let's say you can think about it as a check in the mail once a year for one hundred 669 01:06:30,800 --> 01:06:35,160 and forty bucks and that's just now. 670 01:06:35,160 --> 01:06:37,400 And so it's going to be increasing right. 671 01:06:37,400 --> 01:06:45,400 So you know next year maybe it'll be two hundred bucks and just more and more and more maybe 672 01:06:45,400 --> 01:06:47,800 you know for perpetuity. 673 01:06:47,800 --> 01:06:55,520 So you hear we've done a thing about this this is with measurement error. 674 01:06:55,520 --> 01:06:59,640 This is excluding months worth of data. 675 01:06:59,640 --> 01:07:04,680 So this this number may even be biased downwards. 676 01:07:04,680 --> 01:07:15,960 So I think that's sort of my takeaway for today is by permitting the cannabis industry 677 01:07:15,960 --> 01:07:23,720 in Maryland in Massachusetts. 678 01:07:23,720 --> 01:07:34,080 There has been an economic benefit that I think is quantifiable because it has increased 679 01:07:34,080 --> 01:07:41,520 from what I can tell it appears to have increased GDP by a small amount but I would say non 680 01:07:41,520 --> 01:07:50,800 negligible amount to the portion where no every person in Massachusetts is you know 681 01:07:50,800 --> 01:07:59,560 at least you know on average about you know one hundred and forty or maybe one hundred 682 01:07:59,560 --> 01:08:05,320 maybe we can say there may be estimated estimation errors but maybe between one hundred and two 683 01:08:05,320 --> 01:08:15,160 hundred dollars per year better off than they were without the cannabis industry. 684 01:08:15,160 --> 01:08:22,880 And I mean that's not nothing in my book you know if I'd rather I'd personally rather have 685 01:08:22,880 --> 01:08:28,480 a check in the mail once a year for two hundred bucks than not. 686 01:08:28,480 --> 01:08:31,520 So that's awesome. 687 01:08:31,520 --> 01:08:38,640 And so I think this is my takeaway that you that other states could take to the bank is 688 01:08:38,640 --> 01:08:44,440 okay yes you know you're just looking at maybe the social aspect of this or maybe the health 689 01:08:44,440 --> 01:08:52,080 aspect where maybe some people are being like we talked about earlier they're forced into 690 01:08:52,080 --> 01:09:00,240 the gray illegal economy and that's maybe not so good at a social level or maybe at 691 01:09:00,240 --> 01:09:06,120 a medicinal level people can't get medicine they need or this or that. 692 01:09:06,120 --> 01:09:12,600 So those are different considerations but we look at that we're looking at the economics 693 01:09:12,600 --> 01:09:21,240 here and from an economic perspective you know the appears to be you know the benefit 694 01:09:21,240 --> 01:09:26,400 here and so you know I think it's a consideration that people should should take into consideration 695 01:09:26,400 --> 01:09:34,520 that you know if if you permit cannabis with regulations like they have in Massachusetts 696 01:09:34,520 --> 01:09:40,600 right it's not like they're they're just you know out there just letting anyone do anything 697 01:09:40,600 --> 01:09:48,720 they want there are rules and regulations in Massachusetts and so you know given a permitted 698 01:09:48,720 --> 01:09:58,680 market there's an economic benefit and so I think that ends sort of the point I wanted 699 01:09:58,680 --> 01:10:08,200 to to make today because that that was just sort of what my you know my thinking and sort 700 01:10:08,200 --> 01:10:13,480 of research and studies has sort of led to I think is I think that's that's the main 701 01:10:13,480 --> 01:10:21,680 takeaway is what is the economic impact of permitting these cannabis markets is it big 702 01:10:21,680 --> 01:10:28,280 is it negligible and can we quantify that and I think at least in Massachusetts we made 703 01:10:28,280 --> 01:10:32,200 a rough attempt at doing just that. 704 01:10:32,200 --> 01:10:44,560 So any thoughts questions comments? 705 01:10:44,560 --> 01:10:50,920 Well on that note it was awesome to have you Heather and then Hendo Ko if you're interested 706 01:10:50,920 --> 01:10:57,440 at all I'm curious to hear about you know what perspective you're coming at the group 707 01:10:57,440 --> 01:11:05,440 from because we're happy to have you. 708 01:11:05,440 --> 01:11:13,320 The R programming language okay so we have people in the group that use R so I'll have 709 01:11:13,320 --> 01:11:22,720 to get you in touch with one of our regulars Paul well I don't think there's a meme that 710 01:11:22,720 --> 01:11:32,840 Paul that that R is is almost this poor neglected you know the poor neglected cat and then everybody's 711 01:11:32,840 --> 01:11:39,880 focused on on Python the the cute the cute dog in the corner so I don't think R should 712 01:11:39,880 --> 01:11:47,240 be should be neglected or forgotten about so I'll get you in touch with Paul because 713 01:11:47,240 --> 01:11:58,160 like I said oh well I'll put a link here in the chat so you can join the group here meetup.com 714 01:11:58,160 --> 01:12:08,280 word slash cannabis data science I believe. 715 01:12:08,280 --> 01:12:20,400 Yes here we are here you can find the group put this in the chat yeah and I'll get you 716 01:12:20,400 --> 01:12:26,000 in touch with Paul he uses R and all of this analysis can be done in your favorite programming 717 01:12:26,000 --> 01:12:32,000 language because we're just using API's and calculating basic statistics here right so 718 01:12:32,000 --> 01:12:40,600 we're just reading the data from the API calculating some means some totals some some min some 719 01:12:40,600 --> 01:12:50,160 maxes what have you nothing nothing too fancy and what I think making some groundbreaking 720 01:12:50,160 --> 01:12:54,920 discoveries and that's what we do that's what we do here the cannabis data science group 721 01:12:54,920 --> 01:13:03,960 right we get real public we get real cannabis data we get it into a format that's usable 722 01:13:03,960 --> 01:13:10,400 we poke and prod at it and find all its flaws point those out because as Heather pointed 723 01:13:10,400 --> 01:13:20,440 out that's critical we analyze it talk about it have fun so that's what we're about so 724 01:13:20,440 --> 01:13:29,280 glad that you joined ah in Mexico well we're incredibly happy to have you because we're 725 01:13:29,280 --> 01:13:34,880 trying to you know spread the word about cannabis data science all around the world so the more 726 01:13:34,880 --> 01:13:47,560 the merrier so please share the word and invite your friends family colleagues co-workers 727 01:13:47,560 --> 01:13:54,920 you know share the group and you know everybody's welcome and doesn't matter your background 728 01:13:54,920 --> 01:14:00,360 you know we got a little technical today we got a little programming heavy but whatever 729 01:14:00,360 --> 01:14:07,560 you may be you know data science academics production processing retail whether you're 730 01:14:07,560 --> 01:14:15,800 at the lab anybody touching cannabis data you're welcome join the group and let's have 731 01:14:15,800 --> 01:14:24,080 some fun crunching some numbers so glad to have you all and without further ado I'm going 732 01:14:24,080 --> 01:14:30,000 to go ahead and wrap it up here today so thank you for coming and then next week we'll dive 733 01:14:30,000 --> 01:14:37,120 back into the data we'll look at all the flaws we'll try to redo the analysis just to make 734 01:14:37,120 --> 01:14:42,080 sure it's nice and crispy maybe make some crispy charts and there's a lot more we can 735 01:14:42,080 --> 01:14:48,840 get to right there's a whole slew of data points ahead of us so tune in next week because 736 01:14:48,840 --> 01:15:13,240 we will be on another adventure so until then keep your nose to the grindstone