Letter to Harvard Administrators about strike

HGSU-UAW, my campus union, has provided a simple form for emailing administrators about the upcoming strike deadline and urging them to stop a strike. They have even provided templates for different constituencies (undergraduates, family, faculty) to build on in their letters. This is the letter I wrote:

Dear administrators,

I have no love for going on strike. I’d much rather spend those hours writing my dissertation than yelling about Harvard’s greedy side on the picket line. But it doesn’t look like we have any choice, unless the tenor of bargaining really changes. It’s been well over a year and we are still at huge impasses.

We are not making unreasonable demands. The University of California provides graduate students with dental insurance. Yale provides health insurance coverage for dependents at no additional charge. The University of Washington provides dental, vision, and dependent health insurance to graduate students (though, to be fair, their endowment is much bigger than Harvard’s). An arbitration procedure for harassment and discrimination is how Harvard handles these issues for other employees on campus like dining workers or clerical workers. We are not asking for anything special.

I am proud to be at Harvard, but you shouldn’t have to be a single, healthy, or wealthy to thrive here. We should lead on these issues of equity, not let people slip through the cracks. Our bargaining committee is ready to finish the contract and to make compromises, but for them to make compromises your side has to finally admit that health care, harassment, and wages are within the scope of bargaining. I urge you to make that happen.

Sincerely,
Michael Zoorob (G4, Department of Government)

Dispensary density and sales

Please find replication materials (code and data) reproducing all figures and statements made in this blogpost at this link.

One perennial theme on Brookline’s Townwide Discussion Facebook page is the NETA cannabis dispensary, with a vocal group of perhaps 10 people (in the Facebook group, the underlying constituency must be larger)  raising concerns about nuisance (most often being public urination and public consumption) and traffic. It’s not hard to imagine how having 2500-3000 people coming to a business near your house comes with some problems.

Inevitably, somebody comments that opening more stores will alleviate the problem. NETA, right now the only store in Brookline, absorbs traffic from a huge area in Metro Boston, because so many big, nearby municipalities (Boston, Cambridge, Somerville, Watertown, etc) do not have any stores.

But intriguingly, many folks have responded that opening new stores will not help, and that more stores actually increases sales per store. E.g. one poster responds with “The data says otherwise. Here’s a study showing that cannabis turnover per retail outlet increases as the number of outlets grows.” The study referenced is actually a short report made by a group opposing a once-proposed marijuana dispensary in Brookline’s St. Mary’s neighborhood (the group charmingly called itself “The Coalition to Save St. Mary’s Neighborhood”). The evidence for the claim that more stores increases sales per store comes from two plots. First, plotting the trend in sales per dispensary in Washington state and superimposing the trend in store openings; both appear to be increasing during the first six months of recreational sales (July 2014 to December 2014). Second, plotting the sales at Cannabis City, Seattle’s first marijuana store, and superimposing the trend in store openings. Both appear to be increasing.

This counterintuitive claim (it is not obvious why the opening of competitor Store B would increase demand at Store A) piqued my curiosity, so I downloaded the monthly sales data per store from the state of Washington and tried to investigate.  I merged these data with publicly available business information for licensed cannabis establishments to get store locations. And directly comparing store openings with monthly sales per store, I find evidence for the much more boring story that more stores means less sales per store. Looking at the city of Seattle (one confusing thing in the St Mary’s report is conflating Seattle with the state of Washington), sales per store is inversely correlated with stores open. (The outlier at the bottom left is the first month, July 2014; sales did not start until the middle of that month and were hindered by a product shortage.)

SeattleStoresSalesPerStore

And sales at Seattle’s first Dispensary — Cannabis City — decline considerably as other stores in the city open.

SeattleFirstDispensarySalesCompetition

One can also plot sales at the first shop in Seattle over time. As more stores open (the numbers above each bar are the stores open in Seattle that month), sales at Cannabis City stabilize and then decline. It’s important to remember that the first month is an outlier in part because sales did not begin until July 8th, and the store closed, having run out of stock, just three days later. If not for the product shortage, sales during the first month would have been much higher.

SeattleFirstDispensarySalesCompetitiontsbar

The relationship between store openings and sales per store can also be estimated formally using an econometric model. I estimated three statistical regression models at the city-month level of monthly sales per store on number of stores, with fixed-effects for city (holding time-invariant city-specific factors constant) and month-year (ie., July 2014, August 2014, …, October 2017, holding constant time-varying shocks that affect all cities). Using this framework, I first estimate how monthly sales per store within a city change when an additional store opens in that city. The result, Model 1 in the table below, is that opening an additional store in a city is associated with decreased sales per store of $7100, and statistically different from zero (p < 0.001). Model 2 takes the natural logarithm of sales to estimate the percent change in sales per store when one additional store opens, finding that an additional store is associated with about a 6% decline in sales per store. Third, I directly estimate the elasticity of sales per store on store openings, by taking the natural logarithm of both the dependent and independent variables. This approach usefully captures the fact that an additional store will have greater impact when fewer stores are open and less impact when more stores are open. In the two-way fixed effects model, the estimated elasticity of price per store on store openings is about -0.46, indicating that a 1% increase in stores is associated with a -0.46% decrease in sales per store. This means that a 10% increase in stores is associated with a 4.3% decrease in sales per store:

> 100*(1 - exp(-0.458629*log(1.1)))
[1] 4.277041

And a doubling of stores (a 100% increase) brings a 27.2% decrease in sales per store.

> 100*(1 - exp(-0.458629*log(2)))
[1] 27.23226
Statistical models of store openings on sales per store
Dependent variable:
Sales per Store log(Sales per Store)
Interpretation: Model 1: Dollars Model 2: % Change Model 3: Elasticity
Δ Stores -7,100*** -0.06***
Standard Error (1,877) (0.02)
Δ log(Stores) -0.46***
Standard Error (0.10)
Observations 3,227 3,227 3,227
Adjusted R2 0.83 0.71 0.71
City Fixed Effects Yes Yes Yes
Time Fixed Effects Yes Yes Yes
Note: *p<0.1; **p<0.05; ***p<0.01

 

A couple of notes on the St Mary’s report. Its plots use two y-axes, which is usually not good sign, since these are hard to interpret and often deceptive (see also). The report also truncates the months displayed in a cherrypicked way (e.g. in the first figure plotting sales at the first shop in Seattle, sales decline considerably in the month following the last month shown). That figure is also not transparent about the number of stores that are open. The plot says there was 1 cannabis store open in July 2014 and 18 in August 2014. The City of Seattle did not have 18 stores open in August 2014. Cannabis City remained the only store open in August (in September, just 1 additional store reported sales in Seattle; in October, 4 stores reported sales). Why is there this discrepancy? The report says it is using all the stores in the state: “Volume in that third month of operation, September 2014, was FIVE TIMES the volume the entire first month, even though the state had gone from one legal store to 47.” This is misleading; it is hard to imagine that stores on the border with Canada are competing with those in Seattle.

But the report is not faithfully representing the number of stores operating statewide, either, because 13 stores in Washington reported recreational marijuana sales in July 2014, not one. Cannabis City wasn’t even the biggest retailer in July — there were 5 stores with more sales that month; two in Vancouver, Washington on the Oregon border, two in Bellingham, Washington near Canada, and one in Prosser, Washington. So is the plot of the city or the state? Beyond this, the numbers for store openings I obtain from publicly available sales data on Washington’s LCB website (see replication materials) are not consistent with what is in the St Mary’s report. In August, I count 24 stores with nonzero sales, not 18 as in the St Mary’s report. While the sales totals for Cannabis City are the same between our two analyses, we count a much different number of stores. Their report does not link to any data source, so it is difficult to clarify the origin of this discrepancy. What is incontrovertible is that Cannabis City was not the only recreational marijuana store operating in Washington in July 2014, so the St Mary’s report misleads in saying that just one store was open in July 2014.

storesopenstatewide.png

One final point. The claim that sales increase with more stores is supported by plots of an increasing trend in sales over six months alongside a trend of increasing store openings during this period. This overlooks some important time-varying contextual factors (i.e., confounders). Due to a weak harvest and early-industry regulatory issues, Washington did not have enough recreational marijuana to sell when stores first opened, so store shelves sat empty (here’s another article about this), artificially depressing sales during the first months. As mentioned, Cannabis City, Seattle’s first store, closed after three days because it ran out of products. As the supply chain matured, we would expect more sales to occur, because there was previously a product shortage. This is important context that was missed entirely by the simple plots in the St Mary’s report. The two-way fixed effects approach is useful because it directly models such time-varying shocks.

I have posted all replication data and code reproducing these plots and statements for free access here.

My fentanyl paper was mentioned in Congressional Testimony!

This is a new one for me! My research made it to Congress (in written testimony to a subcommittee). Bryce Pardo, a Rand Analyst, in Testimony presented before the House Committee on Homeland Security, Subcommittee on Intelligence and Counterterrorism and Subcommittee on Border Security, Facilitation, and Operations, on July 25, 2019 twice mentioned my paper about fentanyl and the changing geography of the overdose crisis in the US.

c1c2

(Tiny caveat; the dependent variable in my paper is all overdose deaths, not synthetic opioid overdose deaths, a deliberate choice because of heterogeneity in whether drugs are listed on death certificates).

(Here’s another link to the testimony via the Rand Institute website that might outlast the .gov link)

Norfolk 15th Primary Election (MA state House, 9/4/18)

In the weeks leading up to the primary election, I walked around this state house district and counted/geocoded all the lawn signs for the Norfolk 15th State House race in a smartphone application. (These are both Democrats; no Republican is running). Lawn signs are pretty consistent with vote shares and you can see “homefield advantage” type stuff in both (more signs & more votes in precincts where candidate lives).

norfolk15_lawnsigns.pngnorfolk15_precnorfolk15_prec_stone

MA-7th Democratic Primary (Pressley beats Capuano)

I was able to get a hold of and digitize the Boston precinct results for this race (maybe 189 of the 270 precincts in this race, though I only get 187 to merge with the Boston precinct shapefile [Edit: One of the oddities may be Long Island; the other is Ward 5, Precinct 2a in the unofficial election results]) without too much trouble. And since I have a Boston voterfile on my computer for research, I can link these precinct results to average precinct demographics among registered voters, like age (from birthdate) and predicted race.

The maps  of precinct results show mostly what you’d expect (white Neighborhoods like Charlestown and Brighton’s Oak Square were better for Capuano than the city as a whole; Black neighborhoods like Roxbury/Mattapan/Dorchester’s Fields Corner much better for Pressley). But Pressley did really well everywhere in Boston, and never gets less than a third of the vote (while Capuano gets <15% in some precincts). East Boston shows up as a relative bright spot for Capuano, which seems odd given the Hispanic predominance of that area (though Capuano also won Chelsea and non-Hispanic whites likely comprise a bigger proportion of the electorate than the population).  I can’t tell too much from the turnout map, though maybe it shows that college-heavy precincts (like around BU and Northeastern and maybe those Longwood-area college campuses) have less turnout (which is consistent with what I find with age and regression later).

Precinct-level regressions indicate that precincts with larger proportions of Black voters, Hispanic voters, and younger mean age gave a higher vote share to Pressley. Higher Hispanic percentage predicted lower turnout, precincts with an older average voter had higher turnout; association for Black percentage on turnout was null (no association with turnout conditional on Hispanic percent and mean age).

 

 

 

Precinct-Level Determinants (Boston only)
Dependent variable:
Pressley Vote Share Turnout
Black % 0.375*** -0.002
(0.02) (0.013)
Hispanic % 0.178*** -0.108***
(0.051) (0.034)
Mean Age -1.194*** 0.710***
(0.135) (0.089)
Observations 187 187
R2 0.676 0.289
Adjusted R2 0.671 0.277
F Statistic (df = 3; 183) 127.525*** 24.777***
Note: *p<0.1; **p<0.05; ***p<0.01

Fixing MA’s Good Samaritan Law: Calling 911 to report an overdose can still land you in legal trouble.

Before the legislative session concludes, Beacon Hill lawmakers will pass Governor Baker’s CARE Act, an important bill boosting services connecting overdose patients in the ER to recovery services. But the Care Act suffers from a critical omission: it leaves broken the Commonwealth’s limited Good Samaritan Law, which has failed to protect those who seek help during an overdose. Reforming the law is needed to ensure that patients call 911 in the first place.

The Good Samaritan Law is intended to protect people who call 911 from legal punishment, since studies show that fear of arrest stops bystanders from calling for help.  Politicians around the state have touted the law as doing just that. In May 2016,  Attorney General Healey and Governor Baker unveiled a $250,000 campaign encouraging bystanders who witnessed an overdose to “Make the Right Call.” The state distributed posters promising those who call 911 that “The Law Protects You.”

mtc_c

[Image: MA’s Department of Public Health  “Make the Right Call Poster”. Poster Text: “You might be high. You might be afraid. If you see an overdose call 911. The Law protects you.”]

Unfortunately, this is not true. The Good Samaritan Law that took effect on August 2nd, 2012 only narrowly exempts bystanders who call 911 and those who need medical attention from charge or prosecution for drug possession. What it does not do is protect bystanders or patients from being arrested, charged, or prosecuted for any offense other than drug possession. According to the Network for Public Health Laws, the Good Samaritan Law in Massachusetts—unlike progressive bastions like Nevada and Tennessee—does not preclude  arrest or prosecution for other crimes including, mindbogglingly, possessing drug paraphernalia. Nor does it confer protection from civil asset forfeiture, prosecution under any outstanding warrants, or violations of probation or parole.

Indeed, under the Massachusetts law, a person can still be arrested for simple drug possession, but not charged or prosecuted. The problem is, once a person is arrested for drug possession, that contact with the criminal justice system can lead to interviews, searches, and sanctions for unrelated crimes, including unpaid court fees or parole violations. What’s more, it is up to a prosecutor to determine whether you were a worthy Good Samaritan or not—something which can hinge on whether or not your drug possession was with intent to distribute. In our criminal justice system, such discretion usually leads to unequal outcomes across zipcodes, races, and economic strata.

Moreover, several cases illustrate how calling 911 for an overdose can lead to arrest, despite the Good Samaritan Law. In Attleboro, January 2013, a man who called 911 to report an overdose was arrested on drug possession charges. Though the drug possession charges were dropped under the Good Samaritan Law, he faced a three year sentence on an outstanding warrant in another state. In Swampscott, August 2014, police responding to a heroin overdose 911 call found that the residents were manufacturing cannabis oil (hash) in their house. Detectives shared pictures of the hash with the DEA and then arrested the residents on charges related to hash, heroin, and an unlicensed firearm. In Taunton, March 2016, a man called 911 to seek medical assistance for a woman who overdosed. When police arrived, the caller himself was protected by the Good Samaritan Law. But police arrested another person at the house—who hid under a blanket  in the bathroom—on an outstanding warrant. In Brockton, in April 2017, someone called 911 to report an apparent overdose of an unconscious man. When police arrived, he had woken up. Officers searched his truck and arrested him for drug possession and driving without a license.

Newspapers document these and many, many, other such cases where a witness’s call for help  is met with punishment. These cases of what can go wrong if you call for help fuel the apprehension about police that discourages calling 911 at all, a concern which is acute among Blacks, who may be especially wary of calling the police. (In Massachusetts, overdose mortality is rapidly increasing among Blacks). Undocumented residents and those close to them may also have reason to worry.

Expanding the protections of the Good Samaritan Law is essential to ensuring that medical emergencies are not treated like crime scenes. Though politicians around the Commonwealth promise that the law will protect you if you call 911, the experiences of some of our most vulnerable disagree.

Unions, Right-to-Work, and Occupational Deaths

Happily, my paper “Does ‘right to work’ imperil the right to health? The effect of labour unions on workplace fatalities” is attracting a lot of attention, which makes me feel good as an academic who cares at least a little about producing things that others find valuable (we will see how the sheriff stuff is received, once it’s finally done, though it seems like there’s something both quaint and gripping about them!). Although I wrote that short article in August 2017 — and had really no sense of the Janus case at that time — it’s really seemed to have caught the moment. It helps that there’s yet a non-trivial, vested constituency in organized labor that finds something useful, or reassuring, in these results. Anyway, in response to a journalist query about, essentially, what “14.1%” (the coefficient of the reduced form regression of the RTW variable) means exactly,  I wrote something like this:

predicted_annualfatalities_rtw

There were 138,736 total deaths on the job in the 50 states over 1992-2016 recorded in the Census of Fatal Occupational Injuries. Using a negative binomial regression (with the same predictors as the “reduced form” regression in the paper, but with the number of workers as an offset variable and the state/annual count of workplace deaths as the dependent variable) to model the incidence of occupational deaths,  the expected count of occupational deaths in a state during this time-period is about 118.7 in a right to work state and about 104.9 in an otherwise similar state without RTW (figure above; Stata margins command with coefplot for figure). Reassuringly, this 13.1% increase in RTW states in the negative binomial model is essentially the same result as in the “reduced form” model in the paper (about a 14% increase in RTW states relative to others). The Incident Rate Ratio on the right-to-work coefficient is also  1.1314 — which makes sense, as that is the ratio of the predicted values!

Over this same 1992-2016 period, 529 state-year dyads had right to work legislation (721 did not). Therefore, the model-implied count of occupational fatalities attributable to RTW  (assuming no confounders) is (118.7-104.9)*529 or about 7,300.That is, if no states had RTW over these years, the model implied counterfactual is that about 131,436 occupational deaths would have occurred instead of the observed 138,736 occupational deaths.

Semi-related: Using the original (non-logged) rate of fatalities, here’s a predicted effects plot of occupational fatality rate at different levels of unionization; I shared this on Twitter previously. This was used with an OLS model containing all the parameters of model 3 in the paper (or replication code), but with the non-logged dv (in the replication code labeled “robustness check using original unscaled variable”). This shows how the actual, observed decreases in unionization rates we’ve seen in states like Wisconsin in recent years may translate into meaningful increases in occupational mortality.

marginaleffects_origdv

White Supremacists are the biggest snowflakes in American History

[Cover Image from Jim Crow Songbook, 1847, admonishing Black people not to laugh at “them who happen to be white,” a rather strange play at victimhood in a year when almost all Blacks in America were enslaved.]

Of course, white supremacist prohibited Blacks from voting, holding office, serving on juries, having sex with or marrying whites, attending school with whites, eating at restaurants with whites, sharing street cars with whites (at issue in Plessy v. Ferguson), and so-on and so forth (Oregon just straight up banned Black people for many years). But I have only recently come to learn the depth, breadth, and sheer pettiness of Jim Crow and white supremacists, who must be among the most fragile people the world has ever seen. Deep down this brutal history really isn’t funny, but the fact that white racists felt it necessary to ban these things is at least a little bit funny–These people had really thin skin. Here are just some of the activities that so perturbed the White Supremacists that they sanctioned them with the law (and the lynch):

Like extreme conservatives in the US today, who have passed laws banning CDC research on guns, banning doctors from talking about guns with their patients, banning the promotion of non-binary pronouns, increasing penalties for protesting, and believe football players should face economic reprisals for protest, white supremacists during Jim Crow shared this belief in restricting the rights of people who disagree with them. Mississippi banned “printing, publishing or circulating printed, typewritten or written matter urging or presenting for public acceptance or general information, arguments or suggestions in favor of social equality or of intermarriage between whites and negroes.” Mississippi also had in its vagrancy law (a racist Jim Crow statute that allowed the police to arrest and “lease out” unemployed Blacks, often to plantation owners) a provision banning whites from “assembling themselves” with blacks “on terms of equality.”

 

Cannabis “strain” variation in THC:CBD ratio, by lab

Using the I-502 data from Washington State and the ggridges library in R can make some fun graphics. In my paper with Nick, we show the distribution of logged THC:CBD ratio, by strain, for 30 or so popular strains (according to Leafly review data) for two labs — Confidence and Peak.  Here are those plots for all six of the big labs.

Analytical 360, LLC._ridgeplotConfidence Analytics_ridgeplotGreen Grower Labs_ridgeplotIntegrity Labs_ridgeplotTesting Technologies_ridgeplotPeak Analytics Laboratory Testing Services_ridgeplot