1
00:00:00,000 --> 00:00:07,880
A

2
00:00:07,880 --> 00:00:37,000
Okay welcome to the cannabis data science meetup March 17 2021

3
00:00:37,000 --> 00:00:45,920
21. Visit Can-Lytics if you want more information and you want to contribute to open source

4
00:00:45,920 --> 00:00:56,860
research on cannabis analytics. Today, I thought it would be interesting to look at the rates

5
00:00:56,860 --> 00:01:07,120
of return on capital in the cannabis industry. Last week, we looked at the real wage that

6
00:01:07,120 --> 00:01:15,200
workers can be expected to be paid. So this week, it would be interesting to look at the

7
00:01:15,200 --> 00:01:24,000
other input, which is capital. If you look at an article by Poulsen from 2013, it would

8
00:01:24,000 --> 00:01:35,400
be interesting to look at the current rate of capital. He expects that there are high

9
00:01:35,400 --> 00:01:42,640
rents from capital because you have a semi-legal market. So the states consider it legal, the

10
00:01:42,640 --> 00:01:49,880
federal government still considers it illegal, which results in a high rate of capital because

11
00:01:49,880 --> 00:01:58,880
large banks and investors are often hesitant to lend to cannabis businesses. So when you

12
00:01:58,880 --> 00:02:06,680
start a business, you may need to take out loans. And then investors also want to get

13
00:02:06,680 --> 00:02:14,200
rate of return on their loans. So it's interesting to know what the rate of return is in the

14
00:02:14,200 --> 00:02:24,560
cannabis industry because you may expect it to be high. So we're going to use a similar

15
00:02:24,560 --> 00:02:30,360
production function as we used last week. However, we're going to use an even simpler

16
00:02:30,360 --> 00:02:36,760
production function. So first, we're going to take the simplest model and we're just

17
00:02:36,760 --> 00:02:44,960
going to assume that cannabis sold in a given period, yt, is a function of the number of

18
00:02:44,960 --> 00:02:53,520
plants, kt. And it's augmented by technology and there's a random element. And so that'll

19
00:02:53,520 --> 00:03:02,560
be epsilon, or random shop. So of course, the actual production of cannabis is more

20
00:03:02,560 --> 00:03:10,680
complicated than just cultivating plants and turning that into revenue. However, we're

21
00:03:10,680 --> 00:03:17,320
going to abstract just to create the simplest model and then you can expand on it to make

22
00:03:17,320 --> 00:03:30,760
it more realistic. So if you plot these variables against each other, we can start to see some

23
00:03:30,760 --> 00:03:38,200
positive trends. So if you plot retail revenue against cultivated retail plants, you do see

24
00:03:38,200 --> 00:03:45,320
the positive trend like we would expect. And there may or may not be diminishing marginal

25
00:03:45,320 --> 00:03:54,440
returns. What that simply says is you would expect that the more plants you cultivate,

26
00:03:54,440 --> 00:04:03,880
you're going to get less and less revenue for each additional plant. So as you grow

27
00:04:03,880 --> 00:04:13,120
more plants, the overall cannabis becomes less valuable. And so the average price per

28
00:04:13,120 --> 00:04:23,000
output decreases and you're only going to be able to get so much revenue from the additional

29
00:04:23,000 --> 00:04:34,400
plants cultivated. What's interesting to note is that there seems to be a lot steeper. It

30
00:04:34,400 --> 00:04:40,400
seems that diminishing marginal returns have taken place in the medical cannabis market.

31
00:04:40,400 --> 00:04:50,360
So if you just look at the comparison, you can see that there's between 300 and 350,000

32
00:04:50,360 --> 00:05:00,680
retail plants cultivated. And the timeframe on this is 2014 through 2016. At the same

33
00:05:00,680 --> 00:05:15,120
period, retail plants went from 50,000 to 450,000. So you see high returns in the beginning

34
00:05:15,120 --> 00:05:22,440
and as the market matures, like in the medical industry, the medical markets, you get diminishing

35
00:05:22,440 --> 00:05:30,800
marginal returns. So we're interested in estimating alpha. And we expect that it should be between

36
00:05:30,800 --> 00:05:43,880
zero and one. So we can take the production function from earlier and we can simply take

37
00:05:43,880 --> 00:05:57,320
the log of both sides. And then you can estimate it using a linear regression. Once we estimate

38
00:05:57,320 --> 00:06:07,920
it, the competitive rate of return is equal to the marginal product of capital. So if

39
00:06:07,920 --> 00:06:15,760
you take the derivative...

40
00:07:07,920 --> 00:07:23,480
Are you still there Charles?

41
00:07:23,480 --> 00:07:26,120
Yes. Yeah, you froze up.

42
00:07:26,120 --> 00:07:36,920
Okay. So I must... Well, that's why we're recording. So anyways, we're back now.

43
00:07:36,920 --> 00:07:42,640
So where was I when things cut off?

44
00:07:42,640 --> 00:07:48,960
So let's see, you're taking the log of both sides and you were going to estimate alpha.

45
00:07:48,960 --> 00:07:52,280
And then it was just right after that.

46
00:07:52,280 --> 00:08:01,960
Okay. So we take the log of both sides and then we want to calculate the marginal product

47
00:08:01,960 --> 00:08:19,960
of capital. So capital is KT. Right? So that's our number of plants.

48
00:08:19,960 --> 00:08:33,800
And so we want to estimate alpha. So if we take the log of both sides, we can run a linear

49
00:08:33,800 --> 00:08:43,800
regression of the log of sales. Great. So we've got the log of sales and then this is

50
00:08:43,800 --> 00:08:46,520
going to be a constant.

51
00:08:46,520 --> 00:08:53,320
So your slides aren't showing now.

52
00:08:53,320 --> 00:08:58,320
Okay.

53
00:08:58,320 --> 00:09:05,920
Okay. Yeah.

54
00:09:05,920 --> 00:09:19,280
Okay. So we're back. Okay. So just to kind of get back into things.

55
00:09:19,280 --> 00:09:28,960
Okay. So just sort of to do a recap here after sort of that bumpy transition. Okay. So we've

56
00:09:28,960 --> 00:09:40,480
got the total amount of cannabis sold and that's YT. And that's going to be a function

57
00:09:40,480 --> 00:09:50,560
of the number of plants, KT. So we've got our production function.

58
00:09:50,560 --> 00:10:02,320
We take the log of both sides and we get log of cannabis, log of sales equals a constant

59
00:10:02,320 --> 00:10:11,480
plus alpha log of plants. So we can estimate this with the linear regression. And then

60
00:10:11,480 --> 00:10:21,640
given alpha, we can find the marginal product of capital.

61
00:10:41,480 --> 00:10:48,840
So that's the marginal product of capital. And so that's simply the derivative of the

62
00:10:48,840 --> 00:10:58,960
production function with respect to alpha. And so that would be how productive your capital

63
00:10:58,960 --> 00:11:06,400
is going to be. And then we're going to subtract off depreciation. And for this simple example,

64
00:11:06,400 --> 00:11:14,200
we're going to assume zero depreciation. So in reality, there is depreciation. So once

65
00:11:14,200 --> 00:11:20,760
again, this is the simplest model. And so, and then this is just the simplest way that

66
00:11:20,760 --> 00:11:30,760
you could estimate the rate of return on capital. And just to go ahead and look at the empirical

67
00:11:30,760 --> 00:11:40,640
literature so we can go ahead and get our Bayesian prior. We saw in Poulsen's paper

68
00:11:40,640 --> 00:11:51,880
that the interest rates are typically between 7 and 15%. And so that was from Poulsen's

69
00:11:51,880 --> 00:12:00,520
research. And so we want to look at data that we have and see if we can estimate a marginal

70
00:12:00,520 --> 00:12:10,480
product of capital. So let's go ahead and jump into that and do just that. So I'm going

71
00:12:10,480 --> 00:12:40,160
to share with you the data resources. So here is the Colorado Department of Revenue

72
00:12:40,160 --> 00:12:49,080
annual updates. And so you also have them quarterly as well. And so for these particular

73
00:12:49,080 --> 00:13:08,960
updates, they're simply PDF documents with tables of data. So here you have the number

74
00:13:08,960 --> 00:13:27,720
of stores. You have the occupational license data. You have the number of plants. You have

75
00:13:27,720 --> 00:13:38,640
retail. This is retail plants. But then you also have flowers sold per month. And then

76
00:13:38,640 --> 00:13:46,800
you get down here to break down for the different product types. And then you also have total

77
00:13:46,800 --> 00:13:55,520
sales. So we'll be utilizing that as well. So we're not given the rate of return, but

78
00:13:55,520 --> 00:14:03,540
we're given just these raw data points. And we're going to figure it out. And for future

79
00:14:03,540 --> 00:14:11,280
research, it may be worth looking into the testing data here. So it could be worthwhile

80
00:14:11,280 --> 00:14:25,140
to compare what's going on in the Colorado testing market versus Washington. But these

81
00:14:25,140 --> 00:14:31,280
are the data sources. So they're a little tricky. You have to sort of essentially copy

82
00:14:31,280 --> 00:14:42,200
and paste out of these tables. And I've done that here. And I've currently got through

83
00:14:42,200 --> 00:14:58,480
data through December of 2016. But as you can see, there is data through October of

84
00:14:58,480 --> 00:15:10,680
2020. So for future exercise, and I'll be adding this to the Cannabis Data Science GitHub

85
00:15:10,680 --> 00:15:21,820
repository, but we can essentially get the latest data and calculate the rate of return

86
00:15:21,820 --> 00:15:31,320
up through 2020. However, for this exercise, we'll simply calculate it through 2016, since

87
00:15:31,320 --> 00:15:40,080
that's the data that I've already compiled. So let's go ahead and just jump right into

88
00:15:40,080 --> 00:15:48,580
that. I will be using Spyder in Python. Of course, you can use your favorite programming

89
00:15:48,580 --> 00:15:56,400
language and text editor. However, I find Spyder simple and easy to use. It's not glamorous,

90
00:15:56,400 --> 00:16:04,480
but easy to use. So to do this, we only need three Python packages. We just need numpy,

91
00:16:04,480 --> 00:16:19,220
stdc, stdc and stdc. Next, after importing the packages, you can import the data. And

92
00:16:19,220 --> 00:16:28,640
so I'll just be importing this spreadsheet here, which I'll make available on the GitHub

93
00:16:28,640 --> 00:16:41,720
repository after the presentation. Once you've read in the data, of course, number one rule,

94
00:16:41,720 --> 00:17:00,920
look at the data. So you've got some observations or columns. Essentially, we just have date.

95
00:17:00,920 --> 00:17:07,740
I grabbed the number of licenses for future work. Then we can do number of plants, cultivations

96
00:17:07,740 --> 00:17:16,840
and revenue. These are variables that will be useful. However, as you saw in the reports,

97
00:17:16,840 --> 00:17:26,040
there are many more variables that you could grab to do further analysis. So we've read

98
00:17:26,040 --> 00:17:39,920
in the data. Now, if you go back to our model, we need to calculate the log of sales and

99
00:17:39,920 --> 00:17:53,800
the log of capital. So we can use numpy, np.log and go ahead and create the log of retail

100
00:17:53,800 --> 00:18:07,080
revenue and the log of retail plants. And now we can do a simple ordinary least squares

101
00:18:07,080 --> 00:18:27,920
regression of the log of sales, yt, on a constant, which will be the log of technology, a, in

102
00:18:27,920 --> 00:18:37,040
addition to the log of capital, which is retail plants. So we can go ahead and fit this model

103
00:18:37,040 --> 00:18:54,880
in. What's interesting to observe is for such a bare bones model. So we've got 36 observations.

104
00:18:54,880 --> 00:19:02,240
You do have a decent R squared. So we are explaining a lot of the variation and retail

105
00:19:02,240 --> 00:19:09,480
plants. It's significant to the 5% level. So like I said, there's this model, it's got

106
00:19:09,480 --> 00:19:18,800
its flaws, but there does seem to be correlation between the number of plants grown and the

107
00:19:18,800 --> 00:19:32,200
amount of capital. One should note that obviously, cannabis grown in time t is not sold in time

108
00:19:32,200 --> 00:19:41,040
t. So it may be more realistic to estimate an autoregressive regression, where you would

109
00:19:41,040 --> 00:19:54,520
estimate yt sales on kt minus one, kt minus two, perhaps all the way up to kt minus six

110
00:19:54,520 --> 00:20:03,840
or more. That would simply mean that sales today are dependent on the plants grown six

111
00:20:03,840 --> 00:20:11,640
months ago, or one month ago, or three months ago. So it would be an interesting analysis

112
00:20:11,640 --> 00:20:26,640
to add multiple lagged capital variables and see if that has a more significant effect.

113
00:20:26,640 --> 00:20:41,520
But that's research for the future. So in our regression, if you go back to our presentation,

114
00:20:41,520 --> 00:20:49,320
you see that we're looking for alpha. So we're looking for the coefficient on log t. And

115
00:20:49,320 --> 00:21:10,280
so in that case, that's going to be 0.67. So we can go ahead and grab alpha. And now

116
00:21:10,280 --> 00:21:21,840
that we know alpha, we can now calculate the rate of return on capital, which will be alpha

117
00:21:21,840 --> 00:21:35,220
times kt to the alpha minus one. And we're assuming depreciation is zero. So we can program

118
00:21:35,220 --> 00:21:49,840
that in SPDR. So here we have alpha times kt, which is retail plants, to the alpha minus

119
00:21:49,840 --> 00:22:05,400
one. And I'm multiplying this by 100 because I would like to display r as a percentage.

120
00:22:05,400 --> 00:22:16,160
And remember, r is the real rate of return of capital. And so let's estimate that, and

121
00:22:16,160 --> 00:22:27,040
then we'll talk a little about that. Rule number one about data, look at the data. So

122
00:22:27,040 --> 00:22:38,240
here we have what we have estimated to be the real rate of return of capital in the

123
00:22:38,240 --> 00:22:50,720
cannabis market in Colorado from 2014 to 2016. So what we're essentially estimating is we're

124
00:22:50,720 --> 00:23:08,160
saying that an investor would expect a rate of return of 2.6% in 2014 to about 1% in 2016.

125
00:23:08,160 --> 00:23:27,600
In 2015. And that would be at time t. So that would be their monthly rate of return. So

126
00:23:27,600 --> 00:23:36,160
they would like to get a 2% to 1% rate of return per month. And then this would be if

127
00:23:36,160 --> 00:23:46,800
everything was entirely efficient. But as Olson noted, he observed that interest rates

128
00:23:46,800 --> 00:23:57,040
were typically between 7 and 15%. So this model doesn't take into consideration risk.

129
00:23:57,040 --> 00:24:12,160
So risk is likely driving these interest rates well above their real rate of return. Like

130
00:24:12,160 --> 00:24:23,800
last week, we noted that when we added in our uncertainty that we couldn't be quite

131
00:24:23,800 --> 00:24:38,560
so precise in our estimations. So let's do let's add our confidence intervals. So what

132
00:24:38,560 --> 00:24:55,040
is our confidence interval? Well, remember earlier, we estimated alpha to be 0.6777.

133
00:24:55,040 --> 00:25:04,560
Our 95% confidence interval, so we're 95% sure given the data at hand, that alpha given

134
00:25:04,560 --> 00:25:18,680
the model at hand is between 0.6 about 0.61 and about 0.74. So that's our 95% confidence

135
00:25:18,680 --> 00:25:30,640
interval. And so we can calculate our upper and lower alpha. But those those are upper

136
00:25:30,640 --> 00:25:40,280
and lower alpha. So we can calculate our upper and lower interest rates. And we can plot

137
00:25:40,280 --> 00:25:57,720
them. And so now you see our original estimation with interest rates being roughly 2% and dropping

138
00:25:57,720 --> 00:26:08,600
to 1%. Then you see our lower estimation with interest rates being close to 1% and then

139
00:26:08,600 --> 00:26:20,640
dropping quite low to well below 1%. However, our upper bound under confidence interval

140
00:26:20,640 --> 00:26:35,200
is starting around 6 or 5% and then drops down to a little above 3%. Once again, this

141
00:26:35,200 --> 00:26:46,560
is a crude estimation, we're using plants to proxy capital, and we're excluding labor.

142
00:26:46,560 --> 00:26:55,480
So for future work, you can estimate what's called a Cobb-Douglas production function,

143
00:26:55,480 --> 00:27:11,360
where you augment the production function with labor. So a Cobb-Douglas production function,

144
00:27:11,360 --> 00:27:18,680
you can estimate the same model and add another regressor. And that's just going to be beta

145
00:27:18,680 --> 00:27:35,520
log LT. And so that's going to be your labor. And so you can you can estimate a more realistic

146
00:27:35,520 --> 00:27:47,960
function simply by adding the log of labor to our regression. So I'm going to leave that

147
00:27:47,960 --> 00:27:57,360
as a future exercise. And I'll go ahead and do the code and post it to the GitHub repository,

148
00:27:57,360 --> 00:28:06,480
the Candidates Cannabis Data Science repository. So you can check it out there to see the latest

149
00:28:06,480 --> 00:28:22,180
code after the presentation. So we set out to estimate the rate of return on capital.

150
00:28:22,180 --> 00:28:32,680
We knew that because the cannabis industry is still gray in terms of legality, that there

151
00:28:32,680 --> 00:28:40,440
would probably be a high rate of return on capital. However, our model does not factor

152
00:28:40,440 --> 00:28:52,320
in any sort of legality. Our model is simply a function of cannabis sold and plants. We've

153
00:28:52,320 --> 00:29:03,560
got cannabis sold and plants. Of course, we also have technology and then a shock to explain

154
00:29:03,560 --> 00:29:11,720
the random variation. So given this simple model and ignoring the legality, we saw that

155
00:29:11,720 --> 00:29:33,520
there does appear to be diminishing marginal returns as medical plants increases.

156
00:29:33,520 --> 00:29:39,960
Next we estimated the production function, which was simply the log of sales on the log

157
00:29:39,960 --> 00:29:49,720
of capital. For future work, you could add labor. And then we were able to estimate the

158
00:29:49,720 --> 00:30:05,160
actual rate of return of capital. And so here, based on our research, we can actually estimate

159
00:30:05,160 --> 00:30:16,080
the mean rate of return between 2014 and 2016 in Colorado. And so we're seeing it's as low

160
00:30:16,080 --> 00:30:39,880
as 1.2.6%. So just approximation, we'll say 1.25. So the rate of return is, you know,

161
00:30:39,880 --> 00:30:53,440
the real rate of return could be as low as 1.25% per month. So investors in the long

162
00:30:53,440 --> 00:31:03,920
term would be satisfied with a rate of return that low for that amount of production. And

163
00:31:03,920 --> 00:31:16,400
likewise, businesses would be looking at low interest rates for taking out loans for capital

164
00:31:16,400 --> 00:31:28,440
equipment for their cultivations. What we currently see as far as in 2013 is we saw

165
00:31:28,440 --> 00:31:49,160
that the, so this is real, right? And then here we have the actual. And so the actual

166
00:31:49,160 --> 00:31:59,400
rates of return that we observe are between 7 and 15%. So that means risk is explaining

167
00:31:59,400 --> 00:32:12,040
between, you know, six and, you know, about six and 14% of the rate of return in the cannabis

168
00:32:12,040 --> 00:32:18,080
industry. Also, I would like to mention that this is a monthly rate of return. And this

169
00:32:18,080 --> 00:32:27,000
may be an annual rate of return. So it should be, we may want to convert this to APR to

170
00:32:27,000 --> 00:32:40,880
make a good comparison. However, what you could take away from this is that as cannabis

171
00:32:40,880 --> 00:32:49,280
becomes more mainstream and as additional states continue to legalize, so that's what

172
00:32:49,280 --> 00:33:03,040
we've seen since 2016. So from 2016 to the present, 2021, we've seen more states legalize

173
00:33:03,040 --> 00:33:14,160
cannabis, you know, cannabis production and cannabis markets. So, and it will be interesting

174
00:33:14,160 --> 00:33:22,560
to do research to see if cannabis laws and regulations have loosened in the states with

175
00:33:22,560 --> 00:33:34,960
legalized cannabis. Long story short, early on in cannabis markets, we saw high real rates

176
00:33:34,960 --> 00:33:41,720
of return, I mean, high actual rates of return. However, we estimated the real rate of return

177
00:33:41,720 --> 00:33:50,600
would be low. So we would predict that over time, cannabis businesses may see their real

178
00:33:50,600 --> 00:34:01,680
rates of return decrease. So over time, it may become easier and easier for producers,

179
00:34:01,680 --> 00:34:15,560
cultivators to take out loans, whereas investors are currently making large margins because

180
00:34:15,560 --> 00:34:23,960
they're taking on a high amount of risk. So you would expect the margins for investors

181
00:34:23,960 --> 00:34:40,880
will decrease as the risk decreases. And with that, that brings us to the end of the presentation.

182
00:34:40,880 --> 00:34:57,160
And so I'll go ahead and just cut back to the meeting if you have any questions.

183
00:34:57,160 --> 00:35:02,600
No, that was a really good presentation.

184
00:35:02,600 --> 00:35:13,960
Thank you. Yeah. Exactly. So essentially last week, we looked at the labor input. So we

185
00:35:13,960 --> 00:35:22,360
saw, okay, we estimated the real wage for labor, right? Because essentially, that's

186
00:35:22,360 --> 00:35:28,120
what economics is, you know, that's a purpose of economics is sort of estimating the prices

187
00:35:28,120 --> 00:35:35,800
in different markets. So that's actually the labor market. So the wage is the price in

188
00:35:35,800 --> 00:35:45,480
the labor market. And the rate of return is the price in the capital market. So the wage

189
00:35:45,480 --> 00:35:54,200
is what you pay for labor. And the rate of return is what you pay for your capital input.

190
00:35:54,200 --> 00:36:01,080
In this point, in this example, we proxied it with plants. So we're basically assuming,

191
00:36:01,080 --> 00:36:09,580
okay, for every X amount of plants you grow, you have to invest a certain amount into capital.

192
00:36:09,580 --> 00:36:17,880
And then when you put together labor and plants, you get output. And so in this case, it's

193
00:36:17,880 --> 00:36:24,760
going to be cannabis sold. And so we're also abstracting out the retail component. Because,

194
00:36:24,760 --> 00:36:31,300
of course, the cultivators are producing the raw flour and plant material, some of which

195
00:36:31,300 --> 00:36:37,560
is then manufactured into manufactured goods and edibles, at which point they're then sold

196
00:36:37,560 --> 00:36:47,800
to the consumers for the actual revenue. So our model abstracted away a large portion

197
00:36:47,800 --> 00:36:54,840
of how the industry actually operates. But it's an abstraction. And so there's a lot

198
00:36:54,840 --> 00:37:01,160
of room for improvement. So I think I'll leave it there. And like I said, I'll make the data

199
00:37:01,160 --> 00:37:10,920
available. And so we can add data up through 2020. So that way we can see how these estimations

200
00:37:10,920 --> 00:37:17,000
evolve over time. And we can maybe even compare our predictions to the actuals. So we looked

201
00:37:17,000 --> 00:37:25,120
at data from 2014 to 2016. And we made some hypotheses. I personally predicted that the

202
00:37:25,120 --> 00:37:31,120
real rate of return is going to continue to decrease. So now we can look at the data and

203
00:37:31,120 --> 00:37:32,120
see what happened.

204
00:37:32,120 --> 00:37:42,520
Okay, yeah, that sounds cool. You know, the other thing I was thinking about was price

205
00:37:42,520 --> 00:37:47,720
of items sold compared to number of licenses.

206
00:37:47,720 --> 00:37:59,960
Exactly. So you can gauge sort of how much each company is taking away at the end of

207
00:37:59,960 --> 00:38:00,960
the day.

208
00:38:00,960 --> 00:38:06,960
Yeah, it is competition increases, price goes down.

209
00:38:06,960 --> 00:38:15,360
Exactly. So we may not have the best measure on price. We could proxy it because we have

210
00:38:15,360 --> 00:38:23,000
essentially we have total revenue in Colorado. Well, we may actually have prices in Washington.

211
00:38:23,000 --> 00:38:29,640
Actually, this could that could actually be a good analysis for Washington. Because you're

212
00:38:29,640 --> 00:38:39,960
right, you could I mean, one way to look at it would just be, I guess, like total amount

213
00:38:39,960 --> 00:38:48,840
of revenue per retailer. And then you could see the market share. In Washington state,

214
00:38:48,840 --> 00:38:57,400
you can actually calculate the actual market share per retailer. And you can make estimations

215
00:38:57,400 --> 00:39:14,880
of the market concentration. In fact, have you heard of the Herfindale? It's HHI, the

216
00:39:14,880 --> 00:39:23,720
Herfindale Hirschman index. And so I may I may present that next week, because it's actually

217
00:39:23,720 --> 00:39:34,840
one of the simpler measures of market concentration you can do. And it's easy to calculate if

218
00:39:34,840 --> 00:39:41,100
you have the right data. And in Washington state, we do. So you simply need to know the

219
00:39:41,100 --> 00:39:52,420
market share per retailer. So market, the retailer, A has shares 0.3, retailer B has

220
00:39:52,420 --> 00:40:02,500
market shares 0.5. And then you actually do the squared. So it's just the sum of all shares

221
00:40:02,500 --> 00:40:16,520
squared. So it would be 0.3 squared plus 0.5 squared plus, you know, 99.2 squared. And

222
00:40:16,520 --> 00:40:24,920
you would see, oh, that's a that's an incredibly concentrated market. Because I forget, I think

223
00:40:24,920 --> 00:40:39,240
I'll need to do a bit of research and tell you more about it next week, because I forget

224
00:40:39,240 --> 00:40:46,080
off the top of my head if the lower numbers or higher numbers represent sharper concentration.

225
00:40:46,080 --> 00:41:01,760
But I want to see higher numbers represent larger, because I feel like the if it's close

226
00:41:01,760 --> 00:41:08,920
to one. But anyways, I may touch on that next week, because that's something that I've been

227
00:41:08,920 --> 00:41:16,020
wanting to look at. And you and it seems they that's your idea as well is, it would be interesting

228
00:41:16,020 --> 00:41:23,640
to look at the concentration in the cannabis markets over time to see if it's becoming

229
00:41:23,640 --> 00:41:36,880
more competitive or more concentrated. Yeah, and we can do it statistically and we can

230
00:41:36,880 --> 00:41:45,040
plot it over time. So that way we could even see month by month, the competitiveness and

231
00:41:45,040 --> 00:41:51,720
it's interesting research, because you could start to find trends and you can maybe find

232
00:41:51,720 --> 00:42:00,880
if if any actions. So if you notice any point in time where there was a big change in the

233
00:42:00,880 --> 00:42:07,960
competitiveness, you may be able to pinpoint the piece of legislation or something that

234
00:42:07,960 --> 00:42:20,160
happened that changed the competitiveness of the market. Yeah, yeah. So I think, of course,

235
00:42:20,160 --> 00:42:26,040
I don't think we have time today. But I think for next week, let's look let's look at just

236
00:42:26,040 --> 00:42:33,080
that because that that's the simplest. That's the simplest measure that I can think of to

237
00:42:33,080 --> 00:42:40,940
look at market concentration. In Washington State, we do have the data. It's in that giant

238
00:42:40,940 --> 00:42:49,080
sales data dump. So it's possible. And it's going to be a good excuse to look at the data.

239
00:42:49,080 --> 00:42:56,600
So that's our philosophy here, correct is look at the data. So I think we've got a good

240
00:42:56,600 --> 00:43:06,680
project for next week. Great. All right, Charles, I think I may just go ahead and wrap up a

241
00:43:06,680 --> 00:43:15,080
little early today. Unless you have anything you want to talk about. Yeah, okay. Well,

242
00:43:15,080 --> 00:43:19,240
I think for today, let's just go ahead and wrap up 10 minutes early. I sort of went through

243
00:43:19,240 --> 00:43:28,680
the presentation quickly. I didn't anticipate moving through it so quickly. So I could have

244
00:43:28,680 --> 00:43:34,600
added a bit more detail. And so like I said, I'll add a bit more detail to the actual GitHub

245
00:43:34,600 --> 00:43:42,720
repository. Just a little technical error. Didn't get it committed before the presentation,

246
00:43:42,720 --> 00:43:48,760
but I'll commit it right after. And then, yeah, and then we'll look at market concentration

247
00:43:48,760 --> 00:43:57,240
next week. And, you know, let me know if you have any questions about this economics. Okay.

248
00:43:57,240 --> 00:44:05,920
Cool. Awesome, Charles. And everyone. So thank you for listening in. And I'll see you this

249
00:44:05,920 --> 00:44:22,640
time next week. See you. Bye. Bye, Charles.